WebClient::error() - Code Metrics - Inspection of "Merge pull request #14 from JBleijenberg/master" - vaites/php-apache-tika - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( ab76ce...ee0b8d )

by David

created 2019-02-19 21:03 UTC

WebClient::error() A

↳ Parent: WebClient

Complexity

Conditions	5
Paths	5

Size

Total Lines	27
Code Lines	15

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
cc	5
eloc	15
nc	5
nop	2
dl	0
loc	27
rs	9.4555
c	0
b	0
f	0

<?php

namespace Vaites\ApacheTika\Clients;

use Exception;

use Vaites\ApacheTika\Client;
use Vaites\ApacheTika\Metadata\Metadata;

/**
 * Apache Tika web client
 *
 * @author  David Martínez <[email protected]>
 * @link    http://wiki.apache.org/tika/TikaJAXRS
 * @link    https://tika.apache.org/1.12/formats.html
 */
class WebClient extends Client
{
    const MODE = 'web';

    /**
     * Cached responses to avoid multiple request for the same file
     *
     * @var array
     */
    protected $cache = [];

    /**
     * Apache Tika server host
     *
     * @var string
     */
    protected $host = null;

    /**
     * Apache Tika server port
     *
     * @var int
     */
    protected $port = null;

    /**
     * Number of retries on server error
     *
     * @var int
     */
    protected $retries = 3;

    /**
     * Default cURL options
     *
     * @var array
     */
    protected $options =
    [
        CURLINFO_HEADER_OUT    => true,
        CURLOPT_HTTPHEADER     => [],
        CURLOPT_PUT            => true,
        CURLOPT_RETURNTRANSFER => true,
        CURLOPT_TIMEOUT        => 5
    ];

    /**
     * Configure class and test if server is running
     *
     * @param   string  $host
     * @param   int     $port
     * @param   array   $options
     * @throws  \Exception
     */
    public function __construct($host = null, $port = null, $options = [])
    {
        if(is_string($host) && filter_var($host, FILTER_VALIDATE_URL))
        {
            $this->setUrl($host);
        }
        elseif($host)
        {
            $this->setHost($host);
        }

        if(is_numeric($port))
        {
            $this->setPort($port);
        }

        if(!empty($options))
        {
            $this->setOptions($options);
        }

        $this->setDownloadRemote(true);

        if(self::$check === true)
        {
            $this->check();
        }
    }

    /**
     * Get the base URL
     *
     * @return string
     */
    public function getUrl()
    {
        return sprintf('http://%s:%d', $this->host, $this->port ?: 9998);
    }

    /**
     * Set the host and port using an URL
     *
     * @param   string  $url
     * @return $this
     */
    public function setUrl($url)
    {
        $url = parse_url($url);

        $this->setHost($url['host']);

        if(isset($url['port']))
        {
            $this->setPort($url['port']);
        }

        return $this;
    }

    /**
     * Get the host
     *
     * @return  null|string
     */
    public function getHost()
    {
        return $this->host;
    }

    /**
     * Set the host
     *
     * @param   string  $host
     * @return  $this
     */
    public function setHost($host)
    {
        $this->host = $host;

        return $this;
    }

    /**
     * Get the port
     *
     * @return  null|int
     */
    public function getPort()
    {
        return $this->port;
    }

    /**
     * Set the port
     *
     * @param   int     $port
     * @return  $this
     */
    public function setPort($port)
    {
        $this->port = $port;

        return $this;
    }

    /**
     * Get the number of retries
     *
     * @return  int
     */
    public function getRetries()
    {
        return $this->retries;
    }

    /**
     * Set the number of retries
     *
     * @param   int     $retries
     * @return  $this
     */
    public function setRetries($retries)
    {
        $this->retries = $retries;

        return $this;
    }

    /**
     * Get all the options
     *
     * @return  null|array
     */
    public function getOptions()
    {
        return $this->options;
    }

    /**
     * Get an specified option
     *
     * @param   string  $key
     * @return  mixed
     */
    public function getOption($key)
    {
        return isset($this->options[$key]) ? $this->options[$key] : null;
    }

    /**
     * Set a cURL option to be set with curl_setopt()
     *
     * @link    http://php.net/manual/en/curl.constants.php
     * @link    http://php.net/manual/en/function.curl-setopt.php
     * @param   string  $key
     * @param   mixed   $value
     * @return  $this
     * @throws  \Exception
     */
    public function setOption($key, $value)
    {
        if(in_array($key, [CURLINFO_HEADER_OUT, CURLOPT_PUT, CURLOPT_RETURNTRANSFER]))
        {
            throw new Exception("Value for cURL option $key cannot be modified", 3);
        }

        $this->options[$key] = $value;

        return $this;
    }

    /**
     * Set the cURL options
     *
     * @param   array   $options
     * @return  $this
     * @throws  \Exception
     */
    public function setOptions($options)
    {
        foreach($options as $key => $value)
        {
            $this->setOption($key, $value);
        }

        return $this;
    }

    /**
     * Get the timeout value for cURL
     *
     * @return  int
     */
    public function getTimeout()
    {
        return $this->getOption(CURLOPT_TIMEOUT);
    }

    /**
     * Set the timeout value for cURL
     *
     * @param   int     $value
     * @return  $this
     * @throws  \Exception
     */
    public function setTimeout($value)
    {
        $this->setOption(CURLOPT_TIMEOUT, (int) $value);

        return $this;
    }

    /**
     * Check if server is running
     *
     * @throws \Exception
     */
    public function check()
    {
        if(self::isChecked() === false)
        {
            self::setChecked(true);

            // throws an exception if server is unreachable or can't connect
            $this->request('version');
        }
    }

    /**
     * Configure, make a request and return its results
     *
     * @param   string  $type
     * @param   string  $file
     * @return  string
     * @throws  \Exception
     */
    public function request($type, $file = null)
    {
        static $retries = [];

        // check if not checked
        $this->check();

        // check if is cached
        if(isset($this->cache[sha1($file)][$type]))
        {
            return $this->cache[sha1($file)][$type];
        }
        elseif(!isset($retries[sha1($file)]))
        {
            $retries[sha1($file)] = $this->retries;
        }

        // parameters for cURL request
        list($resource, $headers) = $this->getParameters($type, $file);

        // check the request
        $file = parent::checkRequest($type, $file);

        // cURL options
        $options = $this->getCurlOptions($type, $file);

        // sets headers
        foreach($headers as $header)
        {
            $options[CURLOPT_HTTPHEADER][] = $header;
        }

        // cURL init and options
        $options[CURLOPT_URL] = $this->getUrl() . "/$resource";

        // get the response and the HTTP status code
        list($response, $status) = $this->exec($options);

        if ($file && is_resource($options[CURLOPT_INFILE])){
            fclose($options[CURLOPT_INFILE]);
        }

        // request completed successfully
        if($status == 200)
        {
            if($type == 'meta')
            {
                $response = Metadata::make($response, $file);
            }

            // cache certain responses
            if(in_array($type, ['lang', 'meta']))
            {
                $this->cache[sha1($file)][$type] = $response;
            }
        }
        // request completed successfully but result is empty
        elseif($status == 204)
        {
            $response = null;
        }
        // retry on request failed with error 500
        elseif($status == 500 && $retries[sha1($file)]--)
        {
            $response = $this->request($type, $file);
        }
        // other status code is an error
        else
        {
            $this->error($status, $resource);
        }

        return $response;

    }

    /**
     * Make a request to Apache Tika Server
     *
     * @param   array   $options
     * @return  array
     * @throws  \Exception
     */
    protected function exec(array $options = [])
    {
        // cURL init and options
        $curl = curl_init();

        // we avoid curl_setopt_array($curl, $options) because extrange Windows behaviour (issue #8)
        foreach($options as $option => $value)
        {
            curl_setopt($curl, $option, $value);
        }

        // make the request
        if(is_null($this->callback))
        {
            $this->response = curl_exec($curl);
        }
        else
        {
            $this->response = '';
            curl_exec($curl);
        }

        // exception if cURL fails
        if(curl_errno($curl))
        {
            throw new Exception(curl_error($curl), curl_errno($curl));
        }

        // return the response and the status code
        return [trim($this->response), curl_getinfo($curl, CURLINFO_HTTP_CODE)];
    }

    /**
     * Throws an exception for an error status code
     *
     * @codeCoverageIgnore
     *
     * @param   int       $status
     * @param   string    $resource
     * @throws  \Exception
     */
    protected function error($status, $resource)
    {
        switch($status)
        {
            //  method not allowed
            case 405:
                throw new Exception('Method not allowed', 405);
                break;

            //  unsupported media type
            case 415:
                throw new Exception('Unsupported media type', 415);
                break;

            //  unprocessable entity
            case 422:
                throw new Exception('Unprocessable document', 422);
                break;

            // server error
            case 500:
                throw new Exception('Error while processing document', 500);
                break;

            // unexpected
            default:
                throw new Exception("Unexpected response for /$resource ($status)", 501);
        }
    }

    /**
     * Get the parameters to make the request
     *
     * @link    https://wiki.apache.org/tika/TikaJAXRS#Specifying_a_URL_Instead_of_Putting_Bytes
     * @param   string  $type
     * @param   string  $file
     * @return  array
     * @throws  \Exception
     */
    protected function getParameters($type, $file = null)
    {
        $headers = [];

        if(!empty($file) && preg_match('/^http/', $file))
        {
            $headers[] = "fileUrl:$file";
        }

        switch($type)
        {
            case 'html':
                $resource = 'tika';
                $headers[] = 'Accept: text/html';
                break;

            case 'lang':
                $resource = 'language/stream';
                break;

            case 'mime':
                $name = basename($file);
                $resource = 'detect/stream';
                $headers[] = "Content-Disposition: attachment, filename=$name";
                break;

            case 'meta':
                $resource = 'meta';
                $headers[] = 'Accept: application/json';
                break;

            case 'text':
                $resource = 'tika';
                $headers[] = 'Accept: text/plain';
                break;

            case 'text-main':
                $resource = 'tika/main';
                $headers[] = 'Accept: text/plain';
                break;

            case 'detectors':
            case 'parsers':
            case 'mime-types':
            case 'version':
                $resource = $type;
                break;

            default:
                throw new Exception("Unknown type $type");
        }

        return [$resource, $headers];
    }

    /**
     * Get the cURL options
     *
     * @param   string  $type
     * @param   string  $file
     * @return  array
     * @throws  \Exception
     */
    protected function getCurlOptions($type, $file = null)
    {
        // base options
        $options = $this->options;

        // callback
        if(!is_null($this->callback))
        {
            $callback = $this->callback;

            $options[CURLOPT_WRITEFUNCTION] = function($handler, $data) use($callback)
            {
                $this->response .= $data;

                $callback($data);

                // safe because cURL must receive the number of *bytes* written
                return strlen($data);
            };
        }

        // remote file options
        if($file && preg_match('/^http/', $file))
        {
            //
        }
        // local file options
        elseif($file && file_exists($file) && is_readable($file))
        {
            $options[CURLOPT_INFILE] = fopen($file, 'r');
            $options[CURLOPT_INFILESIZE] = filesize($file);
        }
        // other options for specific requests
        elseif(in_array($type,  ['detectors', 'mime-types', 'parsers', 'version']))
        {
            $options[CURLOPT_PUT] = false;
        }
        // file not accesible
        else
        {
            throw new Exception("File $file can't be opened");
        }

        return $options;
    }
}


1			<?php
2
3			namespace Vaites\ApacheTika\Clients;
4
5			use Exception;
6
7			use Vaites\ApacheTika\Client;
8			use Vaites\ApacheTika\Metadata\Metadata;
9
10			/**
11			* Apache Tika web client
12			*
13			* @author David Martínez <[email protected]>
14			* @link http://wiki.apache.org/tika/TikaJAXRS
15			* @link https://tika.apache.org/1.12/formats.html
16			*/
17			class WebClient extends Client
18			{
19			const MODE = 'web';
20
21			/**
22			* Cached responses to avoid multiple request for the same file
23			*
24			* @var array
25			*/
26			protected $cache = [];
27
28			/**
29			* Apache Tika server host
30			*
31			* @var string
32			*/
33			protected $host = null;
34
35			/**
36			* Apache Tika server port
37			*
38			* @var int
39			*/
40			protected $port = null;
41
42			/**
43			* Number of retries on server error
44			*
45			* @var int
46			*/
47			protected $retries = 3;
48
49			/**
50			* Default cURL options
51			*
52			* @var array
53			*/
54			protected $options =
55			[
56			CURLINFO_HEADER_OUT => true,
57			CURLOPT_HTTPHEADER => [],
58			CURLOPT_PUT => true,
59			CURLOPT_RETURNTRANSFER => true,
60			CURLOPT_TIMEOUT => 5
61			];
62
63			/**
64			* Configure class and test if server is running
65			*
66			* @param string $host
67			* @param int $port
68			* @param array $options
69			* @throws \Exception
70			*/
71			public function __construct($host = null, $port = null, $options = [])
72			{
73			if(is_string($host) && filter_var($host, FILTER_VALIDATE_URL))
74			{
75			$this->setUrl($host);
76			}
77			elseif($host)
78			{
79			$this->setHost($host);
80			}
81
82			if(is_numeric($port))
83			{
84			$this->setPort($port);
85			}
86
87			if(!empty($options))
88			{
89			$this->setOptions($options);
90			}
91
92			$this->setDownloadRemote(true);
93
94			if(self::$check === true)
95			{
96			$this->check();
97			}
98			}
99
100			/**
101			* Get the base URL
102			*
103			* @return string
104			*/
105			public function getUrl()
106			{
107			return sprintf('http://%s:%d', $this->host, $this->port ?: 9998);
108			}
109
110			/**
111			* Set the host and port using an URL
112			*
113			* @param string $url
114			* @return $this
115			*/
116			public function setUrl($url)
117			{
118			$url = parse_url($url);
119
120			$this->setHost($url['host']);
121
122			if(isset($url['port']))
123			{
124			$this->setPort($url['port']);
125			}
126
127			return $this;
128			}
129
130			/**
131			* Get the host
132			*
133			* @return null\|string
134			*/
135			public function getHost()
136			{
137			return $this->host;
138			}
139
140			/**
141			* Set the host
142			*
143			* @param string $host
144			* @return $this
145			*/
146			public function setHost($host)
147			{
148			$this->host = $host;
149
150			return $this;
151			}
152
153			/**
154			* Get the port
155			*
156			* @return null\|int
157			*/
158			public function getPort()
159			{
160			return $this->port;
161			}
162
163			/**
164			* Set the port
165			*
166			* @param int $port
167			* @return $this
168			*/
169			public function setPort($port)
170			{
171			$this->port = $port;
172
173			return $this;
174			}
175
176			/**
177			* Get the number of retries
178			*
179			* @return int
180			*/
181			public function getRetries()
182			{
183			return $this->retries;
184			}
185
186			/**
187			* Set the number of retries
188			*
189			* @param int $retries
190			* @return $this
191			*/
192			public function setRetries($retries)
193			{
194			$this->retries = $retries;
195
196			return $this;
197			}
198
199			/**
200			* Get all the options
201			*
202			* @return null\|array
203			*/
204			public function getOptions()
205			{
206			return $this->options;
207			}
208
209			/**
210			* Get an specified option
211			*
212			* @param string $key
213			* @return mixed
214			*/
215			public function getOption($key)
216			{
217			return isset($this->options[$key]) ? $this->options[$key] : null;
218			}
219
220			/**
221			* Set a cURL option to be set with curl_setopt()
222			*
223			* @link http://php.net/manual/en/curl.constants.php
224			* @link http://php.net/manual/en/function.curl-setopt.php
225			* @param string $key
226			* @param mixed $value
227			* @return $this
228			* @throws \Exception
229			*/
230			public function setOption($key, $value)
231			{
232			if(in_array($key, [CURLINFO_HEADER_OUT, CURLOPT_PUT, CURLOPT_RETURNTRANSFER]))
233			{
234			throw new Exception("Value for cURL option $key cannot be modified", 3);
235			}
236
237			$this->options[$key] = $value;
238
239			return $this;
240			}
241
242			/**
243			* Set the cURL options
244			*
245			* @param array $options
246			* @return $this
247			* @throws \Exception
248			*/
249			public function setOptions($options)
250			{
251			foreach($options as $key => $value)
252			{
253			$this->setOption($key, $value);
254			}
255
256			return $this;
257			}
258
259			/**
260			* Get the timeout value for cURL
261			*
262			* @return int
263			*/
264			public function getTimeout()
265			{
266			return $this->getOption(CURLOPT_TIMEOUT);
267			}
268
269			/**
270			* Set the timeout value for cURL
271			*
272			* @param int $value
273			* @return $this
274			* @throws \Exception
275			*/
276			public function setTimeout($value)
277			{
278			$this->setOption(CURLOPT_TIMEOUT, (int) $value);
279
280			return $this;
281			}
282
283			/**
284			* Check if server is running
285			*
286			* @throws \Exception
287			*/
288			public function check()
289			{
290			if(self::isChecked() === false)
291			{
292			self::setChecked(true);
293
294			// throws an exception if server is unreachable or can't connect
295			$this->request('version');
296			}
297			}
298
299			/**
300			* Configure, make a request and return its results
301			*
302			* @param string $type
303			* @param string $file
304			* @return string
305			* @throws \Exception
306			*/
307			public function request($type, $file = null)
308			{
309			static $retries = [];
310
311			// check if not checked
312			$this->check();
313
314			// check if is cached
315			if(isset($this->cache[sha1($file)][$type]))
316			{
317			return $this->cache[sha1($file)][$type];
318			}
319			elseif(!isset($retries[sha1($file)]))
320			{
321			$retries[sha1($file)] = $this->retries;
322			}
323
324			// parameters for cURL request
325			list($resource, $headers) = $this->getParameters($type, $file);
326
327			// check the request
328			$file = parent::checkRequest($type, $file);
329
330			// cURL options
331			$options = $this->getCurlOptions($type, $file);
332
333			// sets headers
334			foreach($headers as $header)
335			{
336			$options[CURLOPT_HTTPHEADER][] = $header;
337			}
338
339			// cURL init and options
340			$options[CURLOPT_URL] = $this->getUrl() . "/$resource";
341
342			// get the response and the HTTP status code
343			list($response, $status) = $this->exec($options);
344
345			if ($file && is_resource($options[CURLOPT_INFILE])){
346			fclose($options[CURLOPT_INFILE]);
347			}
348
349			// request completed successfully
350			if($status == 200)
351			{
352			if($type == 'meta')
353			{
354			$response = Metadata::make($response, $file);
355			}
356
357			// cache certain responses
358			if(in_array($type, ['lang', 'meta']))
359			{
360			$this->cache[sha1($file)][$type] = $response;
361			}
362			}
363			// request completed successfully but result is empty
364			elseif($status == 204)
365			{
366			$response = null;
367			}
368			// retry on request failed with error 500
369			elseif($status == 500 && $retries[sha1($file)]--)
370			{
371			$response = $this->request($type, $file);
372			}
373			// other status code is an error
374			else
375			{
376			$this->error($status, $resource);
377			}
378
379			return $response;
			0 ignored issues – show Bug Best Practice introduced 2019-01-20 21:26 UTC by Report Bug Copy Issue Report The expression `return $response` also could return the type `Vaites\ApacheTika\Metada...\Metadata\ImageMetadata` which is incompatible with the documented return type `string`. Loading history...
380			}
381
382			/**
383			* Make a request to Apache Tika Server
384			*
385			* @param array $options
386			* @return array
387			* @throws \Exception
388			*/
389			protected function exec(array $options = [])
390			{
391			// cURL init and options
392			$curl = curl_init();
393
394			// we avoid curl_setopt_array($curl, $options) because extrange Windows behaviour (issue #8)
395			foreach($options as $option => $value)
396			{
397			curl_setopt($curl, $option, $value);
398			}
399
400			// make the request
401			if(is_null($this->callback))
402			{
403			$this->response = curl_exec($curl);
404			}
405			else
406			{
407			$this->response = '';
408			curl_exec($curl);
409			}
410
411			// exception if cURL fails
412			if(curl_errno($curl))
413			{
414			throw new Exception(curl_error($curl), curl_errno($curl));
415			}
416
417			// return the response and the status code
418			return [trim($this->response), curl_getinfo($curl, CURLINFO_HTTP_CODE)];
419			}
420
421			/**
422			* Throws an exception for an error status code
423			*
424			* @codeCoverageIgnore
425			*
426			* @param int $status
427			* @param string $resource
428			* @throws \Exception
429			*/
430			protected function error($status, $resource)
431			{
432			switch($status)
433			{
434			// method not allowed
435			case 405:
436			throw new Exception('Method not allowed', 405);
437			break;
438
439			// unsupported media type
440			case 415:
441			throw new Exception('Unsupported media type', 415);
442			break;
443
444			// unprocessable entity
445			case 422:
446			throw new Exception('Unprocessable document', 422);
447			break;
448
449			// server error
450			case 500:
451			throw new Exception('Error while processing document', 500);
452			break;
453
454			// unexpected
455			default:
456			throw new Exception("Unexpected response for /$resource ($status)", 501);
457			}
458			}
459
460			/**
461			* Get the parameters to make the request
462			*
463			* @link https://wiki.apache.org/tika/TikaJAXRS#Specifying_a_URL_Instead_of_Putting_Bytes
464			* @param string $type
465			* @param string $file
466			* @return array
467			* @throws \Exception
468			*/
469			protected function getParameters($type, $file = null)
470			{
471			$headers = [];
472
473			if(!empty($file) && preg_match('/^http/', $file))
474			{
475			$headers[] = "fileUrl:$file";
476			}
477
478			switch($type)
479			{
480			case 'html':
481			$resource = 'tika';
482			$headers[] = 'Accept: text/html';
483			break;
484
485			case 'lang':
486			$resource = 'language/stream';
487			break;
488
489			case 'mime':
490			$name = basename($file);
491			$resource = 'detect/stream';
492			$headers[] = "Content-Disposition: attachment, filename=$name";
493			break;
494
495			case 'meta':
496			$resource = 'meta';
497			$headers[] = 'Accept: application/json';
498			break;
499
500			case 'text':
501			$resource = 'tika';
502			$headers[] = 'Accept: text/plain';
503			break;
504
505			case 'text-main':
506			$resource = 'tika/main';
507			$headers[] = 'Accept: text/plain';
508			break;
509
510			case 'detectors':
511			case 'parsers':
512			case 'mime-types':
513			case 'version':
514			$resource = $type;
515			break;
516
517			default:
518			throw new Exception("Unknown type $type");
519			}
520
521			return [$resource, $headers];
522			}
523
524			/**
525			* Get the cURL options
526			*
527			* @param string $type
528			* @param string $file
529			* @return array
530			* @throws \Exception
531			*/
532			protected function getCurlOptions($type, $file = null)
533			{
534			// base options
535			$options = $this->options;
536
537			// callback
538			if(!is_null($this->callback))
539			{
540			$callback = $this->callback;
541
542			$options[CURLOPT_WRITEFUNCTION] = function($handler, $data) use($callback)
543			{
544			$this->response .= $data;
545
546			$callback($data);
547
548			// safe because cURL must receive the number of bytes written
549			return strlen($data);
550			};
551			}
552
553			// remote file options
554			if($file && preg_match('/^http/', $file))
555			{
556			//
557			}
558			// local file options
559			elseif($file && file_exists($file) && is_readable($file))
560			{
561			$options[CURLOPT_INFILE] = fopen($file, 'r');
562			$options[CURLOPT_INFILESIZE] = filesize($file);
563			}
564			// other options for specific requests
565			elseif(in_array($type, ['detectors', 'mime-types', 'parsers', 'version']))
566			{
567			$options[CURLOPT_PUT] = false;
568			}
569			// file not accesible
570			else
571			{
572			throw new Exception("File $file can't be opened");
573			}
574
575			return $options;
576			}
577			}
578

vaites / php-apache-tika

Push — master ( ab76ce...ee0b8d )

WebClient::error() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like