merged branch gajdaw/finder_search_by_contents (PR symfony#4011)

Commits
-------

218813c [Finder] contains(), notContains()
33e119a Merge branch 'master' of https://github.com/symfony/symfony into finder_search_by_contents
082d86e [Finder] content(), notContent() methods

Discussion
----------

[Finder] search by contents

Bug fix: no
Feature addition: yes
Backwards compatibility break: no
Symfony2 tests pass: yes
Fixes the following tickets: -
Todo: -

Sometimes I need to search for files containing some text:

```
$finder
    ->content('lorem')->notContent('ipsum')
    ->content('/^Begining/m')->notContent('/the end$/m');
```

I don't know how to tests exceptions thrown by `file_get_contents()` calls.

---------------------------------------------------------------------------

by gajdaw at 2012-04-19T15:53:05Z

To keep it as close as possible to `name` and `notName`.

src/Symfony/Component/Finder/Finder.php

@@ -44,6 +44,8 @@ class Finder implements \IteratorAggregate
     private $dirs        = array();
     private $dates       = array();
     private $iterators   = array();
+    private $contains    = array();
+    private $notContains = array();
 
     static private $vcsPatterns = array('.svn', '_svn', 'CVS', '_darcs', '.arch-params', '.monotone', '.bzr', '.git', '.hg');
 
@@ -188,6 +190,49 @@ public function notName($pattern)
         return $this;
     }
 
+    /**
+     * Adds tests that file contents must match.
+     *
+     * Strings or PCRE patterns can be used:
+     *
+     * $finder->contains('Lorem ipsum')
+     * $finder->contains('/Lorem ipsum/i')
+     *
+     * @param  string $pattern A pattern (string or regexp)
+     *
+     * @return Finder The current Finder instance
+     *
+     * @see Symfony\Component\Finder\Iterator\FilecontentFilterIterator
+     */
+    public function contains($pattern)
+    {
+        $this->contains[] = $pattern;
+
+        return $this;
+    }
+
+    /**
+     * Adds tests that file contents must not match.
+     *
+     * Strings or PCRE patterns can be used:
+     *
+     * $finder->notContains('Lorem ipsum')
+     * $finder->notContains('/Lorem ipsum/i')
+
+     *
+     * @param  string $pattern A pattern (string or regexp)
+     *
+     * @return Finder The current Finder instance
+     *
+     * @see Symfony\Component\Finder\Iterator\FilecontentFilterIterator
+     */
+    public function notContains($pattern)
+    {
+        $this->notContains[] = $pattern;
+
+        return $this;
+    }
+
     /**
      * Adds tests for file sizes.
      *
@@ -551,6 +596,10 @@ private function searchInDirectory($dir)
             $iterator = new Iterator\FilenameFilterIterator($iterator, $this->names, $this->notNames);
         }
 
+        if ($this->contains || $this->notContains) {
+            $iterator = new Iterator\FilecontentFilterIterator($iterator, $this->contains, $this->notContains);
+        }
+
         if ($this->sizes) {
             $iterator = new Iterator\SizeRangeFilterIterator($iterator, $this->sizes);
         }

src/Symfony/Component/Finder/Iterator/FilecontentFilterIterator.php

@@ -0,0 +1,83 @@
+<?php
+
+/*
+ * This file is part of the Symfony package.
+ *
+ * (c) Fabien Potencier <[email protected]>
+ *
+ * For the full copyright and license information, please view the LICENSE
+ * file that was distributed with this source code.
+ */
+
+namespace Symfony\Component\Finder\Iterator;
+
+/**
+ * FilecontentFilterIterator filters files by their contents using patterns (regexps or strings).
+ *
+ * @author Fabien Potencier  <[email protected]>
+ * @author Włodzimierz Gajda <[email protected]>
+ */
+class FilecontentFilterIterator extends MultiplePcreFilterIterator
+{
+
+    /**
+     * Filters the iterator values.
+     *
+     * @return Boolean true if the value should be kept, false otherwise
+     */
+    public function accept()
+    {
+        // should at least match one rule
+        if ($this->matchRegexps) {
+            $match = false;
+            foreach ($this->matchRegexps as $regex) {
+                $content = file_get_contents($this->getRealpath());
+                if (false === $content) {
+                    throw new \RuntimeException(sprintf('Error reading file "%s".', $this->getRealpath()));
+                }
+                if (preg_match($regex, $content)) {
+                    $match = true;
+                    break;
+                }
+            }
+        } else {
+            $match = true;
+        }
+
+        // should at least not match one rule to exclude
+        if ($this->noMatchRegexps) {
+            $exclude = false;
+            foreach ($this->noMatchRegexps as $regex) {
+                $content = file_get_contents($this->getRealpath());
+                if (false === $content) {
+                    throw new \RuntimeException(sprintf('Error reading file "%s".', $this->getRealpath()));
+                }
+                if (preg_match($regex, $content)) {
+                    $exclude = true;
+                    break;
+                }
+            }
+        } else {
+            $exclude = false;
+        }
+
+        return $match && !$exclude;
+    }
+
+    /**
+     * Converts string to regexp if necessary.
+     *
+     * @param string $str Pattern: string or regexp
+     *
+     * @return string regexp corresponding to a given string or regexp
+     */
+    protected function toRegex($str)
+    {
+        if (preg_match('/^([^a-zA-Z0-9\\\\]).+?\\1[ims]?$/', $str)) {
+            return $str;
+        }
+
+        return sprintf('/%s/', preg_quote($str, '/'));
+    }
+
+}

src/Symfony/Component/Finder/Iterator/FilenameFilterIterator.php

@@ -18,32 +18,8 @@
  *
  * @author Fabien Potencier <[email protected]>
  */
-class FilenameFilterIterator extends \FilterIterator
+class FilenameFilterIterator extends MultiplePcreFilterIterator
 {
-    private $matchRegexps;
-    private $noMatchRegexps;
-
-    /**
-     * Constructor.
-     *
-     * @param \Iterator $iterator        The Iterator to filter
-     * @param array     $matchPatterns   An array of patterns that need to match
-     * @param array     $noMatchPatterns An array of patterns that need to not match
-     */
-    public function __construct(\Iterator $iterator, array $matchPatterns, array $noMatchPatterns)
-    {
-        $this->matchRegexps = array();
-        foreach ($matchPatterns as $pattern) {
-            $this->matchRegexps[] = $this->toRegex($pattern);
-        }
-
-        $this->noMatchRegexps = array();
-        foreach ($noMatchPatterns as $pattern) {
-            $this->noMatchRegexps[] = $this->toRegex($pattern);
-        }
-
-        parent::__construct($iterator);
-    }
 
     /**
      * Filters the iterator values.
@@ -81,7 +57,17 @@ public function accept()
         return $match && !$exclude;
     }
 
-    private function toRegex($str)
+    /**
+     * Converts glob to regexp.
+     *
+     * PCRE patterns are left unchanged.
+     * Glob strings are transformed with Glob::toRegex().
+     *
+     * @param string $str Pattern: glob or regexp
+     *
+     * @return string regexp corresponding to a given glob or regexp
+     */
+    protected function toRegex($str)
     {
         if (preg_match('/^([^a-zA-Z0-9\\\\]).+?\\1[ims]?$/', $str)) {
             return $str;

src/Symfony/Component/Finder/Iterator/MultiplePcreFilterIterator.php

@@ -0,0 +1,57 @@
+<?php
+
+/*
+ * This file is part of the Symfony package.
+ *
+ * (c) Fabien Potencier <[email protected]>
+ *
+ * For the full copyright and license information, please view the LICENSE
+ * file that was distributed with this source code.
+ */
+
+namespace Symfony\Component\Finder\Iterator;
+
+use Symfony\Component\Finder\Glob;
+
+/**
+ * MultiplePcreFilterIterator filters files using patterns (regexps, globs or strings).
+ *
+ * @author Fabien Potencier <[email protected]>
+ */
+abstract class MultiplePcreFilterIterator extends \FilterIterator
+{
+    protected $matchRegexps;
+    protected $noMatchRegexps;
+
+    /**
+     * Constructor.
+     *
+     * @param \Iterator $iterator        The Iterator to filter
+     * @param array     $matchPatterns   An array of patterns that need to match
+     * @param array     $noMatchPatterns An array of patterns that need to not match
+     */
+    public function __construct(\Iterator $iterator, array $matchPatterns, array $noMatchPatterns)
+    {
+        $this->matchRegexps = array();
+        foreach ($matchPatterns as $pattern) {
+            $this->matchRegexps[] = $this->toRegex($pattern);
+        }
+
+        $this->noMatchRegexps = array();
+        foreach ($noMatchPatterns as $pattern) {
+            $this->noMatchRegexps[] = $this->toRegex($pattern);
+        }
+
+        parent::__construct($iterator);
+    }
+
+    /**
+     * Converts string into regexp.
+     *
+     * @param string $str Pattern
+     *
+     * @return string regexp corresponding to a given string
+     */
+    abstract protected function toRegex($str);
+
+}

src/Symfony/Component/Finder/Tests/FinderTest.php

@@ -330,4 +330,46 @@ protected function toAbsolute($files)
 
         return $f;
     }
+
+    protected function toAbsoluteFixtures($files)
+    {
+        $f = array();
+        foreach ($files as $file) {
+            $f[] = __DIR__.DIRECTORY_SEPARATOR.'Fixtures'.DIRECTORY_SEPARATOR.$file;
+        }
+
+        return $f;
+    }
+
+    /**
+     * @dataProvider getContainsTestData
+     */
+    public function testContains($matchPatterns, $noMatchPatterns, $expected)
+    {
+        $finder = new Finder();
+        $finder->in(__DIR__.DIRECTORY_SEPARATOR.'Fixtures')
+            ->name('*.txt')->sortByName()
+            ->contains($matchPatterns)
+            ->notContains($noMatchPatterns);
+
+        $this->assertIterator($this->toAbsoluteFixtures($expected), $finder);
+    }
+
+    public function getContainsTestData()
+    {
+        return array(
+            array('', '', array()),
+            array('foo', 'bar', array()),
+            array('', 'foobar', array('dolor.txt', 'ipsum.txt', 'lorem.txt')),
+            array('lorem ipsum dolor sit amet', 'foobar', array('lorem.txt')),
+            array('sit', 'bar', array('dolor.txt', 'ipsum.txt', 'lorem.txt')),
+            array('dolor sit amet', '@^L@m', array('dolor.txt', 'ipsum.txt')),
+            array('/^lorem ipsum dolor sit amet$/m', 'foobar', array('lorem.txt')),
+            array('lorem', 'foobar', array('lorem.txt')),
+
+            array('', 'lorem', array('dolor.txt', 'ipsum.txt')),
+            array('ipsum dolor sit amet', '/^IPSUM/m', array('lorem.txt')),
+        );
+    }
+
 }

src/Symfony/Component/Finder/Tests/Fixtures/dolor.txt

@@ -0,0 +1,2 @@
+dolor sit amet
+DOLOR SIT AMET

src/Symfony/Component/Finder/Tests/Fixtures/ipsum.txt

@@ -0,0 +1,2 @@
+ipsum dolor sit amet
+IPSUM DOLOR SIT AMET

src/Symfony/Component/Finder/Tests/Fixtures/lorem.txt

@@ -0,0 +1,2 @@
+lorem ipsum dolor sit amet
+LOREM IPSUM DOLOR SIT AMET

src/Symfony/Component/Finder/Tests/Iterator/FilecontentFilterIteratorTest.php

@@ -0,0 +1,41 @@
+<?php
+
+/*
+ * This file is part of the Symfony package.
+ *
+ * (c) Fabien Potencier <[email protected]>
+ *
+ * For the full copyright and license information, please view the LICENSE
+ * file that was distributed with this source code.
+ */
+
+namespace Symfony\Component\Finder\Tests\Iterator;
+
+use Symfony\Component\Finder\Iterator\FilecontentFilterIterator;
+
+class FilecontentFilterIteratorTest extends IteratorTestCase
+{
+
+    public function testAccept()
+    {
+        $inner = new ContentInnerNameIterator(array('test.txt'));
+
+        $iterator = new FilecontentFilterIterator($inner, array(), array());
+
+        $this->assertIterator(array('test.txt'), $iterator);
+    }
+
+}
+
+class ContentInnerNameIterator extends \ArrayIterator
+{
+    public function current()
+    {
+        return new \SplFileInfo(parent::current());
+    }
+
+    public function getFilename()
+    {
+        return parent::current();
+    }
+}