Add --strip-comments option

Makefile.am

@@ -132,7 +132,7 @@ endif
 
 ### Tests (make check)
 
-TESTS = tests/optionaltest tests/mantest tests/jqtest tests/onigtest tests/shtest tests/utf8test tests/base64test
+TESTS = tests/optionaltest tests/mantest tests/jqtest tests/onigtest tests/shtest tests/utf8test tests/base64test tests/commenttest
 TESTS_ENVIRONMENT = NO_VALGRIND=$(NO_VALGRIND)
 
 # This is a magic make variable that causes it to treat tests/man.test as a
@@ -215,7 +215,8 @@ EXTRA_DIST = $(DOC_FILES) $(man_MANS) $(TESTS) $(TEST_LOG_COMPILER)     \
         tests/optional.test tests/optionaltest				\
         tests/utf8-truncate.jq tests/utf8test				\
         tests/base64.test tests/base64test 				\
-        tests/jq-f-test.sh tests/shtest
+        tests/jq-f-test.sh tests/shtest					\
+        tests/commenttest
 
 DISTCHECK_CONFIGURE_FLAGS=--disable-maintainer-mode --with-oniguruma=builtin
 

src/jv.h

@@ -223,6 +223,7 @@ enum {
   JV_PARSE_SEQ              = 1,
   JV_PARSE_STREAMING        = 2,
   JV_PARSE_STREAM_ERRORS    = 4,
+  JV_PARSE_STRIP_COMMENTS   = 8
 };
 
 jv jv_parse(const char* string);

src/jv_parse.c

@@ -31,6 +31,7 @@ enum last_seen {
 };
 
 struct jv_parser {
+  pfunc (*scan)(struct jv_parser* p, char ch, jv* out);
   const char* curr_buf;
   int curr_buf_length;
   int curr_buf_pos;
@@ -65,9 +66,14 @@ struct jv_parser {
   unsigned int last_ch_was_ws:1;
 };
 
+static pfunc scan_json(struct jv_parser* p, char ch, jv* out);
+static pfunc scan_line_comment(struct jv_parser* p, char ch, jv* out);
+static pfunc scan_slash_comment(struct jv_parser* p, char ch, jv* out);
+static pfunc scan_c_comment(struct jv_parser* p, char ch, jv* out);
 
 static void parser_init(struct jv_parser* p, int flags) {
   p->flags = flags;
+  p->scan = scan_json;
   if ((p->flags & JV_PARSE_STREAMING)) {
     p->path = jv_array();
   } else {
@@ -613,12 +619,42 @@ static int stream_is_top_num(struct jv_parser* p) {
 #define is_top_num(p) \
    (((p)->flags & JV_PARSE_STREAMING) ? stream_is_top_num((p)) : parse_is_top_num((p)))
 
-static pfunc scan(struct jv_parser* p, char ch, jv* out) {
-  p->column++;
-  if (ch == '\n') {
-    p->line++;
-    p->column = 0;
+
+static pfunc scan_line_comment(struct jv_parser* p, char ch, jv* out) {
+  if(ch == '\n')
+    p->scan = scan_json;
+  return OK;
+}
+
+static pfunc scan_c_comment_close(struct jv_parser* p, char ch, jv* out) {
+  if(ch == '/') {
+    p->scan = scan_json;
+  } else {
+    p->scan = scan_c_comment;
   }
+  return OK;
+}
+
+static pfunc scan_c_comment(struct jv_parser* p, char ch, jv* out) {
+  if(ch == '*') {
+    p->scan = scan_c_comment_close;
+  }
+  return OK;
+}
+
+static pfunc scan_slash_comment(struct jv_parser* p, char ch, jv* out) {
+  if(ch == '/') {
+    p->scan = scan_line_comment;
+    return OK;
+  }
+  if(ch == '*') {
+    p->scan = scan_c_comment;
+    return OK;
+  }
+  return "Incomplete comment token; slash must be followed by another slash or asterisk";
+}
+
+static pfunc scan_json(struct jv_parser* p, char ch, jv* out) {
   if (ch == '\036' /* ASCII RS; see draft-ietf-json-sequence-07 */) {
     if (check_truncation(p)) {
       if (check_literal(p) == 0 && is_top_num(p))
@@ -635,9 +671,15 @@ static pfunc scan(struct jv_parser* p, char ch, jv* out) {
     *out = jv_invalid();
     return OK;
   }
+
   presult answer = 0;
   p->last_ch_was_ws = 0;
   if (p->st == JV_PARSER_NORMAL) {
+    if(ch == '/' && (p->flags & JV_PARSE_STRIP_COMMENTS)) {
+      p->scan = scan_slash_comment;
+      return OK;
+    }
+
     chclass cls = classify(ch);
     if (cls == WHITESPACE)
       p->last_ch_was_ws = 1;
@@ -678,6 +720,16 @@ static pfunc scan(struct jv_parser* p, char ch, jv* out) {
   return answer;
 }
 
+static pfunc scan(struct jv_parser* p, char ch, jv* out) {
+  p->column++;
+  if (ch == '\n') {
+    p->line++;
+    p->column = 0;
+  }
+  return p->scan(p, ch, out);
+}
+
+
 struct jv_parser* jv_parser_new(int flags) {
   struct jv_parser* p = jv_mem_alloc(sizeof(struct jv_parser));
   parser_init(p, flags);

src/main.c

@@ -81,6 +81,7 @@ static void usage(int code, int keep_it_short) {
       "  -M               monochrome (don't colorize JSON);\n"
       "  -S               sort keys of objects on output;\n"
       "  --tab            use tabs for indentation;\n"
+      "  --strip-comments accept and strip comments from input;\n"
       "  --arg a v        set variable $a to value <v>;\n"
       "  --argjson a v    set variable $a to JSON value <v>;\n"
       "  --slurpfile a f  set variable $a to an array of JSON texts read from <f>;\n"
@@ -454,6 +455,10 @@ int main(int argc, char* argv[]) {
         parser_flags |= JV_PARSE_STREAMING;
         continue;
       }
+      if (isoption(argv[i], 0, "strip-comments", &short_opts)) {
+        parser_flags |= JV_PARSE_STRIP_COMMENTS;
+        continue;
+      }
       if (isoption(argv[i], 0, "stream-errors", &short_opts)) {
         parser_flags |= JV_PARSE_STREAM_ERRORS;
         continue;

tests/commenttest

@@ -0,0 +1,15 @@
+#!/bin/sh
+
+. "${0%/*}/setup" "$@"
+
+if [ "$(echo '{"hi":"there"/*comment*/}' | $VALGRIND $Q $JQ --strip-comments '.hi')" != '"there"' ]; then
+	echo "C-style comment test failed"
+	exit 1
+fi
+
+if [ "$(printf '{"hi"://comment\n"there"}' | $VALGRIND $Q $JQ --strip-comments '.hi')" != '"there"' ]; then
+	echo "C++-style comment test failed"
+	exit 1
+fi
+
+exit 0