Added support for processing instructions.

4 files changed, 68 insertions, 3 deletions

diff --git a/src/HTML5/Parser/EventHandler.php b/src/HTML5/Parser/EventHandler.php
index bc16cdf..c28d80b 100644
--- a/src/HTML5/Parser/EventHandler.php
+++ b/src/HTML5/Parser/EventHandler.php
@@ -51,7 +51,22 @@ interface EventHandler {
    */
   public function parseError($msg, $line, $col);
 
-  // Do we need...
+  /**
+   * A CDATA section.
+   *
+   * @param string $data
+   *   The unparsed character data.
+   */
   public function cdata($data);
-  // public function processorInstruction();
+  /**
+   * This is a holdover from the XML spec.
+   *
+   * While user agents don't get PIs, server-side does.
+   *
+   * @param string $name
+   *   The name of the processor (e.g. 'php').
+   * @param string $data
+   *   The unparsed data.
+   */
+  public function processingInstruction($name, $data = NULL);
 }
diff --git a/src/HTML5/Parser/Tokenizer.php b/src/HTML5/Parser/Tokenizer.php
index 4f1404f..d79f2c0 100644
--- a/src/HTML5/Parser/Tokenizer.php
+++ b/src/HTML5/Parser/Tokenizer.php
@@ -222,8 +222,8 @@ class Tokenizer {
 
     return $this->markupDeclaration() ||
       $this->endTag() ||
-      $this->tagName() ||
       $this->processingInstruction() ||
+      $this->tagName() ||
       // This always returns false.
       $this->parseError("Illegal tag opening") ||
       $this->characterData();
@@ -784,6 +784,38 @@ class Tokenizer {
    * EventListener::processingInstruction() event.
    */
   protected function processingInstruction() {
+    if ($this->scanner->current() != '?') {
+      return FALSE;
+    }
+
+    $tok = $this->scanner->next();
+    $procName = $this->scanner->getAsciiAlpha();
+    $white = strlen($this->scanner->whitespace());
+
+    // If not a PI, send to bogusComment.
+    if (strlen($procName) == 0 || $white == 0 || $this->scanner->current() == FALSE) {
+      $this->parseError("Expected processing instruction name, got $tok");
+      $this->bogusComment('<?' . $tok . $procName);
+      return TRUE;
+    }
+
+    $data = '';
+    while ($this->scanner->current() != '?' && $this->scanner->peek() != '>') {
+      $data .= $this->scanner->current();
+
+      $tok = $this->scanner->next();
+      if ($tok === FALSE) {
+        $this->parseError("Unexpected EOF in processing instruction.");
+        $this->events->processingInstruction($procName, $data);
+        return TRUE;
+      }
+
+    }
+
+    $this->scanner->next(); // >
+    $this->scanner->next(); // Next token.
+    $this->events->processingInstruction($procName, $data);
+    return TRUE;
   }
 
 
diff --git a/test/HTML5/Parser/EventStack.php b/test/HTML5/Parser/EventStack.php
index e865507..f197855 100644
--- a/test/HTML5/Parser/EventStack.php
+++ b/test/HTML5/Parser/EventStack.php
@@ -71,6 +71,9 @@ class EventStack implements EventHandler {
     $this->store('error', func_get_args());
   }
 
+  public function processingInstruction($name, $data = NULL) {
+    $this->store('pi', func_get_args());
+  }
 
 }
 class EventStackParseError extends \Exception {
diff --git a/test/HTML5/Parser/TokenizerTest.php b/test/HTML5/Parser/TokenizerTest.php
index fb33e37..c4c66e7 100644
--- a/test/HTML5/Parser/TokenizerTest.php
+++ b/test/HTML5/Parser/TokenizerTest.php
@@ -106,6 +106,8 @@ class TokenizerTest extends \HTML5\Tests\TestCase {
       '<!CDATA[[ test ]]>',
       '<![CDATA[',
       '<![CDATA[hellooooo hello',
+      '<? Hello World ?>',
+      '<? Hello World',
     );
     foreach ($bogus as $str) {
       $events = $this->parse($str);
@@ -258,6 +260,19 @@ class TokenizerTest extends \HTML5\Tests\TestCase {
     }
   }
 
+  public function testProcessorInstruction() {
+    $good = array(
+      '<?hph ?>' => 'hph',
+      '<?hph echo "Hello World"; ?>' => array('hph', 'echo "Hello World"; '), 
+      "<?hph \necho 'Hello World';\n?>" => array('hph', "echo 'Hello World';\n"), 
+    );
+    foreach ($good as $test => $expects) {
+      $events = $this->parse($test);
+      $this->assertEquals(2, $events->depth(), "Counting events for '$test': " . print_r($events, TRUE));
+      $this->assertEventEquals('pi', $expects, $events->get(0));
+    }
+  }
+
   public function testText() {
     $good = array(
       'a<br>b',