Added support for processing instructions.

2 files changed, 50 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;
   }