Markdown: correctly group paragraphs regarding lists, quotes and links

spec/std/markdown/markdown_spec.cr

@@ -3,7 +3,7 @@ require "markdown"
 
 private def assert_render(input, output, file = __FILE__, line = __LINE__)
   it "renders #{input.inspect}", file, line do
-    Markdown.to_html(input).should eq(output)
+    Markdown.to_html(input).should eq(output), file, line
   end
 end
 
@@ -70,25 +70,32 @@ describe Markdown do
   assert_render "Hello\n```\nWorld\n```", "<p>Hello</p>\n\n<pre><code>World</code></pre>"
 
   assert_render "> Hello World\n", "<blockquote>Hello World</blockquote>"
+  assert_render "> This spawns\nmultiple\nlines\n\ntext", "<blockquote>This spawns\nmultiple\nlines</blockquote>\n\n<p>text</p>"
 
   assert_render "* Hello", "<ul><li>Hello</li></ul>"
   assert_render "* Hello\n* World", "<ul><li>Hello</li><li>World</li></ul>"
   assert_render "* Hello\n* World\n  * Crystal", "<ul><li>Hello</li><li>World</li><ul><li>Crystal</li></ul></ul>"
   assert_render "* Level1\n  * Level2\n  * Level2\n* Level1", "<ul><li>Level1</li><ul><li>Level2</li><li>Level2</li></ul><li>Level1</li></ul>"
   assert_render "* Level1\n  * Level2\n  * Level2", "<ul><li>Level1</li><ul><li>Level2</li><li>Level2</li></ul></ul>"
-  assert_render "* Hello\nWorld", "<ul><li>Hello</li></ul>\n\n<p>World</p>"
+  assert_render "* Hello\nWorld", "<ul><li>Hello\nWorld</li></ul>"
   assert_render "Params:\n* Foo\n* Bar", "<p>Params:</p>\n\n<ul><li>Foo</li><li>Bar</li></ul>"
 
   assert_render "+ Hello", "<ul><li>Hello</li></ul>"
   assert_render "- Hello", "<ul><li>Hello</li></ul>"
 
   assert_render "* Hello\n+ World\n- Crystal", "<ul><li>Hello</li></ul>\n\n<ul><li>World</li></ul>\n\n<ul><li>Crystal</li></ul>"
 
+  assert_render "* This spawns\nmultiple\nlines\n\ntext", "<ul><li>This spawns\nmultiple\nlines</li></ul>\n\n<p>text</p>"
+  assert_render "* Two\nlines\n* This spawns\nmultiple\nlines\n\ntext", "<ul><li>Two\nlines</li><li>This spawns\nmultiple\nlines</li></ul>\n\n<p>text</p>"
+
   assert_render "1. Hello", "<ol><li>Hello</li></ol>"
   assert_render "2. Hello", "<ol><li>Hello</li></ol>"
   assert_render "01. Hello\n02. World", "<ol><li>Hello</li><li>World</li></ol>"
   assert_render "Params:\n  1. Foo\n  2. Bar", "<p>Params:</p>\n\n<ol><li>Foo</li><li>Bar</li></ol>"
 
+  assert_render "1. This spawns\nmultiple\nlines\n\ntext", "<ol><li>This spawns\nmultiple\nlines</li></ol>\n\n<p>text</p>"
+  assert_render "1. Two\nlines\n1. This spawns\nmultiple\nlines\n\ntext", "<ol><li>Two\nlines</li><li>This spawns\nmultiple\nlines</li></ol>\n\n<p>text</p>"
+
   assert_render "Hello [world](http://example.com)", %(<p>Hello <a href="http://example.com">world</a></p>)
   assert_render "Hello [world](http://example.com)!", %(<p>Hello <a href="http://example.com">world</a>!</p>)
   assert_render "Hello [world **2**](http://example.com)!", %(<p>Hello <a href="http://example.com">world <strong>2</strong></a>!</p>)
@@ -98,6 +105,9 @@ describe Markdown do
 
   assert_render "[![foo](bar)](baz)", %(<p><a href="baz"><img src="bar" alt="foo"/></a></p>)
 
+  assert_render "This [spawns\nmultiple\nlines](http://example.com)\n\ntext",
+    %(<p>This <a href="http://example.com">spawns\nmultiple\nlines</a></p>\n\n<p>text</p>)
+
   assert_render "***", "<hr/>"
   assert_render "---", "<hr/>"
   assert_render "___", "<hr/>"

src/markdown/parser.cr

@@ -1,4 +1,7 @@
 class Markdown::Parser
+  record PrefixHeader, count : Int32
+  record UnorderedList, char : Char
+
   @lines : Array(String)
 
   def initialize(text : String, @renderer : Renderer)
@@ -15,57 +18,82 @@ class Markdown::Parser
   def process_paragraph
     line = @lines[@line]
 
-    if empty? line
+    case item = classify(line)
+    when :empty
       @line += 1
-      return
+    when :header1
+      render_header 1, line, 2
+    when :header2
+      render_header 2, line, 2
+    when PrefixHeader
+      render_prefix_header(item.count, line)
+    when :code
+      render_code
+    when :horizontal_rule
+      render_horizontal_rule
+    when UnorderedList
+      render_unordered_list(item.char)
+    when :fenced_code
+      render_fenced_code
+    when :ordered_list
+      render_ordered_list
+    when :quote
+      render_quote
+    else
+      render_paragraph
+    end
+  end
+
+  def classify(line)
+    if empty? line
+      return :empty
     end
 
     if next_line_is_all?('=')
-      return render_header 1, line, 2
+      return :header1
     end
 
     if next_line_is_all?('-')
-      return render_header 2, line, 2
+      return :header2
     end
 
-    pounds = count_pounds line
-    if pounds
-      return render_prefix_header pounds, line
+    if pounds = count_pounds line
+      return PrefixHeader.new(pounds)
     end
 
     if line.starts_with? "    "
-      return render_code
+      return :code
     end
 
     if horizontal_rule? line
-      return render_horizontal_rule
+      return :horizontal_rule
     end
 
     if starts_with_bullet_list_marker?(line, '*')
-      return render_unordered_list('*')
+      return UnorderedList.new('*')
     end
 
     if starts_with_bullet_list_marker?(line, '+')
-      return render_unordered_list('+')
+      return UnorderedList.new('+')
     end
 
     if starts_with_bullet_list_marker?(line, '-')
-      return render_unordered_list('-')
+      return UnorderedList.new('-')
     end
 
     if starts_with_backticks? line
-      return render_fenced_code
+      return :fenced_code
     end
 
     if starts_with_digits_dot? line
-      return render_ordered_list
+      return :ordered_list
     end
 
     if line.starts_with? ">"
-      return render_quote
+      return :quote
     end
 
-    render_paragraph
+    nil
   end
 
   def render_prefix_header(level, line)
@@ -91,27 +119,9 @@ class Markdown::Parser
   def render_paragraph
     @renderer.begin_paragraph
 
-    while true
-      process_line @lines[@line]
-      @line += 1
-
-      if @line == @lines.size
-        break
-      end
-
-      line = @lines[@line]
-
-      if empty? line
-        @line += 1
-        break
-      end
-
-      if (starts_with_bullet_list_marker?(line) || starts_with_backticks?(line) || starts_with_digits_dot?(line))
-        break
-      end
-
-      newline
-    end
+    join_next_lines continue_on: nil
+    process_line @lines[@line]
+    @line += 1
 
     @renderer.end_paragraph
 
@@ -185,20 +195,11 @@ class Markdown::Parser
   def render_quote
     @renderer.begin_quote
 
-    while true
-      line = @lines[@line]
-
-      break unless line.starts_with? ">"
-
-      @renderer.text line.byte_slice(Math.min(line.bytesize, 2))
-      @line += 1
-
-      if @line == @lines.size
-        break
-      end
+    join_next_lines continue_on: :quote
+    line = @lines[@line]
 
-      newline
-    end
+    @renderer.text line.byte_slice(Math.min(line.bytesize, 2))
+    @line += 1
 
     @renderer.end_quote
 
@@ -209,6 +210,7 @@ class Markdown::Parser
     @renderer.begin_unordered_list
 
     while true
+      join_next_lines continue_on: nil, stop_on: UnorderedList.new(prefix)
       line = @lines[@line]
 
       if empty? line
@@ -251,6 +253,7 @@ class Markdown::Parser
     @renderer.begin_ordered_list
 
     while true
+      join_next_lines continue_on: nil, stop_on: :ordered_list
       line = @lines[@line]
 
       if empty? line
@@ -604,4 +607,37 @@ class Markdown::Parser
   def newline
     @renderer.text "\n"
   end
+
+  # Join this line with next lines if they form a paragraph,
+  # until next lines don't start another entity like a list,
+  # header, etc.
+  def join_next_lines(continue_on = :none, stop_on = :none)
+    start = @line
+    line = @line
+    line += 1
+    while line < @lines.size
+      item = classify(@lines[line])
+
+      case item
+      when continue_on
+        # continue
+      when stop_on
+        line -= 1
+        break
+      when nil
+        # paragraph: continue
+      else
+        line -= 1
+        break
+      end
+
+      line += 1
+    end
+    line -= 1 if line == @lines.size
+
+    if line > start
+      @lines[line] = (start..line).join("\n") { |i| @lines[i] }
+      @line = line
+    end
+  end
 end