Commit: 1ec665f4fddae9acb7f979e2849803b595d108d8 - luce (git)

A Markdown shard for the Crystal programming language

Commit MetaInfo

• Révision: 1ec665f4fddae9acb7f979e2849803b595d108d8 (tree)
• l'heure: 2023-04-07 11:07:45
• Auteur: supercell <stigma@disr...>
• Commiter: supercell

Message de Log

Optimize TableSyntax

Change Summary

• modified: spec/extensions/tables.unit
• modified: spec/gfm/tables_extension.unit
• modified: src/luce/block_parser.cr
• modified: src/luce/block_syntaxes/table_syntax.cr
• modified: src/luce/html_renderer.cr
• modified: tools/gfm_stats.json
• modified: tools/gfm_stats.txt

Modification

--- a/spec/extensions/tables.unit
+++ b/spec/extensions/tables.unit
@@ -4,7 +4,20 @@ head | cells
 body | cells
 
 <<<
-<table><thead><tr><th>head</th><th>cells</th></tr></thead><tbody><tr><td>body</td><td>cells</td></tr></tbody></table>
+<table>
+<thead>
+<tr>
+<th>head</th>
+<th>cells</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>body</td>
+<td>cells</td>
+</tr>
+</tbody>
+</table>
 >>> multiple rows
 head | cells
 -----|------
@@ -12,40 +25,124 @@ body | cells
 more | cells
 
 <<<
-<table><thead><tr><th>head</th><th>cells</th></tr></thead><tbody><tr><td>body</td><td>cells</td></tr><tr><td>more</td><td>cells</td></tr></tbody></table>
+<table>
+<thead>
+<tr>
+<th>head</th>
+<th>cells</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>body</td>
+<td>cells</td>
+</tr>
+<tr>
+<td>more</td>
+<td>cells</td>
+</tr>
+</tbody>
+</table>
 >>> rows wrapped in pipes
 | head | cells |
 |------|-------|
 | body | cells |
 
 <<<
-<table><thead><tr><th>head</th><th>cells</th></tr></thead><tbody><tr><td>body</td><td>cells</td></tr></tbody></table>
+<table>
+<thead>
+<tr>
+<th>head</th>
+<th>cells</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>body</td>
+<td>cells</td>
+</tr>
+</tbody>
+</table>
 >>> rows wrapped in pipes, whitespace alignment row
 | head | cells |
 |  --  |  ---  |
 | body | cells |
 
 <<<
-<table><thead><tr><th>head</th><th>cells</th></tr></thead><tbody><tr><td>body</td><td>cells</td></tr></tbody></table>
+<table>
+<thead>
+<tr>
+<th>head</th>
+<th>cells</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>body</td>
+<td>cells</td>
+</tr>
+</tbody>
+</table>
 >>> rows wrapped in pipes, tabs in whitespace
 |	head	|	cells	|
 |	--	|	---	|
 |	body	|	cells	|
 
 <<<
-<table><thead><tr><th>head</th><th>cells</th></tr></thead><tbody><tr><td>body</td><td>cells</td></tr></tbody></table>
+<table>
+<thead>
+<tr>
+<th>head</th>
+<th>cells</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>body</td>
+<td>cells</td>
+</tr>
+</tbody>
+</table>
 >>> cells with inline syntax
 head `code` | _cells_
 ------------|--------
 *text*      | <span>text</span>
 <<<
-<table><thead><tr><th>head <code>code</code></th><th><em>cells</em></th></tr></thead><tbody><tr><td><em>text</em></td><td><span>text</span></td></tr></tbody></table>
+<table>
+<thead>
+<tr>
+<th>head <code>code</code></th>
+<th><em>cells</em></th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td><em>text</em></td>
+<td><span>text</span></td>
+</tr>
+</tbody>
+</table>
 >>> cells are parsed before inline syntax
 header | _foo | bar_
 -------|------------|---
 text   | text
 <<<
-<table><thead><tr><th>header</th><th>_foo</th><th>bar_</th></tr></thead><tbody><tr><td>text</td><td>text</td><td /></tr></tbody></table>
+<table>
+<thead>
+<tr>
+<th>header</th>
+<th>_foo</th>
+<th>bar_</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>text</td>
+<td>text</td>
+<td></td>
+</tr>
+</tbody>
+</table>
 >>> cells contain reference links
 header | header
 -------|--------
@@ -53,20 +150,64 @@ text   | [link][here]
 
 [here]: http://url
 <<<
-<table><thead><tr><th>header</th><th>header</th></tr></thead><tbody><tr><td>text</td><td><a href="http://url">link</a></td></tr></tbody></table>
+<table>
+<thead>
+<tr>
+<th>header</th>
+<th>header</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>text</td>
+<td><a href="http://url">link</a></td>
+</tr>
+</tbody>
+</table>
 >>> one column tables
 head
 -----|
 body
 <<<
-<table><thead><tr><th>head</th></tr></thead><tbody><tr><td>body</td></tr></tbody></table>
+<table>
+<thead>
+<tr>
+<th>head</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>body</td>
+</tr>
+</tbody>
+</table>
 >>> varying cells per row
 head | foo | bar
 -----|-----|-----
 body
 row with | two cells
 <<<
-<table><thead><tr><th>head</th><th>foo</th><th>bar</th></tr></thead><tbody><tr><td>body</td><td /><td /></tr><tr><td>row with</td><td>two cells</td><td /></tr></tbody></table>
+<table>
+<thead>
+<tr>
+<th>head</th>
+<th>foo</th>
+<th>bar</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>body</td>
+<td></td>
+<td></td>
+</tr>
+<tr>
+<td>row with</td>
+<td>two cells</td>
+<td></td>
+</tr>
+</tbody>
+</table>
 >>> left, center, and right alignment
 head | cells | here
 :----|:-----:|----:
@@ -74,7 +215,27 @@ body | cells | here
 too | many | cells | here
 
 <<<
-<table><thead><tr><th style="text-align: left;">head</th><th style="text-align: center;">cells</th><th style="text-align: right;">here</th></tr></thead><tbody><tr><td style="text-align: left;">body</td><td style="text-align: center;">cells</td><td style="text-align: right;">here</td></tr><tr><td style="text-align: left;">too</td><td style="text-align: center;">many</td><td style="text-align: right;">cells</td></tr></tbody></table>
+<table>
+<thead>
+<tr>
+<th align="left">head</th>
+<th align="center">cells</th>
+<th align="right">here</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td align="left">body</td>
+<td align="center">cells</td>
+<td align="right">here</td>
+</tr>
+<tr>
+<td align="left">too</td>
+<td align="center">many</td>
+<td align="right">cells</td>
+</tr>
+</tbody>
+</table>
 >>> left, center, and right alignment, with whitespace
 head | cells | here
  :-- | :---: | ---:
@@ -82,7 +243,27 @@ body | cells | here
 too | many | cells | here
 
 <<<
-<table><thead><tr><th style="text-align: left;">head</th><th style="text-align: center;">cells</th><th style="text-align: right;">here</th></tr></thead><tbody><tr><td style="text-align: left;">body</td><td style="text-align: center;">cells</td><td style="text-align: right;">here</td></tr><tr><td style="text-align: left;">too</td><td style="text-align: center;">many</td><td style="text-align: right;">cells</td></tr></tbody></table>
+<table>
+<thead>
+<tr>
+<th align="left">head</th>
+<th align="center">cells</th>
+<th align="right">here</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td align="left">body</td>
+<td align="center">cells</td>
+<td align="right">here</td>
+</tr>
+<tr>
+<td align="left">too</td>
+<td align="center">many</td>
+<td align="right">cells</td>
+</tr>
+</tbody>
+</table>
 >>> escape pipe
 | Name     | Character |
 | ---      | ---       |
@@ -90,17 +271,60 @@ too | many | cells | here
 | Pipe     | \|        |
 
 <<<
-<table><thead><tr><th>Name</th><th>Character</th></tr></thead><tbody><tr><td>Backtick</td><td>`</td></tr><tr><td>Pipe</td><td>|</td></tr></tbody></table>
+<table>
+<thead>
+<tr>
+<th>Name</th>
+<th>Character</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>Backtick</td>
+<td>`</td>
+</tr>
+<tr>
+<td>Pipe</td>
+<td>|</td>
+</tr>
+</tbody>
+</table>
 >>> escape pipe, preserve trailing whitespace
 | Name     | Character |
 | ---      | ---       |
 | Pipe     | \| abcdef |
 
 <<<
-<table><thead><tr><th>Name</th><th>Character</th></tr></thead><tbody><tr><td>Pipe</td><td>| abcdef</td></tr></tbody></table>
+<table>
+<thead>
+<tr>
+<th>Name</th>
+<th>Character</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>Pipe</td>
+<td>| abcdef</td>
+</tr>
+</tbody>
+</table>
 >>> trailing whitespace after final pipe
 | Name     | Character | 
 | ---      | ---       |	
 | Pipe     | abcdef    |
 <<<
-<table><thead><tr><th>Name</th><th>Character</th></tr></thead><tbody><tr><td>Pipe</td><td>abcdef</td></tr></tbody></table>
+<table>
+<thead>
+<tr>
+<th>Name</th>
+<th>Character</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>Pipe</td>
+<td>abcdef</td>
+</tr>
+</tbody>
+</table>

--- a/spec/gfm/tables_extension.unit
+++ b/spec/gfm/tables_extension.unit
@@ -3,27 +3,80 @@
 | --- | --- |
 | baz | bim |
 <<<
-<table><thead><tr><th>foo</th><th>bar</th></tr></thead><tbody><tr><td>baz</td><td>bim</td></tr></tbody></table>
+<table>
+<thead>
+<tr>
+<th>foo</th>
+<th>bar</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>baz</td>
+<td>bim</td>
+</tr>
+</tbody>
+</table>
 >>> Tables (extension) - 199
 | abc | defghi |
 :-: | -----------:
 bar | baz
 <<<
-<table><thead><tr><th style="text-align: center;">abc</th><th style="text-align: right;">defghi</th></tr></thead><tbody><tr><td style="text-align: center;">bar</td><td style="text-align: right;">baz</td></tr></tbody></table>
+<table>
+<thead>
+<tr>
+<th align="center">abc</th>
+<th align="right">defghi</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td align="center">bar</td>
+<td align="right">baz</td>
+</tr>
+</tbody>
+</table>
 >>> Tables (extension) - 200
 | f\|oo  |
 | ------ |
 | b `\|` az |
 | b **\|** im |
 <<<
-<table><thead><tr><th>f|oo</th></tr></thead><tbody><tr><td>b <code>|</code> az</td></tr><tr><td>b <strong>|</strong> im</td></tr></tbody></table>
+<table>
+<thead>
+<tr>
+<th>f|oo</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>b <code>|</code> az</td>
+</tr>
+<tr>
+<td>b <strong>|</strong> im</td>
+</tr>
+</tbody>
+</table>
 >>> Tables (extension) - 201
 | abc | def |
 | --- | --- |
 | bar | baz |
 > bar
 <<<
-<table><thead><tr><th>abc</th><th>def</th></tr></thead><tbody><tr><td>bar</td><td>baz</td></tr></tbody></table>
+<table>
+<thead>
+<tr>
+<th>abc</th>
+<th>def</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>bar</td>
+<td>baz</td>
+</tr>
+</tbody>
+</table>
 <blockquote>
 <p>bar</p>
 </blockquote>
@@ -35,14 +88,32 @@ bar
 
 bar
 <<<
-<table><thead><tr><th>abc</th><th>def</th></tr></thead><tbody><tr><td>bar</td><td>baz</td></tr><tr><td>bar</td><td /></tr></tbody></table>
+<table>
+<thead>
+<tr>
+<th>abc</th>
+<th>def</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>bar</td>
+<td>baz</td>
+</tr>
+<tr>
+<td>bar</td>
+<td></td>
+</tr>
+</tbody>
+</table>
 <p>bar</p>
 >>> Tables (extension) - 203
 | abc | def |
 | --- |
 | bar |
 <<<
-<p>| --- |
+<p>| abc | def |
+| --- |
 | bar |</p>
 >>> Tables (extension) - 204
 | abc | def |
@@ -50,9 +121,33 @@ bar
 | bar |
 | bar | baz | boo |
 <<<
-<table><thead><tr><th>abc</th><th>def</th></tr></thead><tbody><tr><td>bar</td><td /></tr><tr><td>bar</td><td>baz</td></tr></tbody></table>
+<table>
+<thead>
+<tr>
+<th>abc</th>
+<th>def</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>bar</td>
+<td></td>
+</tr>
+<tr>
+<td>bar</td>
+<td>baz</td>
+</tr>
+</tbody>
+</table>
 >>> Tables (extension) - 205
 | abc | def |
 | --- | --- |
 <<<
-<table><thead><tr><th>abc</th><th>def</th></tr></thead></table>
+<table>
+<thead>
+<tr>
+<th>abc</th>
+<th>def</th>
+</tr>
+</thead>
+</table>

--- a/src/luce/block_parser.cr
+++ b/src/luce/block_parser.cr
@@ -88,6 +88,10 @@ module Luce
       @pos += 1
     end
 
+    def retreat : Nil
+      @pos -= 1
+    end
+
     def done? : Bool
       @pos >= @lines.size
     end
@@ -106,11 +110,23 @@ module Luce
 
     def parse_lines : Array(Node)
       blocks = [] of Node
+
+      # If the @pos does not change before and after `parse()`, never try to
+      # parse the line at @pos with the same syntax again.
+      # For example, the `TableSyntax` might not advance the @pos in `parse`
+      # method, because the header row does not match the delimiter row in the
+      # number of cells, which makes the table like structure not be recognized.
+      never_match : BlockSyntax? = nil
+
       until done?
         block_syntaxes.each do |syntax|
+          next if never_match == syntax
+
           if syntax.can_parse? self
+            position_before = @pos
             block = syntax.parse self
             blocks << block unless block.nil?
+            never_match = @pos != position_before ? nil : syntax
             break
           end
         end

--- a/src/luce/block_syntaxes/table_syntax.cr
+++ b/src/luce/block_syntaxes/table_syntax.cr
@@ -28,7 +28,10 @@ module Luce
       alignments = parse_alignments(parser.next.not_nil!)
       column_count = alignments.size
       head_row = parse_row(parser, alignments, "th")
-      return nil if head_row.children.not_nil!.size != column_count
+      if head_row.children.not_nil!.size != column_count
+        parser.retreat
+        return nil
+      end
 
       head = Element.new("thead", [head_row.as Node])
 
@@ -42,7 +45,7 @@ module Luce
         if !children.nil?
           while children.size < column_count
             # Insert synthetic empty cells.
-            children << Element.empty("td").as Node
+            children << Element.new("td", [] of Node)
           end
           while children.size > column_count
             children.pop
@@ -62,29 +65,37 @@ module Luce
     end
 
     private def parse_alignments(line : String) : Array(String?)
-      start_index = walk_past_opening_pipe(line)
-
-      end_index = line.size - 1
-      while end_index > 0
-        ch = line.codepoint_at(end_index)
-        if ch == Charcode::PIPE
-          end_index -= 1
-          break
+      columns = [] of String?
+      # Set the value to `true` when hitting a non whitespace character other
+      # than the first pipe character.
+      started = false
+      hit_dash = false
+      alignment : String? = nil
+
+      line.each_codepoint do |char|
+        next if char == Charcode::SPACE || char == Charcode::TAB || (!started && char == Charcode::PIPE)
+        started = true
+
+        if char == Charcode::COLON
+          if hit_dash
+            alignment = alignment == "left" ? "center" : "right"
+          else
+            alignment = "left"
+          end
         end
-        if ch != Charcode::SPACE && ch != Charcode::TAB
-          break
+
+        if char == Charcode::PIPE
+          columns << alignment
+          hit_dash = false
+          alignment = nil
+        else
+          hit_dash = true
         end
-        end_index -= 1
       end
 
-      # Optimization: We walk *line* too many times. One lap should do it.
-      line[start_index..end_index].split('|').map do |column|
-        column = column.strip
-        next "center" if column.starts_with?(":") && column.ends_with?(":")
-        next "left" if column.starts_with?(":")
-        next "right" if column.ends_with?(":")
-        nil
-      end
+      columns << alignment if hit_dash
+
+      columns
     end
 
     # Parses a table row at the current line into a table row element, with
@@ -152,7 +163,7 @@ module Luce
           next
         end
 
-        (row[i].as Element).attributes["style"] = "text-align: #{alignments[i]};"
+        (row[i].as Element).attributes["align"] = "#{alignments[i]}"
         i += 1
       end
 

--- a/src/luce/html_renderer.cr
+++ b/src/luce/html_renderer.cr
@@ -36,6 +36,11 @@ module Luce
     "nav",
     "section",
     "table",
+    "thead",
+    "tbody",
+    "th",
+    "tr",
+    "td",
   ]
 
   protected def self.block_tags : Array(String)

--- a/tools/gfm_stats.json
+++ b/tools/gfm_stats.json
@@ -219,14 +219,14 @@
   "197": "strict"
  },
  "Tables (extension)": {
-  "198": "fail",
-  "199": "fail",
-  "200": "fail",
-  "201": "fail",
-  "202": "fail",
-  "203": "loose",
-  "204": "fail",
-  "205": "fail"
+  "198": "strict",
+  "199": "strict",
+  "200": "strict",
+  "201": "strict",
+  "202": "strict",
+  "203": "strict",
+  "204": "strict",
+  "205": "strict"
  },
  "Block quotes": {
   "206": "strict",

--- a/tools/gfm_stats.txt
+++ b/tools/gfm_stats.txt
@@ -9,7 +9,7 @@
   21 of   28 -  75.0%  Link reference definitions
    8 of    8 - 100.0%  Paragraphs
    1 of    1 - 100.0%  Blank lines
-   1 of    8 -  12.5%  Tables (extension)
+   8 of    8 - 100.0%  Tables (extension)
   23 of   25 -  92.0%  Block quotes
   45 of   48 -  93.8%  List items
   19 of   26 -  73.1%  Lists
@@ -28,5 +28,5 @@
   15 of   15 - 100.0%  Hard line breaks
    2 of    2 - 100.0%  Soft line breaks
    3 of    3 - 100.0%  Textual content
- 614 of  671 -  91.5%  TOTAL
- 561 of  614 -  91.4%  TOTAL Strict
+ 621 of  671 -  92.5%  TOTAL
+ 569 of  621 -  91.6%  TOTAL Strict