Update to 0.29.0.gfm.13

CHANGELOG.md

@@ -1,5 +1,13 @@
 # Changelog
 
+## [v0.23.10] (2023-07-31)
+
+- Update GFM release to [`0.29.0.gfm.12`](https://github.com/github/cmark-gfm/releases/tag/0.29.0.gfm.12) and [`0.29.0.gfm.13`](https://github.com/github/cmark-gfm/releases/tag/0.29.0.gfm.13), thereby [fixing a polynomial time complexity security vulnerability](https://github.com/github/cmark-gfm/security/advisories/GHSA-w4qg-3vf7-m9x5).
+- Of note to users of this library, GFM releases `0.29.0.gfm.12` and `0.29.0.gfm.13` also:
+  - Normalized marker row vs. delimiter row nomenclature ([#273](https://github.com/github/cmark-gfm/pull/273))
+  - Exposed CMARK_NODE_FOOTNOTE_DEFINITION literal value ([#336](https://github.com/github/cmark-gfm/pull/336))
+
+
 ## [v0.23.4](https://github.com/gjtorikian/commonmarker/tree/v0.23.4) (2022-03-03)
 
 [Full Changelog](https://github.com/gjtorikian/commonmarker/compare/v0.23.2...v0.23.4)

ext/commonmarker/autolink.c

@@ -296,7 +296,7 @@ static cmark_node *match(cmark_syntax_extension *ext, cmark_parser *parser,
   // inline was finished in inlines.c.
 }
 
-static bool validate_protocol(char protocol[], uint8_t *data, size_t rewind, size_t max_rewind) {
+static bool validate_protocol(const char protocol[], uint8_t *data, size_t rewind, size_t max_rewind) {
   size_t len = strlen(protocol);
 
   if (len > (max_rewind - rewind)) {

ext/commonmarker/blocks.c

@@ -1217,15 +1217,17 @@ static void open_new_blocks(cmark_parser *parser, cmark_node **container,
                              parser->first_nonspace + 1);
       S_advance_offset(parser, input, input->len - 1 - parser->offset, false);
     } else if (!indented &&
-               parser->options & CMARK_OPT_FOOTNOTES &&
+               (parser->options & CMARK_OPT_FOOTNOTES) &&
+               depth < MAX_LIST_DEPTH &&
                (matched = scan_footnote_definition(input, parser->first_nonspace))) {
       cmark_chunk c = cmark_chunk_dup(input, parser->first_nonspace + 2, matched - 2);
-      cmark_chunk_to_cstr(parser->mem, &c);
 
       while (c.data[c.len - 1] != ']')
         --c.len;
       --c.len;
 
+      cmark_chunk_to_cstr(parser->mem, &c);
+
       S_advance_offset(parser, input, parser->first_nonspace + matched - parser->offset, false);
       *container = add_child(parser, *container, CMARK_NODE_FOOTNOTE_DEFINITION, parser->first_nonspace + matched + 1);
       (*container)->as.literal = c;

ext/commonmarker/commonmark.c

@@ -165,7 +165,7 @@ static int S_render_node(cmark_renderer *renderer, cmark_node *node,
   char fencechar[2] = {'\0', '\0'};
   size_t info_len, code_len;
   char listmarker[LISTMARKER_SIZE];
-  char *emph_delim;
+  const char *emph_delim;
   bool first_in_list_item;
   bufsize_t marker_width;
   bool allow_wrap = renderer->width > 0 && !(CMARK_OPT_NOBREAKS & options) &&

ext/commonmarker/node.c

@@ -377,6 +377,7 @@ const char *cmark_node_get_literal(cmark_node *node) {
   case CMARK_NODE_HTML_INLINE:
   case CMARK_NODE_CODE:
   case CMARK_NODE_FOOTNOTE_REFERENCE:
+  case CMARK_NODE_FOOTNOTE_DEFINITION:
     return cmark_chunk_to_cstr(NODE_MEM(node), &node->as.literal);
 
   case CMARK_NODE_CODE_BLOCK:

ext/commonmarker/node.h

@@ -105,6 +105,7 @@ struct cmark_node {
     cmark_link link;
     cmark_custom custom;
     int html_block_type;
+    int cell_index; // For keeping track of TABLE_CELL table alignments
     void *opaque;
   } as;
 };

ext/commonmarker/table.c

@@ -11,6 +11,9 @@
 #include "table.h"
 #include "cmark-gfm-core-extensions.h"
 
+// Limit to prevent a malicious input from causing a denial of service.
+#define MAX_AUTOCOMPLETED_CELLS 0x80000
+
 // Custom node flag, initialized in `create_table_extension`.
 static cmark_node_internal_flags CMARK_NODE__TABLE_VISITED;
 
@@ -31,6 +34,8 @@ typedef struct {
 typedef struct {
   uint16_t n_columns;
   uint8_t *alignments;
+  int n_rows;
+  int n_nonempty_cells;
 } node_table;
 
 typedef struct {
@@ -83,6 +88,33 @@ static int set_n_table_columns(cmark_node *node, uint16_t n_columns) {
   return 1;
 }
 
+// Increment the number of rows in the table. Also update n_nonempty_cells,
+// which keeps track of the number of cells which were parsed from the
+// input file. (If one of the rows is too short, then the trailing cells
+// are autocompleted. Autocompleted cells are not counted in n_nonempty_cells.)
+// The purpose of this is to prevent a malicious input from generating a very
+// large number of autocompleted cells, which could cause a denial of service
+// vulnerability.
+static int incr_table_row_count(cmark_node *node, int i) {
+  if (!node || node->type != CMARK_NODE_TABLE) {
+    return 0;
+  }
+
+  ((node_table *)node->as.opaque)->n_rows++;
+  ((node_table *)node->as.opaque)->n_nonempty_cells += i;
+  return 1;
+}
+
+// Calculate the number of autocompleted cells.
+static int get_n_autocompleted_cells(cmark_node *node) {
+  if (!node || node->type != CMARK_NODE_TABLE) {
+    return 0;
+  }
+
+  const node_table *nt = (node_table *)node->as.opaque;
+  return (nt->n_columns * nt->n_rows) - nt->n_nonempty_cells;
+}
+
 static uint8_t *get_table_alignments(cmark_node *node) {
   if (!node || node->type != CMARK_NODE_TABLE)
     return 0;
@@ -98,6 +130,23 @@ static int set_table_alignments(cmark_node *node, uint8_t *alignments) {
   return 1;
 }
 
+static uint8_t get_cell_alignment(cmark_node *node) {
+  if (!node || node->type != CMARK_NODE_TABLE_CELL)
+    return 0;
+
+  const uint8_t *alignments = get_table_alignments(node->parent->parent);
+  int i = node->as.cell_index;
+  return alignments[i];
+}
+
+static int set_cell_index(cmark_node *node, int i) {
+  if (!node || node->type != CMARK_NODE_TABLE_CELL)
+    return 0;
+
+  node->as.cell_index = i;
+  return 1;
+}
+
 static cmark_strbuf *unescape_pipes(cmark_mem *mem, unsigned char *string, bufsize_t len)
 {
   cmark_strbuf *res = (cmark_strbuf *)mem->calloc(1, sizeof(cmark_strbuf));
@@ -257,7 +306,7 @@ static cmark_node *try_opening_table_header(cmark_syntax_extension *self,
                                             unsigned char *input, int len) {
   cmark_node *table_header;
   table_row *header_row = NULL;
-  table_row *marker_row = NULL;
+  table_row *delimiter_row = NULL;
   node_table_row *ntr;
   const char *parent_string;
   uint16_t i;
@@ -270,16 +319,16 @@ static cmark_node *try_opening_table_header(cmark_syntax_extension *self,
     return parent_container;
   }
 
-  // Since scan_table_start was successful, we must have a marker row.
-  marker_row = row_from_string(self, parser,
-                               input + cmark_parser_get_first_nonspace(parser),
-                               len - cmark_parser_get_first_nonspace(parser));
+  // Since scan_table_start was successful, we must have a delimiter row.
+  delimiter_row = row_from_string(
+    self, parser, input + cmark_parser_get_first_nonspace(parser),
+    len - cmark_parser_get_first_nonspace(parser));
   // assert may be optimized out, don't rely on it for security boundaries
-  if (!marker_row) {
+  if (!delimiter_row) {
       return parent_container;
   }
-  
-  assert(marker_row);
+
+  assert(delimiter_row);
 
   cmark_arena_push();
 
@@ -289,31 +338,31 @@ static cmark_node *try_opening_table_header(cmark_syntax_extension *self,
   parent_string = cmark_node_get_string_content(parent_container);
   header_row = row_from_string(self, parser, (unsigned char *)parent_string,
                                (int)strlen(parent_string));
-  if (!header_row || header_row->n_columns != marker_row->n_columns) {
-    free_table_row(parser->mem, marker_row);
+  if (!header_row || header_row->n_columns != delimiter_row->n_columns) {
+    free_table_row(parser->mem, delimiter_row);
     free_table_row(parser->mem, header_row);
     cmark_arena_pop();
     parent_container->flags |= CMARK_NODE__TABLE_VISITED;
     return parent_container;
   }
 
   if (cmark_arena_pop()) {
-    marker_row = row_from_string(
+    delimiter_row = row_from_string(
         self, parser, input + cmark_parser_get_first_nonspace(parser),
         len - cmark_parser_get_first_nonspace(parser));
     header_row = row_from_string(self, parser, (unsigned char *)parent_string,
                                  (int)strlen(parent_string));
     // row_from_string can return NULL, add additional check to ensure n_columns match
-    if (!marker_row || !header_row || header_row->n_columns != marker_row->n_columns) {
-        free_table_row(parser->mem, marker_row);
+    if (!delimiter_row || !header_row || header_row->n_columns != delimiter_row->n_columns) {
+        free_table_row(parser->mem, delimiter_row);
         free_table_row(parser->mem, header_row);
         return parent_container;
     }
   }
 
   if (!cmark_node_set_type(parent_container, CMARK_NODE_TABLE)) {
     free_table_row(parser->mem, header_row);
-    free_table_row(parser->mem, marker_row);
+    free_table_row(parser->mem, delimiter_row);
     return parent_container;
   }
 
@@ -326,12 +375,12 @@ static cmark_node *try_opening_table_header(cmark_syntax_extension *self,
   parent_container->as.opaque = parser->mem->calloc(1, sizeof(node_table));
   set_n_table_columns(parent_container, header_row->n_columns);
 
-  // allocate alignments based on marker_row->n_columns
-  // since we populate the alignments array based on marker_row->cells
+  // allocate alignments based on delimiter_row->n_columns
+  // since we populate the alignments array based on delimiter_row->cells
   uint8_t *alignments =
-      (uint8_t *)parser->mem->calloc(marker_row->n_columns, sizeof(uint8_t));
-  for (i = 0; i < marker_row->n_columns; ++i) {
-    node_cell *node = &marker_row->cells[i];
+      (uint8_t *)parser->mem->calloc(delimiter_row->n_columns, sizeof(uint8_t));
+  for (i = 0; i < delimiter_row->n_columns; ++i) {
+    node_cell *node = &delimiter_row->cells[i];
     bool left = node->buf->ptr[0] == ':', right = node->buf->ptr[node->buf->size - 1] == ':';
 
     if (left && right)
@@ -353,25 +402,26 @@ static cmark_node *try_opening_table_header(cmark_syntax_extension *self,
   table_header->as.opaque = ntr = (node_table_row *)parser->mem->calloc(1, sizeof(node_table_row));
   ntr->is_header = true;
 
-  {
-    for (i = 0; i < header_row->n_columns; ++i) {
-      node_cell *cell = &header_row->cells[i];
-      cmark_node *header_cell = cmark_parser_add_child(parser, table_header,
-          CMARK_NODE_TABLE_CELL, parent_container->start_column + cell->start_offset);
-      header_cell->start_line = header_cell->end_line = parent_container->start_line;
-      header_cell->internal_offset = cell->internal_offset;
-      header_cell->end_column = parent_container->start_column + cell->end_offset;
-      cmark_node_set_string_content(header_cell, (char *) cell->buf->ptr);
-      cmark_node_set_syntax_extension(header_cell, self);
-    }
+  for (i = 0; i < header_row->n_columns; ++i) {
+    node_cell *cell = &header_row->cells[i];
+    cmark_node *header_cell = cmark_parser_add_child(parser, table_header,
+                                                     CMARK_NODE_TABLE_CELL, parent_container->start_column + cell->start_offset);
+    header_cell->start_line = header_cell->end_line = parent_container->start_line;
+    header_cell->internal_offset = cell->internal_offset;
+    header_cell->end_column = parent_container->start_column + cell->end_offset;
+    cmark_node_set_string_content(header_cell, (char *) cell->buf->ptr);
+    cmark_node_set_syntax_extension(header_cell, self);
+    set_cell_index(header_cell, i);
   }
 
+  incr_table_row_count(parent_container, i);
+
   cmark_parser_advance_offset(
       parser, (char *)input,
       (int)strlen((char *)input) - 1 - cmark_parser_get_offset(parser), false);
 
   free_table_row(parser->mem, header_row);
-  free_table_row(parser->mem, marker_row);
+  free_table_row(parser->mem, delimiter_row);
   return parent_container;
 }
 
@@ -385,6 +435,10 @@ static cmark_node *try_opening_table_row(cmark_syntax_extension *self,
   if (cmark_parser_is_blank(parser))
     return NULL;
 
+  if (get_n_autocompleted_cells(parent_container) > MAX_AUTOCOMPLETED_CELLS) {
+    return NULL;
+  }
+
   table_row_block =
       cmark_parser_add_child(parser, parent_container, CMARK_NODE_TABLE_ROW,
                              parent_container->start_column);
@@ -412,12 +466,16 @@ static cmark_node *try_opening_table_row(cmark_syntax_extension *self,
       node->end_column = parent_container->start_column + cell->end_offset;
       cmark_node_set_string_content(node, (char *) cell->buf->ptr);
       cmark_node_set_syntax_extension(node, self);
+      set_cell_index(node, i);
     }
 
+    incr_table_row_count(parent_container, i);
+
     for (; i < table_columns; ++i) {
       cmark_node *node = cmark_parser_add_child(
           parser, table_row_block, CMARK_NODE_TABLE_CELL, 0);
       cmark_node_set_syntax_extension(node, self);
+      set_cell_index(node, i);
     }
   }
 
@@ -602,13 +660,7 @@ static const char *xml_attr(cmark_syntax_extension *extension,
                             cmark_node *node) {
   if (node->type == CMARK_NODE_TABLE_CELL) {
     if (cmark_gfm_extensions_get_table_row_is_header(node->parent)) {
-      uint8_t *alignments = get_table_alignments(node->parent->parent);
-      int i = 0;
-      cmark_node *n;
-      for (n = node->parent->first_child; n; n = n->next, ++i)
-        if (n == node)
-          break;
-      switch (alignments[i]) {
+      switch (get_cell_alignment(node)) {
       case 'l': return " align=\"left\"";
       case 'c': return " align=\"center\"";
       case 'r': return " align=\"right\"";
@@ -696,7 +748,6 @@ static void html_render(cmark_syntax_extension *extension,
                         cmark_event_type ev_type, int options) {
   bool entering = (ev_type == CMARK_EVENT_ENTER);
   cmark_strbuf *html = renderer->html;
-  cmark_node *n;
 
   // XXX: we just monopolise renderer->opaque.
   struct html_table_state *table_state =
@@ -745,7 +796,6 @@ static void html_render(cmark_syntax_extension *extension,
       }
     }
   } else if (node->type == CMARK_NODE_TABLE_CELL) {
-    uint8_t *alignments = get_table_alignments(node->parent->parent);
     if (entering) {
       cmark_html_render_cr(html);
       if (table_state->in_table_header) {
@@ -754,12 +804,7 @@ static void html_render(cmark_syntax_extension *extension,
         cmark_strbuf_puts(html, "<td");
       }
 
-      int i = 0;
-      for (n = node->parent->first_child; n; n = n->next, ++i)
-        if (n == node)
-          break;
-
-      switch (alignments[i]) {
+      switch (get_cell_alignment(node)) {
       case 'l': html_table_add_align(html, "left", options); break;
       case 'c': html_table_add_align(html, "center", options); break;
       case 'r': html_table_add_align(html, "right", options); break;

lib/commonmarker/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module CommonMarker
-  VERSION = "0.23.9"
+  VERSION = "0.23.10"
 end

script/update_submodules

@@ -1,6 +1,6 @@
 #!/bin/bash
 
-set -e
+set -euo pipefail
 
 if [ -z "$1" ]; then
   BRANCH="main"
@@ -14,7 +14,10 @@ echo "Checking out cmark-upstream"
 echo "---------------------"
 cd ext/commonmarker/cmark-upstream
 git fetch origin
-git checkout $BRANCH && git pull
+# when checking out a tag, for whatever reason the git pull step fails
+# so we comment it out for now:
+# git checkout $BRANCH && git pull
+git checkout $BRANCH
 sha=`git rev-parse HEAD`
 cd ../../..
 make