commonmarker 0.17.13 → 0.23.4

data/ext/commonmarker/scanners.h

@@ -26,7 +26,6 @@ bufsize_t _scan_link_title(const unsigned char *p);
 bufsize_t _scan_spacechars(const unsigned char *p);
 bufsize_t _scan_atx_heading_start(const unsigned char *p);
 bufsize_t _scan_setext_heading_line(const unsigned char *p);
-bufsize_t _scan_thematic_break(const unsigned char *p);
 bufsize_t _scan_open_code_fence(const unsigned char *p);
 bufsize_t _scan_close_code_fence(const unsigned char *p);
 bufsize_t _scan_entity(const unsigned char *p);
@@ -50,7 +49,6 @@ bufsize_t _scan_footnote_definition(const unsigned char *p);
 #define scan_atx_heading_start(c, n) _scan_at(&_scan_atx_heading_start, c, n)
 #define scan_setext_heading_line(c, n)                                         \
   _scan_at(&_scan_setext_heading_line, c, n)
-#define scan_thematic_break(c, n) _scan_at(&_scan_thematic_break, c, n)
 #define scan_open_code_fence(c, n) _scan_at(&_scan_open_code_fence, c, n)
 #define scan_close_code_fence(c, n) _scan_at(&_scan_close_code_fence, c, n)
 #define scan_entity(c, n) _scan_at(&_scan_entity, c, n)

data/ext/commonmarker/strikethrough.c

@@ -28,7 +28,7 @@ static cmark_node *match(cmark_syntax_extension *self, cmark_parser *parser,
   res->start_column = cmark_inline_parser_get_column(inline_parser) - delims;
 
   if ((left_flanking || right_flanking) &&
-      (!(parser->options & CMARK_OPT_STRIKETHROUGH_DOUBLE_TILDE) || delims == 2)) {
+      (delims == 2 || (!(parser->options & CMARK_OPT_STRIKETHROUGH_DOUBLE_TILDE) && delims == 1))) {
     cmark_inline_parser_push_delimiter(inline_parser, character, left_flanking,
                                        right_flanking, res);
   }
@@ -46,6 +46,9 @@ static delimiter *insert(cmark_syntax_extension *self, cmark_parser *parser,
 
   strikethrough = opener->inl_text;
 
+  if (opener->inl_text->as.literal.len != closer->inl_text->as.literal.len)
+    goto done;
+
   if (!cmark_node_set_type(strikethrough, CMARK_NODE_STRIKETHROUGH))
     goto done;
 

data/ext/commonmarker/syntax_extension.c

@@ -97,6 +97,11 @@ void cmark_syntax_extension_set_latex_render_func(cmark_syntax_extension *extens
   extension->latex_render_func = func;
 }
 
+void cmark_syntax_extension_set_xml_attr_func(cmark_syntax_extension *extension,
+                                              cmark_xml_attr_func func) {
+  extension->xml_attr_func = func;
+}
+
 void cmark_syntax_extension_set_man_render_func(cmark_syntax_extension *extension,
                                                 cmark_common_render_func func) {
   extension->man_render_func = func;
@@ -128,6 +133,11 @@ void *cmark_syntax_extension_get_private(cmark_syntax_extension *extension) {
     return extension->priv;
 }
 
+void cmark_syntax_extension_set_opaque_alloc_func(cmark_syntax_extension *extension,
+                                                  cmark_opaque_alloc_func func) {
+  extension->opaque_alloc_func = func;
+}
+
 void cmark_syntax_extension_set_opaque_free_func(cmark_syntax_extension *extension,
                                                  cmark_opaque_free_func func) {
   extension->opaque_free_func = func;

data/ext/commonmarker/syntax_extension.h

@@ -21,10 +21,12 @@ struct cmark_syntax_extension {
   cmark_common_render_func        commonmark_render_func;
   cmark_common_render_func        plaintext_render_func;
   cmark_common_render_func        latex_render_func;
+  cmark_xml_attr_func             xml_attr_func;
   cmark_common_render_func        man_render_func;
   cmark_html_render_func          html_render_func;
   cmark_html_filter_func          html_filter_func;
   cmark_postprocess_func          postprocess_func;
+  cmark_opaque_alloc_func         opaque_alloc_func;
   cmark_opaque_free_func          opaque_free_func;
   cmark_commonmark_escape_func    commonmark_escape_func;
 };

data/ext/commonmarker/table.c

@@ -9,12 +9,14 @@
 #include "ext_scanners.h"
 #include "strikethrough.h"
 #include "table.h"
+#include "cmark-gfm-core-extensions.h"
 
 cmark_node_type CMARK_NODE_TABLE, CMARK_NODE_TABLE_ROW,
     CMARK_NODE_TABLE_CELL;
 
 typedef struct {
   uint16_t n_columns;
+  int paragraph_offset;
   cmark_llist *cells;
 } table_row;
 
@@ -112,20 +114,40 @@ static cmark_strbuf *unescape_pipes(cmark_mem *mem, unsigned char *string, bufsi
 static table_row *row_from_string(cmark_syntax_extension *self,
                                   cmark_parser *parser, unsigned char *string,
                                   int len) {
+  // Parses a single table row. It has the following form:
+  // `delim? table_cell (delim table_cell)* delim? newline`
+  // Note that cells are allowed to be empty.
+  //
+  // From the GitHub-flavored Markdown specification:
+  //
+  // > Each row consists of cells containing arbitrary text, in which inlines
+  // > are parsed, separated by pipes (|). A leading and trailing pipe is also
+  // > recommended for clarity of reading, and if there’s otherwise parsing
+  // > ambiguity.
+
   table_row *row = NULL;
-  bufsize_t cell_matched, pipe_matched, offset;
+  bufsize_t cell_matched = 1, pipe_matched = 1, offset;
+  int expect_more_cells = 1;
+  int row_end_offset = 0;
+  int int_overflow_abort = 0;
 
   row = (table_row *)parser->mem->calloc(1, sizeof(table_row));
   row->n_columns = 0;
   row->cells = NULL;
 
+  // Scan past the (optional) leading pipe.
   offset = scan_table_cell_end(string, len, 0);
 
-  do {
+  // Parse the cells of the row. Stop if we reach the end of the input, or if we
+  // cannot detect any more cells.
+  while (offset < len && expect_more_cells) {
     cell_matched = scan_table_cell(string, len, offset);
     pipe_matched = scan_table_cell_end(string, len, offset + cell_matched);
 
     if (cell_matched || pipe_matched) {
+      // We are guaranteed to have a cell, since (1) either we found some
+      // content and cell_matched, or (2) we found an empty cell followed by a
+      // pipe.
       cmark_strbuf *cell_buf = unescape_pipes(parser->mem, string + offset,
           cell_matched);
       cmark_strbuf_trim(cell_buf);
@@ -134,23 +156,52 @@ static table_row *row_from_string(cmark_syntax_extension *self,
       cell->buf = cell_buf;
       cell->start_offset = offset;
       cell->end_offset = offset + cell_matched - 1;
+
       while (cell->start_offset > 0 && string[cell->start_offset - 1] != '|') {
         --cell->start_offset;
         ++cell->internal_offset;
       }
+
+      // make sure we never wrap row->n_columns
+      // offset will != len and our exit will clean up as intended
+      if (row->n_columns == UINT16_MAX) {
+          int_overflow_abort = 1;
+          break;
+      }
       row->n_columns += 1;
       row->cells = cmark_llist_append(parser->mem, row->cells, cell);
     }
 
     offset += cell_matched + pipe_matched;
 
-    if (!pipe_matched) {
-      pipe_matched = scan_table_row_end(string, len, offset);
-      offset += pipe_matched;
+    if (pipe_matched) {
+      expect_more_cells = 1;
+    } else {
+      // We've scanned the last cell. Check if we have reached the end of the row
+      row_end_offset = scan_table_row_end(string, len, offset);
+      offset += row_end_offset;
+
+      // If the end of the row is not the end of the input,
+      // the row is not a real row but potentially part of the paragraph
+      // preceding the table.
+      if (row_end_offset && offset != len) {
+        row->paragraph_offset = offset;
+
+        cmark_llist_free_full(parser->mem, row->cells, (cmark_free_func)free_table_cell);
+        row->cells = NULL;
+        row->n_columns = 0;
+
+        // Scan past the (optional) leading pipe.
+        offset += scan_table_cell_end(string, len, offset);
+
+        expect_more_cells = 1;
+      } else {
+        expect_more_cells = 0;
+      }
     }
-  } while ((cell_matched || pipe_matched) && offset < len);
+  }
 
-  if (offset != len || !row->n_columns) {
+  if (offset != len || row->n_columns == 0 || int_overflow_abort) {
     free_table_row(parser->mem, row);
     row = NULL;
   }
@@ -158,12 +209,30 @@ static table_row *row_from_string(cmark_syntax_extension *self,
   return row;
 }
 
+static void try_inserting_table_header_paragraph(cmark_parser *parser,
+                                                 cmark_node *parent_container,
+                                                 unsigned char *parent_string,
+                                                 int paragraph_offset) {
+  cmark_node *paragraph;
+  cmark_strbuf *paragraph_content;
+
+  paragraph = cmark_node_new_with_mem(CMARK_NODE_PARAGRAPH, parser->mem);
+
+  paragraph_content = unescape_pipes(parser->mem, parent_string, paragraph_offset);
+  cmark_strbuf_trim(paragraph_content);
+  cmark_node_set_string_content(paragraph, (char *) paragraph_content->ptr);
+  cmark_strbuf_free(paragraph_content);
+  parser->mem->free(paragraph_content);
+
+  if (!cmark_node_insert_before(parent_container, paragraph)) {
+    parser->mem->free(paragraph);
+  }
+}
+
 static cmark_node *try_opening_table_header(cmark_syntax_extension *self,
                                             cmark_parser *parser,
                                             cmark_node *parent_container,
                                             unsigned char *input, int len) {
-  bufsize_t matched =
-      scan_table_start(input, len, cmark_parser_get_first_nonspace(parser));
   cmark_node *table_header;
   table_row *header_row = NULL;
   table_row *marker_row = NULL;
@@ -171,41 +240,48 @@ static cmark_node *try_opening_table_header(cmark_syntax_extension *self,
   const char *parent_string;
   uint16_t i;
 
-  if (!matched)
-    return parent_container;
-
-  parent_string = cmark_node_get_string_content(parent_container);
-
-  cmark_arena_push();
-
-  header_row = row_from_string(self, parser, (unsigned char *)parent_string,
-                               (int)strlen(parent_string));
-
-  if (!header_row) {
-    free_table_row(parser->mem, header_row);
-    cmark_arena_pop();
+  if (!scan_table_start(input, len, cmark_parser_get_first_nonspace(parser))) {
     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));
-
+  // assert may be optimized out, don't rely on it for security boundaries
+  if (!marker_row) {
+      return parent_container;
+  }
+  
   assert(marker_row);
 
-  if (header_row->n_columns != marker_row->n_columns) {
-    free_table_row(parser->mem, header_row);
+  cmark_arena_push();
+
+  // Check for a matching header row. We call `row_from_string` with the entire
+  // (potentially long) parent container as input, but this should be safe since
+  // `row_from_string` bails out early if it does not find a row.
+  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);
+    free_table_row(parser->mem, header_row);
     cmark_arena_pop();
     return parent_container;
   }
 
   if (cmark_arena_pop()) {
+    marker_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));
-    marker_row = row_from_string(self, parser,
-                                 input + cmark_parser_get_first_nonspace(parser),
-                                 len - cmark_parser_get_first_nonspace(parser));
+    // 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);
+        free_table_row(parser->mem, header_row);
+        return parent_container;
+    }
   }
 
   if (!cmark_node_set_type(parent_container, CMARK_NODE_TABLE)) {
@@ -214,14 +290,19 @@ static cmark_node *try_opening_table_header(cmark_syntax_extension *self,
     return parent_container;
   }
 
-  cmark_node_set_syntax_extension(parent_container, self);
+  if (header_row->paragraph_offset) {
+    try_inserting_table_header_paragraph(parser, parent_container, (unsigned char *)parent_string,
+                                         header_row->paragraph_offset);
+  }
 
+  cmark_node_set_syntax_extension(parent_container, 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
   uint8_t *alignments =
-      (uint8_t *)parser->mem->calloc(header_row->n_columns, sizeof(uint8_t));
+      (uint8_t *)parser->mem->calloc(marker_row->n_columns, sizeof(uint8_t));
   cmark_llist *it = marker_row->cells;
   for (i = 0; it; it = it->next, ++i) {
     node_cell *node = (node_cell *)it->data;
@@ -290,6 +371,12 @@ static cmark_node *try_opening_table_row(cmark_syntax_extension *self,
   row = row_from_string(self, parser, input + cmark_parser_get_first_nonspace(parser),
       len - cmark_parser_get_first_nonspace(parser));
 
+  if (!row) {
+      // clean up the dangling node
+      cmark_node_free(table_row_block);
+      return NULL;
+  }
+
   {
     cmark_llist *tmp;
     int i, table_columns = get_n_table_columns(parent_container);
@@ -488,6 +575,27 @@ static void latex_render(cmark_syntax_extension *extension,
   }
 }
 
+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]) {
+      case 'l': return " align=\"left\"";
+      case 'c': return " align=\"center\"";
+      case 'r': return " align=\"right\"";
+      }
+    }
+  }
+
+  return NULL;
+}
+
 static void man_render(cmark_syntax_extension *extension,
                        cmark_renderer *renderer, cmark_node *node,
                        cmark_event_type ev_type, int options) {
@@ -648,6 +756,16 @@ static void html_render(cmark_syntax_extension *extension,
   }
 }
 
+static void opaque_alloc(cmark_syntax_extension *self, cmark_mem *mem, cmark_node *node) {
+  if (node->type == CMARK_NODE_TABLE) {
+    node->as.opaque = mem->calloc(1, sizeof(node_table));
+  } else if (node->type == CMARK_NODE_TABLE_ROW) {
+    node->as.opaque = mem->calloc(1, sizeof(node_table_row));
+  } else if (node->type == CMARK_NODE_TABLE_CELL) {
+    node->as.opaque = mem->calloc(1, sizeof(node_cell));
+  }
+}
+
 static void opaque_free(cmark_syntax_extension *self, cmark_mem *mem, cmark_node *node) {
   if (node->type == CMARK_NODE_TABLE) {
     free_node_table(mem, node->as.opaque);
@@ -675,8 +793,10 @@ cmark_syntax_extension *create_table_extension(void) {
   cmark_syntax_extension_set_commonmark_render_func(self, commonmark_render);
   cmark_syntax_extension_set_plaintext_render_func(self, commonmark_render);
   cmark_syntax_extension_set_latex_render_func(self, latex_render);
+  cmark_syntax_extension_set_xml_attr_func(self, xml_attr);
   cmark_syntax_extension_set_man_render_func(self, man_render);
   cmark_syntax_extension_set_html_render_func(self, html_render);
+  cmark_syntax_extension_set_opaque_alloc_func(self, opaque_alloc);
   cmark_syntax_extension_set_opaque_free_func(self, opaque_free);
   cmark_syntax_extension_set_commonmark_escape_func(self, escape);
   CMARK_NODE_TABLE = cmark_syntax_extension_add_node(0);
@@ -699,3 +819,30 @@ uint8_t *cmark_gfm_extensions_get_table_alignments(cmark_node *node) {
 
   return ((node_table *)node->as.opaque)->alignments;
 }
+
+int cmark_gfm_extensions_set_table_columns(cmark_node *node, uint16_t n_columns) {
+  return set_n_table_columns(node, n_columns);
+}
+
+int cmark_gfm_extensions_set_table_alignments(cmark_node *node, uint16_t ncols, uint8_t *alignments) {
+  uint8_t *a = (uint8_t *)cmark_node_mem(node)->calloc(1, ncols);
+  memcpy(a, alignments, ncols);
+  return set_table_alignments(node, a);
+}
+
+int cmark_gfm_extensions_get_table_row_is_header(cmark_node *node)
+{
+  if (!node || node->type != CMARK_NODE_TABLE_ROW)
+    return 0;
+
+  return ((node_table_row *)node->as.opaque)->is_header;
+}
+
+int cmark_gfm_extensions_set_table_row_is_header(cmark_node *node, int is_header)
+{
+  if (!node || node->type != CMARK_NODE_TABLE_ROW)
+    return 0;
+
+  ((node_table_row *)node->as.opaque)->is_header = (is_header != 0);
+  return 1;
+}

data/ext/commonmarker/tasklist.c

@@ -0,0 +1,156 @@
+#include "tasklist.h"
+#include <parser.h>
+#include <render.h>
+#include <html.h>
+#include "ext_scanners.h"
+
+typedef enum {
+  CMARK_TASKLIST_NOCHECKED,
+  CMARK_TASKLIST_CHECKED,
+} cmark_tasklist_type;
+
+// Local constants
+static const char *TYPE_STRING = "tasklist";
+
+static const char *get_type_string(cmark_syntax_extension *extension, cmark_node *node) {
+  return TYPE_STRING;
+}
+
+
+// Return 1 if state was set, 0 otherwise
+int cmark_gfm_extensions_set_tasklist_item_checked(cmark_node *node, bool is_checked) {
+  // The node has to exist, and be an extension, and actually be the right type in order to get the value.
+  if (!node || !node->extension || strcmp(cmark_node_get_type_string(node), TYPE_STRING))
+    return 0;
+
+  node->as.list.checked = is_checked;
+  return 1;
+}
+
+bool cmark_gfm_extensions_get_tasklist_item_checked(cmark_node *node) {
+  if (!node || !node->extension || strcmp(cmark_node_get_type_string(node), TYPE_STRING))
+    return false;
+
+  if (node->as.list.checked) {
+    return true;
+  }
+  else {
+    return false;
+  }
+}
+
+static bool parse_node_item_prefix(cmark_parser *parser, const char *input,
+                                   cmark_node *container) {
+  bool res = false;
+
+  if (parser->indent >=
+      container->as.list.marker_offset + container->as.list.padding) {
+    cmark_parser_advance_offset(parser, input, container->as.list.marker_offset +
+                                        container->as.list.padding,
+                     true);
+    res = true;
+  } else if (parser->blank && container->first_child != NULL) {
+    // if container->first_child is NULL, then the opening line
+    // of the list item was blank after the list marker; in this
+    // case, we are done with the list item.
+    cmark_parser_advance_offset(parser, input, parser->first_nonspace - parser->offset,
+                     false);
+    res = true;
+  }
+  return res;
+}
+
+static int matches(cmark_syntax_extension *self, cmark_parser *parser,
+                   unsigned char *input, int len,
+                   cmark_node *parent_container) {
+  return parse_node_item_prefix(parser, (const char*)input, parent_container);
+}
+
+static int can_contain(cmark_syntax_extension *extension, cmark_node *node,
+                       cmark_node_type child_type) {
+  return (node->type == CMARK_NODE_ITEM) ? 1 : 0;
+}
+
+static cmark_node *open_tasklist_item(cmark_syntax_extension *self,
+                                      int indented, cmark_parser *parser,
+                                      cmark_node *parent_container,
+                                      unsigned char *input, int len) {
+  cmark_node_type node_type = cmark_node_get_type(parent_container);
+  if (node_type != CMARK_NODE_ITEM) {
+    return NULL;
+  }
+
+  bufsize_t matched = scan_tasklist(input, len, 0);
+  if (!matched) {
+    return NULL;
+  }
+
+  cmark_node_set_syntax_extension(parent_container, self);
+  cmark_parser_advance_offset(parser, (char *)input, 3, false);
+
+  // Either an upper or lower case X means the task is completed.
+  parent_container->as.list.checked = (strstr((char*)input, "[x]") || strstr((char*)input, "[X]"));
+
+  return NULL;
+}
+
+static void commonmark_render(cmark_syntax_extension *extension,
+                              cmark_renderer *renderer, cmark_node *node,
+                              cmark_event_type ev_type, int options) {
+  bool entering = (ev_type == CMARK_EVENT_ENTER);
+  if (entering) {
+    renderer->cr(renderer);
+    if (node->as.list.checked) {
+      renderer->out(renderer, node, "- [x] ", false, LITERAL);
+    } else {
+      renderer->out(renderer, node, "- [ ] ", false, LITERAL);
+    }
+    cmark_strbuf_puts(renderer->prefix, "  ");
+  } else {
+    cmark_strbuf_truncate(renderer->prefix, renderer->prefix->size - 2);
+    renderer->cr(renderer);
+  }
+}
+
+static void html_render(cmark_syntax_extension *extension,
+                        cmark_html_renderer *renderer, cmark_node *node,
+                        cmark_event_type ev_type, int options) {
+  bool entering = (ev_type == CMARK_EVENT_ENTER);
+  if (entering) {
+    cmark_html_render_cr(renderer->html);
+    cmark_strbuf_puts(renderer->html, "<li");
+    cmark_html_render_sourcepos(node, renderer->html, options);
+    cmark_strbuf_putc(renderer->html, '>');
+    if (node->as.list.checked) {
+      cmark_strbuf_puts(renderer->html, "<input type=\"checkbox\" checked=\"\" disabled=\"\" /> ");
+    } else {
+      cmark_strbuf_puts(renderer->html, "<input type=\"checkbox\" disabled=\"\" /> ");
+    }
+  } else {
+    cmark_strbuf_puts(renderer->html, "</li>\n");
+  }
+}
+
+static const char *xml_attr(cmark_syntax_extension *extension,
+                            cmark_node *node) {
+  if (node->as.list.checked) {
+    return " completed=\"true\"";
+  } else {
+    return " completed=\"false\"";
+  }
+}
+
+cmark_syntax_extension *create_tasklist_extension(void) {
+  cmark_syntax_extension *ext = cmark_syntax_extension_new("tasklist");
+
+  cmark_syntax_extension_set_match_block_func(ext, matches);
+  cmark_syntax_extension_set_get_type_string_func(ext, get_type_string);
+  cmark_syntax_extension_set_open_block_func(ext, open_tasklist_item);
+  cmark_syntax_extension_set_can_contain_func(ext, can_contain);
+  cmark_syntax_extension_set_commonmark_render_func(ext, commonmark_render);
+  cmark_syntax_extension_set_plaintext_render_func(ext, commonmark_render);
+  cmark_syntax_extension_set_html_render_func(ext, html_render);
+  cmark_syntax_extension_set_xml_attr_func(ext, xml_attr);
+
+  return ext;
+}

data/ext/commonmarker/tasklist.h

@@ -0,0 +1,8 @@
+#ifndef TASKLIST_H
+#define TASKLIST_H
+
+#include "cmark-gfm-core-extensions.h"
+
+cmark_syntax_extension *create_tasklist_extension(void);
+
+#endif

data/ext/commonmarker/xml.c

@@ -8,6 +8,7 @@
 #include "node.h"
 #include "buffer.h"
 #include "houdini.h"
+#include "syntax_extension.h"
 
 #define BUFFER_SIZE 100
 
@@ -50,6 +51,12 @@ static int S_render_node(cmark_node *node, cmark_event_type ev_type,
       cmark_strbuf_puts(xml, buffer);
     }
 
+    if (node->extension && node->extension->xml_attr_func) {
+      const char* r = node->extension->xml_attr_func(node->extension, node);
+      if (r != NULL)
+        cmark_strbuf_puts(xml, r);
+    }
+
     literal = false;
 
     switch (node->type) {
@@ -60,7 +67,7 @@ static int S_render_node(cmark_node *node, cmark_event_type ev_type,
     case CMARK_NODE_CODE:
     case CMARK_NODE_HTML_BLOCK:
     case CMARK_NODE_HTML_INLINE:
-      cmark_strbuf_puts(xml, ">");
+      cmark_strbuf_puts(xml, " xml:space=\"preserve\">");
       escape_xml(xml, node->as.literal.data, node->as.literal.len);
       cmark_strbuf_puts(xml, "</");
       cmark_strbuf_puts(xml, cmark_node_get_type_string(node));
@@ -100,7 +107,7 @@ static int S_render_node(cmark_node *node, cmark_event_type ev_type,
         escape_xml(xml, node->as.code.info.data, node->as.code.info.len);
         cmark_strbuf_putc(xml, '"');
       }
-      cmark_strbuf_puts(xml, ">");
+      cmark_strbuf_puts(xml, " xml:space=\"preserve\">");
       escape_xml(xml, node->as.code.literal.data, node->as.code.literal.len);
       cmark_strbuf_puts(xml, "</");
       cmark_strbuf_puts(xml, cmark_node_get_type_string(node));