--full-info-string

Makefile

@@ -1,6 +1,11 @@
 C_SOURCES = $(wildcard ext/commonmarker/*.[ch])
 
-update-c-sources: $(C_SOURCES)
+update-c-sources: build-upstream $(C_SOURCES)
+
+.PHONY: build-upstream
+
+build-upstream:
+	cd ext/commonmarker/cmark-upstream && make
 
 ext/commonmarker/%: ext/commonmarker/cmark-upstream/src/%
 	cp $< $@

README.md

@@ -140,15 +140,16 @@ CommonMarker accepts the same options that CMark does, as symbols. Note that the
 
 ### Render options
 
-| Name               | Description
-| ------------------ | -----------
-| `:DEFAULT`         | The default rendering system.
-| `:GITHUB_PRE_LANG` | Use GitHub-style `<pre lang>` for fenced code blocks.
-| `:HARDBREAKS`      | Treat `\n` as hardbreaks (by adding `<br/>`).
-| `:NOBREAKS`        | Translate `\n` in the source to a single whitespace.
-| `:SAFE`            | Suppress raw HTML and unsafe links.
-| `:SOURCEPOS`       | Include source position in rendered HTML.
-| `:TABLE_PREFER_STYLE_ATTRIBUTES` | Use `style` insted of `align` for table cells
+| Name                             | Description                                                    |
+| ------------------               | -----------                                                    |
+| `:DEFAULT`                       | The default rendering system.                                  |
+| `:GITHUB_PRE_LANG`               | Use GitHub-style `<pre lang>` for fenced code blocks.          |
+| `:HARDBREAKS`                    | Treat `\n` as hardbreaks (by adding `<br/>`).                  |
+| `:NOBREAKS`                      | Translate `\n` in the source to a single whitespace.           |
+| `:SAFE`                          | Suppress raw HTML and unsafe links.                            |
+| `:SOURCEPOS`                     | Include source position in rendered HTML.                      |
+| `:TABLE_PREFER_STYLE_ATTRIBUTES` | Use `style` insted of `align` for table cells                  |
+| `:FULL_INFO_STRING`              | Include full info strings of code blocks in separate attribute |
 
 ### Passing options
 

ext/commonmarker/blocks.c

@@ -755,22 +755,24 @@ static void S_find_first_nonspace(cmark_parser *parser, cmark_chunk *input) {
   char c;
   int chars_to_tab = TAB_STOP - (parser->column % TAB_STOP);
 
-  parser->first_nonspace = parser->offset;
-  parser->first_nonspace_column = parser->column;
-  while ((c = peek_at(input, parser->first_nonspace))) {
-    if (c == ' ') {
-      parser->first_nonspace += 1;
-      parser->first_nonspace_column += 1;
-      chars_to_tab = chars_to_tab - 1;
-      if (chars_to_tab == 0) {
+  if (parser->first_nonspace <= parser->offset) {
+    parser->first_nonspace = parser->offset;
+    parser->first_nonspace_column = parser->column;
+    while ((c = peek_at(input, parser->first_nonspace))) {
+      if (c == ' ') {
+        parser->first_nonspace += 1;
+        parser->first_nonspace_column += 1;
+        chars_to_tab = chars_to_tab - 1;
+        if (chars_to_tab == 0) {
+          chars_to_tab = TAB_STOP;
+        }
+      } else if (c == '\t') {
+        parser->first_nonspace += 1;
+        parser->first_nonspace_column += chars_to_tab;
         chars_to_tab = TAB_STOP;
+      } else {
+        break;
       }
-    } else if (c == '\t') {
-      parser->first_nonspace += 1;
-      parser->first_nonspace_column += chars_to_tab;
-      chars_to_tab = TAB_STOP;
-    } else {
-      break;
     }
   }
 
@@ -1137,6 +1139,7 @@ static void open_new_blocks(cmark_parser *parser, cmark_node **container,
 
       (*container)->internal_offset = matched;
     } else if ((!indented || cont_type == CMARK_NODE_LIST) &&
+	       parser->indent < 4 &&
                (matched = parse_list_marker(
                     parser->mem, input, parser->first_nonspace,
                     (*container)->type == CMARK_NODE_PARAGRAPH, &data))) {
@@ -1372,6 +1375,9 @@ static void S_process_line(cmark_parser *parser, const unsigned char *buffer,
 
   parser->offset = 0;
   parser->column = 0;
+  parser->first_nonspace = 0;
+  parser->first_nonspace_column = 0;
+  parser->indent = 0;
   parser->blank = false;
   parser->partially_consumed_tab = false;
 

ext/commonmarker/cmark.h

@@ -733,6 +733,11 @@ char *cmark_render_latex_with_mem(cmark_node *root, int options, int width, cmar
  */
 #define CMARK_OPT_TABLE_PREFER_STYLE_ATTRIBUTES (1 << 15)
 
+/** Include the remainder of the info string in code blocks in
+ * a separate attribute.
+ */
+#define CMARK_OPT_FULL_INFO_STRING (1 << 16)
+
 /**
  * ## Version information
  */

ext/commonmarker/cmark_extension_api.h

@@ -5,7 +5,7 @@
 extern "C" {
 #endif
 
-#include <cmark.h>
+#include "cmark.h"
 
 struct cmark_renderer;
 struct cmark_html_renderer;

ext/commonmarker/core-extensions.h

@@ -5,7 +5,7 @@
 extern "C" {
 #endif
 
-#include <cmark_extension_api.h>
+#include "cmark_extension_api.h"
 #include "cmarkextensions_export.h"
 #include <stdint.h>
 

ext/commonmarker/html.c

@@ -203,12 +203,20 @@ static int S_render_node(cmark_html_renderer *renderer, cmark_node *node,
         cmark_html_render_sourcepos(node, html, options);
         cmark_strbuf_puts(html, " lang=\"");
         escape_html(html, node->as.code.info.data, first_tag);
+        if (first_tag < node->as.code.info.len && (options & CMARK_OPT_FULL_INFO_STRING)) {
+          cmark_strbuf_puts(html, "\" data-meta=\"");
+          escape_html(html, node->as.code.info.data + first_tag + 1, node->as.code.info.len - first_tag - 1);
+        }
         cmark_strbuf_puts(html, "\"><code>");
       } else {
         cmark_strbuf_puts(html, "<pre");
         cmark_html_render_sourcepos(node, html, options);
         cmark_strbuf_puts(html, "><code class=\"language-");
         escape_html(html, node->as.code.info.data, first_tag);
+        if (first_tag < node->as.code.info.len && (options & CMARK_OPT_FULL_INFO_STRING)) {
+          cmark_strbuf_puts(html, "\" data-meta=\"");
+          escape_html(html, node->as.code.info.data + first_tag + 1, node->as.code.info.len - first_tag - 1);
+        }
         cmark_strbuf_puts(html, "\">");
       }
     }

ext/commonmarker/inlines.c

@@ -972,7 +972,7 @@ static bufsize_t manual_scan_link_url(cmark_chunk *input, bufsize_t offset,
         break;
       } else if (input->data[i] == '\\')
         i += 2;
-      else if (cmark_isspace(input->data[i]) || input->data[i] == '<')
+      else if (input->data[i] == '\n' || input->data[i] == '<')
         return manual_scan_link_url_2(input, offset, output);
       else
         ++i;