Sequence parser: wait for RS on startup (fix #687)

Per-RFC7464.

2 files changed, 37 insertions, 27 deletions

diff --git a/jv_parse.c b/jv_parse.c
index cdffd1c0..b6d6e955 100644
--- a/jv_parse.c
+++ b/jv_parse.c
@@ -77,7 +77,10 @@ static void parser_init(struct jv_parser* p, int flags) {
   p->next = jv_invalid();
   p->tokenbuf = 0;
   p->tokenlen = p->tokenpos = 0;
-  p->st = JV_PARSER_NORMAL;
+  if ((p->flags & JV_PARSE_SEQ))
+    p->st = JV_PARSER_WAITING_FOR_RS;
+  else
+    p->st = JV_PARSER_NORMAL;
   p->eof = 0;
   p->curr_buf = 0;
   p->curr_buf_length = p->curr_buf_pos = p->curr_buf_is_partial = 0;
@@ -720,8 +723,17 @@ jv jv_parser_next(struct jv_parser* p) {
   presult msg = 0;
   while (!msg && p->curr_buf_pos < p->curr_buf_length) {
     ch = p->curr_buf[p->curr_buf_pos++];
-    if (ch != '\036' && p->st == JV_PARSER_WAITING_FOR_RS)
+    if (p->st == JV_PARSER_WAITING_FOR_RS) {
+      if (ch == '\n') {
+        p->line++;
+        p->column = 0;
+      } else {
+        p->column++;
+      }
+      if (ch == '\036')
+        p->st = JV_PARSER_NORMAL;
       continue; // need to resync, wait for RS
+    }
     msg = scan(p, ch, &value);
   }
   if (msg == OK) {
@@ -752,26 +764,26 @@ jv jv_parser_next(struct jv_parser* p) {
     p->eof = 1;
     assert(p->curr_buf_pos == p->curr_buf_length);
     jv_free(value);
-    if (p->st != JV_PARSER_WAITING_FOR_RS) {
-      if (p->st != JV_PARSER_NORMAL) {
-        value = make_error(p, "Unfinished string at EOF at line %d, column %d", p->line, p->column);
-        parser_reset(p);
-        p->st = JV_PARSER_WAITING_FOR_RS;
-        return value;
-      }
-      if ((msg = check_literal(p))) {
-        value = make_error(p, "%s at EOF at line %d, column %d", msg, p->line, p->column);
-        parser_reset(p);
-        p->st = JV_PARSER_WAITING_FOR_RS;
-        return value;
-      }
-      if (((p->flags & JV_PARSE_STREAMING) && p->stacklen != 0) ||
-          (!(p->flags & JV_PARSE_STREAMING) && p->stackpos != 0)) {
-        value = make_error(p, "Unfinished JSON term at EOF at line %d, column %d", p->line, p->column);
-        parser_reset(p);
-        p->st = JV_PARSER_WAITING_FOR_RS;
-        return value;
-      }
+    if (p->st == JV_PARSER_WAITING_FOR_RS)
+      return make_error(p, "Unfinished abandoned text at EOF at line %d, column %d", p->line, p->column);
+    if (p->st != JV_PARSER_NORMAL) {
+      value = make_error(p, "Unfinished string at EOF at line %d, column %d", p->line, p->column);
+      parser_reset(p);
+      p->st = JV_PARSER_WAITING_FOR_RS;
+      return value;
+    }
+    if ((msg = check_literal(p))) {
+      value = make_error(p, "%s at EOF at line %d, column %d", msg, p->line, p->column);
+      parser_reset(p);
+      p->st = JV_PARSER_WAITING_FOR_RS;
+      return value;
+    }
+    if (((p->flags & JV_PARSE_STREAMING) && p->stacklen != 0) ||
+        (!(p->flags & JV_PARSE_STREAMING) && p->stackpos != 0)) {
+      value = make_error(p, "Unfinished JSON term at EOF at line %d, column %d", p->line, p->column);
+      parser_reset(p);
+      p->st = JV_PARSER_WAITING_FOR_RS;
+      return value;
     }
     // 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
diff --git a/tests/run b/tests/run
index b213b565..0dc966f4 100755
--- a/tests/run
+++ b/tests/run
@@ -111,7 +111,6 @@ fi
 ## into tests/all.test
 
 cat > $d/expected <<EOF
-ignoring parse error: Potentially truncated top-level numeric value at line 1, column 2
 ignoring parse error: Truncated value at line 2, column 5
 ignoring parse error: Truncated value at line 2, column 25
 ignoring parse error: Truncated value at line 2, column 41
@@ -120,7 +119,6 @@ printf '1\0362 3\n[0,1\036[4,5]true"ab"{"c":4\036{}{"d":5,"e":6"\036false\n'|$VA
 cmp $d/out $d/expected
 
 cat > $d/expected <<EOF
-ignoring parse error: Potentially truncated top-level numeric value at line 1, column 2
 ignoring parse error: Truncated value at line 2, column 5
 ignoring parse error: Truncated value at line 2, column 25
 ignoring parse error: Truncated value at line 3, column 1
@@ -131,21 +129,21 @@ cmp $d/out $d/expected
 # Note that here jq sees no inputs at all but it still succeeds because
 # --seq ignores parse errors
 cat > $d/expected <<EOF
-ignoring parse error: Unfinished string at EOF at line 1, column 4
+ignoring parse error: Unfinished abandoned text at EOF at line 1, column 4
 EOF
 printf '"foo'|./jq -ce --seq . > $d/out 2>&1
 cmp $d/out $d/expected
 
 # Numeric values truncated by EOF are ignored
 cat > $d/expected <<EOF
-ignoring parse error: Potentially truncated top-level numeric value at EOF at line 1, column 1
+ignoring parse error: Unfinished abandoned text at EOF at line 1, column 1
 EOF
 printf '1'|./jq -ce --seq . > $d/out 2>&1
 cmp $d/out $d/expected
 
 cat > $d/expected <<EOF
 EOF
-printf '1\n'|./jq -ces --seq '. == [1]' >/dev/null 2> $d/out
+printf '1\n'|./jq -cen --seq '[inputs] == []' >/dev/null 2> $d/out
 cmp $d/out $d/expected
 
 ## Test streaming parser