Fix table cannot be recoginsed without empty line (commonmark#154)

* fix(table): recoginse-without-empty-line (commonmark#141)

* fix(table): fix bufsize_t not convert to uint16_t

* fix(table): fix uint16_6 not convert to int

* fix(table): fix uint16_6 not convert to int

* fix(table): clear unused type conversion

* restore whitespace

* Always free `paragraph_content`

`cmark_node_set_string_content` allocates and copies the data in
`paragraph_content` so it is not needed afterwards.

```
=================================================================
==14==ERROR: LeakSanitizer: detected memory leaks

Direct leak of 24 byte(s) in 1 object(s) allocated from:
    0 0x4dd330 in calloc /src/llvm/projects/compiler-rt/lib/asan/asan_malloc_linux.cc:97
    1 0x59e243 in xcalloc /src/octofuzz/src/cmark.c:18:15
    2 0x58fd75 in unescape_pipes /src/octofuzz/extensions/table.c:95:39
    3 0x58fd75 in try_inserting_table_header_paragraph /src/octofuzz/extensions/table.c:187
    4 0x58fd75 in try_opening_table_header /src/octofuzz/extensions/table.c:254
    5 0x58fd75 in try_opening_table_block /src/octofuzz/extensions/table.c:370
    6 0x5b22d5 in open_new_blocks /src/octofuzz/src/blocks.c:1275:27
    7 0x5b22d5 in S_process_line /src/octofuzz/src/blocks.c:1465
    8 0x5aa7f0 in cmark_parser_finish /src/octofuzz/src/blocks.c:1492:5
    9 0x58f2fc in LLVMFuzzerTestOneInput /src/octofuzz/test/cmark-fuzz.c:46:23

Indirect leak of 8 byte(s) in 1 object(s) allocated from:
    0 0x4dd580 in realloc /src/llvm/projects/compiler-rt/lib/asan/asan_malloc_linux.cc:107
    1 0x59e2d3 in xrealloc /src/octofuzz/src/cmark.c:27:19
    2 0x640364 in cmark_strbuf_grow /src/octofuzz/src/buffer.c:57:31
    3 0x640364 in cmark_strbuf_init /src/octofuzz/src/buffer.c:31
    4 0x58fd8b in unescape_pipes /src/octofuzz/extensions/table.c:98:3
    5 0x58fd8b in try_inserting_table_header_paragraph /src/octofuzz/extensions/table.c:187
    6 0x58fd8b in try_opening_table_header /src/octofuzz/extensions/table.c:254
    7 0x58fd8b in try_opening_table_block /src/octofuzz/extensions/table.c:370
    8 0x5b22d5 in open_new_blocks /src/octofuzz/src/blocks.c:1275:27
    9 0x5b22d5 in S_process_line /src/octofuzz/src/blocks.c:1465
    10 0x5aa7f0 in cmark_parser_finish /src/octofuzz/src/blocks.c:1492:5
    11 0x58f2fc in LLVMFuzzerTestOneInput /src/octofuzz/test/cmark-fuzz.c:46:23

SUMMARY: AddressSanitizer: 32 byte(s) leaked in 2 allocation(s).
```

Author: jinhucheung

.gitignore

@@ -32,6 +32,7 @@ bstrlib.txt
 build
 cmark.dSYM/*
 cmark
+.vscode
 
 # Testing and benchmark
 alltests.md

extensions/table.c

@@ -16,6 +16,7 @@ cmark_node_type CMARK_NODE_TABLE, CMARK_NODE_TABLE_ROW,
 
 typedef struct {
   uint16_t n_columns;
+  int paragraph_offset;
   cmark_llist *cells;
 } table_row;
 
@@ -115,6 +116,7 @@ static table_row *row_from_string(cmark_syntax_extension *self,
                                   int len) {
   table_row *row = NULL;
   bufsize_t cell_matched = 1, pipe_matched = 1, offset;
+  int cell_end_offset;
 
   row = (table_row *)parser->mem->calloc(1, sizeof(table_row));
   row->n_columns = 0;
@@ -129,20 +131,32 @@ static table_row *row_from_string(cmark_syntax_extension *self,
     pipe_matched = scan_table_cell_end(string, len, offset + cell_matched);
 
     if (cell_matched || pipe_matched) {
-      cmark_strbuf *cell_buf = unescape_pipes(parser->mem, string + offset,
-          cell_matched);
-      cmark_strbuf_trim(cell_buf);
-
-      node_cell *cell = (node_cell *)parser->mem->calloc(1, sizeof(*cell));
-      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;
+      cell_end_offset = offset + cell_matched - 1;
+
+      if (string[cell_end_offset] == '\n' || string[cell_end_offset] == '\r') {
+        row->paragraph_offset = cell_end_offset;
+
+        cmark_llist_free_full(parser->mem, row->cells, (cmark_free_func)free_table_cell);
+        row->cells = NULL;
+        row->n_columns = 0;
+      } else {
+        cmark_strbuf *cell_buf = unescape_pipes(parser->mem, string + offset,
+            cell_matched);
+        cmark_strbuf_trim(cell_buf);
+
+        node_cell *cell = (node_cell *)parser->mem->calloc(1, sizeof(*cell));
+        cell->buf = cell_buf;
+        cell->start_offset = offset;
+        cell->end_offset = cell_end_offset;
+
+        while (cell->start_offset > 0 && string[cell->start_offset - 1] != '|') {
+          --cell->start_offset;
+          ++cell->internal_offset;
+        }
+
+        row->n_columns += 1;
+        row->cells = cmark_llist_append(parser->mem, row->cells, cell);
       }
-      row->n_columns += 1;
-      row->cells = cmark_llist_append(parser->mem, row->cells, cell);
     }
 
     offset += cell_matched + pipe_matched;
@@ -161,6 +175,26 @@ 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,
@@ -217,6 +251,11 @@ static cmark_node *try_opening_table_header(cmark_syntax_extension *self,
     return parent_container;
   }
 
+  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));

test/extensions.txt

@@ -472,6 +472,32 @@ Here's a link to [Freedom Planet 2][].
 </table>
 ````````````````````````````````
 
+### a table can be recognised when separated from a paragraph of text without an empty line
+
+```````````````````````````````` example
+123
+456
+| a | b |
+| ---| --- |
+d | e
+.
+<p>123
+456</p>
+<table>
+<thead>
+<tr>
+<th>a</th>
+<th>b</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>d</td>
+<td>e</td>
+</tr>
+</tbody>
+</table>
+````````````````````````````````
 
 ## Strikethroughs