code block parsing

Author: ta4ilka69

src/Markdown.hs

@@ -91,8 +91,7 @@ titleAttr title = if T.null title then T.pack "" else T.concat [T.pack " title=\
 
 escapeHtml :: Text -> Text
 escapeHtml =
-  T.replace (T.pack "&") (T.pack "&")
-    . T.replace (T.pack "<") (T.pack "&lt;")
+  T.replace (T.pack "<") (T.pack "&lt;")
     . T.replace (T.pack ">") (T.pack "&gt;")
     . T.replace (T.pack "\"") (T.pack "&quot;")
 

src/Parser.hs

@@ -114,6 +114,9 @@ parseMarkdown = parseLines [] . skipEmptyLines
 parseLines :: [Text] -> [Text] -> [MDElement]
 parseLines acc [] = processBlock (reverse acc)
 parseLines acc (line : lines)
+  | isCodeBlockStart line =
+      let (codeBlock, rest) = parseCodeBlock (line : lines)
+       in processBlock (reverse acc) ++ [codeBlock] ++ parseLines [] rest
   | isTableLine line =
       case identifyTable (line : lines) of
         Just (tableLines, rest) ->
@@ -359,6 +362,7 @@ parseInline text
   | T.null text = []
   | otherwise =
       case T.uncons text of
+        Just ('`', _) -> parseInlineCode text
         Just ('[', _) -> parseLinkOrCheckbox text
         Just ('*', _) -> parseDecoration '*' text
         Just ('_', _) -> parseDecoration '_' text
@@ -516,3 +520,29 @@ parsePlainText :: Text -> [MDElement]
 parsePlainText text =
   let (content, rest) = T.break (`elem` ['*', '_', '~', '<']) text
    in PlainText content : parseInline rest
+
+isCodeBlockStart :: Text -> Bool
+isCodeBlockStart line =
+  T.isPrefixOf (T.pack "```") (T.stripStart line)
+
+parseCodeBlock :: [Text] -> (MDElement, [Text])
+parseCodeBlock (firstLine : rest) =
+  let lang = T.strip $ T.drop 3 $ T.stripStart firstLine
+      (codeLines, remainingLines) = span (not . isCodeBlockStart) rest
+      codeBlockContent = T.unlines codeLines
+   in (CodeBlock codeBlockContent, drop 1 remainingLines)
+parseCodeBlock _ = (CodeBlock T.empty, [])
+
+parseInlineCode :: Text -> [MDElement]
+parseInlineCode text
+  | T.isPrefixOf (T.pack "``") text =
+      let (content, rest) = T.breakOn (T.pack "``") (T.drop 2 text)
+       in if T.isPrefixOf (T.pack "``") rest
+            then InlineCode (T.strip content) : parseInline (T.drop 2 rest)
+            else PlainText (T.pack "``") : parseInline (T.drop 2 text)
+  | T.isPrefixOf (T.pack "`") text =
+      let (content, rest) = T.breakOn (T.pack "`") (T.drop 1 text)
+       in if T.isPrefixOf (T.pack "`") rest
+            then InlineCode (T.strip content) : parseInline (T.drop 1 rest)
+            else PlainText (T.pack "`") : parseInline (T.drop 1 text)
+  | otherwise = [PlainText text]

src/Types.hs

@@ -27,12 +27,12 @@ data MDElement
   | Link
       { linkText :: Text,
         linkUrl :: Text,
-        linkTitle :: Maybe Text -- Optional title attribute
+        linkTitle :: Maybe Text
       }
   | Image Text Text Text
   | UnorderedList [MDElement]
   | OrderedList [MDElement]
-  | ListItem MDElement [MDElement] -- New type to support nested lists
+  | ListItem MDElement [MDElement]
   | CodeBlock Text
   | InlineCode Text
   | Checkbox Bool [MDElement]

test/output.html

@@ -204,12 +204,16 @@ <h2 id="9f9a76a951506992fc919b8ae3f5ae15">Заголовок второго ур
   alt="Изображение title="Логотип Markdown"" />
 </p>
 <p>Функция `print (x)` выводит содержимое переменной ```x```.</p>
-<p>```</p>
-<h1 id="967f745ae03431977fb89c965c02c00c">
-  include <a href="stdio.h">stdio.h</a>
-</h1>
-<p>int main() { printf(&amp;quot;Hello, World!&amp;quot;); return 0; } ```</p>
-<p>```javascript let x = 12; let y = 6; console.log(x + y); ```</p>
+<pre><code>#include &lt;stdio.h&gt;
+int main() {
+   printf(&quot;Hello, World!&quot;);
+   return 0;
+}
+</code></pre>
+<pre><code>let x = 12;
+let y = 6;
+console.log(x + y);
+</code></pre>
 <table>
   <thead>
     <tr>