Moved the scanner to the new Parser InputStream and updated the unit tests to use StringInputStream

4 files changed, 61 insertions, 42 deletions

diff --git a/src/HTML5/Parser/InputStream.php b/src/HTML5/Parser/InputStream.php
index 4c5c075..bb2ca1c 100644
--- a/src/HTML5/Parser/InputStream.php
+++ b/src/HTML5/Parser/InputStream.php
@@ -4,12 +4,12 @@ namespace HTML5\Parser;
 /**
  * Interface for stream readers.
  */
-interface InputStream {
+interface InputStream extends \Iterator {
 
   /**
    * Returns the current line that is being consumed.
    *
-   * TODO: Move this to the tokenizer.
+   * TODO: Move this to the scanner.
    */
   public function currentLine();
 
@@ -18,7 +18,7 @@ interface InputStream {
    *
    * Newlines are column 0. The first char after a newline is column 1.
    *
-   * @TODO Move this to the tokenizer.
+   * @TODO Move this to the scanner.
    *
    * @return int
    *   The column number.
@@ -26,11 +26,6 @@ interface InputStream {
   public function columnOffset();
 
   /**
-   * Retrieve the currently consumed character.
-   */
-  public function char();
-
-  /**
    * Get all characters until EOF.
    *
    * This consumes characters until the EOF.
@@ -74,16 +69,14 @@ interface InputStream {
 
   /**
    * Unconsume one character.
+   *
+   * @param  int $howMany
+   *   The number of characters to move the pointer back.
    */
-  public function unconsume();
+  public function unconsume($howMany = 1);
 
   /**
    * Retrieve the next character without advancing the pointer.
    */
   public function peek();
-
-  /**
-   * Get the position of the reader.
-   */
-  public function position();
 }
diff --git a/src/HTML5/Parser/Scanner.php b/src/HTML5/Parser/Scanner.php
index 10698d7..4716fc4 100644
--- a/src/HTML5/Parser/Scanner.php
+++ b/src/HTML5/Parser/Scanner.php
@@ -11,8 +11,6 @@ class Scanner {
   const CHARS_ALNUM = 'abcdefAghijklmnopqrstuvwxyABCDEFGHIJKLMNOPQRSTUVWXYZ01234567890';
   const CHARS_ALPHA = 'abcdefAghijklmnopqrstuvwxyABCDEFGHIJKLMNOPQRSTUVWXYZ';
 
-
-  protected $char;
   protected $is;
 
   /**
@@ -32,7 +30,7 @@ class Scanner {
    *   The current intiger byte position.
    */
   public function position() {
-    return $this->is->position();
+    return $this->is->key();
   }
 
   /**
@@ -54,8 +52,11 @@ class Scanner {
    *   The next character.
    */
   public function next() {
-    $this->char = $this->is->char();
-    return $this->char;
+    $this->is->next();
+    if ($this->is->valid()) {
+      return $this->is->current();
+    }
+    return FALSE;
   }
 
   /**
@@ -67,7 +68,7 @@ class Scanner {
    *   The current character.
    */
   public function current() {
-    return $this->char;
+    return $this->is->current();
   }
 
   /**
@@ -77,9 +78,7 @@ class Scanner {
    *   The number of characters to move the pointer back.
    */
   public function unconsume($howMany = 1) {
-    for ($i = 0; $i < $howMany; ++$i) {
-      $this->is->unconsume();
-    }
+    $this->is->unconsume($howMany);
   }
 
   /**
diff --git a/src/HTML5/Parser/StringInputStream.php b/src/HTML5/Parser/StringInputStream.php
index 4ceae44..e26bb38 100644
--- a/src/HTML5/Parser/StringInputStream.php
+++ b/src/HTML5/Parser/StringInputStream.php
@@ -269,16 +269,40 @@ class StringInputStream implements InputStream {
   }
 
   /**
-   * Retrieve the currently consumed character.
-   * @note This performs bounds checking
+   * Get the current character.
+   * 
+   * @return string
+   *   The current character.
+   */
+  public function current() {
+    return $this->data[$this->char];
+  }
+
+  /**
+   * Advance the pointer. This is part of the Iterator interface.
+   */
+  public function next() {
+    $this->char++;
+  }
+
+  /**
+   * Rewind to the start of the string.
+   */
+  public function rewind() {
+    $this->char = 0;
+  }
+
+  /**
+   * Is the current pointer location valid.
+   *
+   * @return bool
+   *   Is the current pointer location valid.
    */
-  public function char() {
-    // MPB: This appears to advance the pointer, which is not the same
-    // as "retrieving the currently consumed character". Calling char() 
-    // twice will return two different results.
-    if ($this->char++ < $this->EOF) {
-      return $this->data[$this->char - 1];
+  public function valid() {
+    if ($this->char < $this->EOF) {
+      return TRUE;
     }
+
     return FALSE;
   }
 
@@ -362,22 +386,26 @@ class StringInputStream implements InputStream {
   }
 
   /**
-   * Unconsume one character.
+   * Unconsume characters.
+   *
+   * @param int $howMany
+   *   The number of characters to unconsume.
    */
-  public function unconsume() {
-    if ($this->char > 0 && $this->char <= $this->EOF) {
-      $this->char--;
+  public function unconsume($howMany = 1) {
+    if (($this->char - $howMany) >= 0) {
+      $this->char = $this->char - $howMany;
     }
   }
-  public function unget() {
-    $this->unconsume();
-  }
 
   public function peek() {
-    return $this->data[$this->char + 1];
+    if (($this->char + 1) <= $this->EOF) {
+      return $this->data[$this->char + 1];
+    }
+
+    return FALSE;
   }
 
-  public function position() {
+  public function key() {
     return $this->char;
   }
 }
diff --git a/test/HTML5/ScannerTest.php b/test/HTML5/ScannerTest.php
index 8c341fa..997123c 100644
--- a/test/HTML5/ScannerTest.php
+++ b/test/HTML5/ScannerTest.php
@@ -25,7 +25,6 @@ class ScannerTest extends TestCase {
   public function testNext() {
     $s = new Scanner(new StringInputStream("abc"));
 
-    $this->assertEquals('a', $s->next());
     $this->assertEquals('b', $s->next());
     $this->assertEquals('c', $s->next());
   }
@@ -42,7 +41,7 @@ class ScannerTest extends TestCase {
   public function testPeek() {
     $s = new Scanner(new StringInputStream("abc"));
 
-    // The scanner is currently pointed before a.
+
     $this->assertEquals('b', $s->peek());
 
     $s->next();