various fixes for footnotes

api_test/main.c

@@ -1571,6 +1571,49 @@ static void table_spans(test_batch_runner *runner) {
   }
 }
 
+static void footnote_metadata(test_batch_runner *runner) {
+  static const char markdown[] =
+    "this is a test[^test] and also a test[^other]\n"
+    "\n"
+    "[^test]: still a test\n"
+    "\n"
+    "[^other]: lorem ipsum\n";
+  static const char expected[] =
+    "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
+    "<!DOCTYPE document SYSTEM \"CommonMark.dtd\">\n"
+    "<document xmlns=\"http://commonmark.org/xml/1.0\">\n"
+    "  <paragraph>\n"
+    "    <text xml:space=\"preserve\">this is a test</text>\n"
+    "    <footnote_reference id=\"test\" />\n"
+    "    <text xml:space=\"preserve\"> and also a test</text>\n"
+    "    <footnote_reference id=\"other\" />\n"
+    "  </paragraph>\n"
+    "  <footnote_definition id=\"test\">\n"
+    "    <paragraph>\n"
+    "      <text xml:space=\"preserve\">still a test</text>\n"
+    "    </paragraph>\n"
+    "  </footnote_definition>\n"
+    "  <footnote_definition id=\"other\">\n"
+    "    <paragraph>\n"
+    "      <text xml:space=\"preserve\">lorem ipsum</text>\n"
+    "    </paragraph>\n"
+    "  </footnote_definition>\n"
+    "</document>\n";
+
+  cmark_node *doc =
+      cmark_parse_document(markdown, sizeof(markdown) - 1, CMARK_OPT_FOOTNOTES);
+
+  const char *xml = cmark_render_xml(doc, CMARK_OPT_DEFAULT);
+
+  STR_EQ(runner, xml, expected, "footnotes render correctly in XML");
+
+  cmark_node *test_def = cmark_node_nth_child(doc, 1);
+  STR_EQ(runner, cmark_node_get_footnote_id(test_def), "test", "footnote ID parsed properly");
+
+  cmark_node *other_def = cmark_node_nth_child(doc, 2);
+  STR_EQ(runner, cmark_node_get_footnote_id(other_def), "other", "footnote ID parsed properly");
+}
+
 int main() {
   int retval;
   test_batch_runner *runner = test_batch_runner_new();
@@ -1608,6 +1651,7 @@ int main() {
   verify_custom_attributes_node_with_footnote(runner);
   parser_interrupt(runner);
   table_spans(runner);
+  footnote_metadata(runner);
 
   test_print_summary(runner);
   retval = test_ok(runner) ? 0 : 1;

src/blocks.c

@@ -1249,12 +1249,15 @@ static void open_new_blocks(cmark_parser *parser, cmark_node **container,
                parser->options & CMARK_OPT_FOOTNOTES &&
                (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;
 
+      // Allocate the cstr rendering after the length adjustment so that the
+      // length is accurate when `cmark_node_get_literal` is called.
+      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;

src/include/cmark-gfm.h

@@ -368,6 +368,11 @@ CMARK_GFM_EXPORT const char *cmark_node_get_literal(cmark_node *node);
  */
 CMARK_GFM_EXPORT int cmark_node_get_backtick_count(cmark_node *node);
 
+/** If 'node' is a footnote reference or footnote definition, reutrns the
+    string ID of that footnote, otherwise returns NULL.
+ */
+CMARK_GFM_EXPORT const char *cmark_node_get_footnote_id(cmark_node *node);
+
 /** Sets the string contents of 'node'.  Returns 1 on success,
  * 0 on failure.
  */

src/node.c

@@ -313,6 +313,10 @@ const char *cmark_node_get_type_string(cmark_node *node) {
     return "link";
   case CMARK_NODE_IMAGE:
     return "image";
+  case CMARK_NODE_FOOTNOTE_REFERENCE:
+    return "footnote_reference";
+  case CMARK_NODE_FOOTNOTE_DEFINITION:
+    return "footnote_definition";
   case CMARK_NODE_ATTRIBUTE:
     return "attribute";
   }
@@ -458,6 +462,24 @@ int cmark_node_set_literal(cmark_node *node, const char *content) {
   return 0;
 }
 
+const char *cmark_node_get_footnote_id(cmark_node *node) {
+  if (node == NULL)
+    return NULL;
+
+  switch (node->type) {
+  case CMARK_NODE_FOOTNOTE_DEFINITION:
+    return cmark_chunk_to_cstr(NODE_MEM(node), &node->as.literal);
+
+  case CMARK_NODE_FOOTNOTE_REFERENCE:
+    return cmark_chunk_to_cstr(NODE_MEM(node->parent_footnote_def), &node->parent_footnote_def->as.literal);
+
+  default:
+    break;
+  }
+
+  return NULL;
+}
+
 const char *cmark_node_get_string_content(cmark_node *node) {
   return (char *) node->content.ptr;
 }

src/xml.c

@@ -137,6 +137,16 @@ static int S_render_node(cmark_node *node, cmark_event_type ev_type,
     case CMARK_NODE_ATTRIBUTE:
       // TODO
       break;
+    case CMARK_NODE_FOOTNOTE_DEFINITION:
+      cmark_strbuf_puts(xml, " id=\"");
+      escape_xml(xml, node->as.literal.data, node->as.literal.len);
+      cmark_strbuf_putc(xml, '"');
+      break;
+    case CMARK_NODE_FOOTNOTE_REFERENCE:
+      cmark_strbuf_puts(xml, " id=\"");
+      escape_xml(xml, node->parent_footnote_def->as.literal.data,
+                 node->parent_footnote_def->as.literal.len);
+      cmark_strbuf_putc(xml, '"');
     default:
       break;
     }