URLごにょごにょするやつ

test.php

<?php

$text = <<< EOT
http://example.com/path/to?key=value#hash
http://invalid.example.com/path/to?key=value#hash
EOT;

$validator = new UrlDomainValidator(['example.com']); // register domain to whitelist
$extractor = new UrlExtractor($text);

/* out
  string(41) "http://example.com/path/to?key=value#hash"
*/
$url = $extractor->first();

/* out
  array(2) {
  [0]=>
  string(41) "http://example.com/path/to?key=value#hash"
  [1]=>
  string(49) "http://invalid.example.com/path/to?key=value#hash"
  }
*/
$urls = $extractor->all();

/* out
  <a href="http://example.com/path/to?key=value#hash">http://example.com/path/to?key=value#hash</a
  http://invalid.example.com/path/to?key=value#hash
*/
// URLをリンクにするやつ
$text = $extractor->replace(function($match) use($validator) {
    return $validator->validate($match[0]) ? sprintf('<a href="%s">%s</a>', $match[0], $match[0]) : $match[0];
});

/* out
  string(92) "http://example.com/path/to?key=value#hash
  ****://*******************/path/to?key=value#hash"
*/
// 許可ドメイン以外を含むURLを伏せ字にするやつ
$text = $extractor->replace(function($match) use($validator) {
    $subject = $match[0];
    if (!$validator->validate($subject)) {
        $subject = str_replace($match[2], str_repeat('*', mb_strlen($match[2])), $subject);
        $subject = str_replace($match[3], str_repeat('*', mb_strlen($match[3])), $subject);
    }
    return $subject;
});

UrlDomainValidator.php

<?php

class UrlDomainValidator
{
    protected $domains = [];

    public function __construct(array $domains = [])
    {
        $this->domains = $domains;
    }

    public function validate(string $url): bool
    {
        return $this->contains($url);
    }

    private function contains(string $url): bool
    {
        $parsed = parse_url($url);
        if (!$parsed || !isset($parsed['host'])) {
            return false;
        }
        if (!in_array($parsed['host'], $this->domains, true)) {
            return false;
        }
        return true;
    }
}

UrlDomainValidatorTest.php

<?php

use PHPUnit\Framework\TestCase;

class UrlDomainValidatorTest extends TestCase
{
    /**
     * @dataProvider urlsProvider
     */
    public function test_validate($url, $expected)
    {
        $validator = new UrlDomainValidator(['example.com', 'sample.example.com']);
        $this->assertEquals($expected, $validator->validate($url));
    }

    public function urlsProvider()
    {
        return [
            ['http://example.com/', true],
            ['//example.com/', true],
            ['http://example.com:8080/', true],
            ['http://ba0918:pass@example.com:8080/', true],
            ['http://ba0918:pass@sample.example.com:8080/', true],
            ['http://ba0918:pass@lie.sample.example.com:8080/', false],

            // invalid domain
            ['http://foo.bar.com/path/to', false],
            // invalid url
            ['http:/example.com/', false],
        ];
    }
}

UrlExtractor.php

<?php

class UrlExtractor
{
    private $pattern = '/((http|https):)?(?:\/\/)([-\+;:&@=\$,\.\w_]+)(\/[-\+~%\/\.\w_]*)?(\?[-\+=&;%@\w_]*)?(#[\w]*)?/';
    private $subject;

    public function __construct(string $subject)
    {
        $this->subject = $subject ?? '';
    }

    public function first(): ?string
    {
        $result = $this->extract();
        return $result[0] ?? null;
    }

    public function all(): array
    {
        $result = $this->extractAll();
        return $result[0] ?? [];
    }

    public function replace(callable $callback): string
    {
        return preg_replace_callback($this->pattern, $callback, $this->subject);
    }

    private function extract(): array
    {
        if (preg_match($this->pattern, $this->subject, $matches)) {
            return $matches;
        }
        return [];
    }

    private function extractAll(): array
    {
        if (preg_match_all($this->pattern, $this->subject, $matches)) {
            return $matches;
        }
        return [];
    }
}

UrlExtractorTest.php

<?php

use PHPUnit\Framework\TestCase;

class UrlExtractorTest extends TestCase
{
    /**
     * @dataProvider urlsProvider
     */
    public function testFirst($subject, $expected)
    {
        $extractor = new UrlExtractor($subject);
        $this->assertEquals($expected, $extractor->first());
    }

    public function urlsProvider()
    {
        return [
            /* OK cases*/
            ['//hoge.example.com', '//hoge.example.com'],
            ['asdf//hoge.example.com', '//hoge.example.com'],
            ['http://hoge.example.com', 'http://hoge.example.com'],
            ['http://hoge.example.com/', 'http://hoge.example.com/'],
            ['http://ba0918:pass@hoge.example.com', 'http://ba0918:pass@hoge.example.com'],
            ['http://hoge.example.com/path/to', 'http://hoge.example.com/path/to'],
            ['http://hoge.example.com/path/to/', 'http://hoge.example.com/path/to/'],
            ['http://hoge.example.com/path/to/?key=value', 'http://hoge.example.com/path/to/?key=value'],
            ['http://hoge.example.com/path/to/?key=value&key2=value2', 'http://hoge.example.com/path/to/?key=value&key2=value2'],
            ['http://hoge.example.com/path/to/?key=value&key2=value2#hash', 'http://hoge.example.com/path/to/?key=value&key2=value2#hash'],
            ['http://hoge.example.com/?key=value', 'http://hoge.example.com/?key=value'],
            ['http://hoge.example.com/#hash', 'http://hoge.example.com/#hash'],
            ['http://hoge.example.com/#hash<br>', 'http://hoge.example.com/#hash'],

            /* NG cases */
            // invalid scheme
            ['http:/hoge.example.com/#hash', ''],
            ['invalid-scheme://hoge.example.com/#hash', '//hoge.example.com/#hash'],
            // invalid query string
            ['http://hoge.example.com/&key=value', 'http://hoge.example.com/'],
            // unsupported multibyte domain name
            ['http://日本語.jp/?key=value', ''],
        ];
    }
}