commonmark/cmark

Parser incorrectly reading strong/emph in paragraphs that contain links.

eclbtownsend opened this issue · 2 comments

The following input produces an incorrect syntax tree:

const char* sampleInput = "*text* [link](google.com)";
cmark_node* cmark_root = cmark_parse_document(sampleInput, strlen(sampleInput), CMARK_OPT_DEFAULT);
const char* html = cmark_render_html(cmark_root, 0);

This produces a tree:

<p>*text* <a href="google.com">link</a></p>

I would expect text to have emphasis tags. It seems to only miss the last set of emphasis before the link. try "*text* *text2* [link](google.com)"

Version: "0.30.1"
Platform: Windows, Visual Studio 2017 v15.9.33
config.h:

#ifndef CMARK_CONFIG_H
#define CMARK_CONFIG_H

#ifdef __cplusplus
extern "C" {
#endif

#define HAVE_STDBOOL_H

#ifdef HAVE_STDBOOL_H
 #include <stdbool.h>
#elif !defined(__cplusplus)
 typedef char bool;
#endif

/* #undef HAVE___BUILTIN_EXPECT */

/* #undef HAVE___ATTRIBUTE__ */

#ifdef HAVE___ATTRIBUTE__
 #define CMARK_ATTRIBUTE(list) __attribute__ (list)
#else
 #define CMARK_ATTRIBUTE(list)
#endif

#ifndef CMARK_INLINE
 #if defined(_MSC_VER) && !defined(__cplusplus)
   #define CMARK_INLINE __inline
 #else
   #define CMARK_INLINE inline
 #endif
#endif

/* snprintf and vsnprintf fallbacks for MSVC before 2015,
  due to Valentin Milea http://stackoverflow.com/questions/2915672/
*/

#if defined(_MSC_VER) && _MSC_VER < 1900

#include <stdio.h>
#include <stdarg.h>

#define snprintf c99_snprintf
#define vsnprintf c99_vsnprintf

CMARK_INLINE int c99_vsnprintf(char *outBuf, size_t size, const char *format, va_list ap)
{
   int count = -1;

   if (size != 0)
       count = _vsnprintf_s(outBuf, size, _TRUNCATE, format, ap);
   if (count == -1)
       count = _vscprintf(format, ap);

   return count;
}

CMARK_INLINE int c99_snprintf(char *outBuf, size_t size, const char *format, ...)
{
   int count;
   va_list ap;

   va_start(ap, format);
   count = c99_vsnprintf(outBuf, size, format, ap);
   va_end(ap);

   return count;
}

#endif

#ifdef __cplusplus
}
#endif

#endif
jgm commented

Curious. I can reproduce with the cmark command line tool:

% build/src/cmark
*text* [link](google.com)
^D
<p>*text* <a href="google.com">link</a></p>

And this is a regression -- I tried with an older version and didn't see this.

jgm commented

Bisecting shows that this bug got introduced with commit ed0a4bf (fix for #389).
@nwellnhof can you take a look?