Parser updates.HEADmaster

9 files changed, 59 insertions, 33 deletions

diff --git a/src/Gallic/Parser.php b/src/Gallic/Parser.php
index cc55e1b..2270685 100644
--- a/src/Gallic/Parser.php
+++ b/src/Gallic/Parser.php
@@ -25,6 +25,12 @@ abstract class Gallic_Parser
 {
         abstract function hasValue();
 
+        /**
+         * Returns either:
+         *   false
+         * or
+         *   array(length[, value])
+         */
         abstract protected function _parse($str, $offset, $len);
 
         static function __callStatic($name, array $args)
@@ -45,10 +51,11 @@ abstract class Gallic_Parser
         }
 
         /**
-         * Returns either:
-         *   false
-         * or
-         *   array(length[, value])
+         * @param str
+         *
+         * @return mixed
+         *
+         * @throws Exception
          */
         final function parse($str, $offset = null, $len = null)
         {
@@ -57,6 +64,16 @@ abstract class Gallic_Parser
                 isset($len)
                         or $len = strlen($str);
 
-                return $this->_parse($str, $offset, $len);
+                if (!($match = $this->_parse($str, $offset, $len)))
+                {
+                        throw new Exception();
+                }
+
+                if ($this->hasValue())
+                {
+                        return $match[1];
+                }
+
+                return null;
         }
 }
diff --git a/src/Gallic/Parser/Alternation.php b/src/Gallic/Parser/Alternation.php
index 809e82f..f5028f8 100644
--- a/src/Gallic/Parser/Alternation.php
+++ b/src/Gallic/Parser/Alternation.php
@@ -27,6 +27,13 @@ final class Gallic_Parser_Alternation extends Gallic_Parser_Composite
                                 continue;
                         }
 
+                        if ($child instanceof Gallic_Parser_Epsilon)
+                        {
+                                // This child will always succeed, no one will match after.
+                                $children[] = $child;
+                                break;
+                        }
+
                         $hash = spl_object_hash($child);
                         if (!isset($children[$hash]))
                         {
@@ -52,7 +59,7 @@ final class Gallic_Parser_Alternation extends Gallic_Parser_Composite
         {
                 foreach ($this->_children as $child)
                 {
-                        if ($match = $child->parse($str, $offset, $len))
+                        if ($match = $child->_parse($str, $offset, $len))
                         {
                                 if ($this->hasValue())
                                 {
diff --git a/src/Gallic/Parser/Grammar.php b/src/Gallic/Parser/Grammar.php
index 81427e8..3e5bc28 100644
--- a/src/Gallic/Parser/Grammar.php
+++ b/src/Gallic/Parser/Grammar.php
@@ -40,7 +40,7 @@ final class Gallic_Parser_Grammar extends Gallic_Parser_Composite
 
         protected function _parse($str, $offset, $len)
         {
-                return $this->_children[$this->_top]->parse($str, $offset, $len);
+                return $this->_children[$this->_top]->_parse($str, $offset, $len);
         }
 
         protected function _resolve(Gallic_Parser_Composite $parser)
diff --git a/src/Gallic/Parser/Ignore.php b/src/Gallic/Parser/Ignore.php
index a98d18f..faffeb8 100644
--- a/src/Gallic/Parser/Ignore.php
+++ b/src/Gallic/Parser/Ignore.php
@@ -25,7 +25,7 @@ final class Gallic_Parser_Ignore extends Gallic_Parser_Composite
 
         protected function _parse($str, $offset, $len)
         {
-                if ($match = $this->_children[0]->parse($str, $offset, $len))
+                if ($match = $this->_children[0]->_parse($str, $offset, $len))
                 {
                         unset($match[1]);
                         return $match;
@@ -33,9 +33,4 @@ final class Gallic_Parser_Ignore extends Gallic_Parser_Composite
 
                 return false;
         }
-
-        /**
-         * @var callable
-         */
-        private $_callback;
 }
diff --git a/src/Gallic/Parser/Negation.php b/src/Gallic/Parser/Negation.php
index 02e68e3..9470e7f 100644
--- a/src/Gallic/Parser/Negation.php
+++ b/src/Gallic/Parser/Negation.php
@@ -32,7 +32,7 @@ final class Gallic_Parser_Negation extends Gallic_Parser_Composite
 
         protected function _parse($str, $offset, $len)
         {
-                if ($match = $this->_children[0]->parse($str, $offset, $len))
+                if ($match = $this->_children[0]->_parse($str, $offset, $len))
                 {
                         return false;
                 }
diff --git a/src/Gallic/Parser/Repetition.php b/src/Gallic/Parser/Repetition.php
index 88e9c4d..5bd24e7 100644
--- a/src/Gallic/Parser/Repetition.php
+++ b/src/Gallic/Parser/Repetition.php
@@ -18,13 +18,14 @@ final class Gallic_Parser_Repetition extends Gallic_Parser_Composite
 
         protected function _optimize()
         {
-                parent::_optimize();
+                $child = $this->_children[0]->optimize();
+                $has_no_value = !$child->hasValue();
 
-                // If “max === 0”, equivalent to ε.
-                /*if ($this->_max === 0)
+                // If “max === 0 && $has_no_value”, equivalent to ε.
+                if ($has_no_value && ($this->_max === 0))
                 {
                         return new Gallic_Parser_Epsilon;
-                        }*/
+                }
 
                 /*
                  * If  the cardinalities are  incorrectly set,  this parser  will always
@@ -36,12 +37,18 @@ final class Gallic_Parser_Repetition extends Gallic_Parser_Composite
                         return new Gallic_Parser_Fail;
                 }
 
-                // No need to use a sequence to match exactly once.
-                /*if (($this->_min === 1) && ($this->_max === 1))
+                /*
+                 * To match exactly once, there is no need to use a repetition.
+                 *
+                 * But the equivalence of the replacement  is only true if the child has
+                 * no value.
+                 */
+                if ($has_no_value && ($this->_min === 1) && ($this->_max === 1))
                 {
-                        return $this->_children[0];
-                        }*/
+                        return $child;
+                }
 
+                $this->_children[0] = $child;
                 return $this;
         }
 
@@ -56,7 +63,7 @@ final class Gallic_Parser_Repetition extends Gallic_Parser_Composite
                 }
 
                 while ((!isset($this->_max) || ($n < $this->_max))
-                       && ($match = $this->_children[0]->parse($str, $offset, $len)))
+                       && ($match = $this->_children[0]->_parse($str, $offset, $len)))
                 {
                         ++$n;
                         $offset += $match[0];
diff --git a/src/Gallic/Parser/SList.php b/src/Gallic/Parser/SList.php
index 2793845..40ead21 100644
--- a/src/Gallic/Parser/SList.php
+++ b/src/Gallic/Parser/SList.php
@@ -32,7 +32,7 @@ final class Gallic_Parser_SList extends Gallic_Parser_Composite
                  * once.
                  */
                 if (($item instanceof Gallic_Parser_Fail)
-                    /*|| ($separator instanceof Gallic_Parser_Fail)*/)
+                    || ($separator instanceof Gallic_Parser_Fail))
                 {
                         return $item;
                 }
@@ -63,7 +63,7 @@ final class Gallic_Parser_SList extends Gallic_Parser_Composite
                         $value = array();
                 }
 
-                if ((!$match_i = $item->parse($str, $offset, $len)))
+                if ((!$match_i = $item->_parse($str, $offset, $len)))
                 {
                         return false;
                 }
@@ -74,8 +74,8 @@ final class Gallic_Parser_SList extends Gallic_Parser_Composite
                         $value[] = $match_i[1];
                 }
 
-                while (($match_s = $separator->parse($str, $offset, $len))
-                       && ($match_i = $item->parse($str, $offset + $match_s[0], $len)))
+                while (($match_s = $separator->_parse($str, $offset, $len))
+                       && ($match_i = $item->_parse($str, $offset + $match_s[0], $len)))
                 {
                         $offset += $match_s[0] + $match_i[0];
 
diff --git a/src/Gallic/Parser/Sequence.php b/src/Gallic/Parser/Sequence.php
index 73c44e6..7a22c93 100644
--- a/src/Gallic/Parser/Sequence.php
+++ b/src/Gallic/Parser/Sequence.php
@@ -14,16 +14,16 @@ final class Gallic_Parser_Sequence extends Gallic_Parser_Composite
                 parent::_optimize();
 
                 // An empty sequence will always match.
-                /*if (empty($this->_children))
+                if (empty($this->_children))
                 {
                         return new Gallic_Parser_Epsilon;
-                        }*/
+                }
 
                 // No need to use a sequence for one element.
-                /*if (count($this->_children) === 1)
+                if (count($this->_children) === 1)
                 {
                         return $this->_children[0];
-                        }*/
+                }
 
                 return $this;
         }
@@ -39,7 +39,7 @@ final class Gallic_Parser_Sequence extends Gallic_Parser_Composite
 
                 foreach ($this->_children as $child)
                 {
-                        if (!($match = $child->parse($str, $offset, $len)))
+                        if (!($match = $child->_parse($str, $offset, $len)))
                         {
                                 return false;
                         }
diff --git a/src/Gallic/Parser/Transform.php b/src/Gallic/Parser/Transform.php
index 402621c..d0ec68f 100644
--- a/src/Gallic/Parser/Transform.php
+++ b/src/Gallic/Parser/Transform.php
@@ -28,7 +28,7 @@ final class Gallic_Parser_Transform extends Gallic_Parser_Composite
 
         protected function _parse($str, $offset, $len)
         {
-                if (!($match = $this->_children[0]->parse($str, $offset, $len)))
+                if (!($match = $this->_children[0]->_parse($str, $offset, $len)))
                 {
                         return false;
                 }