updated cmark-gfm from 0.29.0.gfm.9 to .13; bumped up version to 1.9.2 (

#30)
r-lib · Jan 30, 2024 · 969e27e · 969e27e
1 parent 89c6ddc
commit 969e27e
Show file tree

Hide file tree

Showing 15 changed files with 251 additions and 151 deletions.
diff --git a/DESCRIPTION b/DESCRIPTION
@@ -1,7 +1,7 @@
 Package: commonmark
 Type: Package
 Title: High Performance CommonMark and Github Markdown Rendering in R
-Version: 1.9.1
+Version: 1.9.2
 Authors@R: c(
     person("Jeroen", "Ooms", ,"[email protected]", role = c("aut", "cre"),
         comment = c(ORCID = "0000-0002-4035-0289")),

diff --git a/NEWS b/NEWS
@@ -1,3 +1,6 @@
+1.9.2
+  - Update libcmark-gfm to 0.29.0.gfm.13
+
 1.9.0
  - Add parameter 'footnotes' to enable footnote parsing
  - Update libcmark-gfm to 0.29.0.gfm.9

diff --git a/src/cmark/blocks.c b/src/cmark/blocks.c
@@ -27,6 +27,14 @@
 #define CODE_INDENT 4
 #define TAB_STOP 4
 
+/**
+ * Very deeply nested lists can cause quadratic performance issues.
+ * This constant is used in open_new_blocks() to limit the nesting
+ * depth. It is unlikely that a non-contrived markdown document will
+ * be nested this deeply.
+ */
+#define MAX_LIST_DEPTH 100
+
 #ifndef MIN
 #define MIN(x, y) ((x < y) ? x : y)
 #endif
@@ -1119,10 +1127,11 @@ static void open_new_blocks(cmark_parser *parser, cmark_node **container,
   bool has_content;
   int save_offset;
   int save_column;
+  size_t depth = 0;
 
   while (cont_type != CMARK_NODE_CODE_BLOCK &&
          cont_type != CMARK_NODE_HTML_BLOCK) {
-
+    depth++;
     S_find_first_nonspace(parser, input);
     indented = parser->indent >= CODE_INDENT;
 
@@ -1208,22 +1217,25 @@ 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;
 
       (*container)->internal_offset = matched;
     } else if ((!indented || cont_type == CMARK_NODE_LIST) &&
 	       parser->indent < 4 &&
+               depth < MAX_LIST_DEPTH &&
                (matched = parse_list_marker(
                     parser->mem, input, parser->first_nonspace,
                     (*container)->type == CMARK_NODE_PARAGRAPH, &data))) {

diff --git a/src/cmark/cmark-gfm.h b/src/cmark/cmark-gfm.h
@@ -413,6 +413,17 @@ CMARK_GFM_EXPORT int cmark_node_get_list_tight(cmark_node *node);
  */
 CMARK_GFM_EXPORT int cmark_node_set_list_tight(cmark_node *node, int tight);
 
+/**
+ * Returns item index of 'node'. This is only used when rendering output
+ * formats such as commonmark, which need to output the index. It is not
+ * required for formats such as html or latex.
+ */
+CMARK_GFM_EXPORT int cmark_node_get_item_index(cmark_node *node);
+
+/** Sets item index of 'node'. Returns 1 on success, 0 on failure.
+ */
+CMARK_GFM_EXPORT int cmark_node_set_item_index(cmark_node *node, int idx);
+
 /** Returns the info string from a fenced code block.
  */
 CMARK_GFM_EXPORT const char *cmark_node_get_fence_info(cmark_node *node);

diff --git a/src/cmark/commonmark.c b/src/cmark/commonmark.c
@@ -153,23 +153,8 @@ static bool is_autolink(cmark_node *node) {
                   link_text->as.literal.len) == 0);
 }
 
-// if node is a block node, returns node.
-// otherwise returns first block-level node that is an ancestor of node.
-// if there is no block-level ancestor, returns NULL.
-static cmark_node *get_containing_block(cmark_node *node) {
-  while (node) {
-    if (CMARK_NODE_BLOCK_P(node)) {
-      return node;
-    } else {
-      node = node->parent;
-    }
-  }
-  return NULL;
-}
-
 static int S_render_node(cmark_renderer *renderer, cmark_node *node,
                          cmark_event_type ev_type, int options) {
-  cmark_node *tmp;
   int list_number;
   cmark_delim_type list_delim;
   int numticks;
@@ -180,7 +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) &&
@@ -189,14 +174,17 @@ static int S_render_node(cmark_renderer *renderer, cmark_node *node,
   // Don't adjust tight list status til we've started the list.
   // Otherwise we loose the blank line between a paragraph and
   // a following list.
-  if (!(node->type == CMARK_NODE_ITEM && node->prev == NULL && entering)) {
-    tmp = get_containing_block(node);
-    renderer->in_tight_list_item =
-        tmp && // tmp might be NULL if there is no containing block
-        ((tmp->type == CMARK_NODE_ITEM &&
-          cmark_node_get_list_tight(tmp->parent)) ||
-         (tmp && tmp->parent && tmp->parent->type == CMARK_NODE_ITEM &&
-          cmark_node_get_list_tight(tmp->parent->parent)));
+  if (entering) {
+    if (node->parent && node->parent->type == CMARK_NODE_ITEM) {
+      renderer->in_tight_list_item = node->parent->parent->as.list.tight;
+    }
+  } else {
+    if (node->type == CMARK_NODE_LIST) {
+      renderer->in_tight_list_item =
+        node->parent &&
+        node->parent->type == CMARK_NODE_ITEM &&
+        node->parent->parent->as.list.tight;
+    }
   }
 
   if (node->extension && node->extension->commonmark_render_func) {
@@ -234,13 +222,8 @@ static int S_render_node(cmark_renderer *renderer, cmark_node *node,
     if (cmark_node_get_list_type(node->parent) == CMARK_BULLET_LIST) {
       marker_width = 4;
     } else {
-      list_number = cmark_node_get_list_start(node->parent);
+      list_number = cmark_node_get_item_index(node);
       list_delim = cmark_node_get_list_delim(node->parent);
-      tmp = node;
-      while (tmp->prev) {
-        tmp = tmp->prev;
-        list_number += 1;
-      }
       // we ensure a width of at least 4 so
       // we get nice transition from single digits
       // to double
@@ -405,10 +388,12 @@ static int S_render_node(cmark_renderer *renderer, cmark_node *node,
     break;
 
   case CMARK_NODE_STRONG:
-    if (entering) {
-      LIT("**");
-    } else {
-      LIT("**");
+    if (node->parent == NULL || node->parent->type != CMARK_NODE_STRONG) {
+      if (entering) {
+        LIT("**");
+      } else {
+        LIT("**");
+      }
     }
     break;
 

diff --git a/src/cmark/html.c b/src/cmark/html.c
@@ -364,10 +364,12 @@ static int S_render_node(cmark_html_renderer *renderer, cmark_node *node,
     break;
 
   case CMARK_NODE_STRONG:
-    if (entering) {
-      cmark_strbuf_puts(html, "<strong>");
-    } else {
-      cmark_strbuf_puts(html, "</strong>");
+    if (node->parent == NULL || node->parent->type != CMARK_NODE_STRONG) {
+      if (entering) {
+        cmark_strbuf_puts(html, "<strong>");
+      } else {
+        cmark_strbuf_puts(html, "</strong>");
+      }
     }
     break;
 

diff --git a/src/cmark/latex.c b/src/cmark/latex.c
@@ -385,10 +385,12 @@ static int S_render_node(cmark_renderer *renderer, cmark_node *node,
     break;
 
   case CMARK_NODE_STRONG:
-    if (entering) {
-      LIT("\\textbf{");
-    } else {
-      LIT("}");
+    if (node->parent == NULL || node->parent->type != CMARK_NODE_STRONG) {
+      if (entering) {
+        LIT("\\textbf{");
+      } else {
+        LIT("}");
+      }
     }
     break;
 

diff --git a/src/cmark/man.c b/src/cmark/man.c
@@ -74,7 +74,6 @@ static void S_outc(cmark_renderer *renderer, cmark_node *node,
 
 static int S_render_node(cmark_renderer *renderer, cmark_node *node,
                          cmark_event_type ev_type, int options) {
-  cmark_node *tmp;
   int list_number;
   bool entering = (ev_type == CMARK_EVENT_ENTER);
   bool allow_wrap = renderer->width > 0 && !(CMARK_OPT_NOBREAKS & options);
@@ -123,12 +122,7 @@ static int S_render_node(cmark_renderer *renderer, cmark_node *node,
       if (cmark_node_get_list_type(node->parent) == CMARK_BULLET_LIST) {
         LIT("\\[bu] 2");
       } else {
-        list_number = cmark_node_get_list_start(node->parent);
-        tmp = node;
-        while (tmp->prev) {
-          tmp = tmp->prev;
-          list_number += 1;
-        }
+        list_number = cmark_node_get_item_index(node);
         char list_number_s[LIST_NUMBER_SIZE];
         snprintf(list_number_s, LIST_NUMBER_SIZE, "\"%d.\" 4", list_number);
         LIT(list_number_s);
@@ -225,10 +219,12 @@ static int S_render_node(cmark_renderer *renderer, cmark_node *node,
     break;
 
   case CMARK_NODE_STRONG:
-    if (entering) {
-      LIT("\\f[B]");
-    } else {
-      LIT("\\f[]");
+    if (node->parent == NULL || node->parent->type != CMARK_NODE_STRONG) {
+      if (entering) {
+        LIT("\\f[B]");
+      } else {
+        LIT("\\f[]");
+      }
     }
     break;
 

diff --git a/src/cmark/node.c b/src/cmark/node.c
@@ -5,6 +5,16 @@
 #include "node.h"
 #include "syntax_extension.h"
 
+/**
+ * Expensive safety checks are off by default, but can be enabled
+ * by calling cmark_enable_safety_checks().
+ */
+static bool enable_safety_checks = false;
+
+void cmark_enable_safety_checks(bool enable) {
+  enable_safety_checks = enable;
+}
+
 static void S_node_unlink(cmark_node *node);
 
 #define NODE_MEM(node) cmark_node_mem(node)
@@ -70,23 +80,23 @@ bool cmark_node_can_contain_type(cmark_node *node, cmark_node_type child_type) {
 }
 
 static bool S_can_contain(cmark_node *node, cmark_node *child) {
-  cmark_node *cur;
-
   if (node == NULL || child == NULL) {
     return false;
   }
   if (NODE_MEM(node) != NODE_MEM(child)) {
     return 0;
   }
 
-  // Verify that child is not an ancestor of node or equal to node.
-  cur = node;
-  do {
-    if (cur == child) {
-      return false;
-    }
-    cur = cur->parent;
-  } while (cur != NULL);
+  if (enable_safety_checks) {
+    // Verify that child is not an ancestor of node or equal to node.
+    cmark_node *cur = node;
+    do {
+      if (cur == child) {
+        return false;
+      }
+      cur = cur->parent;
+    } while (cur != NULL);
+  }
 
   return cmark_node_can_contain_type(node, (cmark_node_type) child->type);
 }
@@ -367,6 +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:
@@ -554,6 +565,31 @@ int cmark_node_set_list_tight(cmark_node *node, int tight) {
   }
 }
 
+int cmark_node_get_item_index(cmark_node *node) {
+  if (node == NULL) {
+    return 0;
+  }
+
+  if (node->type == CMARK_NODE_ITEM) {
+    return node->as.list.start;
+  } else {
+    return 0;
+  }
+}
+
+int cmark_node_set_item_index(cmark_node *node, int idx) {
+  if (node == NULL || idx < 0) {
+    return 0;
+  }
+
+  if (node->type == CMARK_NODE_ITEM) {
+    node->as.list.start = idx;
+    return 1;
+  } else {
+    return 0;
+  }
+}
+
 const char *cmark_node_get_fence_info(cmark_node *node) {
   if (node == NULL) {
     return NULL;

diff --git a/src/cmark/node.h b/src/cmark/node.h
@@ -82,6 +82,14 @@ struct cmark_node {
 
   cmark_syntax_extension *extension;
 
+  /**
+   * Used during cmark_render() to cache the most recent non-NULL
+   * extension, if you go up the parent chain like this:
+   *
+   * node->parent->...parent->extension
+   */
+  cmark_syntax_extension *ancestor_extension;
+
   union {
     int ref_ix;
     int def_count;
@@ -97,6 +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;
 };
@@ -144,6 +153,13 @@ static CMARK_INLINE bool CMARK_NODE_INLINE_P(cmark_node *node) {
 
 CMARK_GFM_EXPORT bool cmark_node_can_contain_type(cmark_node *node, cmark_node_type child_type);
 
+/**
+ * Enable (or disable) extra safety checks. These extra checks cause
+ * extra performance overhead (in some cases quadratic), so they are only
+ * intended to be used during testing.
+ */
+CMARK_GFM_EXPORT void cmark_enable_safety_checks(bool enable);
+
 #ifdef __cplusplus
 }
 #endif