From b8256cecae96eb37d8275399928c39d7689d3105 Mon Sep 17 00:00:00 2001
From: Jaapio <jaap@ijaap.nl>
Date: Fri, 20 Dec 2019 17:43:34 +0100
Subject: [PATCH] implement glob spec

---
 .../04-sample-phpdoc-layout-using-glob.php    |  35 +++
 phpcs.xml.dist                                |   1 +
 src/Specification/Glob.php                    | 261 ++++++++++++++++++
 ...t.php => FindOnSamplePhpdocLayoutTest.php} |   2 +-
 .../FindOnSamplePhpdocLayoutUsingGlobTest.php |  36 +++
 tests/unit/Specification/GlobTest.php         | 169 ++++++++++++
 6 files changed, 503 insertions(+), 1 deletion(-)
 create mode 100644 examples/04-sample-phpdoc-layout-using-glob.php
 create mode 100644 src/Specification/Glob.php
 rename tests/integration/{FindOnSamplePhpdocLayout.php => FindOnSamplePhpdocLayoutTest.php} (94%)
 create mode 100644 tests/integration/FindOnSamplePhpdocLayoutUsingGlobTest.php
 create mode 100644 tests/unit/Specification/GlobTest.php

diff --git a/examples/04-sample-phpdoc-layout-using-glob.php b/examples/04-sample-phpdoc-layout-using-glob.php
new file mode 100644
index 0000000..99c39b4
--- /dev/null
+++ b/examples/04-sample-phpdoc-layout-using-glob.php
@@ -0,0 +1,35 @@
+<?php
+require_once(__DIR__ . '/../vendor/autoload.php');
+
+use Flyfinder\Specification\Glob;
+use League\Flysystem\Filesystem;
+use League\Flysystem\Adapter\Local;
+use Flyfinder\Finder;
+use Flyfinder\Path;
+use Flyfinder\Specification\IsHidden;
+use Flyfinder\Specification\HasExtension;
+use Flyfinder\Specification\InPath;
+use Flyfinder\Specification\AndSpecification;
+
+// (03-sample-files based on some phpDocumentor2 src files)
+$filesystem = new Filesystem(new Local(__DIR__ . '/03-sample-files'));
+$filesystem->addPlugin(new Finder());
+
+/*
+ * "phpdoc -d src -i src/phpDocumentor/DomainModel"
+ * should result in src/Cilex and src/phpDocumentor/. files being found,
+ * but src/phpDocumentor/DomainModel files being left out
+ */
+$dashDirectoryPath = new Glob('/src/**/*');
+$dashIgnorePath = new InPath(new Path('src/phpDocumentor/DomainModel'));
+$isHidden = new IsHidden();
+$isPhpFile = new HasExtension(['php']);
+$spec = new AndSpecification($dashDirectoryPath, $dashIgnorePath->notSpecification());
+$spec->andSpecification($isHidden->notSpecification());
+$spec->andSpecification($isPhpFile);
+
+$generator = $filesystem->find($spec);
+$result = [];
+foreach($generator as $value) {
+    $result[] = $value;
+}
diff --git a/phpcs.xml.dist b/phpcs.xml.dist
index ad2d53d..d9873c7 100644
--- a/phpcs.xml.dist
+++ b/phpcs.xml.dist
@@ -17,6 +17,7 @@
     <rule ref="Generic.Files.LineLength.TooLong">
         <exclude-pattern>*/src/Finder.php</exclude-pattern>
         <exclude-pattern>*/src/Specification/SpecificationInterface.php</exclude-pattern>
+        <exclude-pattern>*/tests/unit/Specification/GlobTest.php</exclude-pattern>
     </rule>
 
 
diff --git a/src/Specification/Glob.php b/src/Specification/Glob.php
new file mode 100644
index 0000000..e07977a
--- /dev/null
+++ b/src/Specification/Glob.php
@@ -0,0 +1,261 @@
+<?php
+
+declare(strict_types=1);
+
+/**
+ * This file is part of phpDocumentor.
+ *
+ * For the full copyright and license information, please view the LICENSE
+ * file that was distributed with this source code.
+ *
+ * Many thanks to webmozart by providing the original code in webmozart/glob
+ *
+ * @link https://github.com/webmozart/glob/blob/master/src/Glob.php
+ * @link      http://phpdoc.org
+ */
+
+namespace Flyfinder\Specification;
+
+use InvalidArgumentException;
+use function preg_match;
+use function sprintf;
+use function strlen;
+use function strpos;
+
+/**
+ * Glob specification class
+ *
+ * @psalm-immutable
+ */
+final class Glob extends CompositeSpecification
+{
+    /** @var string */
+    private $regex;
+
+    /**
+     * The "static prefix" is the part of the glob up to the first wildcard "*".
+     * If the glob does not contain wildcards, the full glob is returned.
+     *
+     * @var string
+     */
+    private $staticPrefix;
+
+    public function __construct(string $glob)
+    {
+        $this->regex        = self::toRegEx($glob);
+        $this->staticPrefix = self::getStaticPrefix($glob);
+    }
+
+    /**
+     * @inheritDoc
+     */
+    public function isSatisfiedBy(array $value) : bool
+    {
+        //Flysystem paths are not absolute, so make it that way.
+        $path = '/' . $value['path'];
+        if (strpos($path, $this->staticPrefix) !== 0) {
+            return false;
+        }
+
+        if (preg_match($this->regex, $path)) {
+            return true;
+        }
+
+        return false;
+    }
+
+    /**
+     * Returns the static prefix of a glob.
+     *
+     * The "static prefix" is the part of the glob up to the first wildcard "*".
+     * If the glob does not contain wildcards, the full glob is returned.
+     *
+     * @param string $glob The canonical glob. The glob should contain forward
+     *                      slashes as directory separators only. It must not
+     *                      contain any "." or ".." segments.
+     *
+     * @return string The static prefix of the glob.
+     *
+     * @psalm-pure
+     */
+    private static function getStaticPrefix(string $glob) : string
+    {
+        if (strpos($glob, '/') !== 0 && strpos($glob, '://') === false) {
+            throw new InvalidArgumentException(sprintf(
+                'The glob "%s" is not absolute and not a URI.',
+                $glob
+            ));
+        }
+        $prefix = '';
+        $length = strlen($glob);
+        for ($i = 0; $i < $length; ++$i) {
+            $c = $glob[$i];
+            switch ($c) {
+                case '/':
+                    $prefix .= '/';
+                    if (self::isRecursiveWildcard($glob, $i)) {
+                        break 2;
+                    }
+                    break;
+                case '*':
+                case '?':
+                case '{':
+                case '[':
+                    break 2;
+                case '\\':
+                    if (isset($glob[$i + 1])) {
+                        switch ($glob[$i + 1]) {
+                            case '*':
+                            case '?':
+                            case '{':
+                            case '}':
+                            case '[':
+                            case ']':
+                            case '-':
+                            case '^':
+                            case '$':
+                            case '~':
+                            case '\\':
+                                $prefix .= $glob[$i + 1];
+                                ++$i;
+                                break;
+                            default:
+                                $prefix .= '\\';
+                        }
+                    }
+                    break;
+                default:
+                    $prefix .= $c;
+                    break;
+            }
+        }
+        return $prefix;
+    }
+
+    /**
+     * Checks if the current position the glob is start of a Recursive directory wildcard
+     *
+     * @psalm-pure
+     */
+    private static function isRecursiveWildcard(string $glob, int $i) : bool
+    {
+        return isset($glob[$i + 3]) && $glob[$i + 1] . $glob[$i + 2] . $glob[$i + 3] === '**/';
+    }
+
+    /**
+     * Converts a glob to a regular expression.
+     *
+     * @param string $glob The canonical glob. The glob should contain forward
+     *                      slashes as directory separators only. It must not
+     *                      contain any "." or ".." segments.
+     *
+     * @return string The regular expression for matching the glob.
+     *
+     * @psalm-pure
+     */
+    private static function toRegEx(string $glob) : string
+    {
+        $delimiter   = '~';
+        $inSquare    = false;
+        $curlyLevels = 0;
+        $regex       = '';
+        $length      = strlen($glob);
+        for ($i = 0; $i < $length; ++$i) {
+            $c = $glob[$i];
+            switch ($c) {
+                case '.':
+                case '(':
+                case ')':
+                case '|':
+                case '+':
+                case '^':
+                case '$':
+                case $delimiter:
+                    $regex .= '\\' . $c;
+                    break;
+                case '/':
+                    if (self::isRecursiveWildcard($glob, $i)) {
+                        $regex .= '/([^/]+/)*';
+                        $i     += 3;
+                    } else {
+                        $regex .= '/';
+                    }
+                    break;
+                case '*':
+                    $regex .= '[^/]*';
+                    break;
+                case '?':
+                    $regex .= '.';
+                    break;
+                case '{':
+                    $regex .= '(';
+                    ++$curlyLevels;
+                    break;
+                case '}':
+                    if ($curlyLevels > 0) {
+                        $regex .= ')';
+                        --$curlyLevels;
+                    } else {
+                        $regex .= '}';
+                    }
+                    break;
+                case ',':
+                    $regex .= $curlyLevels > 0 ? '|' : ',';
+                    break;
+                case '[':
+                    $regex   .= '[';
+                    $inSquare = true;
+                    if (isset($glob[$i + 1]) && $glob[$i + 1] === '^') {
+                        $regex .= '^';
+                        ++$i;
+                    }
+                    break;
+                case ']':
+                    $regex   .= $inSquare ? ']' : '\\]';
+                    $inSquare = false;
+                    break;
+                case '-':
+                    $regex .= $inSquare ? '-' : '\\-';
+                    break;
+                case '\\':
+                    if (isset($glob[$i + 1])) {
+                        switch ($glob[$i + 1]) {
+                            case '*':
+                            case '?':
+                            case '{':
+                            case '}':
+                            case '[':
+                            case ']':
+                            case '-':
+                            case '^':
+                            case '$':
+                            case '~':
+                            case '\\':
+                                $regex .= '\\' . $glob[$i + 1];
+                                ++$i;
+                                break;
+                            default:
+                                $regex .= '\\\\';
+                        }
+                    }
+                    break;
+                default:
+                    $regex .= $c;
+                    break;
+            }
+        }
+        if ($inSquare) {
+            throw new InvalidArgumentException(sprintf(
+                'Invalid glob: missing ] in %s',
+                $glob
+            ));
+        }
+        if ($curlyLevels > 0) {
+            throw new InvalidArgumentException(sprintf(
+                'Invalid glob: missing } in %s',
+                $glob
+            ));
+        }
+        return $delimiter . '^' . $regex . '$' . $delimiter;
+    }
+}
diff --git a/tests/integration/FindOnSamplePhpdocLayout.php b/tests/integration/FindOnSamplePhpdocLayoutTest.php
similarity index 94%
rename from tests/integration/FindOnSamplePhpdocLayout.php
rename to tests/integration/FindOnSamplePhpdocLayoutTest.php
index 56045e7..3d263d5 100644
--- a/tests/integration/FindOnSamplePhpdocLayout.php
+++ b/tests/integration/FindOnSamplePhpdocLayoutTest.php
@@ -20,7 +20,7 @@
  *
  * @coversNothing
  */
-class FindOnSamplePhpdocLayout extends TestCase
+class FindOnSamplePhpdocLayoutTest extends TestCase
 {
     public function testFindingOnSamplePhpdocLayout() : void
     {
diff --git a/tests/integration/FindOnSamplePhpdocLayoutUsingGlobTest.php b/tests/integration/FindOnSamplePhpdocLayoutUsingGlobTest.php
new file mode 100644
index 0000000..1c67287
--- /dev/null
+++ b/tests/integration/FindOnSamplePhpdocLayoutUsingGlobTest.php
@@ -0,0 +1,36 @@
+<?php
+
+declare(strict_types=1);
+
+/**
+ * This file is part of phpDocumentor.
+ *
+ * For the full copyright and license information, please view the LICENSE
+ * file that was distributed with this source code.
+ *
+ * @link      http://phpdoc.org
+ */
+
+namespace Flyfinder;
+
+use PHPUnit\Framework\TestCase;
+
+/**
+ * Integration test against examples/04-sample-phpdoc-layout-using-glob.php
+ *
+ * @coversNothing
+ */
+class FindOnSamplePhpdocLayoutUsingGlobTest extends TestCase
+{
+    public function testFindingOnSamplePhpdocLayout() : void
+    {
+        $result = [];
+        include __DIR__ . '/../../examples/04-sample-phpdoc-layout-using-glob.php';
+
+        $this->assertCount(4, $result);
+        $this->assertSame('JmsSerializerServiceProvider.php', $result[0]['basename']);
+        $this->assertSame('MonologServiceProvider.php', $result[1]['basename']);
+        $this->assertSame('Application.php', $result[2]['basename']);
+        $this->assertSame('Bootstrap.php', $result[3]['basename']);
+    }
+}
diff --git a/tests/unit/Specification/GlobTest.php b/tests/unit/Specification/GlobTest.php
new file mode 100644
index 0000000..f115562
--- /dev/null
+++ b/tests/unit/Specification/GlobTest.php
@@ -0,0 +1,169 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Flyfinder\Specification;
+
+use Generator;
+use InvalidArgumentException;
+use PHPUnit\Framework\TestCase;
+use function is_array;
+use function sprintf;
+
+/**
+ * @coversDefaultClass \Flyfinder\Specification\Glob
+ * @covers ::<private>
+ * @covers ::isSatisfiedBy
+ * @covers ::__construct
+ */
+final class GlobTest extends TestCase
+{
+    /**
+     * @param mixed[] $file
+     *
+     * @dataProvider matchingPatternFileProvider
+     * @dataProvider matchingPatternFileWithEscapeCharProvider
+     *
+     * @psalm-param array{basename: string, path: string, stream: resource, dirname: string, type: string, extension: string} $file
+     */
+    public function testGlobIsMatching(string $pattern, array $file) : void
+    {
+        $glob = new Glob($pattern);
+
+        $this->assertTrue(
+            $glob->isSatisfiedBy($file),
+            sprintf('Failed: %s to match %s', $pattern, $file['path'])
+        );
+    }
+
+    /**
+     * @param mixed[] $file
+     *
+     * @dataProvider notMatchingPatternFileProvider
+     * @psalm-param array{basename: string, path: string, stream: resource, dirname: string, type: string, extension: string} $file
+     */
+    public function testGlobIsNotMatching(string $pattern, array $file) : void
+    {
+        $glob = new Glob($pattern);
+
+        $this->assertFalse(
+            $glob->isSatisfiedBy($file),
+            sprintf('Failed: %s to match %s', $pattern, $file['path'])
+        );
+    }
+
+    /**
+     * @dataProvider invalidPatternProvider
+     */
+    public function testInvalidGlobThrows(string $pattern) : void
+    {
+        $this->expectException(InvalidArgumentException::class);
+        new Glob($pattern);
+    }
+
+    public function invalidPatternProvider() : Generator
+    {
+        $invalidPatterns = [
+            '[aaa',
+            '{aaa',
+            '{a,{b}',
+            'aaaa', //path must be absolute
+        ];
+
+        foreach ($invalidPatterns as $pattern) {
+            yield $pattern => [$pattern];
+        }
+    }
+
+    public function matchingPatternFileProvider() : Generator
+    {
+        $input = [
+            '/*.php' => 'test.php',
+            '/src/*' => 'src/test.php',
+            '/src/**/*.php' => 'src/subdir/test.php',
+            '/src/**/*' => 'src/subdir/second/test.php',
+            '/src/{subdir,other}/*' => [
+                'src/subdir/test.php',
+                'src/other/test.php',
+            ],
+            '/src/subdir/test-[a-c].php' => [
+                'src/subdir/test-a.php',
+                'src/subdir/test-b.php',
+                'src/subdir/test-c.php',
+            ],
+            '/src/subdir/test-[^a-c].php' => 'src/subdir/test-d.php',
+            '/src/subdir/test-?.php' => [
+                'src/subdir/test-a.php',
+                'src/subdir/test-b.php',
+                'src/subdir/test-c.php',
+                'src/subdir/test-~.php',
+            ],
+            '/src/subdir/test-}.php' => 'src/subdir/test-}.php',
+        ];
+
+        yield from $this->toTestData($input);
+    }
+
+    public function matchingPatternFileWithEscapeCharProvider() : Generator
+    {
+        $escapeChars = [
+            '*',
+            '?',
+            '{',
+            '}',
+            '[',
+            ']',
+            '-',
+            '^',
+            '$',
+            '~',
+            '\\',
+            '\\\\',
+        ];
+
+        foreach ($escapeChars as $char) {
+            $file = sprintf('/src/test\\%s.php', $char);
+            yield $file => [
+                $file,
+                ['path' => sprintf('src/test%s.php', $char)],
+            ];
+        }
+    }
+
+    public function notMatchingPatternFileProvider() : Generator
+    {
+        $input = [
+            '/*.php' => 'test.css',
+            '/src/*' => 'src/subdir/test.php',
+            '/src/**/*.php' => 'src/subdir/test.css',
+            '/src/subdir/test-[a-c].php' => 'src/subdir/test-d.php',
+            '/src/subdir/test-[^a-c].php' => [
+                'src/subdir/test-a.php',
+                'src/subdir/test-b.php',
+                'src/subdir/test-c.php',
+            ],
+            '/src' => 'test/file.php',
+        ];
+
+        yield from $this->toTestData($input);
+    }
+
+    /**
+     * @param mixed[] $input
+     */
+    private function toTestData(array $input) : Generator
+    {
+        foreach ($input as $glob => $path) {
+            if (!is_array($path)) {
+                $path = [$path];
+            }
+
+            foreach ($path as $key => $item) {
+                yield ($key !== 0 ? $key . ' - ' : '') . $glob => [
+                    $glob,
+                    ['path' => $item],
+                ];
+            }
+        }
+    }
+}