From b49053cfab01c2f9c66fcc6204ce38d7b1f4f19a Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Kasper=20Garn=C3=A6s?= <kasper@reload.dk>
Date: Tue, 9 Aug 2011 16:44:26 +0200
Subject: [PATCH] Implemented support for XML

---
 bin/phrocco                 |  2 +-
 lib/adapters/XmlAdapter.php | 71 +++++++++++++++++++++++++++++++++++++
 phrocco.php                 |  3 +-
 3 files changed, 74 insertions(+), 2 deletions(-)
 create mode 100644 lib/adapters/XmlAdapter.php

diff --git a/bin/phrocco b/bin/phrocco
index 127b217..afb902f 100755
--- a/bin/phrocco
+++ b/bin/phrocco
@@ -19,7 +19,7 @@ if(count($argv)>1){
 Command line options:
   -i <input directory>     Where to look for source code files - Searches recursively
   -o <output directory>    Root directory to output to
-  -l <language>            Language to parse. Currently PHP only, more to come.
+  -l <language>            Language to parse. Currently PHP and XML only, more to come.
 
   Examples:
   phrocco -i ./ -o ./docs
diff --git a/lib/adapters/XmlAdapter.php b/lib/adapters/XmlAdapter.php
new file mode 100644
index 0000000..2172e10
--- /dev/null
+++ b/lib/adapters/XmlAdapter.php
@@ -0,0 +1,71 @@
+<?php
+
+/**
+ * XML Adapter Class
+ * Its job is to handle XML code and create an array of
+ * code and comments from a XML file.
+ * @author Kasper Garnæs
+ **/
+
+class XmlAdapter {
+
+  /**
+   * Main parsing method.
+   * Uses the php tokenizer to separate comments from code
+   * @return `array`
+   *    array("code"=>`array of code lines`, "docs"=>`array of doc blocks`)
+   **/
+  public function parse($file) {
+    // Load the file
+    $file = trim(file_get_contents($file));
+    // Determine if we start with a documentation or code block
+    $is_start_doc = '<!--' == substr($file, 0, 4);
+    // Explode by start comment delimiter.
+    // If we start with a comment block every array entry will contain
+    // documentation first followed by code. Otherwise we start with a entry
+    // containing nothing but code followed by documentation-code entries.
+    $file = explode('<!--', $file);
+    // Explode every entry by end comment delimiter.
+    foreach ($file as &$f) {
+      $f = explode('-->', $f);
+    }
+    // Every entry should now be an array containing documentation first and
+    // code second. If we started with a code only block then we add an empty
+    // documentation block to keep this structure.
+    if (!$is_start_doc) {
+      array_unshift($file[0], '');
+    }
+
+    // Separate entries into code and documentation
+    $docs = array();
+    $code = array();
+    for ($i = 0; $i < sizeof($file); $i++) {
+      $docs[] = trim($file[$i][0]);
+      $code[] = $file[$i][1];
+    }
+
+    // Passes code onto pygmentize to add syntax highlighting
+    // Assemble the code into a single string we can pass on to pygmentize.
+    // Each block is separated by a delimiter.
+    $code = implode("\n<!--CODEBLOCK-->\n", $code);
+    $pyg = new Pygment;
+    $code = $pyg->pygmentize("xml", $code);
+    // Separate by syntax highlighted delimiter again
+    $code = explode('<span class="c">&lt;!--CODEBLOCK--&gt;</span>', $code);
+
+    foreach($code as &$val) {
+      $val = rtrim($val);
+      $val = str_replace("\t", " ", $val);
+      $val='<div class="highlight"><pre>'.trim($val, "\n\r").'</pre></div>';
+    }
+
+    // Pass documentation through Markdown
+    foreach($docs as &$doc) {
+      $doc = Markdown($doc);
+    }
+
+    // Our final array of code mapped to comments
+    return array("code"=>$code,"docs"=>$docs);
+  }
+
+}
diff --git a/phrocco.php b/phrocco.php
index 2e1571c..9290920 100755
--- a/phrocco.php
+++ b/phrocco.php
@@ -57,7 +57,8 @@ public function getExtension() {
 class PhroccoGroup {
 
   public $extensions = array(
-    "php" => array("php","phps","phpt")
+    "php" => array("php","phps","phpt"),
+    "xml" => array("xml")
   );
   public $language = "php";
   public $defaults = array(