Skip to content

Commit

Permalink
Fix markdown parsing bug (Issue #503)
Browse files Browse the repository at this point in the history
  • Loading branch information
@michaelrsweet
michaelrsweet committed Sep 13, 2023
1 parent 8d16a96 commit cbf3130
Show file tree
Hide file tree
Showing 4 changed files with 151 additions and 73 deletions.
1 change: 1 addition & 0 deletions CHANGES.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -6,6 +6,7 @@
- Updated PostScript and PDF date/time information to use UTC (Issue #490)
- Fixed multiple conversions of UTF-8 HTML files from the GUI (Issue #496)
- Fixed a compile bug on Solaris (Issue #498)
- Fixed a markdown parsing issue (Issue #503)
- Fixed a crash bug with bad title images (Issue #510)
- Fixed some minor CodeQL warnings.

Expand Down
3 changes: 2 additions & 1 deletion htmldoc/markdown.cxx
View file
Original file line number Diff line number Diff line change
Expand Up @@ -36,14 +36,15 @@ mdReadFile(tree_t *parent, /* I - Parent node */
FILE *fp, /* I - File to read from */
const char *base) /* I - Base path/URL */
{
mmd_t *doc = mmdLoadFile(fp); /* Markdown document */
mmd_t *doc; /* Markdown document */
tree_t *html, /* HTML element */
*head, /* HEAD element */
*temp, /* META/TITLE element */
*body; /* BODY element */
const char *meta; /* Title, author, etc. */


doc = mmdLoadFile(NULL, fp);
html = htmlAddTree(parent, MARKUP_HTML, NULL);
if ((meta = mmdGetMetadata(doc, "lang")) != NULL)
htmlSetVariable(html, (uchar *)"lang", get_text((uchar *)meta));
Expand Down
207 changes: 140 additions & 67 deletions htmldoc/mmd.c
View file
Original file line number Diff line number Diff line change
Expand Up @@ -3,7 +3,7 @@
*
* https://github.com/michaelrsweet/mmd
*
* Copyright © 2017-2021 by Michael R Sweet.
* Copyright © 2017-2023 by Michael R Sweet.
*
* Licensed under Apache License v2.0. See the file "LICENSE" for more
* information.
Expand All @@ -13,9 +13,7 @@
* Define DEBUG to get debug printf messages to stderr.
*/

#ifndef DEBUG
# define DEBUG 0
#endif // !DEBUG
#define DEBUG 0
#if DEBUG > 0
# define DEBUG_printf(...) fprintf(stderr, __VA_ARGS__)
# define DEBUG_puts(s) fputs(s, stderr);
Expand Down Expand Up @@ -69,7 +67,6 @@

#ifdef _WIN32
# define snprintf _snprintf
# define strcasecmp _stricmp
# define strdup _strdup
#endif /* _WIN32 */

Expand Down Expand Up @@ -184,30 +181,17 @@ mmdCopyAllText(mmd_t *node) /* I - Parent node */
*/

textlen = strlen(current->text);
allsize += textlen + (size_t)current->whitespace;
temp = realloc(all, allsize);

if (allsize == 0)
if (!temp)
{
allsize = textlen + (size_t)current->whitespace + 1;
all = malloc(allsize);
allptr = all;

if (!all)
return (NULL);
free(all);
return (NULL);
}
else
{
allsize += textlen + (size_t)current->whitespace;
temp = realloc(all, allsize);

if (!temp)
{
free(all);
return (NULL);
}

allptr = temp + (allptr - all);
all = temp;
}
allptr = temp + (allptr - all);
all = temp;

if (current->whitespace)
*allptr++ = ' ';
Expand Down Expand Up @@ -469,25 +453,29 @@ mmdIsBlock(mmd_t *node) /* I - Node */
* 'mmdLoad()' - Load a markdown file into nodes.
*/

mmd_t * /* O - First node in markdown */
mmdLoad(const char *filename) /* I - File to load */
mmd_t * /* O - Root node in markdown */
mmdLoad(mmd_t *root, /* I - Root node for document or `NULL` for a new document */
const char *filename) /* I - File to load */
{
FILE *fp; /* File */
mmd_t *doc; /* Document */


/*
* Open the file and create an empty document...
* Open the file and load the document...
*/

if ((fp = fopen(filename, "r")) == NULL)
return (NULL);

doc = mmdLoadFile(fp);
root = mmdLoadFile(root, fp);

/*
* Close and return...
*/

fclose(fp);

return (doc);
return (root);
}


Expand All @@ -496,7 +484,8 @@ mmdLoad(const char *filename) /* I - File to load */
*/

mmd_t * /* O - First node in markdown */
mmdLoadFile(FILE *fp) /* I - File to load */
mmdLoadFile(mmd_t *root,
FILE *fp) /* I - File to load */
{
size_t i; /* Looping var */
_mmd_doc_t doc; /* Document */
Expand All @@ -519,14 +508,17 @@ mmdLoadFile(FILE *fp) /* I - File to load */


/*
* Create an empty document...
* Create an empty document as needed...
*/

DEBUG_printf("mmdLoadFile: mmd_options=%d%s%s\n", mmd_options, (mmd_options & MMD_OPTION_METADATA) ? " METADATA" : "", (mmd_options & MMD_OPTION_TABLES) ? " TABLES" : "");

memset(&doc, 0, sizeof(doc));

doc.root = mmd_add(NULL, MMD_TYPE_DOCUMENT, 0, NULL, NULL);
if (root)
doc.root = root;
else
doc.root = mmd_add(NULL, MMD_TYPE_DOCUMENT, 0, NULL, NULL);

if (!doc.root)
return (NULL);
Expand All @@ -545,6 +537,10 @@ mmdLoadFile(FILE *fp) /* I - File to load */
memset(&file, 0, sizeof(file));
file.fp = fp;

#ifdef __clang_analyzer__
memset(line, 0, sizeof(line));
#endif // __clang_analyzer__

while ((lineptr = mmd_read_line(&file, line, sizeof(line))) != NULL)
{
DEBUG_printf("%03d %-12s %s", stackptr->indent, mmd_type_string(stackptr->parent->type) + 9, lineptr);
Expand All @@ -561,7 +557,14 @@ mmdLoadFile(FILE *fp) /* I - File to load */
DEBUG2_printf(" line indent=%d\n", (int)(lineptr - line));
DEBUG2_printf(" stackptr=%d\n", (int)(stackptr - stack));

if (*lineptr == '>' && (lineptr - linestart) < 4)
if (!*lineptr && stackptr->parent->type == MMD_TYPE_TABLE)
{
DEBUG2_puts("END TABLE\n");
stackptr --;
block = NULL;
continue;
}
else if (*lineptr == '>' && (lineptr - linestart) < 4)
{
/*
* Block quote. See if there is an existing blockquote...
Expand Down Expand Up @@ -744,7 +747,7 @@ mmdLoadFile(FILE *fp) /* I - File to load */
stackptr = stack;

mmd_add(stackptr->parent, MMD_TYPE_THEMATIC_BREAK, 0, NULL, NULL);
type = MMD_TYPE_PARAGRAPH;
// type = MMD_TYPE_PARAGRAPH;
block = NULL;
continue;
}
Expand All @@ -766,13 +769,13 @@ mmdLoadFile(FILE *fp) /* I - File to load */
while (stackptr > stack && stackptr->indent > newindent)
stackptr --;

if (stackptr->parent->type == MMD_TYPE_LIST_ITEM && stackptr->indent == newindent)
if (stackptr > stack && stackptr->parent->type == MMD_TYPE_LIST_ITEM && stackptr->indent == newindent)
stackptr --;

if (stackptr->parent->type == MMD_TYPE_ORDERED_LIST && stackptr->indent == newindent)
if (stackptr > stack && stackptr->parent->type == MMD_TYPE_ORDERED_LIST && stackptr->indent == newindent)
stackptr --;

if (stackptr->parent->type == MMD_TYPE_BLOCK_QUOTE && line[0] != '>')
if (stackptr > stack && stackptr->parent->type == MMD_TYPE_BLOCK_QUOTE && line[0] != '>')
stackptr --;

if (stackptr->parent->type != MMD_TYPE_UNORDERED_LIST && stackptr < (stack + sizeof(stack) / sizeof(stack[0]) - 1))
Expand Down Expand Up @@ -1192,6 +1195,67 @@ mmdLoadFile(FILE *fp) /* I - File to load */
}


/*
* 'mmdLoadString()' - Load a markdown string into nodes.
*/

mmd_t * /* O - Root node in markdown */
mmdLoadString(mmd_t *root, /* I - Root node for document or `NULL` for a new document */
const char *s) /* I - String to load */
{
FILE *fp; /* File */


#if _WIN32
/*
* Windows does not provide the POSIX fmemopen, so create a temporary file and
* have MMD read it...
*/

char tempfile[1024]; /* Temporary filename */

if (tempnam_s(tempfile, sizeof(tempfile))
return (root);

if ((fp = fopen(tempfile, "w+")) == NULL)
return (root);

fputs(fp, s);
frewind(fp);

#else // POSIX
/*
* Open the string as a file...
*/

if ((fp = fmemopen((void *)s, strlen(s), "r")) == NULL)
return (root);
#endif // _WIN32

/*
* Load the string...
*/

root = mmdLoadFile(root, fp);

/*
* Close the memory file...
*/

fclose(fp);

#if _WIN32
/*
* Remove the temporary file...
*/

unlink(tempfile);
#endif // _WIN32

return (root);
}


/*
* 'mmdSetOptions()' - Set (enable/disable) support for various markdown options.
*/
Expand Down Expand Up @@ -1552,8 +1616,7 @@ mmd_parse_inline(_mmd_doc_t *doc, /* I - Document */
*lineptr = '\0';
mmd_add(parent, type, whitespace, text, NULL);

text = NULL;
whitespace = 0;
text = NULL;
}

if (!strncmp(lineptr + 1, " \n", 2) && lineptr[3])
Expand Down Expand Up @@ -1602,7 +1665,7 @@ mmd_parse_inline(_mmd_doc_t *doc, /* I - Document */
else if (*lineptr == '[' && type != MMD_TYPE_CODE_TEXT)
{
/*
* Link...
* Link or checkbox...
*/

if (text)
Expand All @@ -1615,38 +1678,48 @@ mmd_parse_inline(_mmd_doc_t *doc, /* I - Document */
whitespace = 0;
}

lineptr = mmd_parse_link(doc, lineptr, &text, &url, &title, &refname);

if (text && *text == '`')
{
char *end = text + strlen(text) - 1;

text ++;
if (end > text && *end == '`')
*end = '\0';

node = mmd_add(parent, MMD_TYPE_CODE_TEXT, whitespace, text, url);
}
else if (text)
if ((mmd_options & MMD_OPTION_TASKS) && (!strncmp(lineptr, "[ ]", 3) || !strncmp(lineptr, "[x]", 3) || !strncmp(lineptr, "[X]", 3)))
{
node = mmd_add(parent, MMD_TYPE_LINKED_TEXT, whitespace, text, url);
if (title)
node->extra = strdup(title);
// Checkbox
mmd_add(parent, MMD_TYPE_CHECKBOX, 0, lineptr[1] == ' ' ? NULL : "x", NULL);
lineptr += 2;
}
else
node = NULL;
{
// Link
lineptr = mmd_parse_link(doc, lineptr, &text, &url, &title, &refname);

DEBUG2_printf("mmd_parse_inline: text=\"%s\", refname=\"%s\", node=%p\n", text, refname, node);
if (text && *text == '`')
{
char *end = text + strlen(text) - 1;

if (refname && node)
mmd_ref_add(doc, node, refname, NULL, title);
text ++;
if (end > text && *end == '`')
*end = '\0';

if (!*lineptr)
return;
node = mmd_add(parent, MMD_TYPE_CODE_TEXT, whitespace, text, url);
}
else if (text)
{
node = mmd_add(parent, MMD_TYPE_LINKED_TEXT, whitespace, text, url);
if (title)
node->extra = strdup(title);
}
else
node = NULL;

text = url = NULL;
whitespace = 0;
lineptr --;
DEBUG2_printf("mmd_parse_inline: text=\"%s\", refname=\"%s\", node=%p\n", text, refname, node);

if (refname && node)
mmd_ref_add(doc, node, refname, NULL, title);

if (!*lineptr)
return;

text = url = NULL;
whitespace = 0;
lineptr --;
}
}
else if (*lineptr == '<' && type != MMD_TYPE_CODE_TEXT && strchr(lineptr + 1, '>'))
{
Expand Down
Loading

0 comments on commit cbf3130

Please sign in to comment.