Version 1.2!

Add information about installing dev environment on OS X

.gitattributes

@@ -0,0 +1,2 @@
+.gitattributes export-ignore
+.gitignore export-ignore

.gitignore

@@ -0,0 +1,15 @@
+*.o
+*~
+
+# Autogenerated
+*.gen.*
+
+# Test binaries
+jv_test
+jv_parse
+parsertest*~
+
+build
+
+# Something delightfully recursive happens otherwise
+docs/content/2.download/source/*

COPYING

@@ -0,0 +1,70 @@
+jq is copyright (C) 2012 Stephen Dolan
+
+Permission is hereby granted, free of charge, to any person obtaining
+a copy of this software and associated documentation files (the
+"Software"), to deal in the Software without restriction, including
+without limitation the rights to use, copy, modify, merge, publish,
+distribute, sublicense, and/or sell copies of the Software, and to
+permit persons to whom the Software is furnished to do so, subject to
+the following conditions:
+
+The above copyright notice and this permission notice shall be
+included in all copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
+LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+
+
+
+jq's documentation (everything found under the docs/ subdirectory in
+the source tree) is licensed under the Creative Commons CC BY 3.0
+license, which can be found at:
+
+         http://creativecommons.org/licenses/by/3.0/
+
+The documentation website includes a copy of Twitter's Boostrap and
+relies on Bonsai, Liquid templates and various other projects, look
+them up for detailed licensing conditions.
+         
+
+         
+jq incorporates David M. Gay's dtoa.c and g_fmt.c, which bear the
+following notices:
+
+dtoa.c:
+The author of this software is David M. Gay.
+
+Copyright (c) 1991, 2000, 2001 by Lucent Technologies.
+
+Permission to use, copy, modify, and distribute this software for any
+purpose without fee is hereby granted, provided that this entire notice
+is included in all copies of any software which is or includes a copy
+or modification of this software and in all copies of the supporting
+documentation for such software.
+
+THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED
+WARRANTY.  IN PARTICULAR, NEITHER THE AUTHOR NOR LUCENT MAKES ANY
+REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY
+OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE.
+
+g_fmt.c:
+The author of this software is David M. Gay.
+
+Copyright (c) 1991, 1996 by Lucent Technologies.
+
+Permission to use, copy, modify, and distribute this software for any
+purpose without fee is hereby granted, provided that this entire notice
+is included in all copies of any software which is or includes a copy
+or modification of this software and in all copies of the supporting
+documentation for such software.
+
+THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED
+WARRANTY.  IN PARTICULAR, NEITHER THE AUTHOR NOR LUCENT MAKES ANY
+REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY
+OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE.
+

JQ.hs

@@ -1,304 +0,0 @@
-module JQ where
-import Text.JSON
-import Text.JSON.String
-import Data.Maybe
-import Data.Char
-import Data.List (sortBy,sort,groupBy,partition,intercalate)
-import Data.Function (on)
-import Data.Ord (comparing)
-import Control.Monad
-import Control.Monad.Writer
-import Control.Monad.List
-import Control.Monad.Reader
-import qualified Data.Map as M
-import qualified Data.Set as S
-import Data.Map ((!))
-import Debug.Trace
-
-type Path = [Either Int String]
-
-type Program = JSValue -> [(JSValue, Path)]
-
-type Filter = JQ JSValue
-newtype Operator = Operator {runOperator:: [Filter] -> Filter}
-
-type JQ = ReaderT (JSValue, M.Map String JSValue, M.Map (String, Int) Operator) (WriterT Path [])
-
-runJQ :: JQ a -> JSValue -> [a]
-runJQ prog val = map fst $ runWriterT $ runReaderT prog (val,M.empty,primitives)
-
-(>|) :: JQ JSValue -> JQ a -> JQ a
-a >| b = do
-  val <- a
-  local (\(v,s,d) -> (val,s,d)) b
-
-setvar name val prog = 
-  local (\(v,s,d) -> (v, M.insert name val s, d)) prog
-getvar name = liftM (! name) $ asks (\(v,s,d) -> s)
-
-input = asks (\(v,s,d) -> v)
-
-collect :: JQ a -> JQ [a]
-collect prog = liftM (map fst) $ collectPaths prog
-  
-collectPaths :: JQ a -> JQ [(a,Path)]
-collectPaths prog = do
-  rd <- ask
-  return $ runWriterT $ runReaderT prog rd
-
-withDefn :: (String, [String], Filter) -> Filter -> Filter
-withDefn (name, formals, body) subexp = 
-  local (\(v,s,d) -> (v,s,M.insert (name,length formals) (Operator func) d)) subexp
-  where
-    func args = local (\(v,s,d) -> (v,M.empty,M.fromList (zip (zip formals (repeat 0)) (map (Operator . const) args)) `M.union` d)) body
-
-subexp :: JQ a -> JQ a
-subexp = censor $ const []
-
-
-yieldPaths :: [(a,Path)] -> JQ a
-yieldPaths ps = ReaderT $ const $ WriterT ps
-
-
-insert :: JQ JSValue -> JSValue -> Path -> JQ JSValue
-insert replace base [] = replace
-insert replace (JSArray values) ((Left n):rest) = do
-  let array = take (max (n+1) (length values)) (values ++ repeat JSNull)
-  replacement <- insert replace (array !! n) rest
-  let (left, (_:right)) = splitAt n array
-  return $ JSArray $ left ++ [replacement] ++ right
-insert replace (JSObject obj) ((Right k):rest) = do
-  let oldval = maybe JSNull id (lookup k $ fromJSObject obj)
-  replacement <- insert replace oldval rest
-  let withoutK = filter ((/= k) . fst) $ fromJSObject obj
-  return $ JSObject $ toJSObject $ (k, replacement):withoutK
-
-insert replace JSNull p@((Right k):rest) = insert replace (JSObject $ toJSObject []) p
-insert replace JSNull p@((Left n):rest)  = insert replace (JSArray []) p
-insert _ base p = error $ "Cannot insert into " ++ intercalate ", " (map (either show show) p) ++ " of " ++ encode base
-
-
-eqj a b = JSBool $ a == b
-
-boolj (JSBool false) = False
-boolj (JSNull) = False
-boolj _ = True
-
-andj a b = JSBool $ boolj a && boolj b
-orj a b = JSBool $ boolj a || boolj b
-
-liftp :: (JSValue -> JSValue) -> JQ JSValue
-liftp f = liftM f input
-
-idp = undefined
-failp t = []
-
-constp :: JSValue -> Program
-constp t t' = idp t
-
-anyj :: [JSValue] -> Bool
-anyj values = any isTrue values
-  where
-    isTrue (JSBool False) = False
-    isTrue (JSNull) = False
-    isTrue _ = True
-
-selectp prog = do
-  match <- collect prog
-  guard $ anyj match
-  input
-
-constStr :: String -> JQ JSValue
-constStr = return . JSString . toJSString
-
-constInt :: Int -> JQ JSValue
-constInt = return . JSRational False . toRational
-
-tr x = trace (show x) x
-
-
-assignp sel replace = do
-  paths <- collectPaths sel
-  t <- input
-  foldM (\base (val,path) -> insert (return val >| replace) base path) t paths
-
-arrayp prog = liftM JSArray $ collect prog
-
-
-childp' :: JSValue -> JQ JSValue
-childp' (JSArray values) = msum [tell [Left i] >> return v | (v,i) <- zip values [0..]]
-childp' (JSObject obj) = msum [tell [Right k] >> return v | (k,v) <- fromJSObject obj]
-childp' _ = mzero
-
-childp = input >>= childp'
-
---findp :: Program -> Program
-findp prog = do
-  found <- collect prog
-  if anyj found then input else childp >| findp prog
-
-groupp prog = do
-  list <- input
-  case list of
-    JSArray values -> do
-      marked <- forM values $ \v -> do 
-        m <- collect (return v >| prog)
-        return (m,v)
-      msum $
-        map (return . JSArray . map snd) $ 
-        groupBy ((==) `on` fst) $ 
-        sortBy (comparing fst) $ 
-        marked
-    _ -> return JSNull
-
-recp prog = do
-  found <- collectPaths prog
-  let (roots,subs) = partition (null . snd) found
-  msum $ 
-    [tell p >> return x | (x,p) <- roots] ++
-    [tell p >> (return x >| recp prog) | (x,p) <- subs]
-
-elsep p1 p2 = do
-  p1' <- collectPaths p1
-  if null p1' then p2 else yieldPaths p1'
-
-fullresultp prog = do
-  res <- collectPaths prog
-  msum [return $ JSObject $ toJSObject $ [("val",a),("path",JSArray $ map fromPath p)] | (a,p) <- res]
-    where
-      fromPath (Left n) = js n
-      fromPath (Right s) = js s
-
-
-withArray f (JSArray values) = JSArray $ f values
-withArray f x = x
-
-withString f (JSString str) = JSString $ toJSString $ f $ fromJSString str
-withString f x = x
-
-{-
-callp "select" [p] = selectp p
-callp "find" [p] = findp p
-callp "sort" [] = liftp (withArray sort)
-callp "group" [p] = groupp p
-callp "rec" [p] = recp p
-callp "empty" [] = mzero
-callp "true" [] = return $ JSBool True
-callp "false" [] = return $ JSBool False
-callp "null" [] = return $ JSNull
-callp "count" [] = liftp countj
-callp "fullresult" [p] = fullresultp p
-callp "uppercase" [] = liftp $ withString $ map toUpper
-callp "lowercase" [] = liftp $ withString $ map toLower
--}
-
-primitives = M.fromList [((name,arglen),Operator func) | 
-                         (name,arglen,func) <- prim]
-  where
-    prim = [("if",1,\[p] -> selectp p),
-            ("find", 1, \[p] -> findp p),
-            ("group", 1, \[p] -> groupp p),
-            ("rec", 1, \[p] -> recp p), 
-            ("true", 0, const $ return $ JSBool True),
-            ("false", 0, const $ return $ JSBool False),
-            ("null", 0, const $ return $ JSNull),
-            ("count", 0, const $ liftp countj),
-            ("fullresult", 1, \[p] -> fullresultp p),
-            ("zip", 0, const $ liftp zipj),
-            ("keys", 0, const $ liftp keysj)
-            ]
-
-callp :: String -> [Filter] -> Filter
-
-callp name args = do
-  (v,s,d) <- ask
-  runOperator (d ! (name, length args)) args
-
-countj (JSArray v) = js$ length v
-countj (JSObject o) = js$ length $ fromJSObject o
-countj _ = js$ (1::Int)
-
-lookupj :: JSValue -> JSValue -> JQ JSValue
-lookupj (JSArray values) (JSRational _ n) = do
-  let idx = round n
-  tell [Left idx]
-  if idx >= 0 && idx < length values
-    then return $ values !! idx
-    else return $ JSNull
-lookupj (JSObject obj) (JSString s) = do
-  tell [Right (fromJSString s)]
-  case (lookup (fromJSString s) (fromJSObject obj)) of
-    Just x -> return x
-    Nothing -> return JSNull
-lookupj JSNull (JSRational _ n) = do
-  tell [Left $ round n]
-  return JSNull
-lookupj JSNull (JSString s) = do
-  tell [Right (fromJSString s)]
-  return JSNull
---lookupj v i = error $ "Cannot get element " ++ encode i ++ " of " ++ encode v
-lookupj _ _ = mzero
-
-
-plusj (JSRational f1 n1) (JSRational f2 n2) = JSRational (f1 || f2) (n1 + n2)
-plusj (JSString s1) (JSString s2) = JSString $ toJSString (fromJSString s1 ++ fromJSString s2)
-plusj (JSArray a1) (JSArray a2) = JSArray $ a1 ++ a2
-plusj (JSObject o1) (JSObject o2) = JSObject $ toJSObject $ o1' ++ fromJSObject o2
-  where
-    newkeys = map fst $ fromJSObject o2
-    o1' = filter (not . (`elem` newkeys) . fst) $ fromJSObject o1
-
-keysj (JSArray v) = js [0..length v - 1]
-keysj (JSObject obj) = js (map fst $ fromJSObject obj)
-keysj _ = JSArray []
-
-zipj jsonValue = result $ tx values ctor
-  where
-    (result, values) = extract packed
-    (packed, ctor) = case jsonValue of
-      JSArray values -> (values, (\vs' -> JSArray $ [v' | Just v' <- vs']))
-      JSObject jsObject -> 
-        let object = fromJSObject jsObject
-            keys = map fst object
-            values = map snd object
-            build vs' = [(k,v') | (k,Just v') <- zip keys vs']
-        in (values, JSObject . toJSObject . build)
-      
-      _ -> error "only arrays and objects may be zipped"
-
-    extract values | all isArray values = (JSArray, [map Just arr | JSArray arr <- values])
-                   | all isObject values = 
-      let objects = [fromJSObject o | JSObject o <- values]
-          keys = S.toList $ S.fromList [k | obj <- objects, (k,_) <- obj]
-          values' :: [[Maybe JSValue]]
-          values' = [[lookup k object | k <- keys] | object <- objects]
-          result r = JSObject $ toJSObject $ zip keys r
-      in (result, values')
-                   | otherwise = error "elements of zipped value must be all objects or all arrays"
-      where
-        isArray (JSArray a) = True
-        isArray _ = False
-        isObject (JSObject o) = True
-        isObject _ = False
-    
-    head' [] = Nothing
-    head' (x:xs) = x
-    tail' [] = []
-    tail' (x:xs) = xs
-    tx values ctor | all null values = []
-                   | otherwise = ctor (map head' values):tx (map tail' values) ctor
-
-js :: JSON a => a -> JSValue
-js = showJSON
-
-index s = do
-  v <- input
-  lookupj v (js s)
-
-
-dictp progs = do
-  liftM (JSObject . toJSObject) $ forM progs $ \(k,v) -> do
-    JSString k' <- subexp k
-    v' <- subexp v
-    return (fromJSString k', v')
-

Lexer.x

@@ -1,102 +0,0 @@
-{
-module Lexer where
-import Control.Monad.Error
-}
-
-%wrapper "monadUserState"
-
-$digit = 0-9
-$alpha = [a-zA-Z_]
-@reserved = "."|"["|"]"|","|":"|"("|")"|"{"|"}"|"|"|"=="|"+"|"="|"$"|"def"|";"|"else"|"and"|"or"|"as"
-@ident = $alpha [$alpha $digit]*
-@string = \" ($printable)* \"
-
-
-tokens :- 
-
-<0> "#" ($printable # [\n\r])* ;
-<0> $white+             ;
-<0> @reserved           { tok TRes }
-<0> @ident              { tok TIdent }
-<0> $digit+             { tok $ TInt . read }
-
-
-<0> \"                  { enterString }
-<string> \"             { leaveString }
-<string> ($printable # [\"\\]) { pushString id }
-<string> \\ [\"\\\/]    { pushString (drop 1) }
-<string> \\ [nrt]       { pushString (escape . drop 1) }
---<string> \\ 'u' [0-9a-fA-F]{4}
---                        { pushString (parseUnicode . drop 2) }
-
--- @string { \s -> TString $ init $ tail s}
-
-
-{
-  
-escape :: String -> String
-escape "r" = "\r"
-escape "n" = "\n"
-escape "t" = "\t"
-
-getState :: Alex AlexState
-getState = Alex $ \s -> Right (s, s)
-
-getUserState :: Alex AlexUserState
-getUserState = liftM alex_ust getState
-
-setUserState :: AlexUserState -> Alex ()
-setUserState s' = Alex $ \s -> Right (s{alex_ust = s'}, ()) 
-
-alexEOF = return $ Nothing
-
-enterString input len = do
-  setUserState []
-  alexSetStartCode string
-  skip input len
-
-pushString f i@(p, _, s) len = do
-  buf <- getUserState
-  setUserState (buf ++ [f $ take len s])
-  skip i len
-  
-leaveString input len = do
-  s <- getUserState
-  alexSetStartCode 0
-  return $ Just $ TString $ concat s
-  
-
-tok f (p,_,s) len = return $ Just $ f (take len s)
-data Token = TRes String | TString String | TIdent String | TInt Int
-
-instance Show Token where
-  show (TRes t) = "token " ++ t
-  show (TString t) = "string " ++ t
-  show (TIdent t) = "identifier " ++ t
-  show (TInt t) = "integer " ++ show t
-
-
-type AlexUserState = [String]
-
-alexInitUserState = undefined
-
-wrapError (Alex scanner) = Alex $ \s -> case scanner s of
-  Left message -> Left (message ++ " at " ++ showpos (alex_pos s))
-    where
-      showpos (AlexPn off line col) = "line " ++ show line ++ ", column " ++ show col
-  x -> x
-
-scanner = do
-  tok <- wrapError alexMonadScan
-  case tok of
-    Nothing -> do
-      s <- getState
-      case alex_scd s of
-        0 -> return []
-        string -> alexError "Unterminated string literal"
-    Just tok -> liftM (tok:) scanner
-
-runLexer :: String -> Either String [Token]
-runLexer input = runAlex input scanner
-
-}

Main.hs

@@ -1,23 +0,0 @@
-import Parser
-import Lexer
-import JQ
-import Text.JSON
-import Text.JSON.String
-import PrettyJSON
-import System.Environment
-import Control.Monad
-import System.IO
-
-parseJS :: String -> JSValue
-parseJS s = case runGetJSON readJSValue s of
-  Left err -> error err
-  Right val -> val
-
-  
-main = do
-  [program] <- getArgs
-  stdlib <- openFile "stdlib.jq" ReadMode >>= hGetContents
-  json <- liftM parseJS $ hGetContents stdin
-  case runLexer (stdlib ++ program) >>= runParser of
-    Left err -> putStrLn err
-    Right program -> mapM_ (putStrLn . show . renderJSON) (runJQ program json)

Makefile

@@ -1,8 +1,102 @@
-jq: *.hs Parser.hs Lexer.hs
-	ghc *.hs -o jq
+CC=gcc
+CFLAGS=-Wextra -Wall -Wno-missing-field-initializers -Wno-unused-parameter -std=gnu99 -ggdb -Wno-unused-function $(EXTRA_CFLAGS)
 
-Parser.hs: Parser.y
-	happy -i Parser.y
+prefix=/usr/local
+mandir=$(prefix)/share/man
 
-Lexer.hs: Lexer.x
-	alex Lexer.x
+
+.PHONY: all clean releasedep tarball install uninstall test releasetag
+all: jq
+
+lexer.gen.c: lexer.l
+	flex -o lexer.gen.c --header-file=lexer.gen.h lexer.l
+lexer.gen.h: lexer.gen.c
+
+parser.gen.c: parser.y lexer.gen.h
+	bison -W -d parser.y -v --report-file=parser.gen.info -o $@
+parser.gen.h: parser.gen.c
+
+jv_utf8_tables.gen.h: gen_utf8_tables.py
+	python $^ > $@
+jv_unicode.c: jv_utf8_tables.gen.h
+
+version.gen.h: VERSION
+	sed 's/.*/#define JQ_VERSION "&"/' $^ > $@
+main.c: version.gen.h
+
+JQ_SRC=parser.gen.c lexer.gen.c opcode.c bytecode.c compile.c execute.c builtin.c jv.c jv_parse.c jv_print.c jv_dtoa.c jv_unicode.c jv_aux.c jv_alloc.c
+
+jq_test: CFLAGS += -DJQ_DEBUG=1
+jq_test: $(JQ_SRC) jq_test.c
+	$(CC) $(CFLAGS) $(CFLAGS_DEBUG) -o $@ $^
+
+jq: CFLAGS += -O -DJQ_DEBUG=0
+jq: $(JQ_SRC) main.c
+	$(CC) $(CFLAGS) $(CFLAGS_OPT) -o $@ $^
+
+test: jq_test
+	valgrind --error-exitcode=1 -q --leak-check=full ./jq_test >/dev/null
+
+BINARIES=jq jq_test
+PLATFORMS=linux32 linux64 osx32 osx64 win32 win64
+
+build/linux32%: CC='x86_64-linux-gnu-gcc -m32'
+build/linux64%: CC='x86_64-linux-gnu-gcc -m64'
+
+# OS X cross compilers can be gotten from
+# https://launchpad.net/~flosoft/+archive/cross-apple
+build/osx32%:   CC='i686-apple-darwin10-gcc -m32'
+build/osx64%:   CC='i686-apple-darwin10-gcc -m64'
+
+# On Debian, you can get windows compilers in the
+# gcc-mingw-w64-i686 and gcc-mingw-w64-x86-64 packages.
+build/win32%:   CC='i686-w64-mingw32-gcc -m32'   EXTRA_CFLAGS=-DJQ_DEFAULT_ENABLE_COLOR=0
+build/win64%:   CC='x86_64-w64-mingw32-gcc -m64' EXTRA_CFLAGS=-DJQ_DEFAULT_ENABLE_COLOR=0
+
+ALL_BINARIES=$(foreach platform, $(PLATFORMS), $(foreach binary, $(BINARIES), build/$(platform)/$(binary)))
+
+$(ALL_BINARIES): build/%:
+	mkdir -p $(@D)
+	make -B $(BINARIES) CC=$(CC)
+	cp $(BINARIES) $(@D)
+
+binaries: $(ALL_BINARIES)
+
+clean:
+	rm -rf build
+	rm -f $(BINARIES) *.gen.*
+
+releasedep: lexer.gen.c parser.gen.c jv_utf8_tables.gen.h
+
+releasetag:
+	git tag -s "jq-$$(cat VERSION)" -m "jq release $$(cat VERSION)"
+
+docs/content/2.download/source/jq.tgz: jq
+	mkdir -p `dirname $@`
+	tar -czvf $@ `git ls-files; ls *.gen.*`
+
+tarball: docs/content/2.download/source/jq.tgz
+
+jq.1: docs/content/3.manual/manual.yml
+	( cd docs; rake manpage; ) > $@
+
+install: jq jq.1
+	install -d -m 0755 $(prefix)/bin
+	install -m 0755 jq $(prefix)/bin
+	install -d -m 0755 $(mandir)/man1
+	install -m 0755 jq.1 $(mandir)/man1
+
+uninstall:
+	rm -vf $(prefix)/bin/jq
+	rm -vf $(mandir)/man1/jq.1
+
+
+www: docs/output www_binaries
+
+docs/output:
+	cd docs; rake build
+
+www_binaries: docs/output binaries
+	$(foreach platform, $(PLATFORMS), $(foreach binary, $(BINARIES), \
+	  mkdir -p docs/output/download/$(platform); \
+	  cp build/$(platform)/$(binary) docs/output/download/$(platform)/$(binary); ))

Parser.y

@@ -1,121 +0,0 @@
-{
-module Parser where
-import Lexer
-import JQ
-import Text.JSON
-import Debug.Trace
-import Data.List
-import Control.Monad.Error
-import Control.Monad.Reader
-
-instance Error (Maybe a) where
-  noMsg = Nothing
-  strMsg = const Nothing
-
-instance (Error a, Error b) => Error (a, b) where
-  noMsg = (noMsg, noMsg)
-  strMsg s = (strMsg s, strMsg s)
-}
-
-%name runParser TopLevel
-%tokentype { Token }
-
-%monad { Either String }
-%error { \t -> throwError $ "parse error: unexpected " ++ if null t then [] else (show $ t) }
-
-%token
-  '|' { TRes "|" }
-  '.' { TRes "." }
-  '[' { TRes "[" }
-  ']' { TRes "]" }
-  '{' { TRes "{" }
-  '}' { TRes "}" }
-  '(' { TRes "(" }
-  ')' { TRes ")" }
-  '$' { TRes "$" }
-  'as' { TRes "as" }
-  ',' { TRes "," }
-  ':' { TRes ":" }
-  '==' { TRes "==" }
-  '=' { TRes "=" }
-  '+' { TRes "+" }
-  'def' { TRes "def" }
-  ';' { TRes ";" }
-  'else' { TRes "else" }
-  'and' { TRes "and" }
-  'or' { TRes "or" }
-  Ident { TIdent $$ }
-  String { TString $$ }
-  Int { TInt $$ }
-
-
-%left 'else'
-%right '|'
-%left '='
-%left ','
-%left 'and' 'or'
-%nonassoc '=='
-%left '+'
-
-%%
-
-TopLevel
-  : Defn TopLevel	{ withDefn $1 $2 }
-  | Exp			{ $1 }
-
-Exp 
-  : Exp 'else' Exp      { $1 `elsep` $3 }
-  | Assign '|' Exp	{ do { v <- snd $1; setvar (fst $1) v $3 } }
-  | Exp '|' Exp         { $1 >| $3 }
-  | Exp ',' Exp         { $1 `mplus` $3 }
-  | Exp 'and' Exp	{ liftM2 andj $1 $3 }
-  | Exp 'or' Exp	{ liftM2 orj $1 $3 }
-  | Exp '=' Exp         { assignp $1 $3 }
-  | Exp '==' Exp        { liftM2 eqj (subexp $1) (subexp $3) }
-  | Exp '+' Exp         { liftM2 plusj (subexp $1) (subexp $3) }
-  | Term                { $1 }
-
-Assign
-  : Term 'as' '$' Ident { ($4, $1) }
-
-Defn
-  : 'def' Ident '=' Exp ';' { ($2, [], $4) }
-  | 'def' Ident '(' ParamList ')' '=' Exp ';' { ($2, $4, $7) }
-
-ParamList
-  :			{ [] }
-  | Ident		{ [$1] }
-  | Ident ';' ParamList { $1:$3 }
-  
-ExpD
-  : ExpD '|' ExpD       { $1 >| $3 }
-  | ExpD '==' ExpD      { liftM2 eqj $1 $3 }
-  | Term                { $1 }
-
-
-Term 
-  : '.'                 { input }
-  | Term '.' Ident      { $1 >| index $3 }
-  | '.' Ident           { index $2 }
-  | String              { constStr $1 }
-  | Term '[' Exp ']'    { do {t <- $1; i <- subexp $3; lookupj t i} }
-  | Term '[' ']'        { $1 >| childp }
-  | '(' Exp ')'         { $2 }
-  | '[' Exp ']'         { arrayp $2 }
-  | '[' ']' 		{ arrayp (callp "empty" []) }
-  | Int                 { constInt $1 }
-  | '{' MkDict '}'      { dictp $2 }
-  | Ident '(' Exp ')'   { callp $1 [$3] }
-  | Ident               { callp $1 [] }
-  | '$' Ident		{ getvar $2 }
-
-MkDict
-  :                     { [] }
-  | MkDictPair          { [$1] }
-  | MkDictPair ',' MkDict { $1:$3 }
-
-MkDictPair
-  : Ident ':' ExpD      { (constStr $1, $3) }
-  | Ident               { (constStr $1, index $1) }
-  | String ':' ExpD     { (constStr $1, $3) }
-  | '(' Exp ')' ':' ExpD{ ($2, $5) }

PrettyJSON.hs

@@ -1,12 +0,0 @@
-module PrettyJSON where
-import Text.JSON
-import Text.PrettyPrint
-
-renderJSON (JSArray vals) = brackets $ fsep $ punctuate comma $ map renderJSON vals
-
-renderJSON (JSObject jsObject) = 
-  let object = fromJSObject jsObject
-  in braces $ fsep $ punctuate comma $ 
-     [hang (renderJSON (JSString $ toJSString $ k) <> colon) 2 (renderJSON v)
-     | (k,v) <- object]
-renderJSON x = text $ encode x

README.md

@@ -0,0 +1,14 @@
+jq
+==
+
+jq is a command-line JSON processor.
+
+If you want to learn to use jq, read the documentation at
+[http://stedolan.github.com/jq](http://stedolan.github.com/jq). This
+documentation is generated from the docs/ folder of this repository.
+
+If you want to hack on jq, feel free, but be warned that its internals
+are not well-documented at the moment. Bring a hard hat and a shovel.
+
+You can find some basic build instructions at
+[http://stedolan.github.com/jq/download](http://stedolan.github.com/jq/download)

VERSION

@@ -0,0 +1 @@
+1.2

builtin.c

@@ -0,0 +1,382 @@
+#include <string.h>
+#include "builtin.h"
+#include "bytecode.h"
+#include "compile.h"
+#include "parser.h"
+#include "locfile.h"
+#include "jv_aux.h"
+
+
+typedef jv (*func_1)(jv);
+typedef jv (*func_2)(jv,jv);
+typedef jv (*func_3)(jv,jv,jv);
+typedef jv (*func_4)(jv,jv,jv,jv);
+typedef jv (*func_5)(jv,jv,jv,jv,jv);
+jv cfunction_invoke(struct cfunction* function, jv input[]) {
+  switch (function->nargs) {
+  case 1: return ((func_1)function->fptr)(input[0]);
+  case 2: return ((func_2)function->fptr)(input[0], input[1]);
+  case 3: return ((func_3)function->fptr)(input[0], input[1], input[2]);
+  case 4: return ((func_4)function->fptr)(input[0], input[1], input[2], input[3]);
+  case 5: return ((func_5)function->fptr)(input[0], input[1], input[2], input[3], input[4]);
+  default: return jv_invalid_with_msg(jv_string("Function takes too many arguments"));
+  }
+}
+
+static jv type_error(jv bad, const char* msg) {
+  jv err = jv_invalid_with_msg(jv_string_fmt("%s %s",
+                                             jv_kind_name(jv_get_kind(bad)),
+                                             msg));
+  jv_free(bad);
+  return err;
+}
+
+static jv type_error2(jv bad1, jv bad2, const char* msg) {
+  jv err = jv_invalid_with_msg(jv_string_fmt("%s and %s %s",
+                                             jv_kind_name(jv_get_kind(bad1)),
+                                             jv_kind_name(jv_get_kind(bad2)),
+                                             msg));
+  jv_free(bad1);
+  jv_free(bad2);
+  return err;
+}
+
+static jv f_plus(jv input, jv a, jv b) {
+  jv_free(input);
+  if (jv_get_kind(a) == JV_KIND_NUMBER && jv_get_kind(b) == JV_KIND_NUMBER) {
+    return jv_number(jv_number_value(a) + 
+                     jv_number_value(b));
+  } else if (jv_get_kind(a) == JV_KIND_STRING && jv_get_kind(b) == JV_KIND_STRING) {
+    return jv_string_concat(a, b);
+  } else if (jv_get_kind(a) == JV_KIND_ARRAY && jv_get_kind(b) == JV_KIND_ARRAY) {
+    return jv_array_concat(a, b);
+  } else if (jv_get_kind(a) == JV_KIND_OBJECT && jv_get_kind(b) == JV_KIND_OBJECT) {
+    return jv_object_merge(a, b);
+  } else {
+    return type_error2(a, b, "cannot be added");
+  }
+}
+
+static jv f_minus(jv input, jv a, jv b) {
+  jv_free(input);
+  if (jv_get_kind(a) == JV_KIND_NUMBER && jv_get_kind(b) == JV_KIND_NUMBER) {
+    return jv_number(jv_number_value(a) - jv_number_value(b));
+  } else if (jv_get_kind(a) == JV_KIND_ARRAY && jv_get_kind(b) == JV_KIND_ARRAY) {
+    jv out = jv_array();
+    for (int i=0; i<jv_array_length(jv_copy(a)); i++) {
+      jv x = jv_array_get(jv_copy(a), i);
+      int include = 1;
+      for (int j=0; j<jv_array_length(jv_copy(b)); j++) {
+        if (jv_equal(jv_copy(x), jv_array_get(jv_copy(b), j))) {
+          include = 0;
+          break;
+        }
+      }
+      if (include)
+        out = jv_array_append(out, jv_copy(x));
+      jv_free(x);
+    }
+    jv_free(a);
+    jv_free(b);
+    return out;
+  } else {
+    return type_error2(a, b, "cannot be subtracted");
+  }
+}
+
+static jv f_multiply(jv input, jv a, jv b) {
+  jv_free(input);
+  if (jv_get_kind(a) == JV_KIND_NUMBER && jv_get_kind(b) == JV_KIND_NUMBER) {
+    return jv_number(jv_number_value(a) * jv_number_value(b));
+  } else {
+    return type_error2(a, b, "cannot be multiplied");
+  }  
+}
+
+static jv f_divide(jv input, jv a, jv b) {
+  jv_free(input);
+  if (jv_get_kind(a) == JV_KIND_NUMBER && jv_get_kind(b) == JV_KIND_NUMBER) {
+    return jv_number(jv_number_value(a) / jv_number_value(b));
+  } else {
+    return type_error2(a, b, "cannot be divided");
+  }  
+}
+
+static jv f_add(jv array) {
+  if (jv_get_kind(array) != JV_KIND_ARRAY) {
+    return type_error(array, "cannot have its elements added");
+  } else if (jv_array_length(jv_copy(array)) == 0) {
+    jv_free(array);
+    return jv_null();
+  } else {
+    jv sum = jv_array_get(jv_copy(array), 0);
+    for (int i = 1; i < jv_array_length(jv_copy(array)); i++) {
+      sum = f_plus(jv_null(), sum, jv_array_get(jv_copy(array), i));
+    }
+    jv_free(array);
+    return sum;
+  }
+}
+
+static jv f_equal(jv input, jv a, jv b) {
+  jv_free(input);
+  return jv_bool(jv_equal(a, b));
+}
+
+static jv f_notequal(jv input, jv a, jv b) {
+  jv_free(input);
+  return jv_bool(!jv_equal(a, b));
+}
+
+enum cmp_op {
+  CMP_OP_LESS,
+  CMP_OP_GREATER,
+  CMP_OP_LESSEQ,
+  CMP_OP_GREATEREQ
+};
+
+static jv order_cmp(jv input, jv a, jv b, enum cmp_op op) {
+  jv_free(input);
+  int r = jv_cmp(a, b);
+  return jv_bool((op == CMP_OP_LESS && r < 0) ||
+                 (op == CMP_OP_LESSEQ && r <= 0) ||
+                 (op == CMP_OP_GREATEREQ && r >= 0) ||
+                 (op == CMP_OP_GREATER && r > 0));
+}
+
+static jv f_less(jv input, jv a, jv b) {
+  return order_cmp(input, a, b, CMP_OP_LESS);
+}
+
+static jv f_greater(jv input, jv a, jv b) {
+  return order_cmp(input, a, b, CMP_OP_GREATER);
+}
+
+static jv f_lesseq(jv input, jv a, jv b) {
+  return order_cmp(input, a, b, CMP_OP_LESSEQ);
+}
+
+static jv f_greatereq(jv input, jv a, jv b) {
+  return order_cmp(input, a, b, CMP_OP_GREATEREQ);
+}
+
+static jv f_contains(jv a, jv b) {
+  jv_kind akind = jv_get_kind(a);
+
+  if (akind == jv_get_kind(b)) {
+    return jv_bool(jv_contains(a, b));
+  } else {
+    return type_error2(a, b, "cannot have their containment checked");
+  }
+}
+
+static jv f_tonumber(jv input) {
+  if (jv_get_kind(input) == JV_KIND_NUMBER) {
+    return input;
+  }
+  if (jv_get_kind(input) == JV_KIND_STRING) {
+    jv parsed = jv_parse(jv_string_value(input));
+    if (!jv_is_valid(parsed) || jv_get_kind(parsed) == JV_KIND_NUMBER) {
+      jv_free(input);
+      return parsed;
+    }
+  }
+  return type_error(input, "cannot be parsed as a number");
+}
+
+static jv f_length(jv input) {
+  if (jv_get_kind(input) == JV_KIND_ARRAY) {
+    return jv_number(jv_array_length(input));
+  } else if (jv_get_kind(input) == JV_KIND_OBJECT) {
+    return jv_number(jv_object_length(input));
+  } else if (jv_get_kind(input) == JV_KIND_STRING) {
+    return jv_number(jv_string_length(input));
+  } else if (jv_get_kind(input) == JV_KIND_NULL) {
+    jv_free(input);
+    return jv_number(0);
+  } else {
+    return type_error(input, "has no length");
+  }
+}
+
+static jv f_tostring(jv input) {
+  if (jv_get_kind(input) == JV_KIND_STRING) {
+    return input;
+  } else {
+    return jv_dump_string(input, 0);
+  }
+}
+
+static jv f_keys(jv input) {
+  if (jv_get_kind(input) == JV_KIND_OBJECT || jv_get_kind(input) == JV_KIND_ARRAY) {
+    return jv_keys(input);
+  } else {
+    return type_error(input, "has no keys");
+  }
+}
+
+static jv f_sort(jv input){
+  if (jv_get_kind(input) == JV_KIND_ARRAY) {
+    return jv_sort(input, jv_copy(input));
+  } else {
+    return type_error(input, "cannot be sorted, as it is not an array");
+  }
+}
+
+static jv f_sort_by_impl(jv input, jv keys) {
+  if (jv_get_kind(input) == JV_KIND_ARRAY && 
+      jv_get_kind(keys) == JV_KIND_ARRAY &&
+      jv_array_length(jv_copy(input)) == jv_array_length(jv_copy(keys))) {
+    return jv_sort(input, keys);
+  } else {
+    return type_error2(input, keys, "cannot be sorted, as they are not both arrays");
+  }
+}
+
+static jv f_group_by_impl(jv input, jv keys) {
+  if (jv_get_kind(input) == JV_KIND_ARRAY && 
+      jv_get_kind(keys) == JV_KIND_ARRAY &&
+      jv_array_length(jv_copy(input)) == jv_array_length(jv_copy(keys))) {
+    return jv_group(input, keys);
+  } else {
+    return type_error2(input, keys, "cannot be sorted, as they are not both arrays");
+  }
+}
+
+static jv minmax_by(jv values, jv keys, int is_min) {
+  if (jv_get_kind(values) != JV_KIND_ARRAY)
+    return type_error2(values, keys, "cannot be iterated over");
+  if (jv_get_kind(keys) != JV_KIND_ARRAY)
+    return type_error2(values, keys, "cannot be iterated over");
+  if (jv_array_length(jv_copy(values)) != jv_array_length(jv_copy(keys)))
+    return type_error2(values, keys, "have wrong length");
+
+  if (jv_array_length(jv_copy(values)) == 0) {
+    jv_free(values);
+    jv_free(keys);
+    return jv_null();
+  }
+  jv ret = jv_array_get(jv_copy(values), 0);
+  jv retkey = jv_array_get(jv_copy(keys), 0);
+  for (int i=1; i<jv_array_length(jv_copy(values)); i++) {
+    jv item = jv_array_get(jv_copy(keys), i);
+    int cmp = jv_cmp(jv_copy(item), jv_copy(retkey));
+    if ((cmp < 0) == (is_min == 1)) {
+      jv_free(retkey);
+      retkey = item;
+      jv_free(ret);
+      ret = jv_array_get(jv_copy(values), i);
+    } else {
+      jv_free(item);
+    }
+  }
+  jv_free(values);
+  jv_free(keys);
+  jv_free(retkey);
+  return ret;
+}
+
+static jv f_min(jv x) {
+  return minmax_by(x, jv_copy(x), 1);
+}
+
+static jv f_max(jv x) {
+  return minmax_by(x, jv_copy(x), 0);
+}
+
+static jv f_min_by_impl(jv x, jv y) {
+  return minmax_by(x, y, 1);
+}
+
+static jv f_max_by_impl(jv x, jv y) {
+  return minmax_by(x, y, 0);
+}
+
+
+static jv f_type(jv input) {
+  jv out = jv_string(jv_kind_name(jv_get_kind(input)));
+  jv_free(input);
+  return out;
+}
+
+static jv f_error(jv input, jv msg) {
+  jv_free(input);
+  msg = f_tostring(msg);
+  return jv_invalid_with_msg(msg);
+}
+
+
+static struct cfunction function_list[] = {
+  {(cfunction_ptr)f_plus, "_plus", 3},
+  {(cfunction_ptr)f_minus, "_minus", 3},
+  {(cfunction_ptr)f_multiply, "_multiply", 3},
+  {(cfunction_ptr)f_divide, "_divide", 3},
+  {(cfunction_ptr)f_tonumber, "tonumber", 1},
+  {(cfunction_ptr)f_tostring, "tostring", 1},
+  {(cfunction_ptr)f_keys, "keys", 1},
+  {(cfunction_ptr)f_equal, "_equal", 3},
+  {(cfunction_ptr)f_notequal, "_notequal", 3},
+  {(cfunction_ptr)f_less, "_less", 3},
+  {(cfunction_ptr)f_greater, "_greater", 3},
+  {(cfunction_ptr)f_lesseq, "_lesseq", 3},
+  {(cfunction_ptr)f_greatereq, "_greatereq", 3},
+  {(cfunction_ptr)f_contains, "contains", 2},
+  {(cfunction_ptr)f_length, "length", 1},
+  {(cfunction_ptr)f_type, "type", 1},
+  {(cfunction_ptr)f_add, "add", 1},
+  {(cfunction_ptr)f_sort, "sort", 1},
+  {(cfunction_ptr)f_sort_by_impl, "_sort_by_impl", 2},
+  {(cfunction_ptr)f_group_by_impl, "_group_by_impl", 2},
+  {(cfunction_ptr)f_min, "min", 1},
+  {(cfunction_ptr)f_max, "max", 1},
+  {(cfunction_ptr)f_min_by_impl, "_min_by_impl", 2},
+  {(cfunction_ptr)f_max_by_impl, "_max_by_impl", 2},
+  {(cfunction_ptr)f_error, "error", 2},
+};
+
+static struct symbol_table cbuiltins = 
+  {function_list, sizeof(function_list)/sizeof(function_list[0])};
+
+typedef block (*bytecoded_builtin)();
+struct bytecoded_builtin { const char* name; block code; };
+static block bind_bytecoded_builtins(block b) {
+  struct bytecoded_builtin builtin_defs[] = {
+    {"empty", gen_op_simple(BACKTRACK)},
+    {"false", gen_const(jv_false())},
+    {"true", gen_const(jv_true())},
+    {"null", gen_const(jv_null())},
+    {"not", gen_condbranch(gen_const(jv_false()),
+                           gen_const(jv_true()))}
+  };
+  block builtins = gen_noop();
+  for (unsigned i=0; i<sizeof(builtin_defs)/sizeof(builtin_defs[0]); i++) {
+    builtins = BLOCK(builtins, gen_function(builtin_defs[i].name, gen_noop(),
+                                            builtin_defs[i].code));
+  }
+  return block_bind(builtins, b, OP_IS_CALL_PSEUDO);
+}
+
+static const char* jq_builtins[] = {
+  "def map(f): [.[] | f];",
+  "def select(f): if f then . else empty end;",
+  "def sort_by(f): _sort_by_impl(map([f]));",
+  "def group_by(f): _group_by_impl(map([f]));",
+  "def unique: group_by(.) | map(.[0]);",
+  "def max_by(f): _max_by_impl(map([f]));",
+  "def min_by(f): _min_by_impl(map([f]));",
+};
+
+
+block builtins_bind(block b) {
+  for (int i=(int)(sizeof(jq_builtins)/sizeof(jq_builtins[0]))-1; i>=0; i--) {
+    struct locfile src;
+    locfile_init(&src, jq_builtins[i], strlen(jq_builtins[i]));
+    block funcs;
+    int nerrors = jq_parse_library(&src, &funcs);
+    assert(!nerrors);
+    b = block_bind(funcs, b, OP_IS_CALL_PSEUDO);
+    locfile_free(&src);
+  }
+  b = bind_bytecoded_builtins(b);
+  return gen_cbinding(&cbuiltins, b);
+}

builtin.h

@@ -0,0 +1,21 @@
+#ifndef BUILTIN_H
+#define BUILTIN_H
+
+#include "compile.h"
+
+block builtins_bind(block);
+
+
+typedef void (*cfunction_ptr)(void);
+
+struct cfunction {
+  cfunction_ptr fptr;
+  const char* name;
+  int nargs;
+};
+
+
+jv cfunction_invoke(struct cfunction* function, jv input[]);
+
+
+#endif

bytecode.c

@@ -0,0 +1,89 @@
+#include <stdio.h>
+#include <stdint.h>
+#include <stdlib.h>
+
+#include "bytecode.h"
+#include "opcode.h"
+#include "jv_alloc.h"
+
+static int bytecode_operation_length(uint16_t* codeptr) {
+  int length = opcode_describe(*codeptr)->length;
+  if (*codeptr == CALL_JQ) {
+    length += codeptr[1] * 2;
+  }
+  return length;
+}
+
+void dump_disassembly(int indent, struct bytecode* bc) {
+  if (bc->nclosures > 0) {
+    printf("%*s[params: %d]\n", indent, "", bc->nclosures);
+  }
+  dump_code(indent, bc);
+  for (int i=0; i<bc->nsubfunctions; i++) {
+    printf("%*ssubfn[%d]:\n", indent, "", i);
+    dump_disassembly(indent+2, bc->subfunctions[i]);
+  }
+}
+
+void dump_code(int indent, struct bytecode* bc) {
+  int pc = 0;
+  while (pc < bc->codelen) {
+    printf("%*s", indent, "");
+    dump_operation(bc, bc->code + pc);
+    printf("\n");
+    pc += bytecode_operation_length(bc->code + pc);
+  }
+}
+
+void dump_operation(struct bytecode* bc, uint16_t* codeptr) {
+  int pc = codeptr - bc->code;
+  printf("%04d ", pc);
+  const struct opcode_description* op = opcode_describe(bc->code[pc++]);
+  printf("%s", op->name);
+  if (op->length > 1) {
+    uint16_t imm = bc->code[pc++];
+    if (op->op == CALL_JQ) {
+      for (int i=0; i<imm+1; i++) {
+        uint16_t level = bc->code[pc++];
+        uint16_t idx = bc->code[pc++];
+        if (idx & ARG_NEWCLOSURE) {
+          printf(" subfn[%d]", idx & ~ARG_NEWCLOSURE);
+        } else {
+          printf(" param[%d]", idx);
+        }
+        if (level) {
+          printf("^%d", level);
+        }
+      }
+    } else if (op->flags & OP_HAS_BRANCH) {
+      printf(" %04d", pc + imm);
+    } else if (op->flags & OP_HAS_CONSTANT) {
+      printf(" ");
+      jv_dump(jv_array_get(jv_copy(bc->constants), imm), 0);
+    } else if (op->flags & OP_HAS_VARIABLE) {
+      uint16_t v = bc->code[pc++];
+      printf(" v%d", v);
+      if (imm) {
+        printf("^%d", imm);
+      }
+    } else {
+      printf(" %d", imm);
+    }
+  }  
+}
+
+void symbol_table_free(struct symbol_table* syms) {
+  jv_mem_free(syms->cfunctions);
+  jv_mem_free(syms);
+}
+
+void bytecode_free(struct bytecode* bc) {
+  jv_mem_free(bc->code);
+  jv_free(bc->constants);
+  for (int i=0; i<bc->nsubfunctions; i++)
+    bytecode_free(bc->subfunctions[i]);
+  if (!bc->parent)
+    symbol_table_free(bc->globals);
+  jv_mem_free(bc->subfunctions);
+  jv_mem_free(bc);
+}

bytecode.h

@@ -0,0 +1,45 @@
+#ifndef BYTECODE_H
+#define BYTECODE_H
+#include <stdint.h>
+
+#include "jv.h"
+#include "opcode.h"
+#include "builtin.h"
+
+#define MAX_CFUNCTION_ARGS 10
+struct symbol_table {
+  struct cfunction* cfunctions;
+  int ncfunctions;
+};
+
+// The bytecode format matters in:
+//   execute.c  - interpreter
+//   compile.c  - compiler
+//   bytecode.c - disassembler
+
+#define ARG_NEWCLOSURE 0x1000
+
+struct bytecode {
+  uint16_t* code;
+  int codelen;
+
+  int nlocals;
+  int nclosures;
+
+  jv constants; // JSON array of constants
+  struct symbol_table* globals;
+
+  struct bytecode** subfunctions;
+  int nsubfunctions;
+
+  struct bytecode* parent;
+};
+
+void dump_disassembly(int, struct bytecode* code);
+void dump_code(int, struct bytecode* code);
+void dump_operation(struct bytecode* bc, uint16_t* op);
+
+void symbol_table_free(struct symbol_table* syms);
+void bytecode_free(struct bytecode* bc);
+
+#endif

compile.c

@@ -0,0 +1,649 @@
+#define _GNU_SOURCE // for strdup
+#include <assert.h>
+#include <string.h>
+#include <stdlib.h>
+#include "opcode.h"
+#include "compile.h"
+#include "bytecode.h"
+#include "locfile.h"
+#include "jv_alloc.h"
+
+/*
+  The intermediate representation for jq filters is as a sequence of
+  struct inst, which form a doubly-linked list via the next and prev
+  pointers. 
+
+  A "block" represents a sequence of "struct inst", which may be
+  empty.
+
+  Blocks are generated by the parser bottom-up, so may have free
+  variables (refer to things not defined). See inst.bound_by and
+  inst.symbol.
+ */
+struct inst {
+  struct inst* next;
+  struct inst* prev;
+
+  opcode op;
+  
+  struct {
+    uint16_t intval;
+    struct inst* target;
+    jv constant;
+    struct cfunction* cfunc;
+  } imm;
+
+  location source;
+
+  // Binding
+  // An instruction requiring binding (for parameters/variables/functions)
+  // is in one of three states:
+  //   inst->bound_by = NULL  - Unbound free variable
+  //   inst->bound_by = inst  - This instruction binds a variable
+  //   inst->bound_by = other - Uses variable bound by other instruction
+  // Unbound instructions (references to other things that may or may not 
+  // exist) are created by "gen_foo_unbound", and bindings are created by
+  // block_bind(definition, body), which binds all instructions in
+  // body which are unboudn and refer to "definition" by name.
+  struct inst* bound_by;
+  char* symbol;
+
+  block subfn;   // used by CLOSURE_CREATE (body of function)
+  block arglist; // used by CLOSURE_CREATE (formals) and CALL_JQ (arguments)
+
+  // This instruction is compiled as part of which function?
+  // (only used during block_compile)
+  struct bytecode* compiled;
+
+  int bytecode_pos; // position just after this insn
+};
+
+static inst* inst_new(opcode op) {
+  inst* i = jv_mem_alloc(sizeof(inst));
+  i->next = i->prev = 0;
+  i->op = op;
+  i->bytecode_pos = -1;
+  i->bound_by = 0;
+  i->symbol = 0;
+  i->subfn = gen_noop();
+  i->arglist = gen_noop();
+  i->source = UNKNOWN_LOCATION;
+  return i;
+}
+
+static void inst_free(struct inst* i) {
+  jv_mem_free(i->symbol);
+  block_free(i->subfn);
+  block_free(i->arglist);
+  if (opcode_describe(i->op)->flags & OP_HAS_CONSTANT) {
+    jv_free(i->imm.constant);
+  }
+  jv_mem_free(i);
+}
+
+static block inst_block(inst* i) {
+  block b = {i,i};
+  return b;
+}
+
+static int block_is_single(block b) {
+  return b.first && b.first == b.last;
+}
+
+static inst* block_take(block* b) {
+  if (b->first == 0) return 0;
+  inst* i = b->first;
+  if (i->next) {
+    i->next->prev = 0;
+    b->first = i->next;
+    i->next = 0;
+  } else {
+    b->first = 0;
+    b->last = 0;
+  }
+  return i;
+}
+
+block gen_location(location loc, block b) {
+  for (inst* i = b.first; i; i = i->next) {
+    if (i->source.start == UNKNOWN_LOCATION.start &&
+        i->source.end == UNKNOWN_LOCATION.end) {
+      i->source = loc;
+    }
+  }
+  return b;
+}
+
+block gen_noop() {
+  block b = {0,0};
+  return b;
+}
+
+block gen_op_simple(opcode op) {
+  assert(opcode_describe(op)->length == 1);
+  return inst_block(inst_new(op));
+}
+
+
+block gen_const(jv constant) {
+  assert(opcode_describe(LOADK)->flags & OP_HAS_CONSTANT);
+  inst* i = inst_new(LOADK);
+  i->imm.constant = constant;
+  return inst_block(i);
+}
+
+block gen_op_target(opcode op, block target) {
+  assert(opcode_describe(op)->flags & OP_HAS_BRANCH);
+  assert(target.last);
+  inst* i = inst_new(op);
+  i->imm.target = target.last;
+  return inst_block(i);
+}
+
+block gen_op_targetlater(opcode op) {
+  assert(opcode_describe(op)->flags & OP_HAS_BRANCH);
+  inst* i = inst_new(op);
+  i->imm.target = 0;
+  return inst_block(i);
+}
+void inst_set_target(block b, block target) {
+  assert(block_is_single(b));
+  assert(opcode_describe(b.first->op)->flags & OP_HAS_BRANCH);
+  assert(target.last);
+  b.first->imm.target = target.last;
+}
+
+block gen_op_var_unbound(opcode op, const char* name) {
+  assert(opcode_describe(op)->flags & OP_HAS_VARIABLE);
+  inst* i = inst_new(op);
+  i->symbol = strdup(name);
+  return inst_block(i);
+}
+
+block gen_op_var_bound(opcode op, block binder) {
+  assert(block_is_single(binder));
+  block b = gen_op_var_unbound(op, binder.first->symbol);
+  b.first->bound_by = binder.first;
+  return b;
+}
+
+block gen_op_block_unbound(opcode op, const char* name) {
+  assert(opcode_describe(op)->flags & OP_IS_CALL_PSEUDO);
+  inst* i = inst_new(op);
+  i->symbol = strdup(name);
+  return inst_block(i);
+}
+
+block gen_op_block_bound(opcode op, block binder) {
+  assert(block_is_single(binder));
+  block b = gen_op_block_unbound(op, binder.first->symbol);
+  b.first->bound_by = binder.first;
+  return b;
+}
+
+static void inst_join(inst* a, inst* b) {
+  assert(a && b);
+  assert(!a->next);
+  assert(!b->prev);
+  a->next = b;
+  b->prev = a;
+}
+
+void block_append(block* b, block b2) {
+  if (b2.first) {
+    if (b->last) {
+      inst_join(b->last, b2.first);
+    } else {
+      b->first = b2.first;
+    }
+    b->last = b2.last;
+  }
+}
+
+block block_join(block a, block b) {
+  block c = a;
+  block_append(&c, b);
+  return c;
+}
+
+int block_has_only_binders(block binders, int bindflags) {
+  bindflags |= OP_HAS_BINDING;
+  for (inst* curr = binders.first; curr; curr = curr->next) {
+    if ((opcode_describe(curr->op)->flags & bindflags) != bindflags) {
+      return 0;
+    }
+  }
+  return 1;
+}
+
+static void block_bind_subblock(block binder, block body, int bindflags) {
+  assert(block_is_single(binder));
+  assert((opcode_describe(binder.first->op)->flags & bindflags) == bindflags);
+  assert(binder.first->symbol);
+  assert(binder.first->bound_by == 0 || binder.first->bound_by == binder.first);
+
+  binder.first->bound_by = binder.first;
+  for (inst* i = body.first; i; i = i->next) {
+    int flags = opcode_describe(i->op)->flags;
+    if ((flags & bindflags) == bindflags &&
+        i->bound_by == 0 &&
+        !strcmp(i->symbol, binder.first->symbol)) {
+      // bind this instruction
+      i->bound_by = binder.first;
+    }
+    // binding recurses into closures
+    block_bind_subblock(binder, i->subfn, bindflags);
+    // binding recurses into argument list
+    block_bind_subblock(binder, i->arglist, bindflags);
+  }
+}
+
+static void block_bind_each(block binder, block body, int bindflags) {
+  assert(block_has_only_binders(binder, bindflags));
+  bindflags |= OP_HAS_BINDING;
+  for (inst* curr = binder.first; curr; curr = curr->next) {
+    block_bind_subblock(inst_block(curr), body, bindflags);
+  }
+}
+
+block block_bind(block binder, block body, int bindflags) {
+  block_bind_each(binder, body, bindflags);
+  return block_join(binder, body);
+}
+
+block gen_function(const char* name, block formals, block body) {
+  block_bind_each(formals, body, OP_IS_CALL_PSEUDO);
+  inst* i = inst_new(CLOSURE_CREATE);
+  i->subfn = body;
+  i->symbol = strdup(name);
+  i->arglist = formals;
+  block b = inst_block(i);
+  block_bind_subblock(b, b, OP_IS_CALL_PSEUDO | OP_HAS_BINDING);
+  return b;
+}
+
+block gen_lambda(block body) {
+  return gen_function("@lambda", gen_noop(), body);
+}
+
+block gen_call(const char* name, block args) {
+  block b = gen_op_block_unbound(CALL_JQ, name);
+  b.first->arglist = args;
+  return b;
+}
+
+
+
+block gen_subexp(block a) {
+  return BLOCK(gen_op_simple(DUP), a, gen_op_simple(SWAP));
+}
+
+block gen_both(block a, block b) {
+  block jump = gen_op_targetlater(JUMP);
+  block fork = gen_op_target(FORK, jump);
+  block c = BLOCK(fork, a, jump, b);
+  inst_set_target(jump, c);
+  return c;
+}
+
+
+block gen_collect(block expr) {
+  block array_var = block_bind(gen_op_var_unbound(STOREV, "collect"),
+                               gen_noop(), OP_HAS_VARIABLE);
+  block c = BLOCK(gen_op_simple(DUP), gen_const(jv_array()), array_var);
+
+  block tail = BLOCK(gen_op_simple(DUP),
+                     gen_op_var_bound(LOADV, array_var),
+                     gen_op_simple(SWAP),
+                     gen_op_simple(APPEND),
+                     gen_op_var_bound(STOREV, array_var),
+                     gen_op_simple(BACKTRACK));
+
+  return BLOCK(c,
+               gen_op_target(FORK, tail),
+               expr, 
+               tail,
+               gen_op_var_bound(LOADV, array_var));
+}
+
+block gen_assign(block expr) {
+  block result_var = block_bind(gen_op_var_unbound(STOREV, "result"),
+                                gen_noop(), OP_HAS_VARIABLE);
+
+  block loop = BLOCK(gen_op_simple(DUP),
+                     expr,
+                     gen_op_var_bound(ASSIGN, result_var),
+                     gen_op_simple(BACKTRACK));
+
+  return BLOCK(gen_op_simple(DUP),
+               result_var,
+               gen_op_target(FORK, loop),
+               loop,
+               gen_op_var_bound(LOADV, result_var));
+}
+
+block gen_definedor(block a, block b) {
+  // var found := false
+  block found_var = block_bind(gen_op_var_unbound(STOREV, "found"),
+                               gen_noop(), OP_HAS_VARIABLE);
+  block init = BLOCK(gen_op_simple(DUP), gen_const(jv_false()), found_var);
+
+  // if found, backtrack. Otherwise execute b
+  block backtrack = gen_op_simple(BACKTRACK);
+  block tail = BLOCK(gen_op_simple(DUP),
+                     gen_op_var_bound(LOADV, found_var),
+                     gen_op_target(JUMP_F, backtrack),
+                     backtrack,
+                     gen_op_simple(POP),
+                     b);
+
+  // try again
+  block if_notfound = gen_op_simple(BACKTRACK);
+
+  // found := true, produce result
+  block if_found = BLOCK(gen_op_simple(DUP),
+                         gen_const(jv_true()),
+                         gen_op_var_bound(STOREV, found_var),
+                         gen_op_target(JUMP, tail));
+
+  return BLOCK(init,
+               gen_op_target(FORK, if_notfound),
+               a,
+               gen_op_target(JUMP_F, if_found),
+               if_found,
+               if_notfound,
+               tail);
+}
+
+block gen_condbranch(block iftrue, block iffalse) {
+  iftrue = BLOCK(iftrue, gen_op_target(JUMP, iffalse));
+  return BLOCK(gen_op_target(JUMP_F, iftrue), iftrue, iffalse);
+}
+
+block gen_and(block a, block b) {
+  // a and b = if a then (if b then true else false) else false
+  return BLOCK(gen_op_simple(DUP), a, 
+               gen_condbranch(BLOCK(gen_op_simple(POP),
+                                    b,
+                                    gen_condbranch(gen_const(jv_true()),
+                                                   gen_const(jv_false()))),
+                              BLOCK(gen_op_simple(POP), gen_const(jv_false()))));
+}
+
+block gen_or(block a, block b) {
+  // a or b = if a then true else (if b then true else false)
+  return BLOCK(gen_op_simple(DUP), a,
+               gen_condbranch(BLOCK(gen_op_simple(POP), gen_const(jv_true())),
+                              BLOCK(gen_op_simple(POP),
+                                    b,
+                                    gen_condbranch(gen_const(jv_true()),
+                                                   gen_const(jv_false())))));
+}
+
+block gen_cond(block cond, block iftrue, block iffalse) {
+  return BLOCK(gen_op_simple(DUP), cond, 
+               gen_condbranch(BLOCK(gen_op_simple(POP), iftrue),
+                              BLOCK(gen_op_simple(POP), iffalse)));
+}
+
+block gen_cbinding(struct symbol_table* t, block code) {
+  for (int cfunc=0; cfunc<t->ncfunctions; cfunc++) {
+    inst* i = inst_new(CLOSURE_CREATE_C);
+    i->imm.cfunc = &t->cfunctions[cfunc];
+    i->symbol = strdup(i->imm.cfunc->name);
+    code = block_bind(inst_block(i), code, OP_IS_CALL_PSEUDO);
+  }
+  return code;
+}
+
+static uint16_t nesting_level(struct bytecode* bc, inst* target) {
+  uint16_t level = 0;
+  assert(bc && target->compiled);
+  while (bc && target->compiled != bc) {
+    level++;
+    bc = bc->parent;
+  }
+  assert(bc && bc == target->compiled);
+  return level;
+}
+
+static int count_cfunctions(block b) {
+  int n = 0;
+  for (inst* i = b.first; i; i = i->next) {
+    if (i->op == CLOSURE_CREATE_C) n++;
+    n += count_cfunctions(i->subfn);
+  }
+  return n;
+}
+
+
+// Expands call instructions into a calling sequence
+static int expand_call_arglist(struct locfile* locations, block* b) {
+  int errors = 0;
+  block ret = gen_noop();
+  for (inst* curr; (curr = block_take(b));) {
+    if (opcode_describe(curr->op)->flags & OP_HAS_BINDING) {
+      if (!curr->bound_by) {
+        locfile_locate(locations, curr->source, "error: %s is not defined", curr->symbol);
+        errors++;
+        // don't process this instruction if it's not well-defined
+        ret = BLOCK(ret, inst_block(curr));
+        continue;
+      }
+    }
+
+    block prelude = gen_noop();
+    if (curr->op == CALL_JQ) {
+      int actual_args = 0, desired_args = 0;
+      // We expand the argument list as a series of instructions
+      switch (curr->bound_by->op) {
+      default: assert(0 && "Unknown function type"); break;
+      case CLOSURE_CREATE: 
+      case CLOSURE_PARAM: {
+        block callargs = gen_noop();
+        for (inst* i; (i = block_take(&curr->arglist));) {
+          assert(opcode_describe(i->op)->flags & OP_IS_CALL_PSEUDO);
+          block b = inst_block(i);
+          switch (i->op) {
+          default: assert(0 && "Unknown type of parameter"); break;
+          case CLOSURE_REF:
+            block_append(&callargs, b);
+            break;
+          case CLOSURE_CREATE:
+            block_append(&prelude, b);
+            block_append(&callargs, gen_op_block_bound(CLOSURE_REF, b));
+            break;
+          }
+          actual_args++;
+        }
+        curr->imm.intval = actual_args;
+        curr->arglist = callargs;
+
+        if (curr->bound_by->op == CLOSURE_CREATE) {
+          for (inst* i = curr->bound_by->arglist.first; i; i = i->next) {
+            assert(i->op == CLOSURE_PARAM);
+            desired_args++;
+          }
+        }
+        break;
+      }
+
+      case CLOSURE_CREATE_C: {
+        for (inst* i; (i = block_take(&curr->arglist)); ) {
+          assert(i->op == CLOSURE_CREATE); // FIXME
+          block body = i->subfn;
+          i->subfn = gen_noop();
+          inst_free(i);
+          // arguments should be pushed in reverse order, prepend them to prelude
+          errors += expand_call_arglist(locations, &body);
+          prelude = BLOCK(gen_subexp(body), prelude);
+          actual_args++;
+        }
+        assert(curr->op == CALL_JQ);
+        curr->op = CALL_BUILTIN;
+        curr->imm.intval = actual_args + 1 /* include the implicit input in arg count */;
+        assert(curr->bound_by->op == CLOSURE_CREATE_C);
+        desired_args = curr->bound_by->imm.cfunc->nargs - 1;
+        assert(!curr->arglist.first);
+        break;
+      }
+      }
+
+      if (actual_args != desired_args) {
+        locfile_locate(locations, curr->source, 
+                       "error: %s arguments to %s (expected %d but got %d)",
+                       actual_args > desired_args ? "too many" : "too few",
+                       curr->symbol, desired_args, actual_args);
+        errors++;
+      }
+
+    }
+    ret = BLOCK(ret, prelude, inst_block(curr));
+  }
+  *b = ret;
+  return errors;
+}
+
+static int compile(struct locfile* locations, struct bytecode* bc, block b) {
+  int errors = 0;
+  int pos = 0;
+  int var_frame_idx = 0;
+  bc->nsubfunctions = 0;
+  errors += expand_call_arglist(locations, &b);
+  if (bc->parent) {
+    // functions should end in a return
+    b = BLOCK(b, gen_op_simple(RET));
+  } else {
+    // the toplevel should YIELD;BACKTRACK; when it finds an answer
+    b = BLOCK(b, gen_op_simple(YIELD), gen_op_simple(BACKTRACK));
+  }
+  for (inst* curr = b.first; curr; curr = curr->next) {
+    if (!curr->next) assert(curr == b.last);
+    int length = opcode_describe(curr->op)->length;
+    if (curr->op == CALL_JQ) {
+      for (inst* arg = curr->arglist.first; arg; arg = arg->next) {
+        length += 2;
+      }
+    }
+    pos += length;
+    curr->bytecode_pos = pos;
+    curr->compiled = bc;
+
+    assert(curr->op != CLOSURE_REF && curr->op != CLOSURE_PARAM);
+
+    if ((opcode_describe(curr->op)->flags & OP_HAS_VARIABLE) &&
+        curr->bound_by == curr) {
+      curr->imm.intval = var_frame_idx++;
+    }
+
+    if (curr->op == CLOSURE_CREATE) {
+      assert(curr->bound_by == curr);
+      curr->imm.intval = bc->nsubfunctions++;
+    }
+    if (curr->op == CLOSURE_CREATE_C) {
+      assert(curr->bound_by == curr);
+      int idx = bc->globals->ncfunctions++;
+      bc->globals->cfunctions[idx] = *curr->imm.cfunc;
+      curr->imm.intval = idx;
+    }
+  }
+  if (bc->nsubfunctions) {
+    bc->subfunctions = jv_mem_alloc(sizeof(struct bytecode*) * bc->nsubfunctions);
+    for (inst* curr = b.first; curr; curr = curr->next) {
+      if (curr->op == CLOSURE_CREATE) {
+        struct bytecode* subfn = jv_mem_alloc(sizeof(struct bytecode));
+        bc->subfunctions[curr->imm.intval] = subfn;
+        subfn->globals = bc->globals;
+        subfn->parent = bc;
+        subfn->nclosures = 0;
+        for (inst* param = curr->arglist.first; param; param = param->next) {
+          assert(param->op == CLOSURE_PARAM);
+          assert(param->bound_by == param);
+          param->imm.intval = subfn->nclosures++;
+          param->compiled = subfn;
+        }
+        errors += compile(locations, subfn, curr->subfn);
+        curr->subfn = gen_noop();
+      }
+    }
+  } else {
+    bc->subfunctions = 0;
+  }
+  bc->codelen = pos;
+  uint16_t* code = jv_mem_alloc(sizeof(uint16_t) * bc->codelen);
+  bc->code = code;
+  pos = 0;
+  jv constant_pool = jv_array();
+  int maxvar = -1;
+  if (!errors) for (inst* curr = b.first; curr; curr = curr->next) {
+    const struct opcode_description* op = opcode_describe(curr->op);
+    if (op->length == 0)
+      continue;
+    code[pos++] = curr->op;
+    assert(curr->op != CLOSURE_REF && curr->op != CLOSURE_PARAM);
+    if (curr->op == CALL_BUILTIN) {
+      assert(curr->bound_by->op == CLOSURE_CREATE_C);
+      assert(!curr->arglist.first);
+      code[pos++] = (uint16_t)curr->imm.intval;
+      code[pos++] = curr->bound_by->imm.intval;
+    } else if (curr->op == CALL_JQ) {
+      assert(curr->bound_by->op == CLOSURE_CREATE ||
+             curr->bound_by->op == CLOSURE_PARAM);
+      code[pos++] = (uint16_t)curr->imm.intval;
+      code[pos++] = nesting_level(bc, curr->bound_by);
+      code[pos++] = curr->bound_by->imm.intval | 
+        (curr->bound_by->op == CLOSURE_CREATE ? ARG_NEWCLOSURE : 0);
+      for (inst* arg = curr->arglist.first; arg; arg = arg->next) {
+        assert(arg->op == CLOSURE_REF && arg->bound_by->op == CLOSURE_CREATE);
+        code[pos++] = nesting_level(bc, arg->bound_by);
+        code[pos++] = arg->bound_by->imm.intval | ARG_NEWCLOSURE;
+      }
+    } else if (op->flags & OP_HAS_CONSTANT) {
+      code[pos++] = jv_array_length(jv_copy(constant_pool));
+      constant_pool = jv_array_append(constant_pool, jv_copy(curr->imm.constant));
+    } else if (op->flags & OP_HAS_VARIABLE) {
+      code[pos++] = nesting_level(bc, curr->bound_by);
+      uint16_t var = (uint16_t)curr->bound_by->imm.intval;
+      code[pos++] = var;
+      if (var > maxvar) maxvar = var;
+    } else if (op->flags & OP_HAS_BRANCH) {
+      assert(curr->imm.target->bytecode_pos != -1);
+      assert(curr->imm.target->bytecode_pos > pos); // only forward branches
+      code[pos] = curr->imm.target->bytecode_pos - (pos + 1);
+      pos++;
+    } else if (op->length > 1) {
+      assert(0 && "codegen not implemented for this operation");
+    }
+  }
+  bc->constants = constant_pool;
+  bc->nlocals = maxvar + 2; // FIXME: frames of size zero?
+  block_free(b);
+  return errors;
+}
+
+int block_compile(block b, struct locfile* locations, struct bytecode** out) {
+  struct bytecode* bc = jv_mem_alloc(sizeof(struct bytecode));
+  bc->parent = 0;
+  bc->nclosures = 0;
+  bc->globals = jv_mem_alloc(sizeof(struct symbol_table));
+  int ncfunc = count_cfunctions(b);
+  bc->globals->ncfunctions = 0;
+  bc->globals->cfunctions = jv_mem_alloc(sizeof(struct cfunction) * ncfunc);
+  int nerrors = compile(locations, bc, b);
+  assert(bc->globals->ncfunctions == ncfunc);
+  if (nerrors > 0) {
+    bytecode_free(bc);
+    *out = 0;
+  } else {
+    *out = bc;
+  }
+  return nerrors;
+}
+
+void block_free(block b) {
+  struct inst* next;
+  for (struct inst* curr = b.first; curr; curr = next) {
+    next = curr->next;
+    inst_free(curr);
+  }
+}

compile.h

@@ -0,0 +1,73 @@
+#ifndef COMPILE_H
+#define COMPILE_H
+#include <stdint.h>
+#include "jv.h"
+#include "opcode.h"
+#include "locfile.h"
+
+struct bytecode;
+struct symbol_table;
+
+struct inst;
+typedef struct inst inst;
+
+
+typedef struct block {
+  inst* first;
+  inst* last;
+} block;
+
+block gen_location(location, block);
+
+block gen_noop();
+block gen_op_simple(opcode op);
+block gen_const(jv constant);
+block gen_op_target(opcode op, block target);
+block gen_op_var_unbound(opcode op, const char* name);
+block gen_op_var_bound(opcode op, block binder);
+block gen_op_block_unbound(opcode op, const char* name);
+
+block gen_function(const char* name, block formals, block body);
+block gen_lambda(block body);
+block gen_call(const char* name, block body);
+block gen_subexp(block a);
+block gen_both(block a, block b);
+block gen_collect(block expr);
+block gen_assign(block expr);
+block gen_definedor(block a, block b);
+block gen_condbranch(block iftrue, block iffalse);
+block gen_and(block a, block b);
+block gen_or(block a, block b);
+
+block gen_cond(block cond, block iftrue, block iffalse);
+
+block gen_cbinding(struct symbol_table* functions, block b);
+
+void block_append(block* b, block b2);
+block block_join(block a, block b);
+int block_has_only_binders(block, int bindflags);
+block block_bind(block binder, block body, int bindflags);
+
+int block_compile(block, struct locfile*, struct bytecode**);
+
+void block_free(block);
+
+
+
+// Here's some horrible preprocessor gunk so that code
+// sequences can be contructed as BLOCK(block1, block2, block3)
+
+#define BLOCK_1(b1) (b1)
+#define BLOCK_2(b1,b2) (block_join((b1),(b2)))
+#define BLOCK_3(b1,b2,b3) (block_join(BLOCK_2(b1,b2),(b3)))
+#define BLOCK_4(b1,b2,b3,b4) (block_join(BLOCK_3(b1,b2,b3),(b4)))
+#define BLOCK_5(b1,b2,b3,b4,b5) (block_join(BLOCK_4(b1,b2,b3,b4),(b5)))
+#define BLOCK_6(b1,b2,b3,b4,b5,b6) (block_join(BLOCK_5(b1,b2,b3,b4,b5),(b6)))
+#define BLOCK_7(b1,b2,b3,b4,b5,b6,b7) (block_join(BLOCK_6(b1,b2,b3,b4,b5,b6),(b7)))
+
+#define BLOCK_IDX(_1,_2,_3,_4,_5,_6,_7,NAME,...) NAME
+#define BLOCK(...) \
+  BLOCK_IDX(__VA_ARGS__, BLOCK_7, BLOCK_6, BLOCK_5, BLOCK_4, BLOCK_3, BLOCK_2, BLOCK_1)(__VA_ARGS__)
+
+
+#endif

docs/.gitignore

@@ -0,0 +1,6 @@
+.DS_Store
+.sass-cache
+output/*
+
+# Autogenerated from public/css/base.scss
+public/css/base.css

docs/Gemfile

@@ -0,0 +1,5 @@
+source :rubygems
+
+gem "bonsai"
+gem "maruku"
+gem "ronn"

docs/Gemfile.lock

@@ -0,0 +1,53 @@
+GEM
+  remote: http://rubygems.org/
+  specs:
+    activesupport (3.2.9)
+      i18n (~> 0.6)
+      multi_json (~> 1.0)
+    addressable (2.3.2)
+    bonsai (1.4.7)
+      activesupport (>= 3.0.3)
+      builder (>= 3.0.0)
+      i18n (>= 0.5.0)
+      launchy (>= 0.3.7)
+      liquid (>= 2.2.2)
+      maruku (>= 0.6.0)
+      rack
+      sass
+      sinatra (>= 1.0)
+      tilt (>= 1.3)
+      watch (>= 0.1.0)
+    builder (3.1.4)
+    hpricot (0.8.6)
+    i18n (0.6.1)
+    launchy (2.1.2)
+      addressable (~> 2.3)
+    liquid (2.4.1)
+    maruku (0.6.1)
+      syntax (>= 1.0.0)
+    multi_json (1.5.0)
+    mustache (0.99.4)
+    rack (1.4.1)
+    rack-protection (1.3.2)
+      rack
+    rdiscount (1.6.8)
+    ronn (0.7.3)
+      hpricot (>= 0.8.2)
+      mustache (>= 0.7.0)
+      rdiscount (>= 1.5.8)
+    sass (3.2.3)
+    sinatra (1.3.3)
+      rack (~> 1.3, >= 1.3.6)
+      rack-protection (~> 1.2)
+      tilt (~> 1.3, >= 1.3.3)
+    syntax (1.0.0)
+    tilt (1.3.3)
+    watch (0.1.0)
+
+PLATFORMS
+  ruby
+
+DEPENDENCIES
+  bonsai
+  maruku
+  ronn

docs/Rakefile

@@ -0,0 +1,110 @@
+require 'bonsai'
+require 'liquid'
+require 'maruku'
+require 'json'
+require 'ronn'
+require 'tempfile'
+
+module ExtraFilters
+  def markdownify(input)
+    Maruku.new(input).to_html
+  end
+
+  def sanitize(input)
+    input.gsub(/[^a-zA-Z0-9_]/,"")
+  end
+
+  def json(input)
+    input.to_json
+  end
+  
+  def unique(input)
+    @n = (@n || 0) + 1
+    input + @n.to_s
+  end
+end
+
+Liquid::Template.register_filter(ExtraFilters)
+
+
+task :serve do
+  begin
+    Bonsai.log "Press Control+C to quit"
+    
+    require 'rack'
+    require 'sinatra'
+    require 'watch'
+    require 'launchy'
+    
+    Bonsai.root_dir = Dir.pwd
+    
+    server = fork {        
+      app = Rack::Builder.app {
+        map "/jq" do
+          use Bonsai::StaticPassThrough, :root => Bonsai.root_dir + "/output", :urls => ["/"]
+        end
+        run Bonsai::DevelopmentServer
+      }
+      Rack::Handler.default.run(app, :Port => 5000) do
+        Launchy.open("http://localhost:5000/jq")
+      end
+    }    
+    Watch.new("{content,templates,public}/**/*") { Bonsai::Exporter.process! }
+  rescue Interrupt
+    Process.kill("QUIT", server)
+    Process.wait(server)
+    exit
+  end
+end
+
+task :build do
+  Bonsai.root_dir = Dir.pwd
+  Bonsai::Exporter.publish!
+end
+
+def load_manual
+  YAML::ENGINE.yamler = 'syck'
+  YAML::load(File.open("content/3.manual/manual.yml"))
+end
+
+task :manpage do
+  Tempfile.open "manpage" do |f|
+    manual = load_manual
+    f.puts manual['manpage_intro']
+    f.puts manual['body']
+    manual['sections'].each do |section|
+      
+      f.puts "## #{section['title'].upcase}\n"
+      f.puts section['body']
+      f.puts ""
+      (section['entries'] || []).each do |entry|
+        f.puts "### #{entry['title']}\n"
+        f.puts entry['body']
+        f.puts ""
+        (entry['examples'] || []).each do |example|
+          f.puts "    jq '#{example['program']}'"
+          f.puts "       #{example['input']}"
+          f.puts "    => #{example['output'].join(", ")}"
+          f.puts
+        end
+      end
+      f.puts ""
+    end
+    f.puts manual['manpage_epilogue']
+    f.close
+    puts Ronn::Document.new(f.path).convert('roff').gsub(/<\/?code>/,"")
+  end  
+end
+
+task :mantests do
+  load_manual['sections'].each do |section|
+    (section['entries'] || []).each do |entry|
+      (entry['examples'] || []).each do |example|
+        puts example['program'].gsub("\n", " ")
+        puts example['input']
+        example['output'].each do |s| puts s end
+        puts
+      end
+    end
+  end
+end

docs/content/1.tutorial/default.yml

@@ -0,0 +1,241 @@
+headline: Tutorial
+body:
+  - text: |
+
+      Twitter has a JSON API, so let's play with that. This URL gets
+      us the last 5 tweets about JSON:
+      
+  - command: "curl 'http://search.twitter.com/search.json?q=json&rpp=5&include_entities=true'"
+    result: |
+      {"completed_in":0.108,"max_id":247677287108067328,"max_id_str":"247677287108067328","next_page":"?page=2&max_id=247677287108067328&q=json&rpp=5&include_entities=1","page":1,"query":"json","refresh_url":"?since_id=247677287108067328&q=json&include_entities=1","results":[{"created_at":"Mon, 17 Sep 2012 12:44:01 +0000","entities":{"hashtags":[],"urls":[{"url":"http:\/\/t.co\/XRvh1ZVw","expanded_url":"http:\/\/jase.im\/Ri7I0M","display_url":"jase.im\/Ri7I0M","indices":[112,132]}],"user_mentions":[{"screen_name":"imagemechanics","name":"Jason Cotterell","id":57271393,"id_str":"57271393","indices":[3,18]}]},"from_user":"_AaronNg","from_user_id":79771704,"from_user_id_str":"79771704","from_user_name":"NgChenChong","geo":null,"id":247677287108067328,"id_str":"247677287108067328","iso_language_code":"en","metadata":{"result_type":"recent"},"profile_image_url":"http:\/\/a0.twimg.com\/profile_images\/2523558403\/ek8mo4j4beq84iw28gjl_normal.jpeg","profile_image_url_https":"https:\/\/si0.twimg.com\/profile_images\/2523558403\/ek8mo4j4beq84iw28gjl_normal.jpeg","source":"<a href="http:\/\/twitter.com\/">web<\/a>","text":"RT @imagemechanics: iPhone 5 website teardown: How Apple compresses video using JPEG, JSON, and <canvas>.\nhttp:\/\/t.co\/XRvh1ZVw","to_user":null,"to_user_id":0,"to_user_id_str":"0","to_user_name":null}, ...
+
+  - text: |
+
+      There's lots of info and no whitespace, so to make it a bit more
+      legible we pipe it through jq, telling jq to just spit the input
+      back at us using the expression `.`:
+
+    
+  - command: "curl 'http://search.twitter.com/search.json?q=json&rpp=5&include_entities=true' | jq '.'"
+    result: |
+
+      {
+        "since_id_str": "0",
+        "since_id": 0,
+        "results_per_page": 5,
+        "completed_in": 0.108,
+        "max_id": 247677287108067330,
+        "max_id_str": "247677287108067328",
+        "next_page": "?page=2&max_id=247677287108067328&q=json&rpp=5&include_entities=1",
+        "page": 1,
+        "query": "json",
+        "refresh_url": "?since_id=247677287108067328&q=json&include_entities=1",
+        "results": [
+          {
+            "from_user_name": "NgChenChong",
+            "from_user_id_str": "79771704",
+            "from_user_id": 79771704,
+            "from_user": "_AaronNg",
+            "iso_language_code": "en",
+            "text": "RT @imagemechanics: iPhone 5 website teardown: How Apple compresses video using JPEG, JSON, and <canvas>.\nhttp://t.co/XRvh1ZVw",
+            "to_user": null
+            /* lots more fields... */
+          },
+          /* lots more results... */
+        ]
+      }
+
+    
+  - text: |
+          
+      Let's pull out the first tweet:
+          
+          
+  - command: "curl 'http://search.twitter.com/search.json?q=json&rpp=5&include_entities=true' | jq '.results[0]'"
+    result: |
+      {
+        "from_user_name": "NgChenChong",
+        "from_user_id_str": "79771704",
+        "from_user_id": 79771704,
+        "from_user": "_AaronNg",
+        "iso_language_code": "en",
+        "text": "RT @imagemechanics: iPhone 5 website teardown: How Apple compresses video using JPEG, JSON, and <canvas>.\nhttp://t.co/XRvh1ZVw",
+        "to_user": null
+        /* lots more fields... */
+      },
+          
+  - text: |
+
+      For the rest of the examples, I'll leave out the `curl` command - it's not going to change.
+
+    
+      There's still a lot of info we don't care about there, so we'll
+      restrict it down to the most interesting fields.
+          
+
+  - command: "jq '.results[0] | {from_user, text}'"
+    result: |
+      {
+        "text": "RT @imagemechanics: iPhone 5 website teardown: How Apple compresses video using JPEG, JSON, and <canvas>.\nhttp://t.co/XRvh1ZVw",
+        "from_user": "_AaronNg"
+      }
+      
+  - text: |
+    
+      The `|` operator in jq feeds the output of one filter
+      (`.results[0]` which gets the first element of the results
+      array) into the input of another (`{from_user, text}` which
+      builds an object of those two fields).
+    
+      Now let's get the rest of the tweets:
+    
+          
+
+  - command: "jq '.results[] | {from_user, text}'"
+    result: |
+      {
+        "text": "RT @imagemechanics: iPhone 5 website teardown: How Apple compresses video using JPEG, JSON, and <canvas>.\nhttp://t.co/XRvh1ZVw",
+        "from_user": "_AaronNg"
+      }
+      {
+        "text": "RT @_kud: iPhone 5 website teardown: How Apple compresses video using JPEG, JSON, and <canvas> -- http://t.co/Lhp92IqD",
+        "from_user": "blouerat"
+      }
+      {
+        "text": "Dynamic Forms Mobile App by pmadiset: Few Forms details are hosted on our server in the form of JSON. This JSON ... http://t.co/7KALdWaX",
+        "from_user": "cebu_iphone"
+      }
+      {
+        "text": "iPhone 5 website insanity - Video compressing using JPEG, JSON, and <canvas> http://t.co/28Jesbio (Oh #Apple, U So #Awesome...)",
+        "from_user": "dieswaytoofast"
+      }
+      {
+        "text": "RT @umutm: A very nice web-based JSON editor - http://t.co/M70snaIf",
+        "from_user": "Leolik"
+      }
+         
+          
+  - text: |
+    
+      `.results[]` returns each element of the results array, one at a
+      time, which are all fed into `{from_user, text}`.
+
+
+      Data in jq is represented as streams of JSON values - every jq
+      expression runs for each value in its input stream, and can
+      produce any number of values to its output stream.
+
+      Streams are serialised by just separating JSON values with
+      whitespace. This is a `cat`-friendly format - you can just join
+      two JSON streams together and get a valid JSON stream.
+    
+      If you want to get the output as a single array, you can tell jq to
+      "collect" all of the answers by wrapping the filter in square
+      brackets:
+
+  - command: "jq '[.results[] | {from_user, text}]'"
+    result: |
+      [
+        {
+          "text": "RT @imagemechanics: iPhone 5 website teardown: How Apple compresses video using JPEG, JSON, and <canvas>.\nhttp://t.co/XRvh1ZVw",
+          "from_user": "_AaronNg"
+        },
+        {
+          "text": "RT @_kud: iPhone 5 website teardown: How Apple compresses video using JPEG, JSON, and <canvas> -- http://t.co/Lhp92IqD",
+          "from_user": "blouerat"
+        },
+        {
+          "text": "Dynamic Forms Mobile App by pmadiset: Few Forms details are hosted on our server in the form of JSON. This JSON ... http://t.co/7KALdWaX",
+          "from_user": "cebu_iphone"
+        },
+        {
+          "text": "iPhone 5 website insanity - Video compressing using JPEG, JSON, and <canvas> http://t.co/28Jesbio (Oh #Apple, U So #Awesome...)",
+          "from_user": "dieswaytoofast"
+        },
+        {
+          "text": "RT @umutm: A very nice web-based JSON editor - http://t.co/M70snaIf",
+          "from_user": "Leolik"
+        }
+      ]
+          
+          
+  - text: |
+
+      - - -
+    
+      Next, let's try getting the URLs out of those API results as
+      well. In each tweet, the Twitter API includes a field called
+      "entities" which looks like this:
+
+          {
+            "user_mentions": [],
+            "urls": [{
+                "indices": [83, 103],
+                "display_url": "bit.ly/StniqT",
+                "expanded_url": "http://bit.ly/StniqT",
+                "url": "http://t.co/28Jesbio"
+            }],
+            "hashtags": [
+              {"indices": [108, 114], "text": "Apple" },
+              {"indices": [121, 129], "text": "Awesome"}
+            ]
+          }
+
+      We want to pull out all of the "url" fields inside that array of url
+      objects, and make a simple list of strings to go along with the
+      "from_user" and "text" fields we already have.
+    
+          
+
+  - command: "jq '.results[] | {from_user, text, urls: [.entities.urls[].url]}'"
+    result: |
+      {
+       "urls": [
+         "http://t.co/XRvh1ZVw"
+       ],
+       "text": "RT @imagemechanics: iPhone 5 website teardown: How Apple compresses video using JPEG, JSON, and <canvas>.\nhttp://t.co/XRvh1ZVw",
+        "from_user": "_AaronNg"
+      }
+      {
+        "urls": [
+          "http://t.co/Lhp92IqD"
+        ],
+        "text": "RT @_kud: iPhone 5 website teardown: How Apple compresses video using JPEG, JSON, and <canvas> -- http://t.co/Lhp92IqD",
+        "from_user": "blouerat"
+      }
+      {
+        "urls": [
+          "http://t.co/7KALdWaX"
+        ],
+        "text": "Dynamic Forms Mobile App by pmadiset: Few Forms details are hosted on our server in the form of JSON. This JSON ... http://t.co/7KALdWaX",
+        "from_user": "cebu_iphone"
+      }
+      {
+        "urls": [
+          "http://t.co/28Jesbio"
+        ],
+        "text": "iPhone 5 website insanity - Video compressing using JPEG, JSON, and <canvas> http://t.co/28Jesbio (Oh #Apple, U So #Awesome...)",
+        "from_user": "dieswaytoofast"
+      }
+      {
+        "urls": [
+          "http://t.co/M70snaIf"
+        ],
+        "text": "RT @umutm: A very nice web-based JSON editor - http://t.co/M70snaIf",
+        "from_user": "Leolik"
+      }
+    
+          
+  - text: |
+    
+      Here we're making an object as before, but this time the urls
+      field is being set to `[.entities.urls[].url]`, which collects
+      all of the URLs defined in the entities object.
+
+  - text: |
+
+      - - -
+
+      Here endeth the tutorial! There's lots more to play with, go
+      read [the manual](../manual) if you're interested, and [download
+      jq](../download) if you haven't already.

docs/content/2.download/default.yml

@@ -0,0 +1,78 @@
+headline: Download jq
+body:
+
+  - text: |
+
+      jq is written in C and has no runtime dependencies, so it should be
+      possible to build it for nearly any platform. Prebuilt binaries are
+      available for Linux, OS X and Windows.
+
+      The binaries should just run, but on OS X and Linux you may need
+      to make them executable first using `chmod +x jq`.
+
+      jq is licensed under the MIT license. For all of the gory
+      details, read the file `COPYING` in the source distribution.
+
+
+      ### Linux
+      
+       * Binaries for [64-bit](linux64/jq) or [32-bit](linux32/jq) systems.
+
+       * For Arch users, a PKGBUILD for jq-1.1 is in the
+         [AUR](https://aur.archlinux.org/packages.php?ID=63849), as
+         well as a PKGBUILD for the HEAD of master
+         ([jq-git](https://aur.archlinux.org/packages.php?ID=63850)).
+         You can install them using
+         [Yaourt](https://wiki.archlinux.org/index.php/Yaourt) or
+         manually by following instructions on [Arch Linux's
+         Wiki](https://wiki.archlinux.org/index.php/Arch_UseRepository).
+
+       * Debian/Ubuntu packages are coming Real Soon Now.
+
+      ### OS X
+
+       * `brew install jq` using [homebrew](http://mxcl.github.com/homebrew/)
+
+       * Or, grab prebuilt [64-bit binaries](osx64/jq) or [32-bit
+         binaries](osx32/jq)
+
+      ### Windows
+
+       * Executables for [64-bit](win64/jq) or [32-bit](win32/jq)
+
+      ### From source on Linux or OS X
+
+          git clone https://github.com/stedolan/jq.git
+          cd jq
+          make && sudo make install
+
+      To build it from a git clone, you'll need to install a few
+      packages first:
+
+       * [Flex](http://www.gnu.org/software/flex/)
+       * [Bison](http://www.gnu.org/software/bison/)
+       * [Python](http://www.python.org)
+       * [GCC](http://gcc.gnu.org)
+       * [Make](http://www.gnu.org/software/make)
+
+      For Linux systems, these will all be in your system's package
+      manager, and if you do development on the machine they're most
+      likely already installed.
+
+      On OS X, these are all included in Apple's command line tools, which
+      can be installed from [Xcode](http://developer.apple.com/technologies/tools/).
+      However, you may find that you need a newer version of Bison than the one provided
+      by Apple. This can be found in [Homebrew](http://mxcl.github.com/homebrew/)
+      or [MacPorts](http://macports.org/).
+
+      `flex` and `bison` are used to generate the lexer and parser for
+      jq, and some python scripts generate the UTF8 encoding tables
+      needed for JSON parsing.
+   
+      #### Building the documentation
+
+      jq's documentation is compiled into static HTML using
+      [Bonsai](http://www.tinytree.info). To view the documentation
+      locally, run `rake serve` from the docs/ subdirectory.
+
+

docs/content/index/index.yml

@@ -0,0 +1,27 @@
+headline: jq
+
+blurb: |
+  
+  jq is a lightweight and flexible command-line JSON processor.
+  
+body1: |
+
+  jq is like `sed` for JSON data - you can use it to slice and filter
+  and map and transform structured data with the same ease that `sed`,
+  `awk`, `grep` and friends let you play with text.
+
+body2: |
+
+  jq is written in portable C, and it has zero runtime
+  dependencies. You can download a single binary, `scp` it to a far away
+  machine, and expect it to work.
+
+body3: |
+
+  jq can mangle the data format that you have into the one that you
+  want with very little effort, and the program to do so is often
+  shorter and simpler than you'd expect.
+  
+tail: |
+  Go read the [tutorial](tutorial) for more, or the [manual](manual)
+  for *way* more.

docs/public/.htaccess

@@ -0,0 +1,28 @@
+DirectoryIndex index.html
+FileETag All
+
+# Compress all static assets
+<IfModule mod_deflate.c>
+	# compress content with type html, text, and css
+  AddOutputFilterByType DEFLATE text/css text/html text/javascript application/javascript application/x-javascript text/js text/plain text/xml
+
+	<IfModule mod_headers.c>
+		# properly handle requests coming from behind proxies
+		Header append Vary User-Agent
+	</IfModule>
+</IfModule>
+
+# Cache, aggressively 
+<IfModule mod_expires.c>
+  ExpiresActive On
+  ExpiresDefault "access plus 10 days"
+
+	<FilesMatch "\.(eot|ttf|otf)$">
+		ExpiresDefault "access plus 10 years"
+	</filesMatch>
+</IfModule>
+
+# Mime-types
+AddType application/vnd.ms-fontobject .eot
+AddType font/ttf .ttf
+AddType font/otf .otf

docs/public/bootstrap/css/bootstrap-responsive.min.css

@@ -0,0 +1,9 @@
+/*!
+ * Bootstrap Responsive v2.1.1
+ *
+ * Copyright 2012 Twitter, Inc
+ * Licensed under the Apache License v2.0
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Designed and built with all the love in the world @twitter by @mdo and @fat.
+ */.clearfix{*zoom:1}.clearfix:before,.clearfix:after{display:table;line-height:0;content:""}.clearfix:after{clear:both}.hide-text{font:0/0 a;color:transparent;text-shadow:none;background-color:transparent;border:0}.input-block-level{display:block;width:100%;min-height:30px;-webkit-box-sizing:border-box;-moz-box-sizing:border-box;box-sizing:border-box}.hidden{display:none;visibility:hidden}.visible-phone{display:none!important}.visible-tablet{display:none!important}.hidden-desktop{display:none!important}.visible-desktop{display:inherit!important}@media(min-width:768px) and (max-width:979px){.hidden-desktop{display:inherit!important}.visible-desktop{display:none!important}.visible-tablet{display:inherit!important}.hidden-tablet{display:none!important}}@media(max-width:767px){.hidden-desktop{display:inherit!important}.visible-desktop{display:none!important}.visible-phone{display:inherit!important}.hidden-phone{display:none!important}}@media(min-width:1200px){.row{margin-left:-30px;*zoom:1}.row:before,.row:after{display:table;line-height:0;content:""}.row:after{clear:both}[class*="span"]{float:left;min-height:1px;margin-left:30px}.container,.navbar-static-top .container,.navbar-fixed-top .container,.navbar-fixed-bottom .container{width:1170px}.span12{width:1170px}.span11{width:1070px}.span10{width:970px}.span9{width:870px}.span8{width:770px}.span7{width:670px}.span6{width:570px}.span5{width:470px}.span4{width:370px}.span3{width:270px}.span2{width:170px}.span1{width:70px}.offset12{margin-left:1230px}.offset11{margin-left:1130px}.offset10{margin-left:1030px}.offset9{margin-left:930px}.offset8{margin-left:830px}.offset7{margin-left:730px}.offset6{margin-left:630px}.offset5{margin-left:530px}.offset4{margin-left:430px}.offset3{margin-left:330px}.offset2{margin-left:230px}.offset1{margin-left:130px}.row-fluid{width:100%;*zoom:1}.row-fluid:before,.row-fluid:after{display:table;line-height:0;content:""}.row-fluid:after{clear:both}.row-fluid [class*="span"]{display:block;float:left;width:100%;min-height:30px;margin-left:2.564102564102564%;*margin-left:2.5109110747408616%;-webkit-box-sizing:border-box;-moz-box-sizing:border-box;box-sizing:border-box}.row-fluid [class*="span"]:first-child{margin-left:0}.row-fluid .span12{width:100%;*width:99.94680851063829%}.row-fluid .span11{width:91.45299145299145%;*width:91.39979996362975%}.row-fluid .span10{width:82.90598290598291%;*width:82.8527914166212%}.row-fluid .span9{width:74.35897435897436%;*width:74.30578286961266%}.row-fluid .span8{width:65.81196581196582%;*width:65.75877432260411%}.row-fluid .span7{width:57.26495726495726%;*width:57.21176577559556%}.row-fluid .span6{width:48.717948717948715%;*width:48.664757228587014%}.row-fluid .span5{width:40.17094017094017%;*width:40.11774868157847%}.row-fluid .span4{width:31.623931623931625%;*width:31.570740134569924%}.row-fluid .span3{width:23.076923076923077%;*width:23.023731587561375%}.row-fluid .span2{width:14.52991452991453%;*width:14.476723040552828%}.row-fluid .span1{width:5.982905982905983%;*width:5.929714493544281%}.row-fluid .offset12{margin-left:105.12820512820512%;*margin-left:105.02182214948171%}.row-fluid .offset12:first-child{margin-left:102.56410256410257%;*margin-left:102.45771958537915%}.row-fluid .offset11{margin-left:96.58119658119658%;*margin-left:96.47481360247316%}.row-fluid .offset11:first-child{margin-left:94.01709401709402%;*margin-left:93.91071103837061%}.row-fluid .offset10{margin-left:88.03418803418803%;*margin-left:87.92780505546462%}.row-fluid .offset10:first-child{margin-left:85.47008547008548%;*margin-left:85.36370249136206%}.row-fluid .offset9{margin-left:79.48717948717949%;*margin-left:79.38079650845607%}.row-fluid .offset9:first-child{margin-left:76.92307692307693%;*margin-left:76.81669394435352%}.row-fluid .offset8{margin-left:70.94017094017094%;*margin-left:70.83378796144753%}.row-fluid .offset8:first-child{margin-left:68.37606837606839%;*margin-left:68.26968539734497%}.row-fluid .offset7{margin-left:62.393162393162385%;*margin-left:62.28677941443899%}.row-fluid .offset7:first-child{margin-left:59.82905982905982%;*margin-left:59.72267685033642%}.row-fluid .offset6{margin-left:53.84615384615384%;*margin-left:53.739770867430444%}.row-fluid .offset6:first-child{margin-left:51.28205128205128%;*margin-left:51.175668303327875%}.row-fluid .offset5{margin-left:45.299145299145295%;*margin-left:45.1927623204219%}.row-fluid .offset5:first-child{margin-left:42.73504273504273%;*margin-left:42.62865975631933%}.row-fluid .offset4{margin-left:36.75213675213675%;*margin-left:36.645753773413354%}.row-fluid .offset4:first-child{margin-left:34.18803418803419%;*margin-left:34.081651209310785%}.row-fluid .offset3{margin-left:28.205128205128204%;*margin-left:28.0987452264048%}.row-fluid .offset3:first-child{margin-left:25.641025641025642%;*margin-left:25.53464266230224%}.row-fluid .offset2{margin-left:19.65811965811966%;*margin-left:19.551736679396257%}.row-fluid .offset2:first-child{margin-left:17.094017094017094%;*maLine truncated

docs/public/bootstrap/css/bootstrap.min.css

@@ -0,0 +1,9 @@
+/*!
+ * Bootstrap v2.1.1
+ *
+ * Copyright 2012 Twitter, Inc
+ * Licensed under the Apache License v2.0
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Designed and built with all the love in the world @twitter by @mdo and @fat.
+ */article,aside,details,figcaption,figure,footer,header,hgroup,nav,section{display:block}audio,canvas,video{display:inline-block;*display:inline;*zoom:1}audio:not([controls]){display:none}html{font-size:100%;-webkit-text-size-adjust:100%;-ms-text-size-adjust:100%}a:focus{outline:thin dotted #333;outline:5px auto -webkit-focus-ring-color;outline-offset:-2px}a:hover,a:active{outline:0}sub,sup{position:relative;font-size:75%;line-height:0;vertical-align:baseline}sup{top:-0.5em}sub{bottom:-0.25em}img{width:auto\9;height:auto;max-width:100%;vertical-align:middle;border:0;-ms-interpolation-mode:bicubic}#map_canvas img{max-width:none}button,input,select,textarea{margin:0;font-size:100%;vertical-align:middle}button,input{*overflow:visible;line-height:normal}button::-moz-focus-inner,input::-moz-focus-inner{padding:0;border:0}button,input[type="button"],input[type="reset"],input[type="submit"]{cursor:pointer;-webkit-appearance:button}input[type="search"]{-webkit-box-sizing:content-box;-moz-box-sizing:content-box;box-sizing:content-box;-webkit-appearance:textfield}input[type="search"]::-webkit-search-decoration,input[type="search"]::-webkit-search-cancel-button{-webkit-appearance:none}textarea{overflow:auto;vertical-align:top}.clearfix{*zoom:1}.clearfix:before,.clearfix:after{display:table;line-height:0;content:""}.clearfix:after{clear:both}.hide-text{font:0/0 a;color:transparent;text-shadow:none;background-color:transparent;border:0}.input-block-level{display:block;width:100%;min-height:30px;-webkit-box-sizing:border-box;-moz-box-sizing:border-box;box-sizing:border-box}body{margin:0;font-family:"Helvetica Neue",Helvetica,Arial,sans-serif;font-size:14px;line-height:20px;color:#333;background-color:#fff}a{color:#08c;text-decoration:none}a:hover{color:#005580;text-decoration:underline}.img-rounded{-webkit-border-radius:6px;-moz-border-radius:6px;border-radius:6px}.img-polaroid{padding:4px;background-color:#fff;border:1px solid #ccc;border:1px solid rgba(0,0,0,0.2);-webkit-box-shadow:0 1px 3px rgba(0,0,0,0.1);-moz-box-shadow:0 1px 3px rgba(0,0,0,0.1);box-shadow:0 1px 3px rgba(0,0,0,0.1)}.img-circle{-webkit-border-radius:500px;-moz-border-radius:500px;border-radius:500px}.row{margin-left:-20px;*zoom:1}.row:before,.row:after{display:table;line-height:0;content:""}.row:after{clear:both}[class*="span"]{float:left;min-height:1px;margin-left:20px}.container,.navbar-static-top .container,.navbar-fixed-top .container,.navbar-fixed-bottom .container{width:940px}.span12{width:940px}.span11{width:860px}.span10{width:780px}.span9{width:700px}.span8{width:620px}.span7{width:540px}.span6{width:460px}.span5{width:380px}.span4{width:300px}.span3{width:220px}.span2{width:140px}.span1{width:60px}.offset12{margin-left:980px}.offset11{margin-left:900px}.offset10{margin-left:820px}.offset9{margin-left:740px}.offset8{margin-left:660px}.offset7{margin-left:580px}.offset6{margin-left:500px}.offset5{margin-left:420px}.offset4{margin-left:340px}.offset3{margin-left:260px}.offset2{margin-left:180px}.offset1{margin-left:100px}.row-fluid{width:100%;*zoom:1}.row-fluid:before,.row-fluid:after{display:table;line-height:0;content:""}.row-fluid:after{clear:both}.row-fluid [class*="span"]{display:block;float:left;width:100%;min-height:30px;margin-left:2.127659574468085%;*margin-left:2.074468085106383%;-webkit-box-sizing:border-box;-moz-box-sizing:border-box;box-sizing:border-box}.row-fluid [class*="span"]:first-child{margin-left:0}.row-fluid .span12{width:100%;*width:99.94680851063829%}.row-fluid .span11{width:91.48936170212765%;*width:91.43617021276594%}.row-fluid .span10{width:82.97872340425532%;*width:82.92553191489361%}.row-fluid .span9{width:74.46808510638297%;*width:74.41489361702126%}.row-fluid .span8{width:65.95744680851064%;*width:65.90425531914893%}.row-fluid .span7{width:57.44680851063829%;*width:57.39361702127659%}.row-fluid .span6{width:48.93617021276595%;*width:48.88297872340425%}.row-fluid .span5{width:40.42553191489362%;*width:40.37234042553192%}.row-fluid .span4{width:31.914893617021278%;*width:31.861702127659576%}.row-fluid .span3{width:23.404255319148934%;*width:23.351063829787233%}.row-fluid .span2{width:14.893617021276595%;*width:14.840425531914894%}.row-fluid .span1{width:6.382978723404255%;*width:6.329787234042553%}.row-fluid .offset12{margin-left:104.25531914893617%;*margin-left:104.14893617021275%}.row-fluid .offset12:first-child{margin-left:102.12765957446808%;*margin-left:102.02127659574467%}.row-fluid .offset11{margin-left:95.74468085106382%;*margin-left:95.6382978723404%}.row-fluid .offset11:first-child{margin-left:93.61702127659574%;*margin-left:93.51063829787232%}.row-fluid .offset10{margin-left:87.23404255319149%;*margin-left:87.12765957446807%}.row-fluid .offset10:first-child{margin-left:85.1063829787234%;*margin-left:84.99999999999999%}.row-fluid .offset9{margin-left:78.72340425531914%;*margin-left:78.61702127659572%}.row-fluid .offset9:first-child{margin-left:76.59574468085106%;*margin-left:76.48936170212764%}.row-fluiLine truncated

docs/public/bootstrap/js/bootstrap.min.js

@@ -0,0 +1,6 @@
+/*!
+* Bootstrap.js by @fat & @mdo
+* Copyright 2012 Twitter, Inc.
+* http://www.apache.org/licenses/LICENSE-2.0.txt
+*/
+!function(e){e(function(){"use strict";e.support.transition=function(){var e=function(){var e=document.createElement("bootstrap"),t={WebkitTransition:"webkitTransitionEnd",MozTransition:"transitionend",OTransition:"oTransitionEnd otransitionend",transition:"transitionend"},n;for(n in t)if(e.style[n]!==undefined)return t[n]}();return e&&{end:e}}()})}(window.jQuery),!function(e){"use strict";var t='[data-dismiss="alert"]',n=function(n){e(n).on("click",t,this.close)};n.prototype.close=function(t){function s(){i.trigger("closed").remove()}var n=e(this),r=n.attr("data-target"),i;r||(r=n.attr("href"),r=r&&r.replace(/.*(?=#[^\s]*$)/,"")),i=e(r),t&&t.preventDefault(),i.length||(i=n.hasClass("alert")?n:n.parent()),i.trigger(t=e.Event("close"));if(t.isDefaultPrevented())return;i.removeClass("in"),e.support.transition&&i.hasClass("fade")?i.on(e.support.transition.end,s):s()},e.fn.alert=function(t){return this.each(function(){var r=e(this),i=r.data("alert");i||r.data("alert",i=new n(this)),typeof t=="string"&&i[t].call(r)})},e.fn.alert.Constructor=n,e(function(){e("body").on("click.alert.data-api",t,n.prototype.close)})}(window.jQuery),!function(e){"use strict";var t=function(t,n){this.$element=e(t),this.options=e.extend({},e.fn.button.defaults,n)};t.prototype.setState=function(e){var t="disabled",n=this.$element,r=n.data(),i=n.is("input")?"val":"html";e+="Text",r.resetText||n.data("resetText",n[i]()),n[i](r[e]||this.options[e]),setTimeout(function(){e=="loadingText"?n.addClass(t).attr(t,t):n.removeClass(t).removeAttr(t)},0)},t.prototype.toggle=function(){var e=this.$element.closest('[data-toggle="buttons-radio"]');e&&e.find(".active").removeClass("active"),this.$element.toggleClass("active")},e.fn.button=function(n){return this.each(function(){var r=e(this),i=r.data("button"),s=typeof n=="object"&&n;i||r.data("button",i=new t(this,s)),n=="toggle"?i.toggle():n&&i.setState(n)})},e.fn.button.defaults={loadingText:"loading..."},e.fn.button.Constructor=t,e(function(){e("body").on("click.button.data-api","[data-toggle^=button]",function(t){var n=e(t.target);n.hasClass("btn")||(n=n.closest(".btn")),n.button("toggle")})})}(window.jQuery),!function(e){"use strict";var t=function(t,n){this.$element=e(t),this.options=n,this.options.slide&&this.slide(this.options.slide),this.options.pause=="hover"&&this.$element.on("mouseenter",e.proxy(this.pause,this)).on("mouseleave",e.proxy(this.cycle,this))};t.prototype={cycle:function(t){return t||(this.paused=!1),this.options.interval&&!this.paused&&(this.interval=setInterval(e.proxy(this.next,this),this.options.interval)),this},to:function(t){var n=this.$element.find(".item.active"),r=n.parent().children(),i=r.index(n),s=this;if(t>r.length-1||t<0)return;return this.sliding?this.$element.one("slid",function(){s.to(t)}):i==t?this.pause().cycle():this.slide(t>i?"next":"prev",e(r[t]))},pause:function(t){return t||(this.paused=!0),this.$element.find(".next, .prev").length&&e.support.transition.end&&(this.$element.trigger(e.support.transition.end),this.cycle()),clearInterval(this.interval),this.interval=null,this},next:function(){if(this.sliding)return;return this.slide("next")},prev:function(){if(this.sliding)return;return this.slide("prev")},slide:function(t,n){var r=this.$element.find(".item.active"),i=n||r[t](),s=this.interval,o=t=="next"?"left":"right",u=t=="next"?"first":"last",a=this,f=e.Event("slide",{relatedTarget:i[0]});this.sliding=!0,s&&this.pause(),i=i.length?i:this.$element.find(".item")[u]();if(i.hasClass("active"))return;if(e.support.transition&&this.$element.hasClass("slide")){this.$element.trigger(f);if(f.isDefaultPrevented())return;i.addClass(t),i[0].offsetWidth,r.addClass(o),i.addClass(o),this.$element.one(e.support.transition.end,function(){i.removeClass([t,o].join(" ")).addClass("active"),r.removeClass(["active",o].join(" ")),a.sliding=!1,setTimeout(function(){a.$element.trigger("slid")},0)})}else{this.$element.trigger(f);if(f.isDefaultPrevented())return;r.removeClass("active"),i.addClass("active"),this.sliding=!1,this.$element.trigger("slid")}return s&&this.cycle(),this}},e.fn.carousel=function(n){return this.each(function(){var r=e(this),i=r.data("carousel"),s=e.extend({},e.fn.carousel.defaults,typeof n=="object"&&n),o=typeof n=="string"?n:s.slide;i||r.data("carousel",i=new t(this,s)),typeof n=="number"?i.to(n):o?i[o]():s.interval&&i.cycle()})},e.fn.carousel.defaults={interval:5e3,pause:"hover"},e.fn.carousel.Constructor=t,e(function(){e("body").on("click.carousel.data-api","[data-slide]",function(t){var n=e(this),r,i=e(n.attr("data-target")||(r=n.attr("href"))&&r.replace(/.*(?=#[^\s]+$)/,"")),s=!i.data("modal")&&e.extend({},i.data(),n.data());i.carousel(s),t.preventDefault()})})}(window.jQuery),!function(e){"use strict";var t=function(t,n){this.$element=e(t),this.options=e.extend({},e.fn.collapse.defaults,n),this.options.parent&&(this.$parent=e(this.options.parent)),this.options.toggle&&this.toggle()};t.prototype={constructor:t,dimension:function(){var e=this.$element.hLine truncated

docs/public/css/base.scss

@@ -0,0 +1,92 @@
+@charset "utf-8";
+body {
+    padding-top: 60px;
+}
+
+footer {
+    background-color: #F5F5F5;
+    padding: 40px 0;
+    height: 100px;
+    margin-top: 40px;
+    color: #999999;
+}
+
+section {
+    padding-top: 30px;
+}
+
+.manual-example {
+    border-top: 1px solid #E5E5E5;
+}
+
+.manual-example td {
+    white-space: pre-wrap;
+    font-family: Monaco, Menlo, Consolas, "Courier New", monospace;
+}
+
+.manual-example td.jqprogram {
+    font-weight: bold;
+}
+
+.manual-example th {
+    text-align: right;
+    padding-right: 10px;
+}
+
+.tutorial-example {
+    position: relative;
+    margin-bottom: 10px;
+}
+
+.tutorial-example pre {
+    margin-bottom: 0px;
+}
+
+.tutorial-example a {
+    position: absolute;
+    top: 0px;
+    right: 0px;
+    padding: 15px 8px;
+    color: #777777;
+    font-weight: bold;
+    line-height: 10px;
+    font-size: 12px;
+    border-left: 1px solid #DDDDDD;
+    bottom: 0;
+    display: block;
+}
+
+
+
+#blurb {
+    padding-top: 40px;
+    padding-right: 50px;
+}
+
+#blurb p {
+    font-size: 20pt;
+}
+
+#blurb .btn-group {
+    margin-top: 20px;
+}
+
+#multiblurb {
+    line-height: 1.7;
+    text-align: center;
+    font-size: 12pt;
+}
+
+#multiblurb code {
+    border: 0;
+    font-size: 12pt;
+}
+
+h3 code {
+    border: 0;
+    font-size: 20px;
+}
+
+footer {
+    text-align: center;
+}

docs/public/robots.txt

@@ -0,0 +1,2 @@
+User-agent: *
+Allow: /

docs/site.yml

@@ -0,0 +1,14 @@
+# The key value pairs found below are available within the templates.
+
+
+jq_version: 1.2
+root: '/jq'
+footer: |
+  
+  This website is made with [Bonsai](http://www.tinytree.info) and
+  [Twitter Bootstrap](http://twitter.github.com/bootstrap/), themed
+  with [Bootswatch](http://bootswatch.com).
+
+  jq is licensed under the MIT license (code) and the
+  [CC-BY-3.0](http://creativecommons.org/licenses/by/3.0/) license
+  (docs).

docs/templates/default.liquid

@@ -0,0 +1,34 @@
+<!DOCTYPE html>
+<html lang="en">
+  {% include "shared/head" %}
+  <body id="{{slug}}">
+    {% include "shared/header" %}	
+    
+    <div class="container">
+        
+      <div class="row">
+        <h1>{{headline}}</h1>
+        {% for item in body %}
+          {% if item.text %}
+            {{ item.text | markdownify }}
+          {% endif %}
+        
+          {% if item.command %}
+
+        
+            {% capture resultID %}{{ "result" | unique}}{% endcapture %}
+            <div class="tutorial-example">
+              <div class="accordion-heading">
+                <pre>{{item.command}}</pre>
+                <a data-toggle="collapse" href="#{{resultID}}">Show result</a>
+              </div>
+              <div id="{{resultID}}" class="accordion-body collapse"><pre>{{item.result}}</pre></div>
+            </div>
+          {% endif %}
+        {% endfor %}
+      </div>
+    </div>
+    {% include "shared/footer" %}
+    
+  </body>
+</html>

docs/templates/index.liquid

@@ -0,0 +1,41 @@
+<!DOCTYPE html>
+<html lang="en">
+  {% include "shared/head" %}
+  <body id="{{slug}}">
+    {% include "shared/header" %}	
+    
+    <div class="container">
+      <div class="row">
+        <div class="hero-unit">
+          <img src="{{root}}/jq.png" style="width: 400px; height: 220px; float: left; margin-right: 60px">
+          <div id="blurb">
+            {{blurb | markdownify}}
+            <div class="btn-group" style="float:right">
+              <a class="btn btn-primary dropdown-toggle" data-toggle="dropdown" href="#">
+                Download jq-{{jq_version}}
+                <span class="caret"></span>
+              </a>
+              <ul class="dropdown-menu">
+                <li><a href="{{root}}/download/linux64/jq">Linux (64-bit)</a></li>
+                <li><a href="{{root}}/download/osx64/jq">OS X (64-bit)</a></li>
+                <li><a href="{{root}}/download">Other platforms and source</a></li>
+              </ul>
+            </div>
+
+          </div>
+          <div style="clear:both;"></div>
+        </div>
+      </div>
+        
+      <div class="row" id="multiblurb">
+	<div class="span4">{{body1 | markdownify}}</div>
+        <div class="span4">{{body2 | markdownify}}</div>
+	<div class="span4">{{body3 | markdownify}}</div>        
+      </div>
+      <div class="row" style="text-align:center; margin-top: 30px">
+        {{tail | markdownify}}
+      </div>
+    </div>
+    {% include "shared/footer" %}    
+  </body>
+</html>

docs/templates/manual.liquid

@@ -0,0 +1,110 @@
+<!DOCTYPE html>
+<html lang="en">
+  {% include "shared/head" %}
+  <body id="{{slug}}" data-spy="scroll" data-target="#navcolumn" data-offset="100">
+    {% include "shared/header" %}
+    <div class="container">
+      <div class="row">
+        <div class="span3" id="navcolumn">
+          <div class="affix">
+          <h4>Contents</h4>
+          <ul class="nav nav-pills nav-stacked">
+            {% for section in sections %}
+              <li {% if forloop.first %} class="active" {% endif %}>
+                <a href="#{{section.title | sanitize}}">{{section.title}}</a>
+              </li>
+            {% endfor %}
+          </ul>
+          <form class="form-search">
+            <script type="text/javascript">
+              var section_map = {
+                {% for section in sections %}
+                  {% for entry in section.entries %}
+                    {{entry.title | json}} : {{entry.title | sanitize | json}},
+                  {% endfor %}
+                  {{section.title | json}} : {{section.title | sanitize | json}}
+                  {% unless forloop.last %},{% endunless %}
+                {% endfor %}
+              };
+              var section_names = $.map(section_map, function(v,k){return k});
+              $(function(){
+                $('#searchbox').typeahead({source: section_names})
+                $('#searchbox').change(function() {
+                  if ($(this).val() in section_map) {
+                    location.hash = '#' + section_map[$(this).val()];
+                  }
+                });
+              });
+            </script>
+            <input type="text"
+                   class="input-medium search-query"
+                   placeholder="Search"
+                   autocomplete="off"
+                   id="searchbox">
+          </form>
+          </div>
+        </div>
+        <div class="span9">
+          <h1>{{headline}}</h1>
+          {{ body | markdownify }}
+          {% for section in sections %}
+            <section id="{{section.title | sanitize}}">
+              <h2>{{section.title}}</h2>
+              {{section.body | markdownify}}
+              {% for entry in section.entries %}
+                <section id="{{entry.title | sanitize}}">
+                <h3>
+                  {{entry.title | markdownify}}
+                  {% if entry.subtitle %}<small>{{entry.subtitle}}</small>{% endif %}
+                </h3>
+                {{entry.body | markdownify}}
+
+                {% if entry.examples %}
+                  <div>
+                    {% capture exampleID %}{{ "" | unique }}{% endcapture %}
+                    <a data-toggle="collapse" href="#example{{exampleID}}">
+                      <i class="icon-chevron-right"></i>
+                      {% if entry.examples[1] %}Examples{%else%}Example{%endif%}
+                    </a>
+                    <div id="example{{exampleID}}" class="collapse">
+                      
+                    {% for example in entry.examples %}
+                      <table class="manual-example">
+                        <tr><th></th><td class="jqprogram">jq '{{example.program}}'</td></tr>
+                        <tr><th>Input</th><td>{{example.input}}</td></tr>
+                        {% unless example.output[0] %}
+                          <tr>
+                            <th>Output</th>
+                            <td><i>none</i></td>
+                          </tr>
+                        {% endunless %}
+                        {% for output in example.output %}
+                          <tr>
+                            {% if forloop.index == 1 %}
+                              <th>Output</th>
+                            {% else %}
+                              <th></th>
+                            {% endif %}
+                            <td>{{output}}</td>
+                          </tr>
+                        {% endfor %}
+                        
+                      </table>
+                    {% endfor %}
+                    </div>
+                  </div>
+                 {% endif %}
+                 </section>
+              {% endfor %}
+            </section>
+          {% endfor %}
+        </div>
+      </div>
+      
+
+    </div>
+      {% include "shared/footer" %}    
+  </body>
+
+
+</html>

docs/templates/shared/_footer.liquid

@@ -0,0 +1,5 @@
+<footer>
+  <div class="container">
+    {{ footer | markdownify }}
+  </div>
+</footer>

docs/templates/shared/_head.liquid

@@ -0,0 +1,12 @@
+<head>
+  <meta charset=UTF-8>
+  <title>{{headline}}</title>
+  <link rel="stylesheet" href="{{root}}/bootstrap/css/bootstrap.css" type="text/css">
+  <link rel="stylesheet" href="{{root}}/css/base.css" type="text/css">
+  <!--[if IE]>
+    <script src="http://html5shiv.googlecode.com/svn/trunk/html5.js"></script>
+  <![endif]-->
+  <script src="//ajax.googleapis.com/ajax/libs/jquery/1.8.1/jquery.min.js"></script>
+  <script src="{{root}}/bootstrap/js/bootstrap.js"></script>
+
+</head>

docs/templates/shared/_header.liquid

@@ -0,0 +1,24 @@
+
+<div class="navbar navbar-inverse navbar-fixed-top">
+  <div class="navbar-inner">
+    <div class="container">
+      <button type="button" class="btn btn-navbar" data-toggle="collapse" data-target=".nav-collapse">
+        <span class="icon-bar"></span>
+        <span class="icon-bar"></span>
+        <span class="icon-bar"></span>
+      </button>
+      <a class="brand" href="{{root}}/">jq</a>
+      <div class="nav-collapse collapse">
+        <ul class="nav">
+          {% for item in navigation %}
+            <li {% if item.permalink == permalink %} class="active" {% endif %}>
+              <a href="{{root}}{{item.permalink}}">{{item.name}}</a>
+            </li>
+          {% endfor %}
+          <li><a href="https://github.com/stedolan/jq/issues">Issues</a></li>
+          <li><a href="https://github.com/stedolan/jq">Source</a></li>
+        </ul>
+      </div>
+    </div>
+  </div>
+</div>

execute.c

@@ -0,0 +1,515 @@
+#include <assert.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdint.h>
+
+#include "execute.h"
+
+#include "opcode.h"
+#include "bytecode.h"
+
+#include "forkable_stack.h"
+#include "frame_layout.h"
+
+#include "jv_alloc.h"
+#include "locfile.h"
+#include "jv.h"
+#include "jv_aux.h"
+#include "parser.h"
+#include "builtin.h"
+
+typedef struct {
+  jv value;
+  int pathidx;
+} stackval;
+
+
+jv* pathbuf;
+int pathsize; // number of allocated elements
+
+int path_push(stackval sv, jv val) {
+  int pos = sv.pathidx;
+  assert(pos <= pathsize);
+  assert(pos >= 0);
+  if (pos == pathsize) {
+    int oldpathsize = pathsize;
+    pathsize = oldpathsize ? oldpathsize * 2 : 100;
+    pathbuf = jv_mem_realloc(pathbuf, sizeof(pathbuf[0]) * pathsize);
+    for (int i=oldpathsize; i<pathsize; i++) {
+      pathbuf[i] = jv_invalid();
+    }
+  }
+  jv_free(pathbuf[pos]);
+  pathbuf[pos] = val;
+  return pos + 1;
+}
+
+stackval stackval_replace(stackval value, jv newjs) {
+  jv_free(value.value);
+  stackval s = {newjs, value.pathidx};
+  return s;
+}
+
+
+// Probably all uses of this function are bugs
+stackval stackval_root(jv v) {
+  stackval s = {v, 0};
+  return s;
+}
+
+struct forkable_stack data_stk;
+typedef struct {
+  FORKABLE_STACK_HEADER;
+  stackval sv;
+} data_stk_elem;
+
+void stack_push(stackval val) {
+  assert(jv_is_valid(val.value));
+  data_stk_elem* s = forkable_stack_push(&data_stk, sizeof(data_stk_elem));
+  s->sv = val;
+}
+
+stackval stack_pop() {
+  data_stk_elem* s = forkable_stack_peek(&data_stk);
+  stackval sv = s->sv;
+  if (!forkable_stack_pop_will_free(&data_stk)) {
+    sv.value = jv_copy(sv.value);
+  }
+  forkable_stack_pop(&data_stk);
+  assert(jv_is_valid(sv.value));
+  return sv;
+}
+
+struct forkable_stack frame_stk;
+
+
+struct forkpoint {
+  FORKABLE_STACK_HEADER;
+  struct forkable_stack_state saved_data_stack;
+  struct forkable_stack_state saved_call_stack;
+};
+
+struct forkable_stack fork_stk;
+
+void stack_save(){
+  struct forkpoint* fork = forkable_stack_push(&fork_stk, sizeof(struct forkpoint));
+  forkable_stack_save(&data_stk, &fork->saved_data_stack);
+  forkable_stack_save(&frame_stk, &fork->saved_call_stack);
+}
+
+void stack_switch() {
+  struct forkpoint* fork = forkable_stack_peek(&fork_stk);
+  forkable_stack_switch(&data_stk, &fork->saved_data_stack);
+  forkable_stack_switch(&frame_stk, &fork->saved_call_stack);
+}
+
+int stack_restore(){
+  while (!forkable_stack_empty(&data_stk) && 
+         forkable_stack_pop_will_free(&data_stk)) {
+    jv_free(stack_pop().value);
+  }
+  while (!forkable_stack_empty(&frame_stk) && 
+         forkable_stack_pop_will_free(&frame_stk)) {
+    frame_pop(&frame_stk);
+  }
+  if (forkable_stack_empty(&fork_stk)) {
+    return 0;
+  } else {
+    struct forkpoint* fork = forkable_stack_peek(&fork_stk);
+    forkable_stack_restore(&data_stk, &fork->saved_data_stack);
+    forkable_stack_restore(&frame_stk, &fork->saved_call_stack);
+    forkable_stack_pop(&fork_stk);
+    return 1;
+  }
+}
+
+static struct closure make_closure(struct forkable_stack* stk, frame_ptr fr, uint16_t* pc) {
+  uint16_t level = *pc++;
+  uint16_t idx = *pc++;
+  fr = frame_get_level(stk, fr, level);
+  if (idx & ARG_NEWCLOSURE) {
+    int subfn_idx = idx & ~ARG_NEWCLOSURE;
+    assert(subfn_idx < frame_self(fr)->bc->nsubfunctions);
+    struct closure cl = {frame_self(fr)->bc->subfunctions[subfn_idx],
+                         forkable_stack_to_idx(stk, fr)};
+    return cl;
+  } else {
+    return *frame_closure_arg(fr, idx);
+  }
+}
+
+void print_error(jv value) {
+  assert(!jv_is_valid(value));
+  jv msg = jv_invalid_get_msg(value);
+  if (jv_get_kind(msg) == JV_KIND_STRING) {
+    fprintf(stderr, "jq: error: %s\n", jv_string_value(msg));
+  }
+  jv_free(msg);
+}
+#define ON_BACKTRACK(op) ((op)+NUM_OPCODES)
+
+jv jq_next() {
+  jv cfunc_input[MAX_CFUNCTION_ARGS];
+
+  assert(!forkable_stack_empty(&frame_stk));
+  uint16_t* pc = *frame_current_retaddr(&frame_stk);
+  frame_pop(&frame_stk);
+  
+  assert(!forkable_stack_empty(&frame_stk));
+
+  int backtracking = 0;
+  while (1) {
+    uint16_t opcode = *pc;
+
+#if JQ_DEBUG
+    dump_operation(frame_current_bytecode(&frame_stk), pc);
+    printf("\t");
+    const struct opcode_description* opdesc = opcode_describe(opcode);
+    data_stk_elem* param;
+    int stack_in = opdesc->stack_in;
+    if (stack_in == -1) stack_in = pc[1];
+    for (int i=0; i<stack_in; i++) {
+      if (i == 0) {
+        param = forkable_stack_peek(&data_stk);
+      } else {
+        printf(" | ");
+        param = forkable_stack_peek_next(&data_stk, param);
+      }
+      if (!param) break;
+      jv_dump(jv_copy(param->sv.value), 0);
+      printf("<%d>", jv_get_refcnt(param->sv.value));
+    }
+
+    if (backtracking) printf("\t<backtracking>");
+
+    printf("\n");
+#endif
+    if (backtracking) {
+      opcode = ON_BACKTRACK(opcode);
+      backtracking = 0;
+    }
+    pc++;
+
+    switch (opcode) {
+    default: assert(0 && "invalid instruction");
+
+    case LOADK: {
+      jv v = jv_array_get(jv_copy(frame_current_bytecode(&frame_stk)->constants), *pc++);
+      assert(jv_is_valid(v));
+      stack_push(stackval_replace(stack_pop(), v));
+      break;
+    }
+
+    case DUP: {
+      stackval v = stack_pop();
+      stackval v2 = v;
+      v2.value = jv_copy(v.value);
+      stack_push(v);
+      stack_push(v2);
+      break;
+    }
+
+    case DUP2: {
+      stackval keep = stack_pop();
+      stackval v = stack_pop();
+      stackval v2 = v;
+      v2.value = jv_copy(v.value);
+      stack_push(v);
+      stack_push(keep);
+      stack_push(v2);
+      break;
+    }
+
+    case SWAP: {
+      stackval a = stack_pop();
+      stackval b = stack_pop();
+      stack_push(a);
+      stack_push(b);
+      break;
+    }
+      
+    case POP: {
+      jv_free(stack_pop().value);
+      break;
+    }
+
+    case APPEND: {
+      // FIXME paths
+      jv v = stack_pop().value;
+      jv array = stack_pop().value;
+      array = jv_array_append(array, v);
+      stack_push(stackval_root(array));
+      break;
+    }
+
+    case INSERT: {
+      stackval stktop = stack_pop();
+      jv v = stack_pop().value;
+      jv k = stack_pop().value;
+      stackval objv = stack_pop();
+      assert(jv_get_kind(k) == JV_KIND_STRING);
+      assert(jv_get_kind(objv.value) == JV_KIND_OBJECT);
+      objv.value = jv_object_set(objv.value, k, v);
+      stack_push(objv);
+      stack_push(stktop);
+      break;
+    }
+
+      // FIXME: loadv/storev may do too much copying/freeing
+    case LOADV: {
+      uint16_t level = *pc++;
+      uint16_t v = *pc++;
+      frame_ptr fp = frame_get_level(&frame_stk, frame_current(&frame_stk), level);
+      jv* var = frame_local_var(fp, v);
+      #if JQ_DEBUG
+      printf("V%d = ", v);
+      jv_dump(jv_copy(*var), 0);
+      printf("\n");
+      #endif
+      stack_push(stackval_replace(stack_pop(), jv_copy(*var)));
+      break;
+    }
+
+    case STOREV: {
+      uint16_t level = *pc++;
+      uint16_t v = *pc++;
+      frame_ptr fp = frame_get_level(&frame_stk, frame_current(&frame_stk), level);
+      jv* var = frame_local_var(fp, v);
+      stackval val = stack_pop();
+      #if JQ_DEBUG
+      printf("V%d = ", v);
+      jv_dump(jv_copy(val.value), 0);
+      printf("\n");
+      #endif
+      jv_free(*var);
+      *var = val.value;
+      break;
+    }
+
+    case ASSIGN: {
+      stackval replacement = stack_pop();
+      stackval path_end = stack_pop();
+      stackval path_start = stack_pop();
+      jv_free(path_end.value);
+      jv_free(path_start.value);
+
+      uint16_t level = *pc++;
+      uint16_t v = *pc++;
+      frame_ptr fp = frame_get_level(&frame_stk, frame_current(&frame_stk), level);
+      jv* var = frame_local_var(fp, v);
+      jv result = jv_insert(*var, replacement.value, pathbuf + path_start.pathidx, path_end.pathidx - path_start.pathidx);
+      if (jv_is_valid(result)) {
+        *var = result;
+      } else {
+        print_error(result);
+        *var = jv_null();
+      }
+      break;
+    }
+
+    case INDEX: {
+      stackval t = stack_pop();
+      jv k = stack_pop().value;
+      int pathidx = path_push(t, jv_copy(k));
+      jv v = jv_lookup(t.value, k);
+      if (jv_is_valid(v)) {
+        stackval sv;
+        sv.value = v;
+        sv.pathidx = pathidx;
+        stack_push(sv);
+      } else {
+        print_error(v);
+        goto do_backtrack;
+      }
+      break;
+    }
+
+
+    case JUMP: {
+      uint16_t offset = *pc++;
+      pc += offset;
+      break;
+    }
+
+    case JUMP_F: {
+      uint16_t offset = *pc++;
+      stackval t = stack_pop();
+      jv_kind kind = jv_get_kind(t.value);
+      if (kind == JV_KIND_FALSE || kind == JV_KIND_NULL) {
+        pc += offset;
+      }
+      stack_push(t); // FIXME do this better
+      break;
+    }
+
+    case EACH: 
+      stack_push(stackval_root(jv_number(-1)));
+      // fallthrough
+    case ON_BACKTRACK(EACH): {
+      int idx = jv_number_value(stack_pop().value);
+      stackval container = stack_pop();
+
+      int keep_going;
+      jv key, value;
+      if (jv_get_kind(container.value) == JV_KIND_ARRAY) {
+        if (opcode == EACH) idx = 0;
+        else idx = idx + 1;
+        keep_going = idx < jv_array_length(jv_copy(container.value));
+        if (keep_going) {
+          key = jv_number(idx);
+          value = jv_array_get(jv_copy(container.value), idx);
+        }
+      } else if (jv_get_kind(container.value) == JV_KIND_OBJECT) {
+        if (opcode == EACH) idx = jv_object_iter(container.value);
+        else idx = jv_object_iter_next(container.value, idx);
+        keep_going = jv_object_iter_valid(container.value, idx);
+        if (keep_going) {
+          key = jv_object_iter_key(container.value, idx);
+          value = jv_object_iter_value(container.value, idx);
+        }
+      } else {
+        assert(opcode == EACH);
+        print_error(jv_invalid_with_msg(jv_string_fmt("Cannot iterate over %s",
+                                                      jv_kind_name(jv_get_kind(container.value)))));
+        keep_going = 0;
+      }
+
+      if (!keep_going) {
+        jv_free(container.value);
+        goto do_backtrack;
+      } else {
+        stack_save();
+        stack_push(container);
+        stack_push(stackval_root(jv_number(idx)));
+        frame_push_backtrack(&frame_stk, pc - 1);
+        stack_switch();
+        
+        stackval sv = {value,
+                       path_push(container, key)};
+        stack_push(sv);
+      }
+      break;
+    }
+
+    do_backtrack:
+    case BACKTRACK: {
+      if (!stack_restore()) {
+        return jv_invalid();
+      }
+      pc = *frame_current_retaddr(&frame_stk);
+      frame_pop(&frame_stk);
+      backtracking = 1;
+      break;
+    }
+
+    case FORK: {
+      stack_save();
+      frame_push_backtrack(&frame_stk, pc - 1);
+      stack_switch();
+      pc++; // skip offset this time
+      break;
+    }
+
+    case ON_BACKTRACK(FORK): {
+      uint16_t offset = *pc++;
+      pc += offset;
+      break;
+    }
+
+    case YIELD: {
+      jv value = stack_pop().value;
+      frame_push_backtrack(&frame_stk, pc);
+      return value;
+    }
+      
+    case CALL_BUILTIN: {
+      int nargs = *pc++;
+      stackval top = stack_pop();
+      cfunc_input[0] = top.value;
+      for (int i = 1; i < nargs; i++) {
+        cfunc_input[i] = stack_pop().value;
+      }
+      struct cfunction* func = &frame_current_bytecode(&frame_stk)->globals->cfunctions[*pc++];
+      top.value = cfunction_invoke(func, cfunc_input);
+      if (jv_is_valid(top.value)) {
+        stack_push(top);
+      } else {
+        print_error(top.value);
+        goto do_backtrack;
+      }
+      break;
+    }
+
+    case CALL_JQ: {
+      uint16_t nclosures = *pc++;
+      uint16_t* retaddr = pc + 2 + nclosures*2;
+      frame_ptr new_frame = frame_push(&frame_stk, 
+                                       make_closure(&frame_stk, frame_current(&frame_stk), pc),
+                                       retaddr);
+      pc += 2;
+      frame_ptr old_frame = forkable_stack_peek_next(&frame_stk, new_frame);
+      assert(nclosures == frame_self(new_frame)->bc->nclosures);
+      for (int i=0; i<nclosures; i++) {
+        *frame_closure_arg(new_frame, i) = make_closure(&frame_stk, old_frame, pc);
+        pc += 2;
+      }
+
+      pc = frame_current_bytecode(&frame_stk)->code;
+      break;
+    }
+
+    case RET: {
+      pc = *frame_current_retaddr(&frame_stk);
+      frame_pop(&frame_stk);
+      break;
+    }
+    }
+  }
+}
+
+
+void jq_init(struct bytecode* bc, jv input) {
+  forkable_stack_init(&data_stk, sizeof(stackval) * 100);
+  forkable_stack_init(&frame_stk, 1024);
+  forkable_stack_init(&fork_stk, 1024);
+  
+  stack_push(stackval_root(input));
+  struct closure top = {bc, -1};
+  frame_push(&frame_stk, top, 0);
+  frame_push_backtrack(&frame_stk, bc->code);
+}
+
+void jq_teardown() {
+  while (stack_restore()) {}
+
+  assert(forkable_stack_empty(&fork_stk));
+  assert(forkable_stack_empty(&data_stk));
+  assert(forkable_stack_empty(&frame_stk));
+  forkable_stack_free(&fork_stk);
+  forkable_stack_free(&data_stk);
+  forkable_stack_free(&frame_stk);
+
+  for (int i=0; i<pathsize; i++) {
+    jv_free(pathbuf[i]);
+  }
+  jv_mem_free(pathbuf);
+  pathbuf = 0;
+  pathsize = 0;
+}
+
+struct bytecode* jq_compile(const char* str) {
+  struct locfile locations;
+  locfile_init(&locations, str, strlen(str));
+  block program;
+  struct bytecode* bc = 0;
+  int nerrors = jq_parse(&locations, &program);
+  if (nerrors == 0) {
+    program = builtins_bind(program);
+    nerrors = block_compile(program, &locations, &bc);
+  }
+  if (nerrors) {
+    fprintf(stderr, "%d compile %s\n", nerrors, nerrors > 1 ? "errors" : "error");
+  }
+  locfile_free(&locations);
+  return bc;
+}

execute.h

@@ -0,0 +1,13 @@
+#ifndef EXECUTE_H
+#define EXECUTE_H
+#include "bytecode.h"
+
+
+struct bytecode* jq_compile(const char* str);
+
+void jq_init(struct bytecode* bc, jv value);
+jv jq_next();
+void jq_teardown();
+
+
+#endif

forkable_stack.h

@@ -0,0 +1,163 @@
+#ifndef FORKABLE_STACK_H
+#define FORKABLE_STACK_H
+#include <stddef.h>
+#include <assert.h>
+#include <string.h>
+#include "jv_alloc.h"
+
+struct forkable_stack_header {
+  /* distance in bytes between this header and the header
+     of the next object on the stack */
+  int next_delta;
+};
+
+#define FORKABLE_STACK_HEADER struct forkable_stack_header fk_header_
+
+struct forkable_stack {
+  // Stack grows down from stk+length
+  char* stk;
+
+  // stk+length is just past end of allocated area
+  // stk = jv_mem_alloc(length)
+  int length;
+
+  // stk+pos is header of top object, or stk+length if empty
+  int pos;
+
+  // everything after-or-including stk+savedlimit must be preserved
+  int savedlimit;
+};
+
+static void forkable_stack_check(struct forkable_stack* s) {
+  assert(s->stk);
+  assert(s->length > 0);
+  assert(s->pos >= 0 && s->pos <= s->length);
+  assert(s->savedlimit >= 0 && s->savedlimit <= s->length);
+}
+
+static int forkable_stack_empty(struct forkable_stack* s) {
+  return s->pos == s->length;
+}
+
+static void forkable_stack_init(struct forkable_stack* s, size_t sz) {
+  s->stk = jv_mem_alloc(sz);
+  s->length = sz;
+  s->pos = s->length;
+  s->savedlimit = s->length;
+  forkable_stack_check(s);
+}
+
+static void forkable_stack_free(struct forkable_stack* s) {
+  assert(forkable_stack_empty(s));
+  assert(s->savedlimit == s->length);
+  jv_mem_free(s->stk);
+  s->stk = 0;
+}
+
+static void* forkable_stack_push(struct forkable_stack* s, size_t sz_size) {
+  assert(sz_size > 0 && sz_size % sizeof(struct forkable_stack_header) == 0);
+  int size = (int)sz_size;
+  forkable_stack_check(s);
+  int curr = s->pos < s->savedlimit ? s->pos : s->savedlimit;
+  if (curr - size < 0) {
+    int oldlen = s->length;
+    s->length = (size + oldlen + 1024) * 2;
+    s->stk = jv_mem_realloc(s->stk, s->length);
+    int shift = s->length - oldlen;
+    memmove(s->stk + shift, s->stk, oldlen);
+    s->pos += shift;
+    s->savedlimit += shift;
+    curr += shift;
+  }
+  int newpos = curr - size;
+  assert(newpos < s->pos);
+  void* ret = (void*)(s->stk + newpos);
+  ((struct forkable_stack_header*)ret)->next_delta = s->pos - newpos;
+  s->pos = newpos;
+  forkable_stack_check(s);
+  return ret;
+}
+
+static void* forkable_stack_peek(struct forkable_stack* s) {
+  assert(!forkable_stack_empty(s));
+  forkable_stack_check(s);
+  return (void*)(s->stk + s->pos);
+}
+
+static void* forkable_stack_peek_next(struct forkable_stack* s, void* top) {
+  forkable_stack_check(s);
+  struct forkable_stack_header* elem = top;
+  int elempos = (char*)top - s->stk;
+  assert(elempos >= 0 && elempos < s->length);
+  assert(elempos + elem->next_delta <= s->length);
+  if (elempos + elem->next_delta < s->length) {
+    return (void*)(s->stk + elempos + elem->next_delta);
+  } else {
+    return 0;
+  }
+}
+
+// Returns 1 if the next forkable_stack_pop will permanently remove an
+// object from the stack (i.e. the top object was not saved with a fork)
+static int forkable_stack_pop_will_free(struct forkable_stack* s) {
+  return s->pos < s->savedlimit ? 1 : 0;
+}
+
+static void forkable_stack_pop(struct forkable_stack* s) {
+  forkable_stack_check(s);
+  struct forkable_stack_header* elem = forkable_stack_peek(s);
+  s->pos += elem->next_delta;
+}
+
+
+
+struct forkable_stack_state {
+  // We save the previous pos, savedlimit as
+  // length-pos, length-savedlimit since these
+  // values are stable across stack reallocations.
+  int prev_pos_delta, prev_limit_delta;
+};
+
+static void forkable_stack_save(struct forkable_stack* s, struct forkable_stack_state* state) {
+  forkable_stack_check(s);
+  state->prev_pos_delta = s->length - s->pos;
+  state->prev_limit_delta = s->length - s->savedlimit;
+  if (s->pos < s->savedlimit) s->savedlimit = s->pos;
+}
+
+static void forkable_stack_switch(struct forkable_stack* s, struct forkable_stack_state* state) {
+  forkable_stack_check(s);
+  int curr_pos = s->pos;
+  s->pos = s->length - state->prev_pos_delta;
+  state->prev_pos_delta = s->length - curr_pos;
+
+  int curr_limit = s->savedlimit;
+  if (curr_pos < curr_limit) s->savedlimit = curr_pos;
+  //state->prevlimit = curr_limit; FIXME clean up
+  forkable_stack_check(s);
+}
+
+static void forkable_stack_restore(struct forkable_stack* s, struct forkable_stack_state* state) {
+  forkable_stack_check(s);
+  assert(s->savedlimit <= s->length - state->prev_pos_delta);
+  assert(s->savedlimit <= s->length - state->prev_limit_delta);
+  s->pos = s->length - state->prev_pos_delta;
+  s->savedlimit = s->length - state->prev_limit_delta;
+  forkable_stack_check(s);
+}
+
+typedef int stack_idx;
+
+static stack_idx forkable_stack_to_idx(struct forkable_stack* s, void* ptr) {
+  char* item = ptr;
+  int pos = item - s->stk;
+  assert(pos >= 0 && pos < s->length);
+  return s->length - pos;
+}
+
+static void* forkable_stack_from_idx(struct forkable_stack* s, stack_idx idx) {
+  assert(idx >= 1 && idx <= s->length);
+  return &s->stk[s->length - idx];
+}
+
+#endif

frame_layout.h

@@ -0,0 +1,118 @@
+#ifndef FRAME_LAYOUT_H
+#include "forkable_stack.h"
+#include "bytecode.h"
+
+struct closure {
+  struct bytecode* bc;
+  stack_idx env;
+};
+
+struct continuation {
+  struct bytecode* bc;
+  stack_idx env;
+  uint16_t* retaddr;
+  // FIXME: probably not necessary as a separate field
+  int is_backtrack_frame;
+};
+
+typedef union frame_elem {
+  FORKABLE_STACK_HEADER;
+  struct continuation cont;
+  struct closure closure;
+  jv jsonval;
+} *frame_ptr;
+
+/*
+ * Frame layout
+ * fr[0] - FORKABLE_STACK_HEADER (next pointer)
+ * fr[1] - self (used to store return addresses, etc)
+ * fr[2...nclosures+2] - closure params
+ * fr[nclosures+2..nclosures+nlocals+2] - local variables
+ */
+
+static int frame_size(struct bytecode* bc) {
+  return sizeof(union frame_elem) * (bc->nclosures + bc->nlocals + 2);
+}
+
+static struct continuation* frame_self(frame_ptr fr) {
+  return &fr[1].cont;
+}
+
+static struct closure* frame_closure_arg(frame_ptr fr, int closure) {
+  assert(closure >= 0);
+  assert(closure < frame_self(fr)->bc->nclosures);
+  return &fr[2+closure].closure;
+}
+
+static jv* frame_local_var(frame_ptr fr, int var) {
+  assert(var >= 0);
+  assert(var < frame_self(fr)->bc->nlocals);
+  return &fr[2 + frame_self(fr)->bc->nclosures + var].jsonval;
+}
+
+
+static frame_ptr frame_current(struct forkable_stack* stk) {
+  frame_ptr fp = forkable_stack_peek(stk);
+  if (forkable_stack_peek_next(stk, fp)) {
+    struct bytecode* bc = frame_self(forkable_stack_peek_next(stk, fp))->bc;
+    assert(frame_self(fp)->retaddr >= bc->code && frame_self(fp)->retaddr < bc->code + bc->codelen);
+  } else {
+    assert(frame_self(fp)->retaddr == 0);
+  }
+  return fp;
+}
+
+static struct bytecode* frame_current_bytecode(struct forkable_stack* stk) {
+  return frame_self(frame_current(stk))->bc;
+}
+static uint16_t** frame_current_retaddr(struct forkable_stack* stk) {
+  return &frame_self(frame_current(stk))->retaddr;
+}
+
+static frame_ptr frame_get_parent(struct forkable_stack* stk, frame_ptr fr) {
+  return forkable_stack_from_idx(stk, frame_self(fr)->env);
+}
+
+static frame_ptr frame_get_level(struct forkable_stack* stk, frame_ptr fr, int level) {
+  for (int i=0; i<level; i++) {
+    fr = frame_get_parent(stk, fr);
+  }
+  return fr;
+}
+
+static frame_ptr frame_push(struct forkable_stack* stk, struct closure cl, uint16_t* retaddr) {
+  frame_ptr fp = forkable_stack_push(stk, frame_size(cl.bc));
+  struct continuation* cc = frame_self(fp);
+  cc->bc = cl.bc;
+  cc->env = cl.env;
+  cc->retaddr = retaddr;
+  cc->is_backtrack_frame = 0;
+  for (int i=0; i<cl.bc->nlocals; i++) {
+    *frame_local_var(fp, i) = jv_invalid();
+  }
+  return fp;
+}
+
+static frame_ptr frame_push_backtrack(struct forkable_stack* stk, uint16_t* retaddr) {
+  struct continuation cc = *frame_self(frame_current(stk));
+  frame_ptr fp = forkable_stack_push(stk, sizeof(union frame_elem) * 2);
+  assert(!cc.is_backtrack_frame);
+  cc.is_backtrack_frame = 1;
+  cc.retaddr = retaddr;
+  *frame_self(fp) = cc;
+  return fp;
+}
+
+static void frame_pop(struct forkable_stack* stk) {
+  frame_ptr fp = frame_current(stk);
+  if (forkable_stack_pop_will_free(stk) && 
+      !frame_self(fp)->is_backtrack_frame) {
+    int nlocals = frame_self(fp)->bc->nlocals;
+    for (int i=0; i<nlocals; i++) {
+      jv_free(*frame_local_var(fp, i));
+    }
+  }
+  forkable_stack_pop(stk);
+}
+
+#endif

gen_utf8_tables.py

@@ -0,0 +1,30 @@
+#!/usr/bin/python
+
+mask = lambda n: (1 << n) - 1
+
+
+def print_table(type, name, t):
+    assert len(t) == 256
+    print("static const %s %s[] =" % (type, name))
+    first = True
+    for i in range(0,len(t),16):
+        print ((" {" if i == 0 else "  ") +
+               ", ".join("0x%02x"%n for n in t[i:i+16]) + 
+               ("," if i + 16 < 256 else "};"))
+
+
+def utf8info(c):
+    if c < 0x80: return 1, mask(7)
+    if 0x80 <= c <= 0xBF: return 255, mask(6)
+    if 0xC0 <= c <= 0xC1: return 0, 0
+    if 0xC2 <= c <= 0xDF: return 2, mask(5)
+    if 0xE0 <= c <= 0xEF: return 3, mask(4)
+    if 0xF0 <= c <= 0xF4: return 4, mask(3)
+    if 0xF4 <= c <= 0xFF: return 0, 0
+        
+table = lambda i: [utf8info(c)[i] for c in range(256)]
+
+print("#define UTF8_CONTINUATION_BYTE ((unsigned char)255)")
+
+print_table("unsigned char", "utf8_coding_length", table(0))
+print_table("unsigned char", "utf8_coding_bits", table(1))

jq_test.c

@@ -0,0 +1,246 @@
+#include <assert.h>
+#include <stdio.h>
+#include <string.h>
+#include <stdlib.h>
+#include "jv.h"
+#include "execute.h"
+
+static void jv_test();
+static void run_jq_tests();
+
+FILE* testdata;
+
+int main(int argc, char* argv[]) {
+  jv_test();
+  if (argc == 1) {
+    testdata = fopen("testdata", "r");
+  } else if (argc == 2) {
+    if (!strcmp(argv[1], "-")) {
+      testdata = stdin;
+    } else {
+      testdata = fopen(argv[1], "r");
+    }
+  } else {
+    printf("usage: %s OR cat testdata | %s - OR %s testdata\n", argv[0], argv[0], argv[0]);
+    return 127;
+  }
+  run_jq_tests();
+  if (testdata != stdin) fclose(testdata);
+}
+
+
+
+
+static int skipline(const char* buf) {
+  int p = 0;
+  while (buf[p] == ' ' || buf[p] == '\t') p++;
+  if (buf[p] == '#' || buf[p] == '\n' || buf[p] == 0) return 1;
+  return 0;
+}
+
+static void run_jq_tests() {
+  char buf[4096];
+  int tests = 0, passed = 0, invalid = 0;
+
+  while (1) {
+    if (!fgets(buf, sizeof(buf), testdata)) break;
+    if (skipline(buf)) continue;
+    if (buf[strlen(buf)-1] == '\n') buf[strlen(buf)-1] = 0;
+    printf("Testing %s\n", buf);
+    int pass = 1;
+    tests++;
+    struct bytecode* bc = jq_compile(buf);
+    if (!bc) {invalid++; continue;}
+#if JQ_DEBUG
+    printf("Disassembly:\n");
+    dump_disassembly(2, bc);
+    printf("\n");
+#endif
+    fgets(buf, sizeof(buf), testdata);
+    jv input = jv_parse(buf);
+    if (!jv_is_valid(input)){ invalid++; continue; }
+    jq_init(bc, input);
+
+    while (fgets(buf, sizeof(buf), testdata)) {
+      if (skipline(buf)) break;
+      jv expected = jv_parse(buf);
+      if (!jv_is_valid(expected)){ invalid++; continue; }
+      jv actual = jq_next();
+      if (!jv_is_valid(actual)) {
+        jv_free(actual);
+        printf("*** Insufficient results\n");
+        pass = 0;
+        break;
+      } else if (!jv_equal(jv_copy(expected), jv_copy(actual))) {
+        printf("*** Expected ");
+        jv_dump(jv_copy(expected), 0);
+        printf(", but got ");
+        jv_dump(jv_copy(actual), 0);
+        printf("\n");
+        pass = 0;
+      }
+      jv as_string = jv_dump_string(jv_copy(expected), rand() & ~JV_PRINT_COLOUR);
+      jv reparsed = jv_parse_sized(jv_string_value(as_string), jv_string_length(jv_copy(as_string)));
+      assert(jv_equal(jv_copy(expected), jv_copy(reparsed)));
+      jv_free(as_string);
+      jv_free(reparsed);
+      jv_free(expected);
+      jv_free(actual);
+    }
+    if (pass) {
+      jv extra = jq_next();
+      if (jv_is_valid(extra)) {
+        printf("*** Superfluous result: ");
+        jv_dump(extra, 0);
+        printf("\n");
+        pass = 0;
+      } else {
+        jv_free(extra);
+      }
+    }
+    jq_teardown();
+    bytecode_free(bc);
+    passed+=pass;
+  }
+  printf("%d of %d tests passed (%d malformed)\n", passed,tests,invalid);
+  if (passed != tests) exit(1);
+}
+
+
+static void jv_test() {
+  /// Arrays and numbers
+  {
+    jv a = jv_array();
+    assert(jv_get_kind(a) == JV_KIND_ARRAY);
+    assert(jv_array_length(jv_copy(a)) == 0);
+    assert(jv_array_length(jv_copy(a)) == 0);
+  
+    a = jv_array_append(a, jv_number(42));
+    assert(jv_array_length(jv_copy(a)) == 1);
+    assert(jv_number_value(jv_array_get(jv_copy(a), 0)) == 42);
+
+    jv a2 = jv_array_append(jv_array(), jv_number(42));
+    assert(jv_equal(jv_copy(a), jv_copy(a)));
+    assert(jv_equal(jv_copy(a2), jv_copy(a2)));
+    assert(jv_equal(jv_copy(a), jv_copy(a2)));
+    assert(jv_equal(jv_copy(a2), jv_copy(a)));
+    jv_free(a2);
+
+    a2 = jv_array_append(jv_array(), jv_number(19));
+    assert(!jv_equal(jv_copy(a), jv_copy(a2)));
+    assert(!jv_equal(jv_copy(a2), jv_copy(a)));
+    jv_free(a2);
+
+
+    assert(a.val.complex.ptr->count == 1);
+    a = jv_array_append(a, jv_copy(a));
+    assert(a.val.complex.ptr->count == 1);
+
+    assert(jv_array_length(jv_copy(a)) == 2);
+    assert(jv_number_value(jv_array_get(jv_copy(a), 0)) == 42);
+
+    for (int i=0; i<10; i++) {
+      jv subarray = jv_array_get(jv_copy(a), 1);
+      assert(jv_get_kind(subarray) == JV_KIND_ARRAY);
+      assert(jv_array_length(jv_copy(subarray)) == 1);
+      assert(jv_number_value(jv_array_get(jv_copy(subarray), 0)) == 42);
+      jv_free(subarray);
+    }
+
+
+    jv subarray = jv_array_get(jv_copy(a), 1);
+    assert(jv_get_kind(subarray) == JV_KIND_ARRAY);
+    assert(jv_array_length(jv_copy(subarray)) == 1);
+    assert(jv_number_value(jv_array_get(jv_copy(subarray), 0)) == 42);
+
+    jv sub2 = jv_copy(subarray);
+    sub2 = jv_array_append(sub2, jv_number(19));
+
+    assert(jv_get_kind(sub2) == JV_KIND_ARRAY);
+    assert(jv_array_length(jv_copy(sub2)) == 2);
+    assert(jv_number_value(jv_array_get(jv_copy(sub2), 0)) == 42);
+    assert(jv_number_value(jv_array_get(jv_copy(sub2), 1)) == 19);
+
+    assert(jv_get_kind(subarray) == JV_KIND_ARRAY);
+    assert(jv_array_length(jv_copy(subarray)) == 1);
+    assert(jv_number_value(jv_array_get(jv_copy(subarray), 0)) == 42);
+
+    jv_free(subarray);
+
+    void* before = sub2.val.complex.ptr;
+    sub2 = jv_array_append(sub2, jv_number(200));
+    void* after = sub2.val.complex.ptr;
+    assert(before == after);
+    jv_free(sub2);
+
+    jv a3 = jv_array_append(jv_copy(a), jv_number(19));
+    assert(jv_array_length(jv_copy(a3)) == 3);
+    assert(jv_number_value(jv_array_get(jv_copy(a3), 0)) == 42);
+    assert(jv_array_length(jv_array_get(jv_copy(a3), 1)) == 1);
+    assert(jv_number_value(jv_array_get(jv_copy(a3), 2)) == 19);
+    jv_free(a3);
+
+
+    assert(jv_array_length(jv_copy(a)) == 2);
+    assert(jv_number_value(jv_array_get(jv_copy(a), 0)) == 42);
+    assert(jv_array_length(jv_array_get(jv_copy(a), 1)) == 1);
+
+
+    jv_dump(jv_copy(a), 0); printf("\n");
+    jv_free(a);
+  }
+
+
+  /// Strings
+  {
+    assert(jv_equal(jv_string("foo"), jv_string_sized("foo", 3)));
+    char nasty[] = "foo\0";
+    jv shortstr = jv_string(nasty), longstr = jv_string_sized(nasty, sizeof(nasty));
+    assert(jv_string_length(shortstr) == (int)strlen(nasty));
+    assert(jv_string_length(longstr) == (int)sizeof(nasty));
+
+  
+    char a1s[] = "hello", a2s[] = "hello", bs[] = "goodbye";
+    jv a1 = jv_string(a1s), a2 = jv_string(a2s), b = jv_string(bs);
+    assert(jv_equal(jv_copy(a1), jv_copy(a2)));
+    assert(jv_equal(jv_copy(a2), jv_copy(a1)));
+    assert(!jv_equal(jv_copy(a1), jv_copy(b)));
+    
+    assert(jv_string_hash(jv_copy(a1)) == jv_string_hash(jv_copy(a1)));
+    assert(jv_string_hash(jv_copy(a1)) == jv_string_hash(jv_copy(a2)));
+    assert(jv_string_hash(jv_copy(b)) != jv_string_hash(jv_copy(a1)));
+    jv_free(a1);
+    jv_free(a2);
+    jv_free(b);
+
+    assert(jv_equal(jv_string("hello42!"), jv_string_fmt("hello%d%s", 42, "!")));
+    char big[20000];
+    for (int i=0; i<(int)sizeof(big); i++) big[i] = 'a';
+    big[sizeof(big)-1] = 0;
+    jv str = jv_string_fmt("%s", big);
+    assert(jv_string_length(jv_copy(str)) == sizeof(big) - 1);
+    assert(!strcmp(big, jv_string_value(str)));
+    jv_free(str);
+  }
+
+  /// Objects
+  {
+    jv o1 = jv_object();
+    o1 = jv_object_set(o1, jv_string("foo"), jv_number(42));
+    o1 = jv_object_set(o1, jv_string("bar"), jv_number(24));
+    assert(jv_number_value(jv_object_get(jv_copy(o1), jv_string("foo"))) == 42);
+    assert(jv_number_value(jv_object_get(jv_copy(o1), jv_string("bar"))) == 24);
+
+    jv o2 = jv_object_set(jv_copy(o1), jv_string("foo"), jv_number(420));
+    o2 = jv_object_set(o2, jv_string("bar"), jv_number(240));
+    assert(jv_number_value(jv_object_get(jv_copy(o1), jv_string("foo"))) == 42);
+    assert(jv_number_value(jv_object_get(jv_copy(o1), jv_string("bar"))) == 24);
+    assert(jv_number_value(jv_object_get(jv_copy(o2), jv_string("foo"))) == 420);
+    jv_free(o1);
+    assert(jv_number_value(jv_object_get(jv_copy(o2), jv_string("bar"))) == 240);
+
+    jv_dump(jv_copy(o2), 0); printf("\n");
+
+    jv_free(o2);
+  }
+}

jv.h

@@ -0,0 +1,146 @@
+#ifndef JV_H
+#define JV_H
+
+#include <stdint.h>
+#include <assert.h>
+#include <stddef.h>
+
+
+
+typedef enum {
+  JV_KIND_INVALID,
+  JV_KIND_NULL,
+  JV_KIND_FALSE,
+  JV_KIND_TRUE,
+  JV_KIND_NUMBER,
+  JV_KIND_STRING,
+  JV_KIND_ARRAY,
+  JV_KIND_OBJECT
+} jv_kind;
+
+typedef struct {
+  size_t count;
+} jv_refcnt;
+
+typedef struct{
+  jv_refcnt* ptr;
+  int i[2];
+} jv_complex;
+
+typedef struct {
+  jv_kind kind;
+  union {
+    double number;
+    jv_complex complex;
+  } val;
+} jv;
+
+/*
+ * All jv_* functions consume (decref) input and produce (incref) output
+ * Except jv_copy
+ */
+
+jv_kind jv_get_kind(jv);
+const char* jv_kind_name(jv_kind);
+static int jv_is_valid(jv x) { return jv_get_kind(x) != JV_KIND_INVALID; }
+
+jv jv_copy(jv);
+void jv_free(jv);
+
+int jv_equal(jv, jv);
+int jv_contains(jv, jv);
+
+jv jv_invalid();
+jv jv_invalid_with_msg(jv);
+jv jv_invalid_get_msg(jv);
+int jv_invalid_has_msg(jv);
+
+
+jv jv_null();
+jv jv_true();
+jv jv_false();
+jv jv_bool(int);
+
+jv jv_number(double);
+double jv_number_value(jv);
+
+jv jv_array();
+jv jv_array_sized(int);
+int jv_array_length(jv);
+jv jv_array_get(jv, int);
+jv jv_array_set(jv, int, jv);
+jv jv_array_append(jv, jv);
+jv jv_array_concat(jv, jv);
+jv jv_array_slice(jv, int, int);
+
+
+jv jv_string(const char*);
+jv jv_string_sized(const char*, int);
+int jv_string_length(jv);
+uint32_t jv_string_hash(jv);
+const char* jv_string_value(jv);
+jv jv_string_concat(jv, jv);
+jv jv_string_fmt(const char*, ...);
+jv jv_string_append_buf(jv a, const char* buf, int len);
+jv jv_string_append_str(jv a, const char* str);
+
+jv jv_object();
+jv jv_object_get(jv object, jv key);
+jv jv_object_set(jv object, jv key, jv value);
+jv jv_object_delete(jv object, jv key);
+int jv_object_length(jv object);
+jv jv_object_merge(jv, jv);
+
+int jv_object_iter(jv);
+int jv_object_iter_next(jv, int);
+int jv_object_iter_valid(jv, int);
+jv jv_object_iter_key(jv, int);
+jv jv_object_iter_value(jv, int);
+#define jv_object_foreach(i,t) \
+  for (int i = jv_object_iter(t);               \
+       jv_object_iter_valid(t, i);              \
+       i = jv_object_iter_next(t, i))           \
+ 
+
+
+int jv_get_refcnt(jv);
+
+enum { JV_PRINT_PRETTY = 1, JV_PRINT_ASCII = 2, JV_PRINT_COLOUR = 4 };
+void jv_dump(jv, int flags);
+jv jv_dump_string(jv, int flags);
+
+jv jv_parse(const char* string);
+jv jv_parse_sized(const char* string, int length);
+
+
+
+
+
+
+
+
+
+
+#endif
+
+
+/*
+
+true/false/null:
+check kind
+
+number:
+introduce/eliminate jv
+to integer
+
+array:
+copy
+free
+slice
+index
+update
+
+updateslice?
+
+
+ */

jv_alloc.c

@@ -0,0 +1,28 @@
+#include <stdlib.h>
+#include <stdio.h>
+#include "jv_alloc.h"
+
+static void memory_exhausted() {
+  fprintf(stderr, "error: cannot allocate memory\n");
+  abort();
+}
+
+void* jv_mem_alloc(size_t sz) {
+  void* p = malloc(sz);
+  if (!p) {
+    memory_exhausted();
+  }
+  return p;
+}
+
+void jv_mem_free(void* p) {
+  free(p);
+}
+
+void* jv_mem_realloc(void* p, size_t sz) {
+  p = realloc(p, sz);
+  if (!p) {
+    memory_exhausted();
+  }
+  return p;
+}

jv_alloc.h

@@ -0,0 +1,10 @@
+#ifndef JV_ALLOC_H
+#define JV_ALLOC_H
+
+#include <stddef.h>
+
+void* jv_mem_alloc(size_t);
+void jv_mem_free(void*);
+__attribute__((warn_unused_result)) void* jv_mem_realloc(void*, size_t);
+
+#endif

jv_aux.c

@@ -0,0 +1,253 @@
+#include "jv_aux.h"
+#include <string.h>
+#include <stdlib.h>
+#include "jv_alloc.h"
+
+jv jv_lookup(jv t, jv k) {
+  jv v;
+  if (jv_get_kind(t) == JV_KIND_OBJECT && jv_get_kind(k) == JV_KIND_STRING) {
+    v = jv_object_get(t, k);
+    if (!jv_is_valid(v)) {
+      jv_free(v);
+      v = jv_null();
+    }
+  } else if (jv_get_kind(t) == JV_KIND_ARRAY && jv_get_kind(k) == JV_KIND_NUMBER) {
+    // FIXME: don't do lookup for noninteger index
+    v = jv_array_get(t, (int)jv_number_value(k));
+    if (!jv_is_valid(v)) {
+      jv_free(v);
+      v = jv_null();
+    }
+  } else if (jv_get_kind(t) == JV_KIND_NULL && 
+             (jv_get_kind(k) == JV_KIND_STRING || jv_get_kind(k) == JV_KIND_NUMBER)) {
+    jv_free(t);
+    jv_free(k);
+    v = jv_null();
+  } else {
+    v = jv_invalid_with_msg(jv_string_fmt("Cannot index %s with %s",
+                                          jv_kind_name(jv_get_kind(t)),
+                                          jv_kind_name(jv_get_kind(k))));
+    jv_free(t);
+    jv_free(k);
+  }
+  return v;
+}
+
+jv jv_modify(jv t, jv k, jv v) {
+  int isnull = jv_get_kind(t) == JV_KIND_NULL;
+  if (jv_get_kind(k) == JV_KIND_STRING && 
+      (jv_get_kind(t) == JV_KIND_OBJECT || isnull)) {
+    if (isnull) t = jv_object();
+    t = jv_object_set(t, k, v);
+  } else if (jv_get_kind(k) == JV_KIND_NUMBER &&
+             (jv_get_kind(t) == JV_KIND_ARRAY || isnull)) {
+    if (isnull) t = jv_array();
+    t = jv_array_set(t, (int)jv_number_value(k), v);
+  } else {
+    jv err = jv_invalid_with_msg(jv_string_fmt("Cannot update field at %s index of %s",
+                                               jv_kind_name(jv_get_kind(t)),
+                                               jv_kind_name(jv_get_kind(v))));
+    jv_free(t);
+    jv_free(k);
+    jv_free(v);
+    t = err;
+  }
+  return t;
+}
+
+jv jv_insert(jv root, jv value, jv* path, int pathlen) {
+  if (pathlen == 0) {
+    jv_free(root);
+    return value;
+  }
+  return jv_modify(root, jv_copy(*path), 
+                   jv_insert(jv_lookup(jv_copy(root), jv_copy(*path)), value, path+1, pathlen-1));
+}
+
+
+static int string_cmp(const void* pa, const void* pb){
+  const jv* a = pa;
+  const jv* b = pb;
+  int lena = jv_string_length(jv_copy(*a));
+  int lenb = jv_string_length(jv_copy(*b));
+  int minlen = lena < lenb ? lena : lenb;
+  int r = memcmp(jv_string_value(*a), jv_string_value(*b), minlen);
+  if (r == 0) r = lena - lenb;
+  return r;
+}
+
+jv jv_keys(jv x) {
+  if (jv_get_kind(x) == JV_KIND_OBJECT) {
+    int nkeys = jv_object_length(jv_copy(x));
+    jv* keys = jv_mem_alloc(sizeof(jv) * nkeys);
+    int kidx = 0;
+    jv_object_foreach(i, x) {
+      keys[kidx++] = jv_object_iter_key(x, i);
+    }
+    qsort(keys, nkeys, sizeof(jv), string_cmp);
+    jv answer = jv_array_sized(nkeys);
+    for (int i = 0; i<nkeys; i++) {
+      answer = jv_array_append(answer, keys[i]);
+    }
+    jv_mem_free(keys);
+    jv_free(x);
+    return answer;
+  } else if (jv_get_kind(x) == JV_KIND_ARRAY) {
+    int n = jv_array_length(x);
+    jv answer = jv_array();
+    for (int i=0; i<n; i++){
+      answer = jv_array_set(answer, i, jv_number(i));
+    }
+    return answer;
+  } else {
+    assert(0 && "jv_keys passed something neither object nor array");
+    return jv_invalid();
+  }
+}
+
+int jv_cmp(jv a, jv b) {
+  if (jv_get_kind(a) != jv_get_kind(b)) {
+    int r = (int)jv_get_kind(a) - (int)jv_get_kind(b);
+    jv_free(a);
+    jv_free(b);
+    return r;
+  }
+  int r = 0;
+  switch (jv_get_kind(a)) {
+  default:
+    assert(0 && "invalid kind passed to jv_cmp");
+  case JV_KIND_NULL:
+  case JV_KIND_FALSE:
+  case JV_KIND_TRUE:
+    // there's only one of each of these values
+    r = 0;
+    break;
+
+  case JV_KIND_NUMBER: {
+    double da = jv_number_value(a), db = jv_number_value(b);
+    
+    // handle NaN as though it were null
+    if (da != da) r = jv_cmp(jv_null(), jv_number(db));
+    else if (db != db) r = jv_cmp(jv_number(da), jv_null());
+    else if (da < db) r = -1;
+    else if (da == db) r = 0;
+    else r = 1;
+    break;
+  }
+
+  case JV_KIND_STRING: {
+    r = string_cmp(&a, &b);
+    break;
+  }
+
+  case JV_KIND_ARRAY: {
+    // Lexical ordering of arrays
+    int i = 0;
+    while (r == 0) {
+      int a_done = i >= jv_array_length(jv_copy(a));
+      int b_done = i >= jv_array_length(jv_copy(b));
+      if (a_done || b_done) {
+        r = b_done - a_done; //suddenly, logic
+        break;
+      }
+      jv xa = jv_array_get(jv_copy(a), i);
+      jv xb = jv_array_get(jv_copy(b), i);
+      r = jv_cmp(xa, xb);
+      i++;
+    }
+    break;
+  }
+
+  case JV_KIND_OBJECT: {
+    jv keys_a = jv_keys(jv_copy(a));
+    jv keys_b = jv_keys(jv_copy(b));
+    r = jv_cmp(jv_copy(keys_a), keys_b);
+    if (r == 0) {
+      for (int i=0; i<jv_array_length(jv_copy(keys_a)); i++) {
+        jv key = jv_array_get(jv_copy(keys_a), i);
+        jv xa = jv_object_get(jv_copy(a), jv_copy(key));
+        jv xb = jv_object_get(jv_copy(b), key);
+        r = jv_cmp(xa, xb);
+        if (r) break;
+      }
+    }
+    jv_free(keys_a);
+    break;
+  }
+  }
+
+  jv_free(a);
+  jv_free(b);
+  return r;
+}
+
+
+struct sort_entry {
+  jv object;
+  jv key;
+};
+
+static int sort_cmp(const void* pa, const void* pb) {
+  const struct sort_entry* a = pa;
+  const struct sort_entry* b = pb;
+  int r = jv_cmp(jv_copy(a->key), jv_copy(b->key));
+  // comparing by address if r == 0 makes the sort stable
+  return r ? r : (int)(a - b);
+}
+
+static struct sort_entry* sort_items(jv objects, jv keys) {
+  assert(jv_get_kind(objects) == JV_KIND_ARRAY);
+  assert(jv_get_kind(keys) == JV_KIND_ARRAY);
+  assert(jv_array_length(jv_copy(objects)) == jv_array_length(jv_copy(keys)));
+  int n = jv_array_length(jv_copy(objects));
+  struct sort_entry* entries = jv_mem_alloc(sizeof(struct sort_entry) * n);
+  for (int i=0; i<n; i++) {
+    entries[i].object = jv_array_get(jv_copy(objects), i);
+    entries[i].key = jv_array_get(jv_copy(keys), i);
+  }
+  jv_free(objects);
+  jv_free(keys);
+  qsort(entries, n, sizeof(struct sort_entry), sort_cmp);
+  return entries;
+}
+
+jv jv_sort(jv objects, jv keys) {
+  assert(jv_get_kind(objects) == JV_KIND_ARRAY);
+  assert(jv_get_kind(keys) == JV_KIND_ARRAY);
+  assert(jv_array_length(jv_copy(objects)) == jv_array_length(jv_copy(keys)));
+  int n = jv_array_length(jv_copy(objects));
+  struct sort_entry* entries = sort_items(objects, keys);
+  jv ret = jv_array();
+  for (int i=0; i<n; i++) {
+    jv_free(entries[i].key);
+    ret = jv_array_set(ret, i, entries[i].object);
+  }
+  jv_mem_free(entries);
+  return ret;
+}
+
+jv jv_group(jv objects, jv keys) {
+  assert(jv_get_kind(objects) == JV_KIND_ARRAY);
+  assert(jv_get_kind(keys) == JV_KIND_ARRAY);
+  assert(jv_array_length(jv_copy(objects)) == jv_array_length(jv_copy(keys)));
+  int n = jv_array_length(jv_copy(objects));
+  struct sort_entry* entries = sort_items(objects, keys);
+  jv ret = jv_array();
+  jv curr_key = entries[0].key;
+  jv group = jv_array_append(jv_array(), entries[0].object);
+  for (int i = 1; i < n; i++) {
+    if (jv_equal(jv_copy(curr_key), jv_copy(entries[i].key))) {
+      jv_free(entries[i].key);
+    } else {
+      jv_free(curr_key);
+      curr_key = entries[i].key;
+      ret = jv_array_append(ret, group);
+      group = jv_array();
+    }
+    group = jv_array_append(group, entries[i].object);
+  }
+  jv_free(curr_key);
+  ret = jv_array_append(ret, group);
+  jv_mem_free(entries);
+  return ret;
+}

jv_aux.h

@@ -0,0 +1,17 @@
+#ifndef JV_AUX_H
+#define JV_AUX_H
+
+#include "jv.h"
+
+jv jv_lookup(jv t, jv k);
+jv jv_modify(jv t, jv k, jv v);
+jv jv_insert(jv root, jv value, jv* path, int pathlen);
+
+jv jv_keys(jv /*object or array*/);
+int jv_cmp(jv, jv);
+
+jv jv_group(jv objects, jv keys);
+jv jv_sort(jv objects, jv keys);
+
+
+#endif

jv_dtoa.h

@@ -0,0 +1,22 @@
+#ifndef JV_DTOA_H
+#define JV_DTOA_H
+#define Kmax 7
+
+struct Bigint;
+struct dtoa_context {
+  struct Bigint *freelist[Kmax+1];
+  struct Bigint *p5s;
+};
+
+void jvp_dtoa_context_init(struct dtoa_context* ctx);
+void jvp_dtoa_context_free(struct dtoa_context* ctx);
+
+double jvp_strtod(struct dtoa_context* C, const char* s, char** se);
+
+
+char* jvp_dtoa(struct dtoa_context* C, double dd, int mode, int ndigits, int *decpt, int *sign, char **rve);
+void jvp_freedtoa(struct dtoa_context* C, char *s);
+
+#define JVP_DTOA_FMT_MAX_LEN 32
+char* jvp_dtoa_fmt(struct dtoa_context* C, register char *b, double x);
+#endif

jv_parse.c

@@ -0,0 +1,439 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include "jv.h"
+#include "jv_dtoa.h"
+#include "jv_parse.h"
+#include "jv_unicode.h"
+#include "jv_alloc.h"
+
+typedef const char* presult;
+
+#define TRY(x) do {presult msg__ = (x); if (msg__) return msg__; } while(0)
+#ifdef __GNUC__
+#define pfunc __attribute__((warn_unused_result)) presult
+#else
+#define pfunc presult
+#endif
+
+void jv_parser_init(struct jv_parser* p) {
+  p->stack = 0;
+  p->stacklen = p->stackpos = 0;
+  p->next = jv_invalid();
+  p->tokenbuf = 0;
+  p->tokenlen = p->tokenpos = 0;
+  p->st = JV_PARSER_NORMAL;
+  p->curr_buf = 0;
+  p->curr_buf_length = p->curr_buf_pos = p->curr_buf_is_partial = 0;
+  p->bom_strip_position = 0;
+  jvp_dtoa_context_init(&p->dtoa);
+}
+
+void jv_parser_free(struct jv_parser* p) {
+  jv_free(p->next);
+  for (int i=0; i<p->stackpos; i++) 
+    jv_free(p->stack[i]);
+  jv_mem_free(p->stack);
+  jv_mem_free(p->tokenbuf);
+  jvp_dtoa_context_free(&p->dtoa);
+}
+
+static pfunc value(struct jv_parser* p, jv val) {
+  if (jv_is_valid(p->next)) return "Expected separator between values";
+  jv_free(p->next);
+  p->next = val;
+  return 0;
+}
+
+static void push(struct jv_parser* p, jv v) {
+  assert(p->stackpos <= p->stacklen);
+  if (p->stackpos == p->stacklen) {
+    p->stacklen = p->stacklen * 2 + 10;
+    p->stack = jv_mem_realloc(p->stack, p->stacklen * sizeof(jv));
+  }
+  assert(p->stackpos < p->stacklen);
+  p->stack[p->stackpos++] = v;
+}
+
+static pfunc token(struct jv_parser* p, char ch) {
+  switch (ch) {
+  case '[':
+    if (jv_is_valid(p->next)) return "Expected separator between values";
+    push(p, jv_array());
+    break;
+
+  case '{':
+    if (jv_is_valid(p->next)) return "Expected separator between values";
+    push(p, jv_object());
+    break;
+
+  case ':':
+    if (!jv_is_valid(p->next)) 
+      return "Expected string key before ':'";
+    if (p->stackpos == 0 || jv_get_kind(p->stack[p->stackpos-1]) != JV_KIND_OBJECT)
+      return "':' not as part of an object";
+    if (jv_get_kind(p->next) != JV_KIND_STRING)
+      return "Object keys must be strings";
+    push(p, p->next);
+    p->next = jv_invalid();
+    break;
+
+  case ',':
+    if (!jv_is_valid(p->next))
+      return "Expected value before ','";
+    if (p->stackpos == 0)
+      return "',' not as part of an object or array";
+    if (jv_get_kind(p->stack[p->stackpos-1]) == JV_KIND_ARRAY) {
+      p->stack[p->stackpos-1] = jv_array_append(p->stack[p->stackpos-1], p->next);
+      p->next = jv_invalid();
+    } else if (jv_get_kind(p->stack[p->stackpos-1]) == JV_KIND_STRING) {
+      assert(p->stackpos > 1 && jv_get_kind(p->stack[p->stackpos-2]) == JV_KIND_OBJECT);
+      p->stack[p->stackpos-2] = jv_object_set(p->stack[p->stackpos-2], 
+                                              p->stack[p->stackpos-1], p->next);
+      p->stackpos--;
+      p->next = jv_invalid();
+    } else {
+      // this case hits on input like {"a", "b"}
+      return "Objects must consist of key:value pairs";
+    }
+    break;
+
+  case ']':
+    if (p->stackpos == 0 || jv_get_kind(p->stack[p->stackpos-1]) != JV_KIND_ARRAY)
+      return "Unmatched ']'";
+    if (jv_is_valid(p->next)) {
+      p->stack[p->stackpos-1] = jv_array_append(p->stack[p->stackpos-1], p->next);
+      p->next = jv_invalid();
+    } else {
+      if (jv_array_length(jv_copy(p->stack[p->stackpos-1])) != 0) {
+        // this case hits on input like [1,2,3,]
+        return "Expected another array element";
+      }
+    }
+    jv_free(p->next);
+    p->next = p->stack[--p->stackpos];
+    break;
+
+  case '}':
+    if (p->stackpos == 0)
+      return "Unmatched '}'";
+    if (jv_is_valid(p->next)) {
+      if (jv_get_kind(p->stack[p->stackpos-1]) != JV_KIND_STRING)
+        return "Objects must consist of key:value pairs";
+      assert(p->stackpos > 1 && jv_get_kind(p->stack[p->stackpos-2]) == JV_KIND_OBJECT);
+      p->stack[p->stackpos-2] = jv_object_set(p->stack[p->stackpos-2], 
+                                              p->stack[p->stackpos-1], p->next);
+      p->stackpos--;
+      p->next = jv_invalid();
+    } else {
+      if (jv_get_kind(p->stack[p->stackpos-1]) != JV_KIND_OBJECT)
+        return "Unmatched '}'";
+      if (jv_object_length(jv_copy(p->stack[p->stackpos-1])) != 0)
+        return "Expected another key-value pair";
+    }
+    jv_free(p->next);
+    p->next = p->stack[--p->stackpos];
+    break;
+  }
+  return 0;
+}
+
+
+static void tokenadd(struct jv_parser* p, char c) {
+  assert(p->tokenpos <= p->tokenlen);
+  if (p->tokenpos == p->tokenlen) {
+    p->tokenlen = p->tokenlen*2 + 256;
+    p->tokenbuf = jv_mem_realloc(p->tokenbuf, p->tokenlen);
+  }
+  assert(p->tokenpos < p->tokenlen);
+  p->tokenbuf[p->tokenpos++] = c;
+}
+
+static int unhex4(char* hex) {
+  int r = 0;
+  for (int i=0; i<4; i++) {
+    char c = *hex++;
+    int n;
+    if ('0' <= c && c <= '9') n = c - '0';
+    else if ('a' <= c && c <= 'f') n = c - 'a' + 10;
+    else if ('A' <= c && c <= 'F') n = c - 'A' + 10;
+    else return -1;
+    r <<= 4;
+    r |= n;
+  }
+  return r;
+}
+
+static pfunc found_string(struct jv_parser* p) {
+  char* in = p->tokenbuf;
+  char* out = p->tokenbuf;
+  char* end = p->tokenbuf + p->tokenpos;
+  
+  while (in < end) {
+    char c = *in++;
+    if (c == '\\') {
+      if (in >= end)
+        return "Expected escape character at end of string";
+      c = *in++;
+      switch (c) {
+      case '\\':
+      case '"':
+      case '/': *out++ = c;    break;
+      case 'b': *out++ = '\b'; break;
+      case 'f': *out++ = '\f'; break;
+      case 't': *out++ = '\t'; break;
+      case 'n': *out++ = '\n'; break;
+      case 'r': *out++ = '\r'; break;
+
+      case 'u':
+        /* ahh, the complicated case */
+        if (in + 4 > end)
+          return "Invalid \\uXXXX escape";
+        int hexvalue = unhex4(in);
+        if (hexvalue < 0)
+          return "Invalid characters in \\uXXXX escape";
+        unsigned long codepoint = (unsigned long)hexvalue;
+        in += 4;
+        if (0xD800 <= codepoint && codepoint <= 0xDBFF) {
+          /* who thought UTF-16 surrogate pairs were a good idea? */
+          if (in + 6 > end || in[0] != '\\' || in[1] != 'u')
+            return "Invalid \\uXXXX\\uXXXX surrogate pair escape";
+          unsigned long surrogate = unhex4(in+2);
+          if (!(0xDC00 <= surrogate && surrogate <= 0xDFFF))
+            return "Invalid \\uXXXX\\uXXXX surrogate pair escape";
+          in += 6;
+          codepoint = 0x10000 + (((codepoint - 0xD800) << 10)
+                                 |(surrogate - 0xDC00));
+        }
+        // FIXME assert valid codepoint
+        out += jvp_utf8_encode(codepoint, out);
+        break;
+
+      default:
+        return "Invalid escape";
+      }
+    } else {
+      *out++ = c;
+    }
+  }
+  TRY(value(p, jv_string_sized(p->tokenbuf, out - p->tokenbuf)));
+  p->tokenpos = 0;
+  return 0;
+}
+
+static pfunc check_literal(struct jv_parser* p) {
+  if (p->tokenpos == 0) return 0;
+
+  const char* pattern = 0;
+  int plen;
+  jv v;
+  switch (p->tokenbuf[0]) {
+  case 't': pattern = "true"; plen = 4; v = jv_true(); break;
+  case 'f': pattern = "false"; plen = 5; v = jv_false(); break;
+  case 'n': pattern = "null"; plen = 4; v = jv_null(); break;
+  }
+  if (pattern) {
+    if (p->tokenpos != plen) return "Invalid literal";
+    for (int i=0; i<plen; i++) 
+      if (p->tokenbuf[i] != pattern[i])
+        return "Invalid literal";
+    TRY(value(p, v));
+  } else {
+    // FIXME: better parser
+    p->tokenbuf[p->tokenpos] = 0; // FIXME: invalid
+    char* end = 0;
+    double d = jvp_strtod(&p->dtoa, p->tokenbuf, &end);
+    if (end == 0 || *end != 0)
+      return "Invalid numeric literal";
+    TRY(value(p, jv_number(d)));
+  }
+  p->tokenpos = 0;
+  return 0;
+}
+
+typedef enum {
+  LITERAL,
+  WHITESPACE,
+  STRUCTURE,
+  QUOTE,
+  INVALID
+} chclass;
+
+static chclass classify(char c) {
+  switch (c) {
+  case ' ':
+  case '\t':
+  case '\r':
+  case '\n':
+    return WHITESPACE;
+  case '"':
+    return QUOTE;
+  case '[':
+  case ',':
+  case ']':
+  case '{':
+  case ':':
+  case '}':
+    return STRUCTURE;
+  default:
+    return LITERAL;
+  }
+}
+
+
+static presult OK = "output produced";
+
+static int check_done(struct jv_parser* p, jv* out) {
+  if (p->stackpos == 0 && jv_is_valid(p->next)) {
+    *out = p->next;
+    p->next = jv_invalid();
+    return 1;
+  } else {
+    return 0;
+  }
+}
+
+static pfunc scan(struct jv_parser* p, char ch, jv* out) {
+  presult answer = 0;
+  if (p->st == JV_PARSER_NORMAL) {
+    chclass cls = classify(ch);
+    if (cls != LITERAL) {
+      TRY(check_literal(p));
+      if (check_done(p, out)) answer = OK;
+    }
+    switch (cls) {
+    case LITERAL:
+      tokenadd(p, ch);
+      break;
+    case WHITESPACE:
+      break;
+    case QUOTE:
+      p->st = JV_PARSER_STRING;
+      break;
+    case STRUCTURE:
+      TRY(token(p, ch));
+      break;
+    case INVALID:
+      return "Invalid character";
+    }
+    if (check_done(p, out)) answer = OK;
+  } else {
+    if (ch == '"' && p->st == JV_PARSER_STRING) {
+      TRY(found_string(p));
+      p->st = JV_PARSER_NORMAL;
+      if (check_done(p, out)) answer = OK;
+    } else {
+      tokenadd(p, ch);
+      if (ch == '\\' && p->st == JV_PARSER_STRING) {
+        p->st = JV_PARSER_STRING_ESCAPE;
+      } else {
+        p->st = JV_PARSER_STRING;
+      }
+    }
+  }
+  return answer;
+}
+
+static unsigned char UTF8_BOM[] = {0xEF,0xBB,0xBF};
+
+void jv_parser_set_buf(struct jv_parser* p, const char* buf, int length, int is_partial) {
+  assert((p->curr_buf == 0 || p->curr_buf_pos == p->curr_buf_length)
+         && "previous buffer not exhausted");
+  while (p->bom_strip_position < sizeof(UTF8_BOM)) {
+    if ((unsigned char)*buf == UTF8_BOM[p->bom_strip_position]) {
+      // matched a BOM character
+      buf++;
+      length--;
+      p->bom_strip_position++;
+    } else {
+      if (p->bom_strip_position == 0) {
+        // no BOM in this document
+        p->bom_strip_position = sizeof(UTF8_BOM);
+      } else {
+        // malformed BOM (prefix present, rest missing)
+        p->bom_strip_position = 0xff;
+      }
+    }
+  }
+  p->curr_buf = buf;
+  p->curr_buf_length = length;
+  p->curr_buf_pos = 0;
+  p->curr_buf_is_partial = is_partial;
+}
+
+jv jv_parser_next(struct jv_parser* p) {
+  assert(p->curr_buf && "a buffer must be provided");
+  if (p->bom_strip_position == 0xff) return jv_invalid_with_msg(jv_string("Malformed BOM"));
+  jv value;
+  presult msg = 0;
+  while (!msg && p->curr_buf_pos < p->curr_buf_length) {
+    char ch = p->curr_buf[p->curr_buf_pos++];
+    msg = scan(p, ch, &value);
+  }
+  if (msg == OK) {
+    return value;
+  } else if (msg) {
+    return jv_invalid_with_msg(jv_string(msg));
+  } else if (p->curr_buf_is_partial) {
+    assert(p->curr_buf_pos == p->curr_buf_length);
+    // need another buffer
+    return jv_invalid();
+  } else {
+    assert(p->curr_buf_pos == p->curr_buf_length);
+    // at EOF
+    if (p->st != JV_PARSER_NORMAL) 
+      return jv_invalid_with_msg(jv_string("Unfinished string"));
+    if ((msg = check_literal(p)))
+      return jv_invalid_with_msg(jv_string(msg));
+    if (p->stackpos != 0)
+      return jv_invalid_with_msg(jv_string("Unfinished JSON term"));
+    // p->next is either invalid (nothing here but no syntax error)
+    // or valid (this is the value). either way it's the thing to return
+    value = p->next;
+    p->next = jv_invalid();
+    return value;
+  }
+}
+
+jv jv_parse_sized(const char* string, int length) {
+  struct jv_parser parser;
+  jv_parser_init(&parser);
+  jv_parser_set_buf(&parser, string, length, 0);
+  jv value = jv_parser_next(&parser);
+  if (jv_is_valid(value)) {
+    jv next = jv_parser_next(&parser);
+    if (jv_is_valid(next)) {
+      // multiple JSON values, we only wanted one
+      jv_free(value);
+      jv_free(next);
+      value = jv_invalid_with_msg(jv_string("Unexpected extra JSON values"));
+    } else if (jv_invalid_has_msg(jv_copy(next))) {
+      // parser error after the first JSON value
+      jv_free(value);
+      value = next;
+    } else {
+      // a single valid JSON value
+      jv_free(next);
+    }
+  } else if (jv_invalid_has_msg(jv_copy(value))) {
+    // parse error, we'll return it
+  } else {
+    // no value at all
+    jv_free(value);
+    value = jv_invalid_with_msg(jv_string("Expected JSON value"));
+  }
+  jv_parser_free(&parser);
+
+  if (!jv_is_valid(value) && jv_invalid_has_msg(jv_copy(value))) {
+    jv msg = jv_invalid_get_msg(value);
+    value = jv_invalid_with_msg(jv_string_fmt("%s (while parsing '%s')",
+                                              jv_string_value(msg),
+                                              string));
+    jv_free(msg);
+  }
+  return value;
+}
+
+jv jv_parse(const char* string) {
+  return jv_parse_sized(string, strlen(string));
+}

jv_parse.h

@@ -0,0 +1,35 @@
+#ifndef JV_PARSE_H
+#define JV_PARSE_H
+#include "jv_dtoa.h"
+struct jv_parser {
+  const char* curr_buf;
+  int curr_buf_length;
+  int curr_buf_pos;
+  int curr_buf_is_partial;
+  unsigned bom_strip_position;
+
+  jv* stack;
+  int stackpos;
+  int stacklen;
+  jv next;
+  
+  char* tokenbuf;
+  int tokenpos;
+  int tokenlen;
+  
+  struct dtoa_context dtoa;
+
+  enum {
+    JV_PARSER_NORMAL,
+    JV_PARSER_STRING,
+    JV_PARSER_STRING_ESCAPE
+  } st;
+};
+
+void jv_parser_init(struct jv_parser* p);
+void jv_parser_free(struct jv_parser* p);
+
+void jv_parser_set_buf(struct jv_parser* p, const char* buf, int length, int is_partial);
+
+jv jv_parser_next(struct jv_parser* p);
+#endif

jv_print.c

@@ -0,0 +1,249 @@
+#include "jv.h"
+#include <stdio.h>
+#include <float.h>
+#include <string.h>
+
+#include "jv_dtoa.h"
+#include "jv_unicode.h"
+
+#define ESC "\033"
+#define COL(c) (ESC "[" c "m")
+#define COLRESET (ESC "[0m")
+
+// Colour table. See http://en.wikipedia.org/wiki/ANSI_escape_code#Colors
+// for how to choose these.
+static jv_kind colour_kinds[] = 
+  {JV_KIND_NULL,   JV_KIND_FALSE, JV_KIND_TRUE, JV_KIND_NUMBER,
+   JV_KIND_STRING, JV_KIND_ARRAY, JV_KIND_OBJECT};
+static const char* colours[] =
+  {COL("30;1"),    COL("0"),      COL("0"),     COL("0"),
+   COL("32"),      COL("37"),     COL("37")};
+#define FIELD_COLOUR COL("34;1")
+
+static void put_buf(const char* s, int len, FILE* fout, jv* strout) {
+  if (strout) {
+    *strout = jv_string_append_buf(*strout, s, len);
+  } else {
+    fwrite(s, 1, len, fout);
+  }
+}
+
+static void put_char(char c, FILE* fout, jv* strout) {
+  put_buf(&c, 1, fout, strout);
+}
+
+static void put_str(const char* s, FILE* fout, jv* strout) {
+  put_buf(s, strlen(s), fout, strout);
+}
+
+static void put_space(int n, FILE* fout, jv* strout) {
+  while (n--) {
+    put_char(' ', fout, strout);
+  }
+}
+
+static void jvp_dump_string(jv str, int ascii_only, FILE* F, jv* S) {
+  assert(jv_get_kind(str) == JV_KIND_STRING);
+  const char* i = jv_string_value(str);
+  const char* end = i + jv_string_length(jv_copy(str));
+  const char* cstart;
+  int c = 0;
+  char buf[32];
+  put_char('"', F, S);
+  while ((i = jvp_utf8_next((cstart = i), end, &c))) {
+    assert(c != -1);
+    int unicode_escape = 0;
+    if (0x20 <= c && c <= 0x7E) {
+      // printable ASCII
+      if (c == '"' || c == '\\') {
+        put_char('\\', F, S);
+      }
+      put_char(c, F, S);
+    } else if (c < 0x20 || c == 0x7F) {
+      // ASCII control character
+      switch (c) {
+      case '\b':
+        put_char('\\', F, S);
+        put_char('b', F, S);
+        break;
+      case '\t':
+        put_char('\\', F, S);
+        put_char('t', F, S);
+        break;
+      case '\r':
+        put_char('\\', F, S);
+        put_char('r', F, S);
+        break;
+      case '\n':
+        put_char('\\', F, S);
+        put_char('n', F, S);
+        break;
+      case '\f':
+        put_char('\\', F, S);
+        put_char('f', F, S);
+        break;
+      default:
+        unicode_escape = 1;
+        break;
+      }
+    } else {
+      if (ascii_only) {
+        unicode_escape = 1;
+      } else {
+        put_buf(cstart, i - cstart, F, S);
+      }
+    }
+    if (unicode_escape) {
+      if (c <= 0xffff) {
+        sprintf(buf, "\\u%04x", c);
+      } else {
+        c -= 0x10000;
+        sprintf(buf, "\\u%04x\\u%04x", 
+                0xD800 | ((c & 0xffc00) >> 10),
+                0xDC00 | (c & 0x003ff));
+      }
+      put_str(buf, F, S);
+    }
+  }
+  assert(c != -1);
+  put_char('"', F, S);
+}
+
+enum { INDENT = 2 };
+
+static void jv_dump_term(struct dtoa_context* C, jv x, int flags, int indent, FILE* F, jv* S) {
+  char buf[JVP_DTOA_FMT_MAX_LEN];
+  const char* colour = 0;
+  if (flags & JV_PRINT_COLOUR) {
+    for (unsigned i=0; i<sizeof(colour_kinds)/sizeof(colour_kinds[0]); i++) {
+      if (jv_get_kind(x) == colour_kinds[i]) {
+        colour = colours[i];
+        put_str(colour, F, S);
+        break;
+      }
+    }
+  }
+  switch (jv_get_kind(x)) {
+  case JV_KIND_INVALID:
+    assert(0 && "Invalid value");
+    break;
+  case JV_KIND_NULL:
+    put_str("null", F, S);
+    break;
+  case JV_KIND_FALSE:
+    put_str("false", F, S);
+    break;
+  case JV_KIND_TRUE:
+    put_str("true", F, S);
+    break;
+  case JV_KIND_NUMBER: {
+    double d = jv_number_value(x);
+    if (d != d) {
+      // JSON doesn't have NaN, so we'll render it as "null"
+      put_str("null", F, S);
+    } else {
+      // Normalise infinities to something we can print in valid JSON
+      if (d > DBL_MAX) d = DBL_MAX;
+      if (d < -DBL_MAX) d = -DBL_MAX;
+      put_str(jvp_dtoa_fmt(C, buf, d), F, S);
+    }
+    break;
+  }
+  case JV_KIND_STRING:
+    jvp_dump_string(x, flags & JV_PRINT_ASCII, F, S);
+    break;
+  case JV_KIND_ARRAY: {
+    if (jv_array_length(jv_copy(x)) == 0) {
+      put_str("[]", F, S);
+      break;
+    }
+    put_str("[", F, S);
+    if (flags & JV_PRINT_PRETTY) {
+      put_char('\n', F, S);
+      put_space(indent+INDENT, F, S);
+    }
+    for (int i=0; i<jv_array_length(jv_copy(x)); i++) {
+      if (i!=0) {
+        if (flags & JV_PRINT_PRETTY) {
+          put_str(",\n", F, S);
+          put_space(indent+INDENT, F, S);
+        } else {
+          put_str(",", F, S);
+        }
+      }
+      jv_dump_term(C, jv_array_get(jv_copy(x), i), flags, indent + INDENT, F, S);
+      if (colour) put_str(colour, F, S);
+    }
+    if (flags & JV_PRINT_PRETTY) {
+      put_char('\n', F, S);
+      put_space(indent, F, S);
+    }
+    if (colour) put_str(colour, F, S);
+    put_char(']', F, S);
+    break;
+  }
+  case JV_KIND_OBJECT: {
+    if (jv_object_length(jv_copy(x)) == 0) {
+      put_str("{}", F, S);
+      break;
+    }
+    put_char('{', F, S);
+    if (flags & JV_PRINT_PRETTY) {
+      put_char('\n', F, S);
+      put_space(indent+INDENT, F, S);
+    }
+    int first = 1;
+    jv_object_foreach(i, x) {
+      if (!first) {
+        if (flags & JV_PRINT_PRETTY){
+          put_str(",\n", F, S);
+          put_space(indent+INDENT, F, S);
+        } else {
+          put_str(",", F, S);
+        }
+      }
+      if (colour) put_str(COLRESET, F, S);
+
+      first = 0;
+      jv key = jv_object_iter_key(x, i);
+      if (colour) put_str(FIELD_COLOUR, F, S);
+      jvp_dump_string(key, flags & JV_PRINT_ASCII, F, S);
+      jv_free(key);
+      if (colour) put_str(COLRESET, F, S);
+
+      if (colour) put_str(colour, F, S);
+      put_str((flags & JV_PRINT_PRETTY) ? ": " : ":", F, S);
+      if (colour) put_str(COLRESET, F, S);
+      
+      jv_dump_term(C, jv_object_iter_value(x, i), flags, indent + INDENT, F, S);
+      if (colour) put_str(colour, F, S);
+    }
+    if (flags & JV_PRINT_PRETTY) {
+      put_char('\n', F, S);
+      put_space(indent, F, S);
+    }
+    if (colour) put_str(colour, F, S);
+    put_char('}', F, S);
+  }
+  }
+  jv_free(x);
+  if (colour) {
+    put_str(COLRESET, F, S);
+  }
+}
+
+void jv_dump(jv x, int flags) {
+  struct dtoa_context C;
+  jvp_dtoa_context_init(&C);
+  jv_dump_term(&C, x, flags, 0, stdout, 0);
+  jvp_dtoa_context_free(&C);
+}
+
+jv jv_dump_string(jv x, int flags) {
+  struct dtoa_context C;
+  jvp_dtoa_context_init(&C);
+  jv s = jv_string("");
+  jv_dump_term(&C, x, flags, 0, 0, &s);
+  jvp_dtoa_context_free(&C);
+  return s;
+}

jv_unicode.c

@@ -0,0 +1,64 @@
+#include <stdio.h>
+#include <assert.h>
+#include "jv_unicode.h"
+#include "jv_utf8_tables.gen.h"
+
+const char* jvp_utf8_next(const char* in, const char* end, int* codepoint) {
+  if (in == end) {
+    codepoint = 0;
+    return 0;
+  }
+  unsigned char first = (unsigned char)in[0];
+  int length = utf8_coding_length[first];
+  if (length == 0 || length == UTF8_CONTINUATION_BYTE || in + length > end) {
+    *codepoint = -1;
+    return 0;
+  }
+  *codepoint = ((unsigned)in[0]) & utf8_coding_bits[first];
+  for (int i=1; i<length; i++) {
+    int ch = (unsigned char)in[i];
+    if (utf8_coding_length[(unsigned char)in[i]] != UTF8_CONTINUATION_BYTE){
+      *codepoint = -1;
+      return 0;
+    }
+    *codepoint = (*codepoint << 6) | (ch & 0x3f);
+  }
+  return in + length;
+}
+
+int jvp_utf8_verify(const char* in, const char* end) {
+  int codepoint = 0;
+  while ((in = jvp_utf8_next(in, end, &codepoint))) {
+    if (codepoint == -1) return 0;
+  }
+  return codepoint != -1;
+}
+
+int jvp_utf8_encode_length(int codepoint) {
+  if (codepoint <= 0x7F) return 1;
+  else if (codepoint <= 0x7FF) return 2;
+  else if (codepoint <= 0xFFFF) return 3;
+  else return 4;
+}
+
+int jvp_utf8_encode(int codepoint, char* out) {
+  assert(codepoint >= 0 && codepoint <= 0x10FFFF);
+  char* start = out;
+  if (codepoint <= 0x7F) {
+    *out++ = codepoint;
+  } else if (codepoint <= 0x7FF) {
+    *out++ = 0xC0 + ((codepoint & 0x7C0) >> 6);
+    *out++ = 0x80 + ((codepoint & 0x03F));
+  } else if(codepoint <= 0xFFFF) {
+    *out++ = 0xE0 + ((codepoint & 0xF000) >> 12);
+    *out++ = 0x80 + ((codepoint & 0x0FC0) >> 6);
+    *out++ = 0x80 + ((codepoint & 0x003F));
+  } else {
+    *out++ = 0xF0 + ((codepoint & 0x1C0000) >> 18);
+    *out++ = 0x80 + ((codepoint & 0x03F000) >> 12);
+    *out++ = 0x80 + ((codepoint & 0x000FC0) >> 6);
+    *out++ = 0x80 + ((codepoint & 0x00003F));
+  }
+  assert(out - start == jvp_utf8_encode_length(codepoint));
+  return out - start;
+}

jv_unicode.h

@@ -0,0 +1,11 @@
+#ifndef JV_UNICODE_H
+#define JV_UNICODE_H
+
+const char* jvp_utf8_next(const char* in, const char* end, int* codepoint);
+
+
+int jvp_utf8_decode_length(char startchar);
+
+int jvp_utf8_encode_length(int codepoint);
+int jvp_utf8_encode(int codepoint, char* out);
+#endif

lexer.l

@@ -0,0 +1,167 @@
+%{
+#include "jv_alloc.h"
+#include "compile.h"
+
+struct lexer_param;
+
+#include "parser.gen.h"  /* Generated by bison. */
+
+#define YY_USER_ACTION                           \
+  do {                                           \
+    yylloc->start = yyget_extra(yyscanner);      \
+    yylloc->end = yylloc->start + yyleng;        \
+    yyset_extra(yylloc->end, yyscanner);         \
+  } while (0);
+
+%}
+
+%s IN_PAREN
+%s IN_BRACKET
+%s IN_BRACE
+%s IN_QQINTERP
+%x IN_QQSTRING
+%{
+  static int enter(int opening, int state, yyscan_t yyscanner);
+  static int try_exit(int closing, int state, yyscan_t yyscanner);
+%}
+
+%option noyywrap nounput noinput nodefault
+%option noyyalloc noyyrealloc noyyfree
+%option reentrant
+%option extra-type="int"
+%option bison-bridge bison-locations
+%option prefix="jq_yy"
+%option stack
+
+
+%%
+
+"#"[^\r\n]* { /* comments */ }
+
+"!=" { return NEQ; }
+"==" { return EQ; }
+"as" { return AS; }
+"def" { return DEF; }
+"if" { return IF; }
+"then" { return THEN; }
+"else" { return ELSE; }
+"elif" { return ELSE_IF; }
+"and" { return AND; }
+"or" { return OR; }
+"end" { return END; }
+"//" { return DEFINEDOR; }
+"|=" { return SETPIPE; }
+"+=" { return SETPLUS; }
+"-=" { return SETMINUS; }
+"*=" { return SETMULT; }
+"/=" { return SETDIV; }
+"//=" { return SETDEFINEDOR; }
+"<=" { return LESSEQ; }
+">=" { return GREATEREQ; }
+"."|"="|";"|","|":"|"|"|"+"|"-"|"*"|"/"|"\$"|"<"|">" { return yytext[0];}
+
+"["|"{"|"(" {
+  return enter(yytext[0], YY_START, yyscanner);
+}
+
+"]"|"}"|")" {
+  return try_exit(yytext[0], YY_START, yyscanner);
+}
+
+
+-?[0-9.]+([eE][+-]?[0-9]+)? { 
+   yylval->literal = jv_parse_sized(yytext, yyleng); return LITERAL; 
+}
+
+"\"" {
+  yy_push_state(IN_QQSTRING, yyscanner);
+  return QQSTRING_START;
+}
+
+<IN_QQSTRING>{
+  "\\(" {
+    return enter(QQSTRING_INTERP_START, YY_START, yyscanner);
+  }
+  "\"" {
+    yy_pop_state(yyscanner);
+    return QQSTRING_END;
+  }
+  (\\[^u(]|\\u[a-zA-Z0-9]{0,4})+ {
+    /* pass escapes to the json parser */
+    jv escapes = jv_string_fmt("\"%.*s\"", yyleng, yytext);
+    yylval->literal = jv_parse_sized(jv_string_value(escapes), jv_string_length(jv_copy(escapes)));
+    jv_free(escapes);
+    return QQSTRING_TEXT;
+  }
+  [^\\\"]+ {
+    yylval->literal = jv_string_sized(yytext, yyleng);
+    return QQSTRING_TEXT;
+  }
+  . {
+    return INVALID_CHARACTER;
+  }
+}
+  
+
+[a-zA-Z_][a-zA-Z_0-9]*  { yylval->literal = jv_string(yytext); return IDENT;}
+
+[ \n\t]+  {}
+
+. { return INVALID_CHARACTER; }
+
+%%
+/* perhaps these should be calls... */
+/*
+"true" { return TRUE; }
+"false" { return FALSE; }
+"null" { return NULL; }
+*/
+static int try_exit(int c, int state, yyscan_t yyscanner) {
+  char match = 0;
+  int ret;
+  switch (state) {
+  case IN_PAREN: match = ret = ')'; break;
+  case IN_BRACKET: match = ret = ']'; break;
+  case IN_BRACE: match = ret = '}'; break;
+
+  case IN_QQINTERP:
+    match = ')'; 
+    ret = QQSTRING_INTERP_END;
+    break;
+
+  default:
+    // may not be the best error to give
+    return INVALID_CHARACTER;
+  }
+  assert(match);
+  if (match == c) {
+    yy_pop_state(yyscanner);
+    return ret;
+  } else {
+    // FIXME: should we pop? Give a better error at least
+    return INVALID_CHARACTER;
+  }
+}
+
+static int enter(int c, int currstate, yyscan_t yyscanner) {
+  int state = 0;
+  switch (c) {
+  case '(': state = IN_PAREN; break;
+  case '[': state = IN_BRACKET; break;
+  case '{': state = IN_BRACE; break;
+  case QQSTRING_INTERP_START: state = IN_QQINTERP; break;
+  }
+  assert(state);
+  yy_push_state(state, yyscanner);
+  return c;
+}
+
+void* yyalloc(size_t sz, void* extra) {
+  return jv_mem_alloc(sz);
+}
+void* yyrealloc(void* p, size_t sz, void* extra) {
+  return jv_mem_realloc(p, sz);
+}
+void yyfree(void* p, void* extra) {
+  jv_mem_free(p);
+}

locfile.h

@@ -0,0 +1,79 @@
+#ifndef _LOCFILE_H
+#define _LOCFILE_H
+#include <stdlib.h>
+#include <stdio.h>
+#include <assert.h>
+#include <stdarg.h>
+#include "jv_alloc.h"
+typedef struct {
+  int start, end;
+} location;
+
+static const location UNKNOWN_LOCATION = {-1, -1};
+
+struct locfile {
+  const char* data;
+  int length;
+  int* linemap;
+  int nlines;
+};
+
+static void locfile_init(struct locfile* l, const char* data, int length) {
+  l->data = data;
+  l->length = length;
+  l->nlines = 1;
+  for (int i=0; i<length; i++) {
+    if (data[i] == '\n') l->nlines++;
+  }
+  l->linemap = jv_mem_alloc(sizeof(int) * (l->nlines + 1));
+  l->linemap[0] = 0;
+  int line = 1;
+  for (int i=0; i<length; i++) {
+    if (data[i] == '\n') {
+      l->linemap[line] = i;
+      line++;
+    }
+  }
+  l->linemap[l->nlines] = length;
+}
+
+static void locfile_free(struct locfile* l) {
+  jv_mem_free(l->linemap);
+}
+
+static int locfile_get_line(struct locfile* l, int pos) {
+  assert(pos < l->length);
+  int line = 0;
+  while (l->linemap[line+1] < pos) line++;
+  assert(line < l->nlines);
+  return line;
+}
+
+static int locfile_line_length(struct locfile* l, int line) {
+  assert(line < l->nlines);
+  return l->linemap[line+1] - l->linemap[line];
+}
+
+static void locfile_locate(struct locfile* l, location loc, const char* fmt, ...) {
+  va_list fmtargs;
+  va_start(fmtargs, fmt);
+  vfprintf(stderr, fmt, fmtargs);
+  va_end(fmtargs);
+  fprintf(stderr, "\n");
+  if (loc.start == -1) {
+    fprintf(stderr, "<unknown location>\n");
+    return;
+  }
+  int startline = locfile_get_line(l, loc.start);
+  int offset = l->linemap[startline];
+  fprintf(stderr, "%.*s\n", locfile_line_length(l, startline), l->data + offset);
+  fprintf(stderr, "%*s", loc.start - offset, "");
+  for (int i = loc.start; 
+       i < loc.end && i < offset + locfile_line_length(l, startline);
+       i++){
+    fprintf(stderr, "^");
+  }
+  fprintf(stderr, "\n");
+}
+
+#endif

main.c

@@ -0,0 +1,269 @@
+#define _POSIX_SOURCE
+#include <stdio.h>
+#include <errno.h>
+#include <string.h>
+#include <ctype.h>
+#include <unistd.h>
+#include "compile.h"
+#include "builtin.h"
+#include "jv.h"
+#include "jv_parse.h"
+#include "locfile.h"
+#include "parser.h"
+#include "execute.h"
+#include "jv_alloc.h"
+#include "version.gen.h"
+
+static const char* progname;
+
+static void usage() {
+  fprintf(stderr, "\njq - commandline JSON processor [version %s]\n", JQ_VERSION);
+  fprintf(stderr, "Usage: %s [options] <jq filter> [file...]\n\n", progname);
+  fprintf(stderr, "For a description of the command line options and\n");
+  fprintf(stderr, "how to write jq filters (and why you might want to)\n");
+  fprintf(stderr, "see the jq manpage, or the online documentation at\n");
+  fprintf(stderr, "http://stedolan.github.com/jq\n\n");
+  exit(1);
+}
+
+static void die() {
+  fprintf(stderr, "Use %s --help for help with command-line options,\n", progname);
+  fprintf(stderr, "or see the jq documentation at http://stedolan.github.com/jq\n");
+  exit(1);
+}
+
+
+
+
+static int isoptish(const char* text) {
+  return text[0] == '-' && (text[1] == '-' || isalpha(text[1]));
+}
+
+static int isoption(const char* text, char shortopt, const char* longopt) {
+  if (text[0] != '-') return 0;
+  if (strlen(text) == 2 && text[1] == shortopt) return 1;
+  if (text[1] == '-' && !strcmp(text+2, longopt)) return 1;
+  return 0;
+}
+
+enum {
+  SLURP = 1,
+  RAW_INPUT = 2,
+  PROVIDE_NULL = 4,
+
+  RAW_OUTPUT = 8,
+  COMPACT_OUTPUT = 16,
+  ASCII_OUTPUT = 32,
+  COLOUR_OUTPUT = 64,
+  NO_COLOUR_OUTPUT = 128,
+
+  FROM_FILE = 256,
+};
+static int options = 0;
+static struct bytecode* bc;
+
+static void process(jv value) {
+  jq_init(bc, value);
+  jv result;
+  while (jv_is_valid(result = jq_next())) {
+    if ((options & RAW_OUTPUT) && jv_get_kind(result) == JV_KIND_STRING) {
+      fwrite(jv_string_value(result), 1, jv_string_length(jv_copy(result)), stdout);
+    } else {
+      int dumpopts;
+#ifdef JQ_DEFAULT_ENABLE_COLOR
+      dumpopts = JQ_DEFAULT_ENABLE_COLOR ? JV_PRINT_COLOUR : 0;
+#else
+      dumpopts = isatty(fileno(stdout)) ? JV_PRINT_COLOUR : 0;
+#endif
+      if (!(options & COMPACT_OUTPUT)) dumpopts |= JV_PRINT_PRETTY;
+      if (options & ASCII_OUTPUT) dumpopts |= JV_PRINT_ASCII;
+      if (options & COLOUR_OUTPUT) dumpopts |= JV_PRINT_COLOUR;
+      if (options & NO_COLOUR_OUTPUT) dumpopts &= ~JV_PRINT_COLOUR;
+      jv_dump(result, dumpopts);
+    }
+    printf("\n");
+  }
+  jv_free(result);
+  jq_teardown();
+}
+
+static jv slurp_file(const char* filename) {
+  FILE* file = fopen(filename, "r");
+  if (!file) {
+    return jv_invalid_with_msg(jv_string_fmt("Could not open %s: %s",
+                                             filename,
+                                             strerror(errno)));
+  }
+  jv data = jv_string("");
+  while (!feof(file) && !ferror(file)) {
+    char buf[4096];
+    size_t n = fread(buf, 1, sizeof(buf), file);
+    data = jv_string_concat(data, jv_string_sized(buf, (int)n));
+  }
+  int badread = ferror(file);
+  fclose(file);
+  if (badread) {
+    jv_free(data);
+    return jv_invalid_with_msg(jv_string_fmt("Error reading from %s",
+                                             filename));
+  }
+  return data;
+}
+
+FILE* current_input;
+const char** input_filenames;
+int ninput_files;
+int next_input_idx;
+static int read_more(char* buf, size_t size) {
+  while (!current_input || feof(current_input)) {
+    if (current_input) {
+      fclose(current_input);
+      current_input = 0;
+    }
+    if (next_input_idx == ninput_files) {
+      return 0;
+    }
+    if (!strcmp(input_filenames[next_input_idx], "-")) {
+      current_input = stdin;
+    } else {
+      current_input = fopen(input_filenames[next_input_idx], "r");
+    }
+    if (!current_input) {
+      fprintf(stderr, "%s: %s: %s\n", progname, input_filenames[next_input_idx], strerror(errno));
+    }
+    next_input_idx++;
+  }
+
+  if (!fgets(buf, sizeof(buf), current_input)) buf[0] = 0;
+  return 1;
+}
+
+int main(int argc, char* argv[]) {
+  if (argc) progname = argv[0];
+
+  const char* program = 0;
+  input_filenames = jv_mem_alloc(sizeof(const char*) * argc);
+  ninput_files = 0;
+  int further_args_are_files = 0;
+  for (int i=1; i<argc; i++) {
+    if (further_args_are_files) {
+      input_filenames[ninput_files++] = argv[i];
+    } else if (!strcmp(argv[i], "--")) {
+      if (!program) usage();
+      further_args_are_files = 1;
+    } else if (!isoptish(argv[i])) {
+      if (program) {
+        input_filenames[ninput_files++] = argv[i];
+      } else {
+        program = argv[i];
+      }
+    } else if (isoption(argv[i], 's', "slurp")) {
+      options |= SLURP;
+    } else if (isoption(argv[i], 'r', "raw-output")) {
+      options |= RAW_OUTPUT;
+    } else if (isoption(argv[i], 'c', "compact-output")) {
+      options |= COMPACT_OUTPUT;
+    } else if (isoption(argv[i], 'C', "color-output")) {
+      options |= COLOUR_OUTPUT;
+    } else if (isoption(argv[i], 'M', "monochrome-output")) {
+      options |= NO_COLOUR_OUTPUT;
+    } else if (isoption(argv[i], 'a', "ascii-output")) {
+      options |= ASCII_OUTPUT;
+    } else if (isoption(argv[i], 'R', "raw-input")) {
+      options |= RAW_INPUT;
+    } else if (isoption(argv[i], 'n', "null-input")) {
+      options |= PROVIDE_NULL;
+    } else if (isoption(argv[i], 'f', "from-file")) {
+      options |= FROM_FILE;
+    } else if (isoption(argv[i], 'h', "help")) {
+      usage();
+    } else if (isoption(argv[i], 'V', "version")) {
+      fprintf(stderr, "jq version %s\n", JQ_VERSION);
+      return 0;
+    } else {
+      fprintf(stderr, "%s: Unknown option %s\n", progname, argv[i]);
+      die();
+    }
+  }
+  if (!program) usage();
+  if (ninput_files == 0) current_input = stdin;
+
+  if ((options & PROVIDE_NULL) && (options & (RAW_INPUT | SLURP))) {
+    fprintf(stderr, "%s: --null-input cannot be used with --raw-input or --slurp\n", progname);
+    die();
+  }
+  
+  if (options & FROM_FILE) {
+    jv data = slurp_file(program);
+    if (!jv_is_valid(data)) {
+      data = jv_invalid_get_msg(data);
+      fprintf(stderr, "%s: %s\n", progname, jv_string_value(data));
+      jv_free(data);
+      return 1;
+    }
+    bc = jq_compile(jv_string_value(data));
+    jv_free(data);
+  } else {
+    bc = jq_compile(program);
+  }
+  if (!bc) return 1;
+
+#if JQ_DEBUG
+  dump_disassembly(0, bc);
+  printf("\n");
+#endif
+
+  if (options & PROVIDE_NULL) {
+    process(jv_null());
+  } else {
+    jv slurped;
+    if (options & SLURP) {
+      if (options & RAW_INPUT) {
+        slurped = jv_string("");
+      } else {
+        slurped = jv_array();
+      }
+    }
+    struct jv_parser parser;
+    jv_parser_init(&parser);
+    char buf[4096];
+    while (read_more(buf, sizeof(buf))) {
+      if (options & RAW_INPUT) {
+        int len = strlen(buf);
+        if (len > 0) {
+          if (options & SLURP) {
+            slurped = jv_string_concat(slurped, jv_string(buf));
+          } else {
+            if (buf[len-1] == '\n') buf[len-1] = 0;
+            process(jv_string(buf));
+          }
+        }
+      } else {
+        jv_parser_set_buf(&parser, buf, strlen(buf), !feof(stdin));
+        jv value;
+        while (jv_is_valid((value = jv_parser_next(&parser)))) {
+          if (options & SLURP) {
+            slurped = jv_array_append(slurped, value);
+          } else {
+            process(value);
+          }
+        }
+        if (jv_invalid_has_msg(jv_copy(value))) {
+          jv msg = jv_invalid_get_msg(value);
+          fprintf(stderr, "parse error: %s\n", jv_string_value(msg));
+          jv_free(msg);
+          break;
+        } else {
+          jv_free(value);
+        }
+      }
+    }
+    jv_parser_free(&parser);
+    if (options & SLURP) {
+      process(slurped);
+    }
+  }
+  jv_mem_free(input_filenames);
+  bytecode_free(bc);
+  return 0;
+}

opcode.c

@@ -0,0 +1,31 @@
+#include "opcode.h"
+
+// flags, length
+#define NONE 0, 1
+#define CONSTANT OP_HAS_CONSTANT, 2
+#define VARIABLE (OP_HAS_VARIABLE | OP_HAS_BINDING), 3
+#define BRANCH OP_HAS_BRANCH, 2
+#define CFUNC (OP_HAS_CFUNC | OP_HAS_BINDING), 3
+#define UFUNC (OP_HAS_UFUNC | OP_HAS_BINDING | OP_IS_CALL_PSEUDO), 4
+#define DEFINITION (OP_IS_CALL_PSEUDO | OP_HAS_BINDING), 0
+#define CLOSURE_REF_IMM (OP_IS_CALL_PSEUDO | OP_HAS_BINDING), 2
+
+#define OP(name, imm, in, out) \
+  {name, #name, imm, in, out},
+
+static const struct opcode_description opcode_descriptions[] = {
+#include "opcode_list.h"
+};
+
+static const struct opcode_description invalid_opcode_description = {
+  -1, "#INVALID", 0, 0, 0, 0
+};
+
+
+const struct opcode_description* opcode_describe(opcode op) {
+  if ((int)op >= 0 && (int)op < NUM_OPCODES) {
+    return &opcode_descriptions[op];
+  } else {
+    return &invalid_opcode_description;
+  }
+}

opcode.h

@@ -0,0 +1,41 @@
+#ifndef OPCODE_H
+#define OPCODE_H
+#include <assert.h>
+
+typedef enum {
+#define OP(name, imm, in, out) name,
+#include "opcode_list.h"
+#undef OP
+} opcode;
+
+enum {
+  NUM_OPCODES = 
+#define OP(name, imm, in, out) +1
+#include "opcode_list.h"
+#undef OP
+};
+
+enum {
+  OP_HAS_CONSTANT = 2,
+  OP_HAS_VARIABLE = 4,
+  OP_HAS_BRANCH = 8,
+  OP_HAS_CFUNC = 32,
+  OP_HAS_UFUNC = 64,
+  OP_IS_CALL_PSEUDO = 128,
+  OP_HAS_BINDING = 1024,
+};
+struct opcode_description {
+  opcode op;
+  const char* name;
+
+  int flags;
+
+  // length in 16-bit units
+  int length;
+
+  int stack_in, stack_out;
+};
+
+const struct opcode_description* opcode_describe(opcode op);
+
+#endif

opcode_list.h

@@ -0,0 +1,28 @@
+OP(LOADK, CONSTANT, 1, 1)
+OP(DUP,   NONE,     1, 2)
+OP(DUP2,  NONE,     2, 3)
+OP(SWAP,  NONE,     2, 2)
+OP(POP,   NONE,     1, 0)
+OP(LOADV, VARIABLE, 1, 1)
+OP(STOREV, VARIABLE, 1, 0)
+OP(INDEX, NONE,     2, 1)
+OP(YIELD, NONE, 1, 0)
+OP(EACH,  NONE,     1, 1)
+OP(FORK,  BRANCH,   0, 0)
+OP(JUMP,  BRANCH,   0, 0)
+OP(JUMP_F,BRANCH,   1, 0)
+OP(BACKTRACK, NONE, 0, 0)
+OP(APPEND, NONE,    2, 1)
+OP(INSERT, NONE,    4, 2)
+
+OP(ASSIGN, VARIABLE, 3, 0)
+
+OP(CALL_BUILTIN, CFUNC, -1, 1)
+
+OP(CALL_JQ, UFUNC, 1, 1)
+OP(RET, NONE, 1, 1)
+
+OP(CLOSURE_PARAM, DEFINITION, 0, 0)
+OP(CLOSURE_REF, CLOSURE_REF_IMM, 0, 0)
+OP(CLOSURE_CREATE, DEFINITION, 0, 0)
+OP(CLOSURE_CREATE_C, DEFINITION, 0, 0)

parser.h

@@ -0,0 +1,7 @@
+#ifndef PARSER_H
+#define PARSER_H
+
+int jq_parse(struct locfile* source, block* answer);
+int jq_parse_library(struct locfile* locations, block* answer);
+
+#endif

parser.y

@@ -0,0 +1,460 @@
+%{
+#include <stdio.h>
+#include <string.h>
+#include "compile.h"
+#include "jv_alloc.h"
+#define YYMALLOC jv_mem_alloc
+#define YYFREE jv_mem_free
+
+struct lexer_param;
+
+%}
+%code requires {
+#include "locfile.h"
+#define YYLTYPE location
+#define YYLLOC_DEFAULT(Loc, Rhs, N)             \
+  do {                                          \
+    if (N) {                                    \
+      (Loc).start = YYRHSLOC(Rhs, 1).start;     \
+      (Loc).end = YYRHSLOC(Rhs, N).end;         \
+    } else {                                    \
+      (Loc).start = YYRHSLOC(Rhs, 0).end;       \
+      (Loc).end = YYRHSLOC(Rhs, 0).end;         \
+    }                                           \
+  } while (0)
+ }
+
+%locations
+%error-verbose
+%define api.pure
+%union {
+  jv literal;
+  block blk;
+}
+
+%destructor { jv_free($$); } <literal>
+%destructor { block_free($$); } <blk>
+
+%parse-param {block* answer}
+%parse-param {int* errors}
+%parse-param {struct locfile* locations}
+%parse-param {struct lexer_param* lexer_param_ptr}
+%lex-param {block* answer}
+%lex-param {int* errors}
+%lex-param {struct locfile* locations}
+%lex-param {struct lexer_param* lexer_param_ptr}
+
+
+%token INVALID_CHARACTER
+%token <literal> IDENT
+%token <literal> LITERAL
+%token EQ "=="
+%token NEQ "!="
+%token DEFINEDOR "//"
+%token AS "as"
+%token DEF "def"
+%token IF "if"
+%token THEN "then"
+%token ELSE "else"
+%token ELSE_IF "elif"
+%token END "end"
+%token AND "and"
+%token OR "or"
+%token SETPIPE "|="
+%token SETPLUS "+="
+%token SETMINUS "-="
+%token SETMULT "*="
+%token SETDIV "/="
+%token SETDEFINEDOR "//="
+%token LESSEQ "<="
+%token GREATEREQ ">="
+
+%token QQSTRING_START
+%token <literal> QQSTRING_TEXT
+%token QQSTRING_INTERP_START
+%token QQSTRING_INTERP_END
+%token QQSTRING_END
+
+ /* revolting hack */
+%left ';'
+%right '|'
+%left ','
+%right "//"
+%nonassoc '=' SETPIPE SETPLUS SETMINUS SETMULT SETDIV SETDEFINEDOR
+%left OR
+%left AND
+%nonassoc NEQ EQ '<' '>' LESSEQ GREATEREQ
+%left '+' '-'
+%left '*' '/'
+
+
+%type <blk> Exp Term MkDict MkDictPair ExpD ElseBody QQString FuncDef FuncDefs String
+%{
+#include "lexer.gen.h"
+struct lexer_param {
+  yyscan_t lexer;
+};
+#define FAIL(loc, msg)                                             \
+  do {                                                             \
+    location l = loc;                                              \
+    yyerror(&l, answer, errors, locations, lexer_param_ptr, msg);  \
+    /*YYERROR*/;                                                   \
+  } while (0)
+
+void yyerror(YYLTYPE* loc, block* answer, int* errors, 
+             struct locfile* locations, struct lexer_param* lexer_param_ptr, const char *s){
+  (*errors)++;
+  locfile_locate(locations, *loc, "error: %s", s);
+}
+
+int yylex(YYSTYPE* yylval, YYLTYPE* yylloc, block* answer, int* errors, 
+          struct locfile* locations, struct lexer_param* lexer_param_ptr) {
+  yyscan_t lexer = lexer_param_ptr->lexer;
+  while (1) {
+    int tok = jq_yylex(yylval, yylloc, lexer);
+    if (tok == INVALID_CHARACTER) {
+      FAIL(*yylloc, "Invalid character");
+    } else {
+      if ((tok == LITERAL || tok == QQSTRING_TEXT) && !jv_is_valid(yylval->literal)) {
+        jv msg = jv_invalid_get_msg(jv_copy(yylval->literal));
+        if (jv_get_kind(msg) == JV_KIND_STRING) {
+          FAIL(*yylloc, jv_string_value(msg));
+        } else {
+          FAIL(*yylloc, "Invalid literal");
+        }
+        jv_free(msg);
+        jv_free(yylval->literal);
+        yylval->literal = jv_null();
+      }
+      return tok;
+    }
+  }
+}
+
+static block gen_dictpair(block k, block v) {
+  return BLOCK(gen_subexp(k), gen_subexp(v), gen_op_simple(INSERT));
+}
+
+static block gen_index(block obj, block key) {
+  return BLOCK(gen_subexp(key), obj, gen_op_simple(INDEX));
+}
+
+static block gen_binop(block a, block b, int op) {
+  const char* funcname = 0;
+  switch (op) {
+  case '+': funcname = "_plus"; break;
+  case '-': funcname = "_minus"; break;
+  case '*': funcname = "_multiply"; break;
+  case '/': funcname = "_divide"; break;
+  case EQ: funcname = "_equal"; break;
+  case NEQ: funcname = "_notequal"; break;
+  case '<': funcname = "_less"; break;
+  case '>': funcname = "_greater"; break;
+  case LESSEQ: funcname = "_lesseq"; break;
+  case GREATEREQ: funcname = "_greatereq"; break;
+  }
+  assert(funcname);
+
+  return gen_call(funcname, BLOCK(gen_lambda(a), gen_lambda(b)));
+}
+
+static block gen_format(block a) {
+  return BLOCK(a, gen_call("tostring", gen_noop()));
+}
+ 
+static block gen_update(block a, block op, int optype) {
+  if (optype) {
+    op = gen_binop(gen_noop(), op, optype);
+  }
+  return gen_assign(BLOCK(a, gen_op_simple(DUP), op));
+}
+
+%}
+
+%%
+TopLevel:
+Exp {
+  *answer = $1;
+} |
+FuncDefs {
+  *answer = $1;
+} 
+
+FuncDefs:
+/* empty */ {
+  $$ = gen_noop();
+} |
+FuncDef FuncDefs {
+  $$ = block_join($1, $2);
+}
+
+Exp:
+FuncDef Exp %prec ';' {
+  $$ = block_bind($1, $2, OP_IS_CALL_PSEUDO);
+} |
+
+Term "as" '$' IDENT '|' Exp {
+  $$ = BLOCK(gen_op_simple(DUP), $1, 
+             block_bind(gen_op_var_unbound(STOREV, jv_string_value($4)), 
+                        $6, OP_HAS_VARIABLE));
+  jv_free($4);
+} |
+
+"if" Exp "then" Exp ElseBody {
+  $$ = gen_cond($2, $4, $5);
+} |
+"if" Exp error {
+  FAIL(@$, "Possibly unterminated 'if' statment");
+  $$ = $2;
+} |
+
+Exp '=' Exp {
+  $$ = gen_assign(BLOCK(gen_op_simple(DUP), $3, gen_op_simple(SWAP), $1, gen_op_simple(SWAP)));
+} |
+
+Exp "or" Exp {
+  $$ = gen_or($1, $3);
+} | 
+
+Exp "and" Exp {
+  $$ = gen_and($1, $3);
+} |
+
+Exp "//" Exp {
+  $$ = gen_definedor($1, $3);
+} |
+
+Exp "//=" Exp {
+  $$ = gen_update($1, gen_definedor(gen_noop(), $3), 0);
+} |
+
+Exp "|=" Exp {
+  $$ = gen_update($1, $3, 0);
+} |
+
+Exp '|' Exp { 
+  $$ = block_join($1, $3); 
+} |
+
+Exp ',' Exp { 
+  $$ = gen_both($1, $3); 
+} |
+
+Exp '+' Exp {
+  $$ = gen_binop($1, $3, '+');
+} |
+
+Exp "+=" Exp {
+  $$ = gen_update($1, $3, '+');
+} |
+
+Exp '-' Exp {
+  $$ = gen_binop($1, $3, '-');
+} |
+
+Exp "-=" Exp {
+  $$ = gen_update($1, $3, '-');
+} |
+
+Exp '*' Exp {
+  $$ = gen_binop($1, $3, '*');
+} |
+
+Exp "*=" Exp {
+  $$ = gen_update($1, $3, '*');
+} |
+
+Exp '/' Exp {
+  $$ = gen_binop($1, $3, '/');
+} |
+
+Exp "/=" Exp {
+  $$ = gen_update($1, $3, '/');
+} |
+
+Exp "==" Exp {
+  $$ = gen_binop($1, $3, EQ);
+} |
+
+Exp "!=" Exp {
+  $$ = gen_binop($1, $3, NEQ);
+} |
+
+Exp '<' Exp {
+  $$ = gen_binop($1, $3, '<');
+} |
+
+Exp '>' Exp {
+  $$ = gen_binop($1, $3, '>');
+} |
+
+Exp "<=" Exp {
+  $$ = gen_binop($1, $3, LESSEQ);
+} |
+
+Exp ">=" Exp {
+  $$ = gen_binop($1, $3, GREATEREQ);
+} |
+
+Term { 
+  $$ = $1; 
+}
+
+FuncDef:
+"def" IDENT ':' Exp ';' {
+  $$ = gen_function(jv_string_value($2), gen_noop(), $4);
+  jv_free($2);
+} |
+
+"def" IDENT '(' IDENT ')' ':' Exp ';' {
+  $$ = gen_function(jv_string_value($2), 
+                    gen_op_block_unbound(CLOSURE_PARAM, jv_string_value($4)), 
+                    $7);
+  jv_free($2);
+  jv_free($4);
+}
+
+
+String:
+QQSTRING_START QQString QQSTRING_END {
+  $$ = $2;
+}
+
+QQString:
+/* empty */ {
+  $$ = gen_const(jv_string(""));
+} |
+QQString QQSTRING_TEXT {
+  $$ = gen_binop($1, gen_const($2), '+');
+} |
+QQString QQSTRING_INTERP_START Exp QQSTRING_INTERP_END {
+  $$ = gen_binop($1, gen_format($3), '+');
+}
+
+
+ElseBody:
+"elif" Exp "then" Exp ElseBody {
+  $$ = gen_cond($2, $4, $5);
+} |
+"else" Exp "end" {
+  $$ = $2;
+}
+
+ExpD:
+ExpD '|' ExpD { 
+  $$ = block_join($1, $3);
+} |
+
+Term {
+  $$ = $1;
+}
+
+
+Term:
+'.' {
+  $$ = gen_noop(); 
+} |
+Term '.' IDENT {
+  $$ = gen_index($1, gen_const($3)); 
+} |
+'.' IDENT { 
+  $$ = gen_index(gen_noop(), gen_const($2)); 
+} |
+/* FIXME: string literals */
+Term '[' Exp ']' {
+  $$ = gen_index($1, $3); 
+} |
+Term '[' ']' {
+  $$ = block_join($1, gen_op_simple(EACH)); 
+} |
+LITERAL {
+  $$ = gen_const($1); 
+} |
+String {
+  $$ = $1;
+} |
+'(' Exp ')' { 
+  $$ = $2; 
+} | 
+'[' Exp ']' { 
+  $$ = gen_collect($2); 
+} |
+'[' ']' { 
+  $$ = gen_const(jv_array()); 
+} |
+'{' MkDict '}' { 
+  $$ = BLOCK(gen_subexp(gen_const(jv_object())), $2, gen_op_simple(POP));
+} |
+'$' IDENT {
+  $$ = gen_location(@$, gen_op_var_unbound(LOADV, jv_string_value($2)));
+  jv_free($2);
+} | 
+IDENT {
+  $$ = gen_location(@$, gen_call(jv_string_value($1), gen_noop()));
+  jv_free($1);
+} |
+IDENT '(' Exp ')' {
+  $$ = gen_call(jv_string_value($1), gen_lambda($3));
+  $$ = gen_location(@1, $$);
+  jv_free($1);
+} |
+'(' error ')' { $$ = gen_noop(); } |
+'[' error ']' { $$ = gen_noop(); } |
+Term '[' error ']' { $$ = $1; } |
+'{' error '}' { $$ = gen_noop(); }
+
+MkDict:
+{ 
+  $$=gen_noop(); 
+} |
+ MkDictPair { $$ = $1; }
+| MkDictPair ',' MkDict { $$=block_join($1, $3); }
+| error ',' MkDict { $$ = $3; }
+
+MkDictPair
+: IDENT ':' ExpD { 
+  $$ = gen_dictpair(gen_const($1), $3);
+ }
+| String ':' ExpD {
+  $$ = gen_dictpair($1, $3);
+  }
+| String {
+  $$ = gen_dictpair($1, BLOCK(gen_op_simple(POP), gen_op_simple(DUP2),
+                              gen_op_simple(DUP2), gen_op_simple(INDEX)));
+  }
+| IDENT {
+  $$ = gen_dictpair(gen_const(jv_copy($1)),
+                    gen_index(gen_noop(), gen_const($1)));
+  }
+| '(' Exp ')' ':' ExpD {
+  $$ = gen_dictpair($2, $5);
+  }
+| '(' error ')' ':' ExpD { $$ = $5; }
+%%
+
+int jq_parse(struct locfile* locations, block* answer) {
+  struct lexer_param scanner;
+  YY_BUFFER_STATE buf;
+  jq_yylex_init_extra(0, &scanner.lexer);
+  buf = jq_yy_scan_bytes(locations->data, locations->length, scanner.lexer);
+  int errors = 0;
+  *answer = gen_noop();
+  yyparse(answer, &errors, locations, &scanner);
+  jq_yy_delete_buffer(buf, scanner.lexer);
+  jq_yylex_destroy(scanner.lexer);
+  if (errors > 0) {
+    block_free(*answer);
+    *answer = gen_noop();
+  }
+  return errors;
+}
+
+int jq_parse_library(struct locfile* locations, block* answer) {
+  int errs = jq_parse(locations, answer);
+  if (errs) return errs;
+  if (!block_has_only_binders(*answer, OP_IS_CALL_PSEUDO)) {
+    locfile_locate(locations, UNKNOWN_LOCATION, "error: library should only have function definitions, not a main expression");
+    return 1;
+  }
+  return 0;
+}

stdlib.jq

@@ -1,11 +0,0 @@
-def map(f) = [.[] | f];
-
-def first = .[0];
-# def last = .[count-1];
-def next = .[count];
-
-# ([])[] would be a decent definition of "empty"
-# except ([]) is defined as syntactic sugar for empty
-def empty = {}[];
-
-def sort = [group(.) | .[]];

testdata

@@ -0,0 +1,475 @@
+# Tests are groups of three lines: program, input, expected output
+# Blank lines and lines starting with # are ignored
+
+#
+# Simple value tests to check parser. Input is irrelevant
+#
+
+true
+null
+true
+
+false
+null
+false
+
+null
+42
+null
+
+1
+null
+1
+
+# FIXME: much more number testing needed
+
+{}
+null
+{}
+
+[]
+null
+[]
+
+# The input line starts with a 0xFEFF (byte order mark) codepoint
+# No, there is no reason to have a byte order mark in UTF8 text.
+# But apparently people do, so jq shouldn't break on it.
+.
+"byte order mark"
+"byte order mark"
+
+# We test escapes by matching them against Unicode codepoints
+# FIXME: more tests needed for weird unicode stuff (e.g. utf16 pairs)
+"Aa\r\n\t\b\f\u03bc"
+null
+"Aa\u000d\u000a\u0009\u0008\u000c\u03bc"
+
+.
+"Aa\r\n\t\b\f\u03bc"
+"Aa\u000d\u000a\u0009\u0008\u000c\u03bc"
+
+"inter\("pol" + "ation")"
+null
+"interpolation"
+
+#
+# Dictionary construction syntax
+#
+
+{a: 1}
+null
+{"a":1}
+
+{a,b,(.d):.a,e:.b}
+{"a":1, "b":2, "c":3, "d":"c"}
+{"a":1, "b":2, "c":1, "e":2}
+
+{"a",b,"a$\(1+1)"}
+{"a":1, "b":2, "c":3, "a$2":4}
+{"a":1, "b":2, "a$2":4}
+
+#
+# Field access, piping
+#
+
+.foo
+{"foo": 42, "bar": 43}
+42
+
+.foo | .bar
+{"foo": {"bar": 42}, "bar": "badvalue"}
+42
+
+.foo.bar
+{"foo": {"bar": 42}, "bar": "badvalue"}
+42
+
+.foo_bar
+{"foo_bar": 2}
+2
+
+.["foo"].bar
+{"foo": {"bar": 42}, "bar": "badvalue"}
+42
+
+
+#
+# Multiple outputs, iteration
+#
+
+.[]
+[1,2,3]
+1
+2
+3
+
+1,1
+[]
+1
+1
+
+1,.
+[]
+1
+[]
+
+[.]
+[2]
+[[2]]
+
+[[2]]
+[3]
+[[2]]
+
+[{}]
+[2]
+[{}]
+
+[.[]]
+["a"]
+["a"]
+
+[(.,1),((.,.[]),(2,3))]
+["a","b"]
+[["a","b"],1,["a","b"],"a","b",2,3]
+
+[([5,5][]),.,.[]]
+[1,2,3]
+[5,5,[1,2,3],1,2,3]
+
+{x: (1,2)},{x:3} | .x
+null
+1
+2
+3
+
+#
+# Variables
+#
+
+1 as $x | 2 as $y | [$x,$y,$x]
+null
+[1,2,1]
+
+[1,2,3][] as $x | [[4,5,6,7][$x]]
+null
+[5]
+[6]
+[7]
+
+42 as $x | . | . | . + 432 | $x + 1
+34324
+43
+
+1 as $x | [$x,$x,$x as $x | $x]
+null
+[1,1,1]
+
+# [.,(.[] | {x:.},.),.,.[]]
+
+#
+# Builtin functions
+#
+
+1+1
+null
+2
+
+1+1
+"wtasdf"
+2.0
+
+1e+0+0.001e3
+"I wonder what this will be?"
+20e-1
+
+.+4
+15
+19.0
+
+[1,2,3] + [.]
+null
+[1,2,3,null]
+
+{"a":1} + {"b":2} + {"c":3}
+"asdfasdf"
+{"a":1, "b":2, "c":3}
+
+"asdf" + "jkl;" + . + . + .
+"some string"
+"asdfjkl;some stringsome stringsome string"
+
+"\u0000\u0020\u0000" + .
+"\u0000\u0020\u0000"
+"\u0000 \u0000\u0000 \u0000"
+
+42 - .
+11
+31
+
+[1,2,3,4,1] - [.,3]
+1
+[2,4]
+
+[10 * 20, 20 / .]
+4
+[200, 5]
+
+1 + 2 * 2 + 10 / 2
+null
+10
+
+[16 / 4 / 2, 16 / 4 * 2, 16 - 4 - 2, 16 - 4 + 2]
+null
+[2, 8, 10, 14]
+
+1 + tonumber + ("10" | tonumber)
+4
+15
+
+[{"a":42},.object,10,.num,false,true,null,"b",[1,4]] | .[] as $x | [$x == .[]]
+{"object": {"a":42}, "num":10.0}
+[true,  true,  false, false, false, false, false, false, false]
+[true,  true,  false, false, false, false, false, false, false]
+[false, false, true,  true,  false, false, false, false, false]
+[false, false, true,  true,  false, false, false, false, false]
+[false, false, false, false, true,  false, false, false, false]
+[false, false, false, false, false, true,  false, false, false]
+[false, false, false, false, false, false, true,  false, false]
+[false, false, false, false, false, false, false, true,  false]
+[false, false, false, false, false, false, false, false, true ]
+
+[.[] | length]
+[[], {}, [1,2], {"a":42}, "asdf"]
+[0, 0, 2, 1, 4]
+
+map(keys)
+[{}, {"abcd":1,"abc":2,"abcde":3}, {"x":1, "z": 3, "y":2}]
+[[], ["abc","abcd","abcde"], ["x","y","z"]]
+
+[1,2,empty,3,empty,4]
+null
+[1,2,3,4]
+
+map(add)
+[[], [1,2,3], ["a","b","c"], [[3],[4,5],[6]], [{"a":1}, {"b":2}, {"a":3}]]
+[null, 6, "abc", [3,4,5,6], {"a":3, "b": 2}]
+
+#
+# User-defined functions
+# Oh god.
+#
+
+def f: . + 1; def g: def g: . + 100; f | g | f; (f | g), g
+3.0
+106.0
+105.0
+
+def f: (1000,2000); f
+123412345
+1000
+2000
+
+([1,2] + [4,5])
+[1,2,3]
+[1,2,4,5]
+
+true
+[1]
+true
+
+null,1,null
+"hello"
+null
+1
+null
+
+[1,2,3]
+[5,6]
+[1,2,3]
+
+def f(x): x | x; f([.], . + [42])
+[1,2,3]
+[[[1,2,3]]]
+[[1,2,3],42]
+[[1,2,3,42]]
+[1,2,3,42,42]
+
+# test closures and lexical scoping
+def id(x):x; 2000 as $x | def f(x):1 as $x | id([$x, x, x]); def g(x): 100 as $x | f($x,$x+x); g($x)
+"more testing"
+[1,100,2100.0,100,2100.0]
+
+# test backtracking through function calls and returns
+# this test is *evil*
+[[20,10][1,0] as $x | def f: (100,200) as $y | def g: [$x + $y, .]; . + $x | g; f[0] | [f][0][1] | f]
+999999999
+[[110.0, 130.0], [210.0, 130.0], [110.0, 230.0], [210.0, 230.0], [120.0, 160.0], [220.0, 160.0], [120.0, 260.0], [220.0, 260.0]]
+
+# test recursion
+def fac: if . == 1 then 1 else . * (. - 1 | fac) end; [.[] | fac]
+[1,2,3,4]
+[1,2,6,24]
+
+# test stack overflow and reallocation
+# this test is disabled for now, it takes a realllllly long time.
+# def f: if length > 1000 then . else .+[1]|f end; f | length
+# []
+# 1001
+
+#
+# Assignment
+#
+.message = "goodbye"
+{"message": "hello"}
+{"message": "goodbye"}
+
+.foo = .bar
+{"bar":42}
+{"foo":42, "bar":42}
+
+.foo |= .+1
+{"foo": 42}
+{"foo": 43}
+
+.[] += 2, .[] *= 2, .[] -= 2, .[] /= 2
+[1,3,5]
+[3,5,7]
+[2,6,10]
+[-1,1,3]
+[0.5, 1.5, 2.5]
+
+.[0].a |= {"old":., "new":(.+1)}
+[{"a":1,"b":2}]
+[{"a":{"old":1, "new":2},"b":2}]
+
+def inc(x): x |= .+1; inc(.[].a)
+[{"a":1,"b":2},{"a":2,"b":4},{"a":7,"b":8}]
+[{"a":2,"b":2},{"a":3,"b":4},{"a":8,"b":8}]
+
+.[2][3] = 1
+[4]
+[4, null, [null, null, null, 1]]
+
+.foo[2].bar = 1
+{"foo":[11], "bar":42}
+{"foo":[11,null,{"bar":1}], "bar":42}
+
+#
+# Conditionals
+#
+
+[.[] | if .foo then "yep" else "nope" end]
+[{"foo":0},{"foo":1},{"foo":[]},{"foo":true},{"foo":false},{"foo":null},{"foo":"foo"},{}]
+["yep","yep","yep","yep","nope","nope","yep","nope"]
+
+[.[] | if .baz then "strange" elif .foo then "yep" else "nope" end]
+[{"foo":0},{"foo":1},{"foo":[]},{"foo":true},{"foo":false},{"foo":null},{"foo":"foo"},{}]
+["yep","yep","yep","yep","nope","nope","yep","nope"]
+
+
+# FIXME: define/test behaviour of 'if (.foo,.bar) then A else B end'
+
+[.[] | [.foo[] // .bar]]
+[{"foo":[1,2], "bar": 42}, {"foo":[1], "bar": null}, {"foo":[null,false,3], "bar": 18}, {"foo":[], "bar":42}, {"foo": [null,false,null], "bar": 41}]
+[[1,2], [1], [3], [42], [41]]
+
+# FIXME: behaviour of update operators
+# .[] //= .[0]
+# ["hello",true,false,[false],null]
+# ["hello",true,"hello",[false],"hello"]
+
+.[] | [.[0] and .[1], .[0] or .[1]]
+[[true,[]], [false,1], [42,null], [null,false]]
+[true,true]
+[false,true]
+[false,true]
+[false,false]
+
+[.[] | not]
+[1,0,false,null,true,"hello"]
+[false,false,true,true,false,false]
+
+# Check numeric comparison binops
+[10 > 0, 10 > 10, 10 > 20, 10 < 0, 10 < 10, 10 < 20]
+{}
+[true,false,false,false,false,true]
+
+[10 >= 0, 10 >= 10, 10 >= 20, 10 <= 0, 10 <= 10, 10 <= 20]
+{}
+[true,true,false,false,true,true]
+
+# And some in/equality tests
+[ 10 == 10, 10 != 10, 10 != 11, 10 == 11]
+{}
+[true,false,true,false]
+
+["hello" == "hello", "hello" != "hello", "hello" == "world", "hello" != "world" ]
+{}
+[true,false,false,true]
+
+[[1,2,3] == [1,2,3], [1,2,3] != [1,2,3], [1,2,3] == [4,5,6], [1,2,3] != [4,5,6]]
+{}
+[true,false,false,true]
+
+[{"foo":42} == {"foo":42},{"foo":42} != {"foo":42}, {"foo":42} != {"bar":42}, {"foo":42} == {"bar":42}]
+{}
+[true,false,true,false]
+
+# ugly complicated thing
+[{"foo":[1,2,{"bar":18},"world"]} == {"foo":[1,2,{"bar":18},"world"]},{"foo":[1,2,{"bar":18},"world"]} == {"foo":[1,2,{"bar":19},"world"]}]
+{}
+[true,false]
+
+# containment operator
+[("foo" | contains("foo")), ("foobar" | contains("foo")), ("foo" | contains("foobar"))]
+{}
+[true, true, false]
+
+map(.[1] as $needle | .[0] | contains($needle))
+[[[],[]], [[1,2,3], [1,2]], [[1,2,3], [3,1]], [[1,2,3], [4]], [[1,2,3], [1,4]]]
+[true, true, true, false, false]
+
+map(.[1] as $needle | .[0] | contains($needle))
+[[["foobar", "foobaz"], ["baz", "bar"]], [["foobar", "foobaz"], ["foo"]], [["foobar", "foobaz"], ["blap"]]]
+[true, true, false]
+
+[({foo: 12, bar:13} | contains({foo: 12})), ({foo: 12} | contains({})), ({foo: 12, bar:13} | contains({baz:14}))]
+{}
+[true, true, false]
+
+{foo: {baz: 12, blap: {bar: 13}}, bar: 14} | contains({bar: 14, foo: {blap: {}}})
+{}
+true
+
+{foo: {baz: 12, blap: {bar: 13}}, bar: 14} | contains({bar: 14, foo: {blap: {bar: 14}}})
+{}
+false
+
+sort
+[42,[2,5,3,11],10,{"a":42,"b":2},{"a":42},true,2,[2,6],"hello",null,[2,5,6],{"a":[],"b":1},"abc","ab",[3,10],{},false,"abcd",null]
+[null,null,false,true,2,10,42,"ab","abc","abcd","hello",[2,5,3,11],[2,5,6],[2,6],[3,10],{},{"a":42},{"a":42,"b":2},{"a":[],"b":1}]
+
+(sort_by(.b) | sort_by(.a)), sort_by(.a, .b), sort_by(.b, .c), group_by(.b), group_by(.a + .b - .c == 2)
+[{"a": 1, "b": 4, "c": 14}, {"a": 4, "b": 1, "c": 3}, {"a": 1, "b": 4, "c": 3}, {"a": 0, "b": 2, "c": 43}]
+[{"a": 0, "b": 2, "c": 43}, {"a": 1, "b": 4, "c": 14}, {"a": 1, "b": 4, "c": 3}, {"a": 4, "b": 1, "c": 3}]
+[{"a": 0, "b": 2, "c": 43}, {"a": 1, "b": 4, "c": 14}, {"a": 1, "b": 4, "c": 3}, {"a": 4, "b": 1, "c": 3}]
+[{"a": 4, "b": 1, "c": 3}, {"a": 0, "b": 2, "c": 43}, {"a": 1, "b": 4, "c": 3}, {"a": 1, "b": 4, "c": 14}]
+[[{"a": 4, "b": 1, "c": 3}], [{"a": 0, "b": 2, "c": 43}], [{"a": 1, "b": 4, "c": 14}, {"a": 1, "b": 4, "c": 3}]]
+[[{"a": 1, "b": 4, "c": 14}, {"a": 0, "b": 2, "c": 43}], [{"a": 4, "b": 1, "c": 3}, {"a": 1, "b": 4, "c": 3}]]
+
+unique
+[1,2,5,3,5,3,1,3]
+[1,2,3,5]
+
+[min, max, min_by(.[1]), max_by(.[1]), min_by(.[2]), max_by(.[2])]
+[[4,2,"a"],[3,1,"a"],[2,4,"a"],[1,3,"a"]]
+[[1,3,"a"],[4,2,"a"],[3,1,"a"],[2,4,"a"],[4,2,"a"],[1,3,"a"]]
+
+[min,max,min_by(.),max_by(.)]
+[]
+[null,null,null,null]
+
+.foo[.baz]
+{"foo":{"bar":4},"baz":"bar"}
+4