From 937fb16d5868c12b2066ac23ce164b35f91f89cd Mon Sep 17 00:00:00 2001
From: chriszielinski <chrisz@berkeley.edu>
Date: Fri, 26 Apr 2019 10:21:05 -0700
Subject: [PATCH 01/27] Add `cmark_strbuf_remove` that removes a subrange of
 characters.

---
 src/buffer.c |  5 +++++
 src/buffer.h | 10 ++++++++++
 2 files changed, 15 insertions(+)
 mode change 100644 => 100755 src/buffer.c
 mode change 100644 => 100755 src/buffer.h

diff --git a/src/buffer.c b/src/buffer.c
old mode 100644
new mode 100755
index c7934e57d..9c7ecae49
--- a/src/buffer.c
+++ b/src/buffer.c
@@ -242,6 +242,11 @@ void cmark_strbuf_trim(cmark_strbuf *buf) {
   cmark_strbuf_rtrim(buf);
 }
 
+void cmark_strbuf_remove(cmark_strbuf *buf, bufsize_t start_offset, bufsize_t len) {
+  memmove(buf->ptr + start_offset, buf->ptr + start_offset + len, buf->size - (start_offset + len));
+  buf->size -= len;
+}
+
 // Destructively modify string, collapsing consecutive
 // space and newline characters into a single space.
 void cmark_strbuf_normalize_whitespace(cmark_strbuf *s) {
diff --git a/src/buffer.h b/src/buffer.h
old mode 100644
new mode 100755
index b85bb4406..20093c969
--- a/src/buffer.h
+++ b/src/buffer.h
@@ -103,6 +103,16 @@ void cmark_strbuf_rtrim(cmark_strbuf *buf);
 CMARK_GFM_EXPORT
 void cmark_strbuf_trim(cmark_strbuf *buf);
 
+/**
+ Removes the characters in the given range.
+
+ @param buf The string buffer.
+ @param start_offset The starting character offset.
+ @param len The length of characters to remove.
+ */
+CMARK_GFM_EXPORT
+void cmark_strbuf_remove(cmark_strbuf *buf, bufsize_t start_offset, bufsize_t len);
+
 CMARK_GFM_EXPORT
 void cmark_strbuf_normalize_whitespace(cmark_strbuf *s);
 

From 0fea1ddd7230ae020aec99ce23b48423645eb1ac Mon Sep 17 00:00:00 2001
From: chriszielinski <chrisz@berkeley.edu>
Date: Fri, 26 Apr 2019 10:29:53 -0700
Subject: [PATCH 02/27] Fix source positions for inlines inside inconsistently
 indented blocks.

---
 src/blocks.c  | 35 +++++++++++++++++++++++
 src/inlines.c | 78 +++++++++++++++++++++++++++++++++++++++++++++++----
 2 files changed, 107 insertions(+), 6 deletions(-)
 mode change 100644 => 100755 src/blocks.c
 mode change 100644 => 100755 src/inlines.c

diff --git a/src/blocks.c b/src/blocks.c
old mode 100644
new mode 100755
index 53e882f19..0c497fb2e
--- a/src/blocks.c
+++ b/src/blocks.c
@@ -207,8 +207,37 @@ static void add_line(cmark_node *node, cmark_chunk *ch, cmark_parser *parser) {
       cmark_strbuf_putc(&node->content, ' ');
     }
   }
+
+  // If inserting the initial line to the node...
+  if (node->content.size == 0
+    // OR the node is a code block...
+    || node->type == CMARK_NODE_CODE_BLOCK
+    // OR the node is a HTML block.
+    || node->type == CMARK_NODE_HTML_BLOCK) {
+
+    // Then do not insert the leading trivia.
   cmark_strbuf_put(&node->content, ch->data + parser->offset,
                    ch->len - parser->offset);
+  } else {
+    // Special case for maintaining the source position of block quotes
+    // as they can be lazy (i.e. the block quote marker can be omitted).
+    //
+    // The simple solution is to replace any block quote markers (">")
+    // present in the leading trivia with whitespace.
+    //
+    // Note: Using `parser->offset` and not `parser->first_nonspace`
+    // because the latter encompasses the former with the addition of
+    // whitespace (which we are not interested in).
+    assert(parser->offset <= parser->first_nonspace);
+    for (int i = 0; i < parser->offset; i++) {
+      if (peek_at(ch, i) == '>')
+        ch->data[i] = ' ';
+    }
+
+    // Otherwise, do not remove leading trivia for appends (i.e. lines
+    // other than the first).
+    cmark_strbuf_put(&node->content, ch->data, ch->len);
+  }
 }
 
 static void remove_trailing_blank_lines(cmark_strbuf *ln) {
@@ -266,6 +295,12 @@ static bool resolve_reference_link_definitions(
 
     chunk.data += pos;
     chunk.len -= pos;
+
+    // Leading whitespace is not stripped.
+    while (cmark_isspace(peek_at(&chunk, 0))) {
+      chunk.data += 1;
+      chunk.len -= 1;
+    }
   }
   cmark_strbuf_drop(node_content, (node_content->size - chunk.len));
   return !is_blank(&b->content, 0);
diff --git a/src/inlines.c b/src/inlines.c
old mode 100644
new mode 100755
index c21430bde..b90427e7f
--- a/src/inlines.c
+++ b/src/inlines.c
@@ -325,10 +325,10 @@ static bufsize_t scan_to_closing_backticks(subject *subj,
 // spaces, then removing a single leading + trailing space,
 // unless the code span consists entirely of space characters.
 static void S_normalize_code(cmark_strbuf *s) {
-  bufsize_t r, w;
+  bufsize_t r, w, last_char_after_nl;
   bool contains_nonspace = false;
 
-  for (r = 0, w = 0; r < s->size; ++r) {
+  for (r = 0, w = 0, last_char_after_nl = 0; r < s->size; ++r) {
     switch (s->ptr[r]) {
     case '\r':
       if (s->ptr[r + 1] != '\n') {
@@ -337,8 +337,25 @@ static void S_normalize_code(cmark_strbuf *s) {
       break;
     case '\n':
       s->ptr[w++] = ' ';
+      last_char_after_nl = w;
+      break;
+    case ' ':
+      s->ptr[w++] = s->ptr[r];
       break;
     default:
+      if (last_char_after_nl) {
+        // Remove leading whitespace.
+        bufsize_t remove_len = r - last_char_after_nl;
+
+        if (remove_len) {
+          cmark_strbuf_remove(s, last_char_after_nl, remove_len);
+          w -= remove_len;
+          r -= remove_len;
+        }
+
+        last_char_after_nl = 0;
+      }
+
       s->ptr[w++] = s->ptr[r];
     }
     if (s->ptr[r] != ' ') {
@@ -346,6 +363,20 @@ static void S_normalize_code(cmark_strbuf *s) {
     }
   }
 
+  if (last_char_after_nl) {
+    // Remove leading whitespace. Only reach here if the closing backquote
+    // delimiter is on its own line.
+    bufsize_t remove_len = r - last_char_after_nl;
+
+    if (remove_len) {
+      cmark_strbuf_remove(s, last_char_after_nl, remove_len);
+      w -= remove_len;
+      r -= remove_len;
+    }
+
+    last_char_after_nl = 0;
+  }
+
   // begins and ends with space?
   if (contains_nonspace &&
       s->ptr[0] == ' ' && s->ptr[w - 1] == ' ') {
@@ -361,13 +392,15 @@ static void S_normalize_code(cmark_strbuf *s) {
 // Parse backtick code section or raw backticks, return an inline.
 // Assumes that the subject has a backtick at the current position.
 static cmark_node *handle_backticks(subject *subj, int options) {
+  // Save the current source position in case of need to rewind.
+  bufsize_t subjpos = subj->pos;
   cmark_chunk openticks = take_while(subj, isbacktick);
   bufsize_t startpos = subj->pos;
   bufsize_t endpos = scan_to_closing_backticks(subj, openticks.len);
 
   if (endpos == 0) {      // not found
     subj->pos = startpos; // rewind
-    return make_str(subj, subj->pos, subj->pos, openticks);
+    return make_str(subj, subjpos, subjpos, openticks);
   } else {
     cmark_strbuf buf = CMARK_BUF_INIT(subj->mem);
 
@@ -797,6 +830,10 @@ static cmark_node *handle_backslash(cmark_parser *parser, subject *subj) {
     advance(subj);
     return make_str(subj, subj->pos - 2, subj->pos - 1, cmark_chunk_dup(&subj->input, subj->pos - 1, 1));
   } else if (!is_eof(subj) && skip_line_end(subj)) {
+    // Adjust the subject source position state.
+    ++subj->line;
+    subj->column_offset = -subj->pos;
+
     return make_linebreak(subj->mem);
   } else {
     return make_str(subj, subj->pos - 1, subj->pos - 1, cmark_chunk_literal("\\"));
@@ -877,7 +914,8 @@ static cmark_node *handle_pointy_brace(subject *subj, int options) {
     contents = cmark_chunk_dup(&subj->input, subj->pos, matchlen - 1);
     subj->pos += matchlen;
 
-    return make_autolink(subj, subj->pos - 1 - matchlen, subj->pos - 1, contents, 0);
+    return make_autolink(subj, subj->pos + subj->column_offset - 1 - matchlen,
+                         subj->pos + subj->column_offset - 1, contents, 0);
   }
 
   // next try to match an email autolink
@@ -886,7 +924,8 @@ static cmark_node *handle_pointy_brace(subject *subj, int options) {
     contents = cmark_chunk_dup(&subj->input, subj->pos, matchlen - 1);
     subj->pos += matchlen;
 
-    return make_autolink(subj, subj->pos - 1 - matchlen, subj->pos - 1, contents, 1);
+    return make_autolink(subj, subj->pos + subj->column_offset - 1 - matchlen,
+                         subj->pos + subj->column_offset - 1, contents, 1);
   }
 
   // finally, try to match an html tag
@@ -1163,7 +1202,8 @@ static cmark_node *handle_close_bracket(cmark_parser *parser, subject *subj) {
   inl = make_simple(subj->mem, is_image ? CMARK_NODE_IMAGE : CMARK_NODE_LINK);
   inl->as.link.url = url;
   inl->as.link.title = title;
-  inl->start_line = inl->end_line = subj->line;
+  inl->start_line = opener->inl_text->start_line;
+  inl->end_line = subj->line;
   inl->start_column = opener->inl_text->start_column;
   inl->end_column = subj->pos + subj->column_offset + subj->block_offset;
   cmark_node_insert_before(opener->inl_text, inl);
@@ -1304,10 +1344,21 @@ static int parse_inline(cmark_parser *parser, subject *subj, cmark_node *parent,
   cmark_chunk contents;
   unsigned char c;
   bufsize_t startpos, endpos;
+  int saved_block_offset = subj->block_offset;
+
   c = peek_char(subj);
   if (c == 0) {
     return 0;
   }
+
+  // If NOT the subject's initial line...
+  if (subj->column_offset != 0) {
+    // Reset the block offset. The line's leading trivia was not trimmed,
+    // so the source position will be computed appropriately without the
+    // block offset.
+    subj->block_offset = 0;
+  }
+
   switch (c) {
   case '\r':
   case '\n':
@@ -1370,12 +1421,27 @@ static int parse_inline(cmark_parser *parser, subject *subj, cmark_node *parent,
       cmark_chunk_rtrim(&contents);
     }
 
+    // If not the initial line (in the subject) AND at the beginning of another line.
+    if (subj->column_offset != 0 && startpos + subj->column_offset == 0) {
+      // Trim leading whitespace.
+      bufsize_t before_trim = contents.len;
+      cmark_chunk_ltrim(&contents);
+
+      if (contents.len == 0)
+        break; // The contents were only whitespaces.
+
+      // Update the start source position.
+      startpos += before_trim - contents.len;
+    }
+
     new_inl = make_str(subj, startpos, endpos - 1, contents);
   }
   if (new_inl != NULL) {
     cmark_node_append_child(parent, new_inl);
   }
 
+  subj->block_offset = saved_block_offset;
+
   return 1;
 }
 

From e523c7380eec9d6cc4b2a8b253e521ef1a6df91f Mon Sep 17 00:00:00 2001
From: chriszielinski <chrisz@berkeley.edu>
Date: Fri, 26 Apr 2019 10:38:37 -0700
Subject: [PATCH 03/27] Add three additional source position tests.

---
 api_test/main.c | 129 ++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 129 insertions(+)
 mode change 100644 => 100755 api_test/main.c

diff --git a/api_test/main.c b/api_test/main.c
old mode 100644
new mode 100755
index 62006eaa9..bd7dd7450
--- a/api_test/main.c
+++ b/api_test/main.c
@@ -994,6 +994,7 @@ static void test_pathological_regressions(test_batch_runner *runner) {
 }
 
 static void source_pos(test_batch_runner *runner) {
+  {
   static const char markdown[] =
     "# Hi *there*.\n"
     "\n"
@@ -1057,6 +1058,107 @@ static void source_pos(test_batch_runner *runner) {
   free(xml);
   cmark_node_free(doc);
 }
+  {
+    static const char markdown[] =
+    "1.  **Start condition:**  line begins with the string `<script`,\n"
+    "     `<pre`, or `<style` (case-insensitive), followed by whitespace,\n"
+    "the string `>`, or the end of the line.\\\n"
+    "    **End condition:**  line contains an end tag\n"
+    "  `</script>`, `</pre>`, or `</style>` (case-insensitive; it\n"
+    "   need not match the start tag).\n";
+
+    cmark_node *doc = cmark_parse_document(markdown, sizeof(markdown) - 1, CMARK_OPT_DEFAULT);
+    char *xml = cmark_render_xml(doc, CMARK_OPT_DEFAULT | CMARK_OPT_SOURCEPOS);
+    STR_EQ(runner, xml, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
+           "<!DOCTYPE document SYSTEM \"CommonMark.dtd\">\n"
+           "<document sourcepos=\"1:1-6:33\" xmlns=\"http://commonmark.org/xml/1.0\">\n"
+           "  <list sourcepos=\"1:1-6:33\" type=\"ordered\" start=\"1\" delim=\"period\" tight=\"true\">\n"
+           "    <item sourcepos=\"1:1-6:33\">\n"
+           "      <paragraph sourcepos=\"1:5-6:33\">\n"
+           "        <strong sourcepos=\"1:5-1:24\">\n"
+           "          <text sourcepos=\"1:7-1:22\" xml:space=\"preserve\">Start condition:</text>\n"
+           "        </strong>\n"
+           "        <text sourcepos=\"1:25-1:54\" xml:space=\"preserve\">  line begins with the string </text>\n"
+           "        <code sourcepos=\"1:56-1:62\" xml:space=\"preserve\">&lt;script</code>\n"
+           "        <text sourcepos=\"1:64-1:64\" xml:space=\"preserve\">,</text>\n"
+           "        <softbreak />\n"
+           "        <code sourcepos=\"2:7-2:10\" xml:space=\"preserve\">&lt;pre</code>\n"
+           "        <text sourcepos=\"2:12-2:16\" xml:space=\"preserve\">, or </text>\n"
+           "        <code sourcepos=\"2:18-2:23\" xml:space=\"preserve\">&lt;style</code>\n"
+           "        <text sourcepos=\"2:25-2:68\" xml:space=\"preserve\"> (case-insensitive), followed by whitespace,</text>\n"
+           "        <softbreak />\n"
+           "        <text sourcepos=\"3:1-3:11\" xml:space=\"preserve\">the string </text>\n"
+           "        <code sourcepos=\"3:13-3:13\" xml:space=\"preserve\">&gt;</code>\n"
+           "        <text sourcepos=\"3:15-3:39\" xml:space=\"preserve\">, or the end of the line.</text>\n"
+           "        <linebreak />\n"
+           "        <strong sourcepos=\"4:5-4:22\">\n"
+           "          <text sourcepos=\"4:7-4:20\" xml:space=\"preserve\">End condition:</text>\n"
+           "        </strong>\n"
+           "        <text sourcepos=\"4:23-4:48\" xml:space=\"preserve\">  line contains an end tag</text>\n"
+           "        <softbreak />\n"
+           "        <code sourcepos=\"5:4-5:12\" xml:space=\"preserve\">&lt;/script&gt;</code>\n"
+           "        <text sourcepos=\"5:14-5:15\" xml:space=\"preserve\">, </text>\n"
+           "        <code sourcepos=\"5:17-5:22\" xml:space=\"preserve\">&lt;/pre&gt;</code>\n"
+           "        <text sourcepos=\"5:24-5:28\" xml:space=\"preserve\">, or </text>\n"
+           "        <code sourcepos=\"5:30-5:37\" xml:space=\"preserve\">&lt;/style&gt;</code>\n"
+           "        <text sourcepos=\"5:39-5:60\" xml:space=\"preserve\"> (case-insensitive; it</text>\n"
+           "        <softbreak />\n"
+           "        <text sourcepos=\"6:4-6:33\" xml:space=\"preserve\">need not match the start tag).</text>\n"
+           "      </paragraph>\n"
+           "    </item>\n"
+           "  </list>\n"
+           "</document>\n",
+           "list (with EOL backslash) sourcepos are as expected");
+    free(xml);
+    cmark_node_free(doc);
+  }
+  {
+    static const char markdown[] =
+    "> The overriding design goal for Markdown's formatting syntax is\n"
+    "  > to make it as **readable as possible**. The idea is that a\n"
+    "> Markdown-formatted document should be publishable as-is, as\n"
+    "  > plain text, without *looking like* it's been marked up with tags\n"
+    " > or formatting instructions.\n"
+    "> (<http://daringfireball.net/projects/markdown/>)\n";
+
+    cmark_node *doc = cmark_parse_document(markdown, sizeof(markdown) - 1, CMARK_OPT_DEFAULT);
+    char *xml = cmark_render_xml(doc, CMARK_OPT_DEFAULT | CMARK_OPT_SOURCEPOS);
+    STR_EQ(runner, xml, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
+           "<!DOCTYPE document SYSTEM \"CommonMark.dtd\">\n"
+           "<document sourcepos=\"1:1-6:50\" xmlns=\"http://commonmark.org/xml/1.0\">\n"
+           "  <block_quote sourcepos=\"1:1-6:50\">\n"
+           "    <paragraph sourcepos=\"1:3-6:50\">\n"
+           "      <text sourcepos=\"1:3-1:64\" xml:space=\"preserve\">The overriding design goal for Markdown's formatting syntax is</text>\n"
+           "      <softbreak />\n"
+           "      <text sourcepos=\"2:5-2:18\" xml:space=\"preserve\">to make it as </text>\n"
+           "      <strong sourcepos=\"2:19-2:42\">\n"
+           "        <text sourcepos=\"2:21-2:40\" xml:space=\"preserve\">readable as possible</text>\n"
+           "      </strong>\n"
+           "      <text sourcepos=\"2:43-2:62\" xml:space=\"preserve\">. The idea is that a</text>\n"
+           "      <softbreak />\n"
+           "      <text sourcepos=\"3:3-3:61\" xml:space=\"preserve\">Markdown-formatted document should be publishable as-is, as</text>\n"
+           "      <softbreak />\n"
+           "      <text sourcepos=\"4:5-4:24\" xml:space=\"preserve\">plain text, without </text>\n"
+           "      <emph sourcepos=\"4:25-4:38\">\n"
+           "        <text sourcepos=\"4:26-4:37\" xml:space=\"preserve\">looking like</text>\n"
+           "      </emph>\n"
+           "      <text sourcepos=\"4:39-4:68\" xml:space=\"preserve\"> it's been marked up with tags</text>\n"
+           "      <softbreak />\n"
+           "      <text sourcepos=\"5:4-5:30\" xml:space=\"preserve\">or formatting instructions.</text>\n"
+           "      <softbreak />\n"
+           "      <text sourcepos=\"6:3-6:3\" xml:space=\"preserve\">(</text>\n"
+           "      <link sourcepos=\"6:4-6:49\" destination=\"http://daringfireball.net/projects/markdown/\" title=\"\">\n"
+           "        <text sourcepos=\"6:-283-6:-240\" xml:space=\"preserve\">http://daringfireball.net/projects/markdown/</text>\n"
+           "      </link>\n"
+           "      <text sourcepos=\"6:50-6:50\" xml:space=\"preserve\">)</text>\n"
+           "    </paragraph>\n"
+           "  </block_quote>\n"
+           "</document>\n",
+           "inconsistently indented blockquote sourcepos are as expected");
+    free(xml);
+    cmark_node_free(doc);
+  }
+}
 
 static void source_pos_inlines(test_batch_runner *runner) {
   {
@@ -1103,6 +1205,33 @@ static void source_pos_inlines(test_batch_runner *runner) {
     free(xml);
     cmark_node_free(doc);
   }
+  {
+    static const char markdown[] =
+    "This link will have two [soft \n"
+    "line\n"
+    " breaks](https://commonmark.org).";
+
+    cmark_node *doc = cmark_parse_document(markdown, sizeof(markdown) - 1, CMARK_OPT_DEFAULT);
+    char *xml = cmark_render_xml(doc, CMARK_OPT_DEFAULT | CMARK_OPT_SOURCEPOS);
+    STR_EQ(runner, xml, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
+           "<!DOCTYPE document SYSTEM \"CommonMark.dtd\">\n"
+           "<document sourcepos=\"1:1-3:33\" xmlns=\"http://commonmark.org/xml/1.0\">\n"
+           "  <paragraph sourcepos=\"1:1-3:33\">\n"
+           "    <text sourcepos=\"1:1-1:26\" xml:space=\"preserve\">This link will have three </text>\n"
+           "    <link sourcepos=\"1:27-3:32\" destination=\"https://commonmark.org\" title=\"\">\n"
+           "      <text sourcepos=\"1:28-1:32\" xml:space=\"preserve\">soft</text>\n"
+           "      <softbreak />\n"
+           "      <text sourcepos=\"2:1-2:4\" xml:space=\"preserve\">line</text>\n"
+           "      <softbreak />\n"
+           "      <text sourcepos=\"3:2-3:7\" xml:space=\"preserve\">breaks</text>\n"
+           "    </link>\n"
+           "    <text sourcepos=\"3:33-3:33\" xml:space=\"preserve\">.</text>\n"
+           "  </paragraph>\n"
+           "</document>\n",
+           "autolink sourcepos are as expected");
+    free(xml);
+    cmark_node_free(doc);
+  }
 }
 
 static void ref_source_pos(test_batch_runner *runner) {

From 6df4f1f556d1a35c7b1e7afe8f64f4d729d02a0a Mon Sep 17 00:00:00 2001
From: chriszielinski <chrisz@berkeley.edu>
Date: Fri, 26 Apr 2019 12:07:29 -0700
Subject: [PATCH 04/27] Fix outdated expected test result.

---
 api_test/main.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/api_test/main.c b/api_test/main.c
index bd7dd7450..8cc1b52ba 100755
--- a/api_test/main.c
+++ b/api_test/main.c
@@ -1217,9 +1217,9 @@ static void source_pos_inlines(test_batch_runner *runner) {
            "<!DOCTYPE document SYSTEM \"CommonMark.dtd\">\n"
            "<document sourcepos=\"1:1-3:33\" xmlns=\"http://commonmark.org/xml/1.0\">\n"
            "  <paragraph sourcepos=\"1:1-3:33\">\n"
-           "    <text sourcepos=\"1:1-1:26\" xml:space=\"preserve\">This link will have three </text>\n"
-           "    <link sourcepos=\"1:27-3:32\" destination=\"https://commonmark.org\" title=\"\">\n"
-           "      <text sourcepos=\"1:28-1:32\" xml:space=\"preserve\">soft</text>\n"
+           "    <text sourcepos=\"1:1-1:24\" xml:space=\"preserve\">This link will have two </text>\n"
+           "    <link sourcepos=\"1:25-3:32\" destination=\"https://commonmark.org\" title=\"\">\n"
+           "      <text sourcepos=\"1:26-1:30\" xml:space=\"preserve\">soft</text>\n"
            "      <softbreak />\n"
            "      <text sourcepos=\"2:1-2:4\" xml:space=\"preserve\">line</text>\n"
            "      <softbreak />\n"

From 891db7009dc9eeacbb83e6ba51e7fbd60d6b0c06 Mon Sep 17 00:00:00 2001
From: chriszielinski <chrisz@berkeley.edu>
Date: Tue, 30 Apr 2019 19:37:52 -0700
Subject: [PATCH 05/27] Fix autolink source position.

---
 src/inlines.c | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)

diff --git a/src/inlines.c b/src/inlines.c
index b90427e7f..9471396f5 100755
--- a/src/inlines.c
+++ b/src/inlines.c
@@ -150,8 +150,8 @@ static CMARK_INLINE cmark_node *make_autolink(subject *subj,
   link->as.link.url = cmark_clean_autolink(subj->mem, &url, is_email);
   link->as.link.title = cmark_chunk_literal("");
   link->start_line = link->end_line = subj->line;
-  link->start_column = start_column + 1;
-  link->end_column = end_column + 1;
+  link->start_column = subj->column_offset + subj->block_offset + start_column + 1;
+  link->end_column = subj->column_offset + subj->block_offset + end_column + 1;
   cmark_node_append_child(link, make_str_with_entities(subj, start_column + 1, end_column - 1, &url));
   return link;
 }
@@ -914,8 +914,7 @@ static cmark_node *handle_pointy_brace(subject *subj, int options) {
     contents = cmark_chunk_dup(&subj->input, subj->pos, matchlen - 1);
     subj->pos += matchlen;
 
-    return make_autolink(subj, subj->pos + subj->column_offset - 1 - matchlen,
-                         subj->pos + subj->column_offset - 1, contents, 0);
+    return make_autolink(subj, subj->pos - 1 - matchlen, subj->pos - 1, contents, 0);
   }
 
   // next try to match an email autolink
@@ -924,8 +923,7 @@ static cmark_node *handle_pointy_brace(subject *subj, int options) {
     contents = cmark_chunk_dup(&subj->input, subj->pos, matchlen - 1);
     subj->pos += matchlen;
 
-    return make_autolink(subj, subj->pos + subj->column_offset - 1 - matchlen,
-                         subj->pos + subj->column_offset - 1, contents, 1);
+    return make_autolink(subj, subj->pos - 1 - matchlen, subj->pos - 1, contents, 1);
   }
 
   // finally, try to match an html tag

From 9160d605718160596f8a166b99f3ecb7b61b90cb Mon Sep 17 00:00:00 2001
From: chriszielinski <chrisz@berkeley.edu>
Date: Tue, 30 Apr 2019 19:39:16 -0700
Subject: [PATCH 06/27] Fix expected autolink test fixture, and add an
 additional autolink source position test case.

---
 api_test/main.c | 30 +++++++++++++++++++++++++++++-
 1 file changed, 29 insertions(+), 1 deletion(-)

diff --git a/api_test/main.c b/api_test/main.c
index 8cc1b52ba..c9fd849ef 100755
--- a/api_test/main.c
+++ b/api_test/main.c
@@ -1148,7 +1148,7 @@ static void source_pos(test_batch_runner *runner) {
            "      <softbreak />\n"
            "      <text sourcepos=\"6:3-6:3\" xml:space=\"preserve\">(</text>\n"
            "      <link sourcepos=\"6:4-6:49\" destination=\"http://daringfireball.net/projects/markdown/\" title=\"\">\n"
-           "        <text sourcepos=\"6:-283-6:-240\" xml:space=\"preserve\">http://daringfireball.net/projects/markdown/</text>\n"
+           "        <text sourcepos=\"6:5-6:48\" xml:space=\"preserve\">http://daringfireball.net/projects/markdown/</text>\n"
            "      </link>\n"
            "      <text sourcepos=\"6:50-6:50\" xml:space=\"preserve\">)</text>\n"
            "    </paragraph>\n"
@@ -1232,6 +1232,34 @@ static void source_pos_inlines(test_batch_runner *runner) {
     free(xml);
     cmark_node_free(doc);
   }
+  {
+    static const char markdown[] =
+    " 1. <http://www.google.com>\n"
+    "    <http://www.google.com>\n";
+
+    cmark_node *doc = cmark_parse_document(markdown, sizeof(markdown) - 1, CMARK_OPT_DEFAULT);
+    char *xml = cmark_render_xml(doc, CMARK_OPT_DEFAULT | CMARK_OPT_SOURCEPOS);
+    STR_EQ(runner, xml, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
+           "<!DOCTYPE document SYSTEM \"CommonMark.dtd\">\n"
+           "<document sourcepos=\"1:1-2:27\" xmlns=\"http://commonmark.org/xml/1.0\">\n"
+           "  <list sourcepos=\"1:2-2:27\" type=\"ordered\" start=\"1\" delim=\"period\" tight=\"true\">\n"
+           "    <item sourcepos=\"1:2-2:27\">\n"
+           "      <paragraph sourcepos=\"1:5-2:27\">\n"
+           "        <link sourcepos=\"1:5-1:27\" destination=\"http://www.google.com\" title=\"\">\n"
+           "          <text sourcepos=\"1:6-1:26\" xml:space=\"preserve\">http://www.google.com</text>\n"
+           "        </link>\n"
+           "        <softbreak />\n"
+           "        <link sourcepos=\"2:5-2:27\" destination=\"http://www.google.com\" title=\"\">\n"
+           "          <text sourcepos=\"2:6-2:26\" xml:space=\"preserve\">http://www.google.com</text>\n"
+           "        </link>\n"
+           "      </paragraph>\n"
+           "    </item>\n"
+           "  </list>\n"
+           "</document>\n",
+           "autolink (in list) sourcepos are as expected");
+    free(xml);
+    cmark_node_free(doc);
+  }
 }
 
 static void ref_source_pos(test_batch_runner *runner) {

From 469031e0b7e8ad996e5e165183448fdcf62c9dd3 Mon Sep 17 00:00:00 2001
From: chriszielinski <chrisz@berkeley.edu>
Date: Thu, 2 May 2019 19:00:17 -0700
Subject: [PATCH 07/27] Fix source position of setext headings.

---
 src/blocks.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/src/blocks.c b/src/blocks.c
index 0c497fb2e..2925ae653 100755
--- a/src/blocks.c
+++ b/src/blocks.c
@@ -323,8 +323,7 @@ static cmark_node *finalize(cmark_parser *parser, cmark_node *b) {
     b->end_line = parser->line_number;
     b->end_column = parser->last_line_length;
   } else if (S_type(b) == CMARK_NODE_DOCUMENT ||
-             (S_type(b) == CMARK_NODE_CODE_BLOCK && b->as.code.fenced) ||
-             (S_type(b) == CMARK_NODE_HEADING && b->as.heading.setext)) {
+             (S_type(b) == CMARK_NODE_CODE_BLOCK && b->as.code.fenced)) {
     b->end_line = parser->line_number;
     b->end_column = parser->curline.size;
     if (b->end_column && parser->curline.ptr[b->end_column - 1] == '\n')

From 51cd4f5731ee8c839dda97a40f63e84d9864178b Mon Sep 17 00:00:00 2001
From: chriszielinski <chrisz@berkeley.edu>
Date: Thu, 2 May 2019 19:00:50 -0700
Subject: [PATCH 08/27] Add setext heading test case.

---
 api_test/main.c | 22 ++++++++++++++++++++++
 1 file changed, 22 insertions(+)

diff --git a/api_test/main.c b/api_test/main.c
index c9fd849ef..baa10e626 100755
--- a/api_test/main.c
+++ b/api_test/main.c
@@ -1260,6 +1260,28 @@ static void source_pos_inlines(test_batch_runner *runner) {
     free(xml);
     cmark_node_free(doc);
   }
+  {
+    static const char markdown[] =
+    "Level 1 Heading\n"
+    "===============\n"
+    "A paragraph.\n";
+
+    cmark_node *doc = cmark_parse_document(markdown, sizeof(markdown) - 1, CMARK_OPT_DEFAULT);
+    char *xml = cmark_render_xml(doc, CMARK_OPT_DEFAULT | CMARK_OPT_SOURCEPOS);
+    STR_EQ(runner, xml, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
+           "<!DOCTYPE document SYSTEM \"CommonMark.dtd\">\n"
+           "<document sourcepos=\"1:1-3:12\" xmlns=\"http://commonmark.org/xml/1.0\">\n"
+           "  <heading sourcepos=\"1:1-2:15\" level=\"1\">\n"
+           "    <text sourcepos=\"1:1-1:15\" xml:space=\"preserve\">Level 1 Heading</text>\n"
+           "  </heading>\n"
+           "  <paragraph sourcepos=\"3:1-3:12\">\n"
+           "    <text sourcepos=\"3:1-3:12\" xml:space=\"preserve\">A paragraph.</text>\n"
+           "  </paragraph>\n"
+           "</document>\n",
+           "heading sourcepos are as expected");
+    free(xml);
+    cmark_node_free(doc);
+  }
 }
 
 static void ref_source_pos(test_batch_runner *runner) {

From 4ef606429a7d9fc8b2055793bb270faf985b8d74 Mon Sep 17 00:00:00 2001
From: chriszielinski <chrisz@berkeley.edu>
Date: Fri, 3 May 2019 13:43:43 -0700
Subject: [PATCH 09/27] Fix source position for ATX-style headings.

---
 src/blocks.c | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/src/blocks.c b/src/blocks.c
index 2925ae653..84c0f85fd 100755
--- a/src/blocks.c
+++ b/src/blocks.c
@@ -318,6 +318,10 @@ static cmark_node *finalize(cmark_parser *parser, cmark_node *b) {
          CMARK_NODE__OPEN); // shouldn't call finalize on closed blocks
   b->flags &= ~CMARK_NODE__OPEN;
 
+  if (S_type(b) == CMARK_NODE_HEADING && !b->as.heading.setext) {
+    parser->last_line_length += b->end_column;
+  }
+
   if (parser->curline.size == 0) {
     // end of input - line number has not been incremented
     b->end_line = parser->line_number;
@@ -1426,7 +1430,10 @@ static void add_text_to_container(cmark_parser *parser, cmark_node *container,
     } else if (accepts_lines(S_type(container))) {
       if (S_type(container) == CMARK_NODE_HEADING &&
           container->as.heading.setext == false) {
+        bufsize_t original_len = input->len;
         chop_trailing_hashtags(input);
+        // Substract one to exclude the trailing newline.
+        container->end_column += original_len - input->len - 1;
       }
       S_advance_offset(parser, input, parser->first_nonspace - parser->offset,
                        false);

From 035c392f6107866367c2210716da1a0da49ef2fd Mon Sep 17 00:00:00 2001
From: chriszielinski <chrisz@berkeley.edu>
Date: Fri, 3 May 2019 13:44:20 -0700
Subject: [PATCH 10/27] Add ATX-style heading source position test case.

---
 api_test/main.c | 24 ++++++++++++++++++++++++
 1 file changed, 24 insertions(+)

diff --git a/api_test/main.c b/api_test/main.c
index baa10e626..741979461 100755
--- a/api_test/main.c
+++ b/api_test/main.c
@@ -1282,6 +1282,30 @@ static void source_pos_inlines(test_batch_runner *runner) {
     free(xml);
     cmark_node_free(doc);
   }
+  {
+    static const char markdown[] =
+    "# This is an H1 #\n"
+    "\n"
+    "> # Header 1 #\n";
+
+    cmark_node *doc = cmark_parse_document(markdown, sizeof(markdown) - 1, CMARK_OPT_DEFAULT);
+    char *xml = cmark_render_xml(doc, CMARK_OPT_DEFAULT | CMARK_OPT_SOURCEPOS);
+    STR_EQ(runner, xml, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
+           "<!DOCTYPE document SYSTEM \"CommonMark.dtd\">\n"
+           "<document sourcepos=\"1:1-3:14\" xmlns=\"http://commonmark.org/xml/1.0\">\n"
+           "  <heading sourcepos=\"1:1-1:17\" level=\"1\">\n"
+           "    <text sourcepos=\"1:3-1:15\" xml:space=\"preserve\">This is an H1</text>\n"
+           "  </heading>\n"
+           "  <block_quote sourcepos=\"3:1-3:14\">\n"
+           "    <heading sourcepos=\"3:3-3:14\" level=\"1\">\n"
+           "      <text sourcepos=\"3:5-3:12\" xml:space=\"preserve\">Header 1</text>\n"
+           "    </heading>\n"
+           "  </block_quote>\n"
+           "</document>\n",
+           "atx heading sourcepos are as expected");
+    free(xml);
+    cmark_node_free(doc);
+  }
 }
 
 static void ref_source_pos(test_batch_runner *runner) {

From 92444d7d4637018dd7bfd5140af71ef598b7cae7 Mon Sep 17 00:00:00 2001
From: chriszielinski <chrisz@berkeley.edu>
Date: Mon, 6 May 2019 10:28:34 -0700
Subject: [PATCH 11/27] Fix HTMl block source position.

---
 src/blocks.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/src/blocks.c b/src/blocks.c
index 84c0f85fd..82704500a 100755
--- a/src/blocks.c
+++ b/src/blocks.c
@@ -327,7 +327,8 @@ static cmark_node *finalize(cmark_parser *parser, cmark_node *b) {
     b->end_line = parser->line_number;
     b->end_column = parser->last_line_length;
   } else if (S_type(b) == CMARK_NODE_DOCUMENT ||
-             (S_type(b) == CMARK_NODE_CODE_BLOCK && b->as.code.fenced)) {
+             (S_type(b) == CMARK_NODE_CODE_BLOCK && b->as.code.fenced) ||
+             (S_type(b) == CMARK_NODE_HTML_BLOCK)) {
     b->end_line = parser->line_number;
     b->end_column = parser->curline.size;
     if (b->end_column && parser->curline.ptr[b->end_column - 1] == '\n')

From 21571967c2b8524e359f2f4851a94b1b4da5921a Mon Sep 17 00:00:00 2001
From: chriszielinski <chrisz@berkeley.edu>
Date: Mon, 6 May 2019 10:34:40 -0700
Subject: [PATCH 12/27] Add HTML block source position test case.

---
 api_test/main.c | 48 ++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 48 insertions(+)

diff --git a/api_test/main.c b/api_test/main.c
index 741979461..138e1fb49 100755
--- a/api_test/main.c
+++ b/api_test/main.c
@@ -1158,6 +1158,54 @@ static void source_pos(test_batch_runner *runner) {
     free(xml);
     cmark_node_free(doc);
   }
+  {
+    static const char markdown[] =
+    "\n"
+    "\n"
+    "<pre lang=\"no-highlight\"><code>```javascript\n"
+    "var s = \"JavaScript syntax highlighting\";\n"
+    "alert(s);\n"
+    "```\n"
+    " \n"
+    "```python\n"
+    "s = \"Python syntax highlighting\"\n"
+    "print s\n"
+    "```\n"
+    " \n"
+    "```\n"
+    "No language indicated, so no syntax highlighting. \n"
+    "But let's throw in a &lt;b&gt;tag&lt;/b&gt;.\n"
+    "```\n"
+    "</code></pre>\n"
+    "\n"
+    "\n";
+
+    cmark_node *doc = cmark_parse_document(markdown, sizeof(markdown) - 1, CMARK_OPT_DEFAULT);
+    char *xml = cmark_render_xml(doc, CMARK_OPT_DEFAULT | CMARK_OPT_SOURCEPOS);
+    STR_EQ(runner, xml, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
+           "<!DOCTYPE document SYSTEM \"CommonMark.dtd\">\n"
+           "<document sourcepos=\"1:1-19:0\" xmlns=\"http://commonmark.org/xml/1.0\">\n"
+           "  <html_block sourcepos=\"3:1-17:13\" xml:space=\"preserve\">&lt;pre lang=&quot;no-highlight&quot;&gt;&lt;code&gt;```javascript\n"
+           "var s = &quot;JavaScript syntax highlighting&quot;;\n"
+           "alert(s);\n"
+           "```\n"
+           " \n"
+           "```python\n"
+           "s = &quot;Python syntax highlighting&quot;\n"
+           "print s\n"
+           "```\n"
+           " \n"
+           "```\n"
+           "No language indicated, so no syntax highlighting. \n"
+           "But let's throw in a &amp;lt;b&amp;gt;tag&amp;lt;/b&amp;gt;.\n"
+           "```\n"
+           "&lt;/code&gt;&lt;/pre&gt;\n"
+           "</html_block>\n"
+           "</document>\n",
+           "html block sourcepos are as expected");
+    free(xml);
+    cmark_node_free(doc);
+  }
 }
 
 static void source_pos_inlines(test_batch_runner *runner) {

From 79bf93b7cfb998e939aae74c88d37494b7505662 Mon Sep 17 00:00:00 2001
From: chriszielinski <chrisz@berkeley.edu>
Date: Mon, 6 May 2019 10:47:21 -0700
Subject: [PATCH 13/27] Fix thematic break source position.

---
 src/blocks.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/src/blocks.c b/src/blocks.c
index 82704500a..5e222d97d 100755
--- a/src/blocks.c
+++ b/src/blocks.c
@@ -318,6 +318,11 @@ static cmark_node *finalize(cmark_parser *parser, cmark_node *b) {
          CMARK_NODE__OPEN); // shouldn't call finalize on closed blocks
   b->flags &= ~CMARK_NODE__OPEN;
 
+  if (S_type(b) == CMARK_NODE_THEMATIC_BREAK) {
+    // Already been "finalized".
+    return parent;
+  }
+
   if (S_type(b) == CMARK_NODE_HEADING && !b->as.heading.setext) {
     parser->last_line_length += b->end_column;
   }
@@ -1220,6 +1225,10 @@ static void open_new_blocks(cmark_parser *parser, cmark_node **container,
       // it's only now that we know the line is not part of a setext heading:
       *container = add_child(parser, *container, CMARK_NODE_THEMATIC_BREAK,
                              parser->first_nonspace + 1);
+      // A thematic break can only be on a single line, so we can set the
+      // end source position here.
+      (*container)->end_line = parser->line_number;
+      (*container)->end_column = input->len - 1;
       S_advance_offset(parser, input, input->len - 1 - parser->offset, false);
     } else if (!indented &&
                parser->options & CMARK_OPT_FOOTNOTES &&

From 53041c309cc9eeacbedae343724ea59833c85634 Mon Sep 17 00:00:00 2001
From: chriszielinski <chrisz@berkeley.edu>
Date: Mon, 6 May 2019 10:47:57 -0700
Subject: [PATCH 14/27] Add thematic break source position test case.

---
 api_test/main.c | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)

diff --git a/api_test/main.c b/api_test/main.c
index 138e1fb49..37cb6360e 100755
--- a/api_test/main.c
+++ b/api_test/main.c
@@ -1206,6 +1206,24 @@ static void source_pos(test_batch_runner *runner) {
     free(xml);
     cmark_node_free(doc);
   }
+  {
+    static const char markdown[] =
+    "---\n"
+    "\n"
+    "\n"
+    "\n";
+
+    cmark_node *doc = cmark_parse_document(markdown, sizeof(markdown) - 1, CMARK_OPT_DEFAULT);
+    char *xml = cmark_render_xml(doc, CMARK_OPT_DEFAULT | CMARK_OPT_SOURCEPOS);
+    STR_EQ(runner, xml, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
+           "<!DOCTYPE document SYSTEM \"CommonMark.dtd\">\n"
+           "<document sourcepos=\"1:1-4:0\" xmlns=\"http://commonmark.org/xml/1.0\">\n"
+           "  <thematic_break sourcepos=\"1:1-1:3\" />\n"
+           "</document>\n",
+           "thematic break sourcepos are as expected");
+    free(xml);
+    cmark_node_free(doc);
+  }
 }
 
 static void source_pos_inlines(test_batch_runner *runner) {

From be7f9c0b4068cdee016f9ddd2226c6ffd9a1ebe7 Mon Sep 17 00:00:00 2001
From: chriszielinski <chrisz@berkeley.edu>
Date: Tue, 7 May 2019 17:41:25 -0700
Subject: [PATCH 15/27] Fix ending source position for lists and list items.

---
 src/blocks.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/src/blocks.c b/src/blocks.c
index 5e222d97d..aee29c937 100755
--- a/src/blocks.c
+++ b/src/blocks.c
@@ -327,7 +327,11 @@ static cmark_node *finalize(cmark_parser *parser, cmark_node *b) {
     parser->last_line_length += b->end_column;
   }
 
-  if (parser->curline.size == 0) {
+  if ((S_type(b) == CMARK_NODE_ITEM || S_type(b) == CMARK_NODE_LIST)
+      && b->last_child) {
+    b->end_line = b->last_child->end_line;
+    b->end_column = b->last_child->end_column;
+  } else if (parser->curline.size == 0) {
     // end of input - line number has not been incremented
     b->end_line = parser->line_number;
     b->end_column = parser->last_line_length;
@@ -1300,6 +1304,8 @@ static void open_new_blocks(cmark_parser *parser, cmark_node **container,
       // add the list item
       *container = add_child(parser, *container, CMARK_NODE_ITEM,
                              parser->first_nonspace + 1);
+//      (*container)->end_line = parser->line_number;
+//      (*container)->end_column = parser->column;
       /* TODO: static */
       memcpy(&((*container)->as.list), data, sizeof(*data));
       parser->mem->free(data);

From d2721bf8ba455e522fad5bb5e6efd3411d82d020 Mon Sep 17 00:00:00 2001
From: chriszielinski <chrisz@berkeley.edu>
Date: Tue, 7 May 2019 18:32:14 -0700
Subject: [PATCH 16/27] Add list/list item source position test case.

---
 api_test/main.c | 47 ++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 46 insertions(+), 1 deletion(-)

diff --git a/api_test/main.c b/api_test/main.c
index 37cb6360e..b03d0ee9b 100755
--- a/api_test/main.c
+++ b/api_test/main.c
@@ -1035,7 +1035,7 @@ static void source_pos(test_batch_runner *runner) {
                       "  </paragraph>\n"
                       "  <block_quote sourcepos=\"6:1-10:20\">\n"
                       "    <list sourcepos=\"6:3-10:20\" type=\"ordered\" start=\"1\" delim=\"period\" tight=\"false\">\n"
-                      "      <item sourcepos=\"6:3-8:1\">\n"
+                        "      <item sourcepos=\"6:3-7:10\">\n"
                       "        <paragraph sourcepos=\"6:6-7:10\">\n"
                       "          <text sourcepos=\"6:6-6:10\" xml:space=\"preserve\">Okay.</text>\n"
                       "          <softbreak />\n"
@@ -1224,6 +1224,51 @@ static void source_pos(test_batch_runner *runner) {
     free(xml);
     cmark_node_free(doc);
   }
+  {
+    static const char markdown[] =
+    "1. List 1, Item A.\n"
+    "    1. List 2, Item A.\n"
+    "       Second line.\n"
+    "\n"
+    "2. List 1, Item B.\n"
+    "   Second line.\n"
+    "\n"
+    "\n"
+    "\n";
+
+    cmark_node *doc = cmark_parse_document(markdown, sizeof(markdown) - 1, CMARK_OPT_DEFAULT);
+    char *xml = cmark_render_xml(doc, CMARK_OPT_DEFAULT | CMARK_OPT_SOURCEPOS);
+    STR_EQ(runner, xml, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
+           "<!DOCTYPE document SYSTEM \"CommonMark.dtd\">\n"
+           "<document sourcepos=\"1:1-9:0\" xmlns=\"http://commonmark.org/xml/1.0\">\n"
+           "  <list sourcepos=\"1:1-6:15\" type=\"ordered\" start=\"1\" delim=\"period\" tight=\"false\">\n"
+           "    <item sourcepos=\"1:1-3:19\">\n"
+           "      <paragraph sourcepos=\"1:4-1:18\">\n"
+           "        <text sourcepos=\"1:4-1:18\" xml:space=\"preserve\">List 1, Item A.</text>\n"
+           "      </paragraph>\n"
+           "      <list sourcepos=\"2:5-3:19\" type=\"ordered\" start=\"1\" delim=\"period\" tight=\"true\">\n"
+           "        <item sourcepos=\"2:5-3:19\">\n"
+           "          <paragraph sourcepos=\"2:8-3:19\">\n"
+           "            <text sourcepos=\"2:8-2:22\" xml:space=\"preserve\">List 2, Item A.</text>\n"
+           "            <softbreak />\n"
+           "            <text sourcepos=\"3:8-3:19\" xml:space=\"preserve\">Second line.</text>\n"
+           "          </paragraph>\n"
+           "        </item>\n"
+           "      </list>\n"
+           "    </item>\n"
+           "    <item sourcepos=\"5:1-6:15\">\n"
+           "      <paragraph sourcepos=\"5:4-6:15\">\n"
+           "        <text sourcepos=\"5:4-5:18\" xml:space=\"preserve\">List 1, Item B.</text>\n"
+           "        <softbreak />\n"
+           "        <text sourcepos=\"6:4-6:15\" xml:space=\"preserve\">Second line.</text>\n"
+           "      </paragraph>\n"
+           "    </item>\n"
+           "  </list>\n"
+           "</document>\n",
+           "list sourcepos are as expected");
+    free(xml);
+    cmark_node_free(doc);
+  }
 }
 
 static void source_pos_inlines(test_batch_runner *runner) {

From 94dba88e6dcfe66e0f691e3cb9529d1666d142f9 Mon Sep 17 00:00:00 2001
From: chriszielinski <chrisz@berkeley.edu>
Date: Tue, 7 May 2019 18:43:41 -0700
Subject: [PATCH 17/27] Remove commented out code.

---
 src/blocks.c | 2 --
 1 file changed, 2 deletions(-)

diff --git a/src/blocks.c b/src/blocks.c
index aee29c937..b052b2adc 100755
--- a/src/blocks.c
+++ b/src/blocks.c
@@ -1304,8 +1304,6 @@ static void open_new_blocks(cmark_parser *parser, cmark_node **container,
       // add the list item
       *container = add_child(parser, *container, CMARK_NODE_ITEM,
                              parser->first_nonspace + 1);
-//      (*container)->end_line = parser->line_number;
-//      (*container)->end_column = parser->column;
       /* TODO: static */
       memcpy(&((*container)->as.list), data, sizeof(*data));
       parser->mem->free(data);

From 99621a33510a02b0783c7300915c90ef40f92458 Mon Sep 17 00:00:00 2001
From: chriszielinski <chrisz@berkeley.edu>
Date: Tue, 7 May 2019 19:27:22 -0700
Subject: [PATCH 18/27] Correct list source position.

---
 src/blocks.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/src/blocks.c b/src/blocks.c
index b052b2adc..64d6e2df5 100755
--- a/src/blocks.c
+++ b/src/blocks.c
@@ -331,6 +331,12 @@ static cmark_node *finalize(cmark_parser *parser, cmark_node *b) {
       && b->last_child) {
     b->end_line = b->last_child->end_line;
     b->end_column = b->last_child->end_column;
+
+    if (S_type(b) == CMARK_NODE_ITEM && b->parent) {
+      // The finalization order is not deterministic...
+      b->parent->end_line = b->end_line;
+      b->parent->end_column = b->end_column;
+    }
   } else if (parser->curline.size == 0) {
     // end of input - line number has not been incremented
     b->end_line = parser->line_number;

From 25709117bc4929bff02d7d89dcd298822456b4a3 Mon Sep 17 00:00:00 2001
From: chriszielinski <chrisz@berkeley.edu>
Date: Wed, 15 May 2019 08:22:51 -0700
Subject: [PATCH 19/27] Fix source position for HTML blocks without a matching
 end condition.

---
 src/blocks.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/src/blocks.c b/src/blocks.c
index 64d6e2df5..89a3d9867 100755
--- a/src/blocks.c
+++ b/src/blocks.c
@@ -343,7 +343,7 @@ static cmark_node *finalize(cmark_parser *parser, cmark_node *b) {
     b->end_column = parser->last_line_length;
   } else if (S_type(b) == CMARK_NODE_DOCUMENT ||
              (S_type(b) == CMARK_NODE_CODE_BLOCK && b->as.code.fenced) ||
-             (S_type(b) == CMARK_NODE_HTML_BLOCK)) {
+             (S_type(b) == CMARK_NODE_HTML_BLOCK && b->end_line == b->start_line && b->end_column == 0)) {
     b->end_line = parser->line_number;
     b->end_column = parser->curline.size;
     if (b->end_column && parser->curline.ptr[b->end_column - 1] == '\n')
@@ -1400,6 +1400,12 @@ static void add_text_to_container(cmark_parser *parser, cmark_node *container,
   } else { // not a lazy continuation
     // Finalize any blocks that were not matched and set cur to container:
     while (parser->current != last_matched_container) {
+      if (S_type(parser->current) == CMARK_NODE_HTML_BLOCK) {
+        // Edge case: Closing an HTML block without a matching end condition.
+        parser->current->end_line = parser->line_number - 1;
+        parser->current->end_column = parser->last_line_length;
+      }
+
       parser->current = finalize(parser, parser->current);
       assert(parser->current != NULL);
     }

From 062d4616bbc559af5d404576cee50f2d4baeb0c5 Mon Sep 17 00:00:00 2001
From: chriszielinski <chrisz@berkeley.edu>
Date: Wed, 15 May 2019 08:23:57 -0700
Subject: [PATCH 20/27] Add test case for source position of a HTML block
 without a matching end condition.

---
 api_test/main.c | 33 +++++++++++++++++++++++++++++++++
 1 file changed, 33 insertions(+)

diff --git a/api_test/main.c b/api_test/main.c
index b03d0ee9b..e13937500 100755
--- a/api_test/main.c
+++ b/api_test/main.c
@@ -1269,6 +1269,39 @@ static void source_pos(test_batch_runner *runner) {
     free(xml);
     cmark_node_free(doc);
   }
+  {
+    static const char markdown[] =
+    "* List 1, item A.\n"
+    "\n"
+    "  <script>\n"
+    "- List 2, item A.\n";
+
+    cmark_node *doc = cmark_parse_document(markdown, sizeof(markdown) - 1, CMARK_OPT_DEFAULT);
+    char *xml = cmark_render_xml(doc, CMARK_OPT_DEFAULT | CMARK_OPT_SOURCEPOS);
+    STR_EQ(runner, xml, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
+           "<!DOCTYPE document SYSTEM \"CommonMark.dtd\">\n"
+           "<document sourcepos=\"1:1-4:17\" xmlns=\"http://commonmark.org/xml/1.0\">\n"
+           "  <list sourcepos=\"1:1-3:10\" type=\"bullet\" tight=\"false\">\n"
+           "    <item sourcepos=\"1:1-3:10\">\n"
+           "      <paragraph sourcepos=\"1:3-1:17\">\n"
+           "        <text sourcepos=\"1:3-1:17\" xml:space=\"preserve\">List 1, item A.</text>\n"
+           "      </paragraph>\n"
+           "      <html_block sourcepos=\"3:3-3:10\" xml:space=\"preserve\">&lt;script&gt;\n"
+           "</html_block>\n"
+           "    </item>\n"
+           "  </list>\n"
+           "  <list sourcepos=\"4:1-4:17\" type=\"bullet\" tight=\"true\">\n"
+           "    <item sourcepos=\"4:1-4:17\">\n"
+           "      <paragraph sourcepos=\"4:3-4:17\">\n"
+           "        <text sourcepos=\"4:3-4:17\" xml:space=\"preserve\">List 2, item A.</text>\n"
+           "      </paragraph>\n"
+           "    </item>\n"
+           "  </list>\n"
+           "</document>\n",
+           "sourcepos of HTML block without end condition is as expected");
+    free(xml);
+    cmark_node_free(doc);
+  }
 }
 
 static void source_pos_inlines(test_batch_runner *runner) {

From a4222f5f585a6fef7f6cd5893ad4764d15d1efac Mon Sep 17 00:00:00 2001
From: chriszielinski <chrisz@berkeley.edu>
Date: Wed, 15 May 2019 08:32:27 -0700
Subject: [PATCH 21/27] Fix line length.

---
 src/blocks.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/src/blocks.c b/src/blocks.c
index 89a3d9867..bad07a11d 100755
--- a/src/blocks.c
+++ b/src/blocks.c
@@ -343,7 +343,8 @@ static cmark_node *finalize(cmark_parser *parser, cmark_node *b) {
     b->end_column = parser->last_line_length;
   } else if (S_type(b) == CMARK_NODE_DOCUMENT ||
              (S_type(b) == CMARK_NODE_CODE_BLOCK && b->as.code.fenced) ||
-             (S_type(b) == CMARK_NODE_HTML_BLOCK && b->end_line == b->start_line && b->end_column == 0)) {
+             (S_type(b) == CMARK_NODE_HTML_BLOCK
+              && b->end_line == b->start_line && b->end_column == 0)) {
     b->end_line = parser->line_number;
     b->end_column = parser->curline.size;
     if (b->end_column && parser->curline.ptr[b->end_column - 1] == '\n')

From 538ff216b7dcbfac7b0e0206085814dc784065f0 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Martin=20=C4=8C=C3=AD=C5=BEek?= <martin.cizek@orchitech.cz>
Date: Fri, 10 Apr 2020 01:34:10 +0200
Subject: [PATCH 22/27] Fix soourcepos for strikethrough.

---
 extensions/strikethrough.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/extensions/strikethrough.c b/extensions/strikethrough.c
index 8145d23b0..8c99cbe7f 100644
--- a/extensions/strikethrough.c
+++ b/extensions/strikethrough.c
@@ -24,7 +24,7 @@ static cmark_node *match(cmark_syntax_extension *self, cmark_parser *parser,
 
   res = cmark_node_new_with_mem(CMARK_NODE_TEXT, parser->mem);
   cmark_node_set_literal(res, buffer);
-  res->start_line = res->end_line = cmark_inline_parser_get_line(inline_parser);
+  res->start_line = cmark_inline_parser_get_line(inline_parser);
   res->start_column = cmark_inline_parser_get_column(inline_parser) - delims;
 
   if ((left_flanking || right_flanking) &&
@@ -64,6 +64,7 @@ static delimiter *insert(cmark_syntax_extension *self, cmark_parser *parser,
     tmp = next;
   }
 
+  strikethrough->end_line = closer->inl_text->start_line;
   strikethrough->end_column = closer->inl_text->start_column + closer->inl_text->as.literal.len - 1;
   cmark_node_free(closer->inl_text);
 

From ce4359222f1f35edc9de918f804fa7fec7a5b885 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Martin=20=C4=8C=C3=AD=C5=BEek?= <martin.cizek@orchitech.cz>
Date: Fri, 10 Apr 2020 23:11:46 +0200
Subject: [PATCH 23/27] Add test for broken multiline strikethrough. Restore
 adapted extension sourcepos tests unfortunately removed with #49.

---
 api_test/main.c | 98 ++++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 97 insertions(+), 1 deletion(-)

diff --git a/api_test/main.c b/api_test/main.c
index e13937500..8a793c35b 100755
--- a/api_test/main.c
+++ b/api_test/main.c
@@ -6,6 +6,7 @@
 #include "cmark-gfm.h"
 #include "node.h"
 #include "../extensions/cmark-gfm-core-extensions.h"
+#include "registry.h"
 
 #include "harness.h"
 #include "cplusplus.h"
@@ -1035,7 +1036,7 @@ static void source_pos(test_batch_runner *runner) {
                       "  </paragraph>\n"
                       "  <block_quote sourcepos=\"6:1-10:20\">\n"
                       "    <list sourcepos=\"6:3-10:20\" type=\"ordered\" start=\"1\" delim=\"period\" tight=\"false\">\n"
-                        "      <item sourcepos=\"6:3-7:10\">\n"
+                      "      <item sourcepos=\"6:3-7:10\">\n"
                       "        <paragraph sourcepos=\"6:6-7:10\">\n"
                       "          <text sourcepos=\"6:6-6:10\" xml:space=\"preserve\">Okay.</text>\n"
                       "          <softbreak />\n"
@@ -1476,6 +1477,97 @@ static void ref_source_pos(test_batch_runner *runner) {
   cmark_node_free(doc);
 }
 
+static void ext_source_pos(test_batch_runner *runner) {
+  static const char *extensions[3] = {
+    "strikethrough",
+    "table",
+    "autolink",
+  };
+
+  static const char markdown[] =
+    "Hi ~~friend~~ and ~~other\n"
+    "friend~~.\n"
+    "\n"
+    "> www.github.com\n"
+    "\n"
+    "1. | a | b | *c* |\n"
+    "   | - | - | --: |\n"
+    "   | 1 | 2 | ~3~ |\n";
+
+  int options = CMARK_OPT_DEFAULT | CMARK_OPT_SOURCEPOS;
+  cmark_parser *parser = cmark_parser_new(options);
+
+  for (int i = 0; i < (int)(sizeof(extensions) / sizeof(*extensions)); ++i) {
+    cmark_syntax_extension *ext = cmark_find_syntax_extension(extensions[i]);
+    cmark_parser_attach_syntax_extension(parser, ext);
+  }
+
+  cmark_parser_feed(parser, markdown, sizeof(markdown) - 1);
+
+  cmark_node *doc = cmark_parser_finish(parser);
+  char *xml = cmark_render_xml(doc, options);
+  STR_EQ(runner, xml, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
+                      "<!DOCTYPE document SYSTEM \"CommonMark.dtd\">\n"
+                      "<document sourcepos=\"1:1-8:18\" xmlns=\"http://commonmark.org/xml/1.0\">\n"
+                      "  <paragraph sourcepos=\"1:1-2:9\">\n"
+                      "    <text sourcepos=\"1:1-1:3\" xml:space=\"preserve\">Hi </text>\n"
+                      "    <strikethrough sourcepos=\"1:4-1:13\">\n"
+                      "      <text sourcepos=\"1:6-1:11\" xml:space=\"preserve\">friend</text>\n"
+                      "    </strikethrough>\n"
+                      "    <text sourcepos=\"1:14-1:18\" xml:space=\"preserve\"> and </text>\n"
+                      "    <strikethrough sourcepos=\"1:19-2:8\">\n"
+                      "      <text sourcepos=\"1:21-1:25\" xml:space=\"preserve\">other</text>\n"
+                      "      <softbreak />\n"
+                      "      <text sourcepos=\"2:1-2:6\" xml:space=\"preserve\">friend</text>\n"
+                      "    </strikethrough>\n"
+                      "    <text sourcepos=\"2:9-2:9\" xml:space=\"preserve\">.</text>\n"
+                      "  </paragraph>\n"
+                      "  <block_quote sourcepos=\"4:1-4:16\">\n"
+                      "    <paragraph sourcepos=\"4:3-4:16\">\n"
+                      "      <link sourcepos=\"4:2-4:16\" destination=\"http://www.github.com\" title=\"\">\n"
+                      "        <text sourcepos=\"4:2-4:16\" xml:space=\"preserve\">www.github.com</text>\n"
+                      "      </link>\n"
+                      "    </paragraph>\n"
+                      "  </block_quote>\n"
+                      "  <list sourcepos=\"6:1-8:18\" type=\"ordered\" start=\"1\" delim=\"period\" tight=\"true\">\n"
+                      "    <item sourcepos=\"6:1-8:18\">\n"
+                      "      <table sourcepos=\"6:4-8:18\">\n"
+                      "        <table_header sourcepos=\"6:4-6:18\">\n"
+                      "          <table_cell sourcepos=\"6:5-6:7\">\n"
+                      "            <text sourcepos=\"6:6-6:6\" xml:space=\"preserve\">a</text>\n"
+                      "          </table_cell>\n"
+                      "          <table_cell sourcepos=\"6:9-6:11\">\n"
+                      "            <text sourcepos=\"6:10-6:10\" xml:space=\"preserve\">b</text>\n"
+                      "          </table_cell>\n"
+                      "          <table_cell sourcepos=\"6:13-6:17\" align=\"right\">\n"
+                      "            <emph sourcepos=\"6:14-6:16\">\n"
+                      "              <text sourcepos=\"6:15-6:15\" xml:space=\"preserve\">c</text>\n"
+                      "            </emph>\n"
+                      "          </table_cell>\n"
+                      "        </table_header>\n"
+                      "        <table_row sourcepos=\"8:4-8:18\">\n"
+                      "          <table_cell sourcepos=\"8:5-8:7\">\n"
+                      "            <text sourcepos=\"8:6-8:6\" xml:space=\"preserve\">1</text>\n"
+                      "          </table_cell>\n"
+                      "          <table_cell sourcepos=\"8:9-8:11\">\n"
+                      "            <text sourcepos=\"8:10-8:10\" xml:space=\"preserve\">2</text>\n"
+                      "          </table_cell>\n"
+                      "          <table_cell sourcepos=\"8:13-8:17\">\n"
+                      "            <strikethrough sourcepos=\"8:14-8:16\">\n"
+                      "              <text sourcepos=\"8:15-8:15\" xml:space=\"preserve\">3</text>\n"
+                      "            </strikethrough>\n"
+                      "          </table_cell>\n"
+                      "        </table_row>\n"
+                      "      </table>\n"
+                      "    </item>\n"
+                      "  </list>\n"
+                      "</document>\n",
+         "sourcepos are as expected");
+  free(xml);
+  cmark_node_free(doc);
+  cmark_parser_free(parser);
+}
+
 int main() {
   int retval;
   test_batch_runner *runner = test_batch_runner_new();
@@ -1507,6 +1599,10 @@ int main() {
   source_pos_inlines(runner);
   ref_source_pos(runner);
 
+  cmark_gfm_core_extensions_ensure_registered();
+  ext_source_pos(runner);
+  cmark_release_plugins();
+
   test_print_summary(runner);
   retval = test_ok(runner) ? 0 : 1;
   free(runner);

From de09b9c23a7cbed1c9ba2cc4a9d183c04070a4de Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Martin=20=C4=8C=C3=AD=C5=BEek?= <martin.cizek@orchitech.cz>
Date: Fri, 10 Apr 2020 23:38:01 +0200
Subject: [PATCH 24/27] Add test catching sourcepos bugs related to tables with
 a paragraph rejected from header (sourcepos of such paragraph, the table and
 all nested elements are broken).

---
 api_test/main.c     | 48 +++++++++++++++++++++++++++++++++++++++++++++
 test/extensions.txt | 17 ++++++++++++++++
 2 files changed, 65 insertions(+)

diff --git a/api_test/main.c b/api_test/main.c
index 8a793c35b..572f251e5 100755
--- a/api_test/main.c
+++ b/api_test/main.c
@@ -1568,6 +1568,53 @@ static void ext_source_pos(test_batch_runner *runner) {
   cmark_parser_free(parser);
 }
 
+static void ext_source_pos_corners(test_batch_runner *runner) {
+  static const char *extensions[3] = {
+    "strikethrough",
+    "table",
+    "autolink",
+  };
+
+  // a paragraph inserted by the table extension
+  static const char markdown[] =
+    "| paragraph |\n" // +6
+    "| header |\n" //+3
+    "| --- |\n";
+
+  int options = CMARK_OPT_DEFAULT | CMARK_OPT_SOURCEPOS;
+  cmark_parser *parser = cmark_parser_new(options);
+
+  for (int i = 0; i < (int)(sizeof(extensions) / sizeof(*extensions)); ++i) {
+    cmark_syntax_extension *ext = cmark_find_syntax_extension(extensions[i]);
+    cmark_parser_attach_syntax_extension(parser, ext);
+  }
+
+  cmark_parser_feed(parser, markdown, sizeof(markdown) - 1);
+
+  cmark_node *doc = cmark_parser_finish(parser);
+  char *xml = cmark_render_xml(doc, options);
+  STR_EQ(runner, xml,
+    "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
+    "<!DOCTYPE document SYSTEM \"CommonMark.dtd\">\n"
+    "<document sourcepos=\"1:1-3:7\" xmlns=\"http://commonmark.org/xml/1.0\">\n"
+    "  <paragraph sourcepos=\"1:1-1:13\">\n"
+    "    <text sourcepos=\"1:1-1:13\" xml:space=\"preserve\">| paragraph |</text>\n"
+    "  </paragraph>\n"
+    "  <table sourcepos=\"2:1-3:7\">\n"
+    "    <table_header sourcepos=\"2:1-2:10\">\n"
+    "      <table_cell sourcepos=\"2:2-2:9\">\n"
+    "        <text sourcepos=\"2:3-2:8\" xml:space=\"preserve\">header</text>\n"
+    "      </table_cell>\n"
+    "    </table_header>\n"
+    "  </table>\n"
+    "</document>\n",
+    "sourcepos are as expected");
+  free(xml);
+  cmark_node_free(doc);
+  cmark_parser_free(parser);
+}
+
+
 int main() {
   int retval;
   test_batch_runner *runner = test_batch_runner_new();
@@ -1601,6 +1648,7 @@ int main() {
 
   cmark_gfm_core_extensions_ensure_registered();
   ext_source_pos(runner);
+  ext_source_pos_corners(runner);
   cmark_release_plugins();
 
   test_print_summary(runner);
diff --git a/test/extensions.txt b/test/extensions.txt
index 0d9993782..a7d90042a 100644
--- a/test/extensions.txt
+++ b/test/extensions.txt
@@ -499,6 +499,23 @@ d | e
 </table>
 ````````````````````````````````
 
+### a table can be recognised when there is a header-like line before
+
+```````````````````````````````` example
+| a |
+| b |
+| --- |
+.
+<p>| a |</p>
+<table>
+<thead>
+<tr>
+<th>b</th>
+</tr>
+</thead>
+</table>
+````````````````````````````````
+
 ## Strikethroughs
 
 A well-formed strikethrough.

From 27ff0481ec4c185bb6c6ec26b960268373c96d98 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Martin=20=C4=8C=C3=AD=C5=BEek?= <martin.cizek@orchitech.cz>
Date: Fri, 10 Apr 2020 23:39:55 +0200
Subject: [PATCH 25/27] Fix sourcepos bugs related to table headers with a
 rejected paragraph.

---
 extensions/table.c | 114 +++++++++++++++++++++++++++++++++++++++++----
 1 file changed, 105 insertions(+), 9 deletions(-)

diff --git a/extensions/table.c b/extensions/table.c
index a5bb44067..7e890fdaf 100644
--- a/extensions/table.c
+++ b/extensions/table.c
@@ -202,14 +202,95 @@ 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) {
+static void count_header_paragraph_newlines(unsigned char *content,
+                                            bufsize_t size,
+                                            int *lpad_nls,
+                                            int *lpad_last_line_columns,
+                                            int *trimmed_nls,
+                                            int *trimmed_last_line_columns,
+                                            int *rpad_nls) {
+  bufsize_t i = 0;
+
+  *lpad_nls = 0;
+  *lpad_last_line_columns = 0;
+  while (i < size && cmark_isspace(content[i])) {
+    if (content[i] == '\n') {
+      ++*lpad_nls;
+      *lpad_last_line_columns = 0;
+    } else {
+      ++*lpad_last_line_columns;
+    }
+    ++i;
+  }
+
+  *rpad_nls = 0;
+  while (size > i && cmark_isspace(content[size - 1])) {
+    if (content[size - 1] == '\n') {
+      ++*rpad_nls;
+    }
+    --size;
+  }
+
+  *trimmed_nls = 0;
+  *trimmed_last_line_columns = 0;
+  while (i < size) {
+    if (content[i] == '\n') {
+      ++*trimmed_nls;
+      *trimmed_last_line_columns = 0;
+    } else {
+      ++*trimmed_last_line_columns;
+    }
+    ++i;
+  }
+}
+
+static void advance_sourcepos_to_table_start(unsigned char *p,
+                                             int *start_line, int *start_column) {
+  bufsize_t i = 0;
+  int line = *start_line;
+  int column = *start_column;
+
+  while (p[i] && p[i] != '|') {
+    if (p[i] == '\n') {
+      ++line;
+      column = 1;
+    } else {
+      ++column;
+    }
+    ++i;
+  }
+
+  if (p[i] == '|') {
+    *start_line = line;
+    *start_column = column;
+  }
+}
+
+static cmark_node *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;
+  int lpad_nls = 0;
+  int lpad_last_line_columns = 0;
+  int trimmed_nls = 0;
+  int trimmed_last_line_columns = 0;
+  int rpad_nls = 0;
 
   paragraph = cmark_node_new_with_mem(CMARK_NODE_PARAGRAPH, parser->mem);
+  if (parser->options & CMARK_OPT_SOURCEPOS) {
+    count_header_paragraph_newlines(parent_string, paragraph_offset,
+      &lpad_nls, &lpad_last_line_columns,
+      &trimmed_nls, &trimmed_last_line_columns,
+      &rpad_nls);
+    paragraph->start_line = parent_container->start_line + lpad_nls;
+    paragraph->start_column = lpad_nls > 0
+      ? lpad_last_line_columns
+      : parent_container->start_column + lpad_last_line_columns;
+    paragraph->end_line = parent_container->start_line + lpad_nls + trimmed_nls;
+    paragraph->end_column = trimmed_last_line_columns;
+  }
 
   paragraph_content = unescape_pipes(parser->mem, parent_string, paragraph_offset);
   cmark_strbuf_trim(paragraph_content);
@@ -219,7 +300,15 @@ static void try_inserting_table_header_paragraph(cmark_parser *parser,
 
   if (!cmark_node_insert_before(parent_container, paragraph)) {
     parser->mem->free(paragraph);
+    return NULL;
   }
+  if (parser->options & CMARK_OPT_SOURCEPOS) {
+      parent_container->start_line += lpad_nls + trimmed_nls + rpad_nls;
+      parent_container->start_column = 1;
+      advance_sourcepos_to_table_start(parent_string + paragraph_offset,
+        &parent_container->start_line, &parent_container->start_column);
+  }
+  return paragraph;
 }
 
 static cmark_node *try_opening_table_header(cmark_syntax_extension *self,
@@ -231,6 +320,7 @@ static cmark_node *try_opening_table_header(cmark_syntax_extension *self,
   table_row *marker_row = NULL;
   node_table_row *ntr;
   const char *parent_string;
+  int start_column_offset = 0;
   uint16_t i;
 
   if (!scan_table_start(input, len, cmark_parser_get_first_nonspace(parser))) {
@@ -273,8 +363,11 @@ static cmark_node *try_opening_table_header(cmark_syntax_extension *self,
   }
 
   if (header_row->paragraph_offset) {
-    try_inserting_table_header_paragraph(parser, parent_container, (unsigned char *)parent_string,
-                                         header_row->paragraph_offset);
+    cmark_node *paragraph = try_inserting_table_header_paragraph(parser,
+      parent_container, (unsigned char *)parent_string, header_row->paragraph_offset);
+    if (paragraph) {
+      start_column_offset = header_row->paragraph_offset + 1;
+    }
   }
 
   cmark_node_set_syntax_extension(parent_container, self);
@@ -301,7 +394,8 @@ static cmark_node *try_opening_table_header(cmark_syntax_extension *self,
       cmark_parser_add_child(parser, parent_container, CMARK_NODE_TABLE_ROW,
                              parent_container->start_column);
   cmark_node_set_syntax_extension(table_header, self);
-  table_header->end_column = parent_container->start_column + (int)strlen(parent_string) - 2;
+  table_header->end_column =
+    parent_container->start_column + (int)strlen(parent_string) - 2 - start_column_offset;
   table_header->start_line = table_header->end_line = parent_container->start_line;
 
   table_header->as.opaque = ntr = (node_table_row *)parser->mem->calloc(1, sizeof(node_table_row));
@@ -313,10 +407,12 @@ static cmark_node *try_opening_table_header(cmark_syntax_extension *self,
     for (tmp = header_row->cells; tmp; tmp = tmp->next) {
       node_cell *cell = (node_cell *) tmp->data;
       cmark_node *header_cell = cmark_parser_add_child(parser, table_header,
-          CMARK_NODE_TABLE_CELL, parent_container->start_column + cell->start_offset);
+          CMARK_NODE_TABLE_CELL,
+          parent_container->start_column + cell->start_offset - start_column_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;
+      header_cell->end_column =
+        parent_container->start_column + cell->end_offset - start_column_offset;
       cmark_node_set_string_content(header_cell, (char *) cell->buf->ptr);
       cmark_node_set_syntax_extension(header_cell, self);
     }

From a2792cfeff8376fa38319eaaab4046e35b4058bc Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Martin=20=C4=8C=C3=AD=C5=BEek?= <martin.cizek@orchitech.cz>
Date: Sat, 11 Apr 2020 02:37:51 +0200
Subject: [PATCH 26/27] Add sourcepos to softbreak and linebreak nodes if
 SOURCEPOS requested on parser.

---
 api_test/main.c | 77 ++++++++++++++++++++++++++++---------------------
 src/inlines.c   | 31 +++++++++++++-------
 2 files changed, 65 insertions(+), 43 deletions(-)

diff --git a/api_test/main.c b/api_test/main.c
index 572f251e5..b4502a05f 100755
--- a/api_test/main.c
+++ b/api_test/main.c
@@ -1272,13 +1272,14 @@ static void source_pos(test_batch_runner *runner) {
   }
   {
     static const char markdown[] =
-    "* List 1, item A.\n"
-    "\n"
-    "  <script>\n"
-    "- List 2, item A.\n";
-
-    cmark_node *doc = cmark_parse_document(markdown, sizeof(markdown) - 1, CMARK_OPT_DEFAULT);
-    char *xml = cmark_render_xml(doc, CMARK_OPT_DEFAULT | CMARK_OPT_SOURCEPOS);
+      "* List 1, item A.\n"
+      "\n"
+      "  <script>\n"
+      "- List 2, item A.\n";
+    int options = CMARK_OPT_DEFAULT | CMARK_OPT_SOURCEPOS;
+
+    cmark_node *doc = cmark_parse_document(markdown, sizeof(markdown) - 1, options);
+    char *xml = cmark_render_xml(doc, options);
     STR_EQ(runner, xml, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
            "<!DOCTYPE document SYSTEM \"CommonMark.dtd\">\n"
            "<document sourcepos=\"1:1-4:17\" xmlns=\"http://commonmark.org/xml/1.0\">\n"
@@ -1306,22 +1307,30 @@ static void source_pos(test_batch_runner *runner) {
 }
 
 static void source_pos_inlines(test_batch_runner *runner) {
+  int options = CMARK_OPT_DEFAULT | CMARK_OPT_SOURCEPOS;
   {
     static const char markdown[] =
       "*first*\n"
-      "second\n";
+      "second  \n"
+      "third\\\n"
+      "fourth\n";
 
-    cmark_node *doc = cmark_parse_document(markdown, sizeof(markdown) - 1, CMARK_OPT_DEFAULT);
-    char *xml = cmark_render_xml(doc, CMARK_OPT_DEFAULT | CMARK_OPT_SOURCEPOS);
+    cmark_node *doc = cmark_parse_document(markdown, sizeof(markdown) - 1, options);
+    char *xml = cmark_render_xml(doc, options);
     STR_EQ(runner, xml, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
                         "<!DOCTYPE document SYSTEM \"CommonMark.dtd\">\n"
-                        "<document sourcepos=\"1:1-2:6\" xmlns=\"http://commonmark.org/xml/1.0\">\n"
-                        "  <paragraph sourcepos=\"1:1-2:6\">\n"
+                        "<document sourcepos=\"1:1-4:6\" xmlns=\"http://commonmark.org/xml/1.0\">\n"
+                        "  <paragraph sourcepos=\"1:1-4:6\">\n"
                         "    <emph sourcepos=\"1:1-1:7\">\n"
                         "      <text sourcepos=\"1:2-1:6\" xml:space=\"preserve\">first</text>\n"
                         "    </emph>\n"
-                        "    <softbreak />\n"
-                        "    <text sourcepos=\"2:1-2:6\" xml:space=\"preserve\">second</text>\n"
+                        "    <softbreak sourcepos=\"1:8-2:0\" />\n"
+                        // two-space hard break considered a part of text - correct?
+                        "    <text sourcepos=\"2:1-2:8\" xml:space=\"preserve\">second</text>\n"
+                        "    <linebreak sourcepos=\"2:7-3:0\" />\n"
+                        "    <text sourcepos=\"3:1-3:5\" xml:space=\"preserve\">third</text>\n"
+                        "    <linebreak sourcepos=\"3:6-4:0\" />\n"
+                        "    <text sourcepos=\"4:1-4:6\" xml:space=\"preserve\">fourth</text>\n"
                         "  </paragraph>\n"
                         "</document>\n",
                         "sourcepos are as expected");
@@ -1333,15 +1342,15 @@ static void source_pos_inlines(test_batch_runner *runner) {
       "*first\n"
       "second*\n";
 
-    cmark_node *doc = cmark_parse_document(markdown, sizeof(markdown) - 1, CMARK_OPT_DEFAULT);
-    char *xml = cmark_render_xml(doc, CMARK_OPT_DEFAULT | CMARK_OPT_SOURCEPOS);
+    cmark_node *doc = cmark_parse_document(markdown, sizeof(markdown) - 1, options);
+    char *xml = cmark_render_xml(doc, options);
     STR_EQ(runner, xml, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
                         "<!DOCTYPE document SYSTEM \"CommonMark.dtd\">\n"
                         "<document sourcepos=\"1:1-2:7\" xmlns=\"http://commonmark.org/xml/1.0\">\n"
                         "  <paragraph sourcepos=\"1:1-2:7\">\n"
                         "    <emph sourcepos=\"1:1-2:7\">\n"
                         "      <text sourcepos=\"1:2-1:6\" xml:space=\"preserve\">first</text>\n"
-                        "      <softbreak />\n"
+                        "      <softbreak sourcepos=\"1:7-2:0\" />\n"
                         "      <text sourcepos=\"2:1-2:6\" xml:space=\"preserve\">second</text>\n"
                         "    </emph>\n"
                         "  </paragraph>\n"
@@ -1356,8 +1365,8 @@ static void source_pos_inlines(test_batch_runner *runner) {
     "line\n"
     " breaks](https://commonmark.org).";
 
-    cmark_node *doc = cmark_parse_document(markdown, sizeof(markdown) - 1, CMARK_OPT_DEFAULT);
-    char *xml = cmark_render_xml(doc, CMARK_OPT_DEFAULT | CMARK_OPT_SOURCEPOS);
+    cmark_node *doc = cmark_parse_document(markdown, sizeof(markdown) - 1, options);
+    char *xml = cmark_render_xml(doc, options);
     STR_EQ(runner, xml, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
            "<!DOCTYPE document SYSTEM \"CommonMark.dtd\">\n"
            "<document sourcepos=\"1:1-3:33\" xmlns=\"http://commonmark.org/xml/1.0\">\n"
@@ -1365,9 +1374,9 @@ static void source_pos_inlines(test_batch_runner *runner) {
            "    <text sourcepos=\"1:1-1:24\" xml:space=\"preserve\">This link will have two </text>\n"
            "    <link sourcepos=\"1:25-3:32\" destination=\"https://commonmark.org\" title=\"\">\n"
            "      <text sourcepos=\"1:26-1:30\" xml:space=\"preserve\">soft</text>\n"
-           "      <softbreak />\n"
+           "      <softbreak sourcepos=\"1:31-2:0\" />\n"
            "      <text sourcepos=\"2:1-2:4\" xml:space=\"preserve\">line</text>\n"
-           "      <softbreak />\n"
+           "      <softbreak sourcepos=\"2:5-3:0\" />\n"
            "      <text sourcepos=\"3:2-3:7\" xml:space=\"preserve\">breaks</text>\n"
            "    </link>\n"
            "    <text sourcepos=\"3:33-3:33\" xml:space=\"preserve\">.</text>\n"
@@ -1382,8 +1391,8 @@ static void source_pos_inlines(test_batch_runner *runner) {
     " 1. <http://www.google.com>\n"
     "    <http://www.google.com>\n";
 
-    cmark_node *doc = cmark_parse_document(markdown, sizeof(markdown) - 1, CMARK_OPT_DEFAULT);
-    char *xml = cmark_render_xml(doc, CMARK_OPT_DEFAULT | CMARK_OPT_SOURCEPOS);
+    cmark_node *doc = cmark_parse_document(markdown, sizeof(markdown) - 1, options);
+    char *xml = cmark_render_xml(doc, options);
     STR_EQ(runner, xml, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
            "<!DOCTYPE document SYSTEM \"CommonMark.dtd\">\n"
            "<document sourcepos=\"1:1-2:27\" xmlns=\"http://commonmark.org/xml/1.0\">\n"
@@ -1393,7 +1402,7 @@ static void source_pos_inlines(test_batch_runner *runner) {
            "        <link sourcepos=\"1:5-1:27\" destination=\"http://www.google.com\" title=\"\">\n"
            "          <text sourcepos=\"1:6-1:26\" xml:space=\"preserve\">http://www.google.com</text>\n"
            "        </link>\n"
-           "        <softbreak />\n"
+           "        <softbreak sourcepos=\"1:28-2:0\" />\n"
            "        <link sourcepos=\"2:5-2:27\" destination=\"http://www.google.com\" title=\"\">\n"
            "          <text sourcepos=\"2:6-2:26\" xml:space=\"preserve\">http://www.google.com</text>\n"
            "        </link>\n"
@@ -1411,8 +1420,8 @@ static void source_pos_inlines(test_batch_runner *runner) {
     "===============\n"
     "A paragraph.\n";
 
-    cmark_node *doc = cmark_parse_document(markdown, sizeof(markdown) - 1, CMARK_OPT_DEFAULT);
-    char *xml = cmark_render_xml(doc, CMARK_OPT_DEFAULT | CMARK_OPT_SOURCEPOS);
+    cmark_node *doc = cmark_parse_document(markdown, sizeof(markdown) - 1, options);
+    char *xml = cmark_render_xml(doc, options);
     STR_EQ(runner, xml, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
            "<!DOCTYPE document SYSTEM \"CommonMark.dtd\">\n"
            "<document sourcepos=\"1:1-3:12\" xmlns=\"http://commonmark.org/xml/1.0\">\n"
@@ -1433,8 +1442,8 @@ static void source_pos_inlines(test_batch_runner *runner) {
     "\n"
     "> # Header 1 #\n";
 
-    cmark_node *doc = cmark_parse_document(markdown, sizeof(markdown) - 1, CMARK_OPT_DEFAULT);
-    char *xml = cmark_render_xml(doc, CMARK_OPT_DEFAULT | CMARK_OPT_SOURCEPOS);
+    cmark_node *doc = cmark_parse_document(markdown, sizeof(markdown) - 1, options);
+    char *xml = cmark_render_xml(doc, options);
     STR_EQ(runner, xml, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
            "<!DOCTYPE document SYSTEM \"CommonMark.dtd\">\n"
            "<document sourcepos=\"1:1-3:14\" xmlns=\"http://commonmark.org/xml/1.0\">\n"
@@ -1454,13 +1463,15 @@ static void source_pos_inlines(test_batch_runner *runner) {
 }
 
 static void ref_source_pos(test_batch_runner *runner) {
+  int options = CMARK_OPT_DEFAULT | CMARK_OPT_SOURCEPOS;
+
   static const char markdown[] =
     "Let's try [reference] links.\n"
     "\n"
     "[reference]: https://github.com (GitHub)\n";
 
-  cmark_node *doc = cmark_parse_document(markdown, sizeof(markdown) - 1, CMARK_OPT_DEFAULT);
-  char *xml = cmark_render_xml(doc, CMARK_OPT_DEFAULT | CMARK_OPT_SOURCEPOS);
+  cmark_node *doc = cmark_parse_document(markdown, sizeof(markdown) - 1, options);
+  char *xml = cmark_render_xml(doc, options);
   STR_EQ(runner, xml, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
                       "<!DOCTYPE document SYSTEM \"CommonMark.dtd\">\n"
                       "<document sourcepos=\"1:1-3:40\" xmlns=\"http://commonmark.org/xml/1.0\">\n"
@@ -1483,6 +1494,7 @@ static void ext_source_pos(test_batch_runner *runner) {
     "table",
     "autolink",
   };
+  int options = CMARK_OPT_DEFAULT | CMARK_OPT_SOURCEPOS;
 
   static const char markdown[] =
     "Hi ~~friend~~ and ~~other\n"
@@ -1494,7 +1506,6 @@ static void ext_source_pos(test_batch_runner *runner) {
     "   | - | - | --: |\n"
     "   | 1 | 2 | ~3~ |\n";
 
-  int options = CMARK_OPT_DEFAULT | CMARK_OPT_SOURCEPOS;
   cmark_parser *parser = cmark_parser_new(options);
 
   for (int i = 0; i < (int)(sizeof(extensions) / sizeof(*extensions)); ++i) {
@@ -1517,7 +1528,7 @@ static void ext_source_pos(test_batch_runner *runner) {
                       "    <text sourcepos=\"1:14-1:18\" xml:space=\"preserve\"> and </text>\n"
                       "    <strikethrough sourcepos=\"1:19-2:8\">\n"
                       "      <text sourcepos=\"1:21-1:25\" xml:space=\"preserve\">other</text>\n"
-                      "      <softbreak />\n"
+                      "      <softbreak sourcepos=\"1:26-2:0\" />\n"
                       "      <text sourcepos=\"2:1-2:6\" xml:space=\"preserve\">friend</text>\n"
                       "    </strikethrough>\n"
                       "    <text sourcepos=\"2:9-2:9\" xml:space=\"preserve\">.</text>\n"
@@ -1574,6 +1585,7 @@ static void ext_source_pos_corners(test_batch_runner *runner) {
     "table",
     "autolink",
   };
+  int options = CMARK_OPT_DEFAULT | CMARK_OPT_SOURCEPOS;
 
   // a paragraph inserted by the table extension
   static const char markdown[] =
@@ -1581,7 +1593,6 @@ static void ext_source_pos_corners(test_batch_runner *runner) {
     "| header |\n" //+3
     "| --- |\n";
 
-  int options = CMARK_OPT_DEFAULT | CMARK_OPT_SOURCEPOS;
   cmark_parser *parser = cmark_parser_new(options);
 
   for (int i = 0; i < (int)(sizeof(extensions) / sizeof(*extensions)); ++i) {
diff --git a/src/inlines.c b/src/inlines.c
index 9471396f5..e6b466d40 100755
--- a/src/inlines.c
+++ b/src/inlines.c
@@ -830,11 +830,17 @@ static cmark_node *handle_backslash(cmark_parser *parser, subject *subj) {
     advance(subj);
     return make_str(subj, subj->pos - 2, subj->pos - 1, cmark_chunk_dup(&subj->input, subj->pos - 1, 1));
   } else if (!is_eof(subj) && skip_line_end(subj)) {
+    cmark_node *res = make_linebreak(subj->mem);
+
+    if (parser->options & CMARK_OPT_SOURCEPOS) {
+      res->start_line = subj->line;
+      res->start_column = subj->pos - 1 + subj->column_offset + subj->block_offset;
+      res->end_line = subj->line + 1;
+    }
     // Adjust the subject source position state.
     ++subj->line;
     subj->column_offset = -subj->pos;
-
-    return make_linebreak(subj->mem);
+    return res;
   } else {
     return make_str(subj, subj->pos - 1, subj->pos - 1, cmark_chunk_literal("\\"));
   }
@@ -1241,8 +1247,9 @@ static cmark_node *handle_close_bracket(cmark_parser *parser, subject *subj) {
 
 // Parse a hard or soft linebreak, returning an inline.
 // Assumes the subject has a cr or newline at the current position.
-static cmark_node *handle_newline(subject *subj) {
+static cmark_node *handle_newline(subject *subj, int options) {
   bufsize_t nlpos = subj->pos;
+  cmark_node *res;
   // skip over cr, crlf, or lf:
   if (peek_at(subj, subj->pos) == '\r') {
     advance(subj);
@@ -1250,16 +1257,20 @@ static cmark_node *handle_newline(subject *subj) {
   if (peek_at(subj, subj->pos) == '\n') {
     advance(subj);
   }
+  res = nlpos > 1 && peek_at(subj, nlpos - 1) == ' ' && peek_at(subj, nlpos - 2) == ' '
+    ? make_linebreak(subj->mem)
+    : make_softbreak(subj->mem);
+  if (options & CMARK_OPT_SOURCEPOS) {
+    res->start_line = subj->line;
+    res->start_column = nlpos + subj->column_offset + subj->block_offset;
+    res->start_column += res->type == CMARK_NODE_LINEBREAK ? -1 : 1;
+    res->end_line = subj->line + 1;
+  }
   ++subj->line;
   subj->column_offset = -subj->pos;
   // skip spaces at beginning of line
   skip_spaces(subj);
-  if (nlpos > 1 && peek_at(subj, nlpos - 1) == ' ' &&
-      peek_at(subj, nlpos - 2) == ' ') {
-    return make_linebreak(subj->mem);
-  } else {
-    return make_softbreak(subj->mem);
-  }
+  return res;
 }
 
 // "\r\n\\`&_*[]<!"
@@ -1360,7 +1371,7 @@ static int parse_inline(cmark_parser *parser, subject *subj, cmark_node *parent,
   switch (c) {
   case '\r':
   case '\n':
-    new_inl = handle_newline(subj);
+    new_inl = handle_newline(subj, options);
     break;
   case '`':
     new_inl = handle_backticks(subj, options);

From 80548df2536ce219bb9528c9dc12d2ccb2853359 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Martin=20=C4=8C=C3=AD=C5=BEek?= <martin.cizek@orchitech.cz>
Date: Tue, 4 May 2021 23:51:40 +0200
Subject: [PATCH 27/27] Fix shifted paragraph table offset introduced in
 ddf21bb.

---
 extensions/table.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/extensions/table.c b/extensions/table.c
index 7e890fdaf..9fb1012d5 100644
--- a/extensions/table.c
+++ b/extensions/table.c
@@ -366,7 +366,7 @@ static cmark_node *try_opening_table_header(cmark_syntax_extension *self,
     cmark_node *paragraph = try_inserting_table_header_paragraph(parser,
       parent_container, (unsigned char *)parent_string, header_row->paragraph_offset);
     if (paragraph) {
-      start_column_offset = header_row->paragraph_offset + 1;
+      start_column_offset = header_row->paragraph_offset;
     }
   }