Client::getMainText() - Code Metrics - vaites/php-apache-tika - Measure and Improve Code Quality continuously with Scrutinizer

Client::getMainText() A
last analyzed 2025-03-16 10:38 UTC

↳ Parent: Client

Complexity

Conditions	2
Paths	2

Size

Total Lines	8
Code Lines	3

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
cc	2
eloc	3
c	0
b	0
f	0
nc	2
nop	3
dl	0
loc	8
rs	10

<?php

namespace Vaites\ApacheTika;

use Closure;
use Exception;
use stdClass;

use Vaites\ApacheTika\Clients\CLIClient;
use Vaites\ApacheTika\Clients\WebClient;
use Vaites\ApacheTika\Metadata\Metadata;
use Vaites\ApacheTika\Metadata\MetadataInterface;

/**
 * Apache Tika client interface
 *
 * @author  David Martínez <[email protected]>
 * @link    https://tika.apache.org/1.24/formats.html
 */
abstract class Client
{
    protected const MODE = null;

    /**
     * Checked flag
     *
     * @var bool
     */
    protected $checked = false;

    /**
     * Response using callbacks
     *
     * @var string
     */
    protected $response = null;

    /**
     * Platform (unix or win)
     *
     * @var string
     */
    protected $platform = null;

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

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

    /**
     * Text encoding
     *
     * @var string|null
     */
    protected $encoding = null;

    /**
     * Callback called on secuential read
     *
     * @var callable|null
     */
    protected $callback = null;

    /**
     * Enable or disable appending when using callback
     *
     * @var bool
     */
    protected $callbackAppend = true;

    /**
     * Size of chunks for callback
     *
     * @var int
     */
    protected $chunkSize = 1048576;

    /**
     * Remote download flag
     *
     * @var bool
     */
    protected $downloadRemote = false;

    /**
     * Configure client
     */
    public function __construct()
    {
        $this->platform = defined('PHP_WINDOWS_VERSION_MAJOR') ? 'win' : 'unix';
    }

    /**
     * Get a class instance throwing an exception if check fails
     *
     * @param string|null     $param1   path or host
     * @param string|int|null $param2   Java binary path or port for web client
     * @param array           $options  options for cURL request
     * @param bool            $check    check JAR file or server connection
     * @return \Vaites\ApacheTika\Clients\CLIClient|\Vaites\ApacheTika\Clients\WebClient
     * @throws \Exception
     */
    public static function make(?string $param1 = null, $param2 = null, array $options = [], bool $check = true): Client
    {
        if(preg_match('/\.jar$/', func_get_arg(0)))
        {
            $path = $param1 ? (string) $param1 : null;
            $java = $param2 ? (string) $param2 : null;

            return new CLIClient($path, $java, $check);
        }
        else
        {
            $host = $param1 ? (string) $param1 : null;
            $port = $param2 ? (int) $param2 : null;

            return new WebClient($host, $port, $options, $check);
        }
    }

    /**
     * Get a class instance delaying the check
     *
     * @param string|null $param1 path or host
     * @param int|null    $param2 Java binary path or port for web client
     * @param array       $options options for cURL request
     * @return \Vaites\ApacheTika\Clients\CLIClient|\Vaites\ApacheTika\Clients\WebClient
     * @throws \Exception
     */
    public static function prepare($param1 = null, $param2 = null, $options = []): Client
    {
        return self::make($param1, $param2, $options, false);
    }

    /**
     * Get the encoding
     */
    public function getEncoding(): ?string
    {
        return $this->encoding;
    }

    /**
     * Set the encoding
     *
     * @throws \Exception
     */
    public function setEncoding(string $encoding): self
    {
        if(!empty($encoding))
        {
            $this->encoding = $encoding;
        }
        else
        {
            throw new Exception('Invalid encoding');
        }

        return $this;
    }

    /**
     * Get the callback
     */
    public function getCallback(): ?callable
    {
        return $this->callback;
    }

    /**
     * Set the callback (callable or closure) for call on secuential read
     *
     * @throws \Exception
     */
    public function setCallback(callable $callback, bool $append = true): self
    {
        if($callback instanceof Closure || is_array($callback))
        {
            $this->callbackAppend = (bool) $append;
            $this->callback = $callback;
        }
        elseif(is_string($callback))
        {
            $this->callbackAppend = (bool) $append;
            $this->callback = function($chunk) use ($callback)
            {
                return call_user_func_array($callback, [$chunk]);
            };
        }
        else
        {
            throw new Exception('Invalid callback');
        }

        return $this;
    }

    /**
     * Get the chunk size
     */
    public function getChunkSize(): int
    {
        return $this->chunkSize;
    }

    /**
     * Set the chunk size for secuential read
     *
     * @throws \Exception
     */
    public function setChunkSize(int $size): self
    {
        if(static::MODE == 'cli')

        {
            $this->chunkSize = $size;
        }
        else
        {
            throw new Exception('Chunk size is not supported on web mode');
        }

        return $this;
    }

    /**
     * Get the remote download flag
     */
    public function getDownloadRemote(): bool
    {
        return $this->downloadRemote;
    }

    /**
     * Set the remote download flag
     */
    public function setDownloadRemote(bool $download): self
    {
        $this->downloadRemote = (bool) $download;

        return $this;
    }

    /**
     * Gets file metadata
     *
     * @throws \Exception
     */
    public function getMetadata(string $file): MetadataInterface
    {
        $response = $this->parseJsonResponse($this->request('meta', $file) ?: 'ERROR');

        if($response instanceof stdClass === false)
        {
            throw new Exception("Unexpected metadata response for $file");
        }

        return Metadata::make($response, $file);
    }

    /**
     * Gets recursive file metadata where the returned array indexes are the file name.
     *
     * Example: for a sample.zip with an example.doc file, the return array looks like if be defined as:
     *
     *  [
     *      'sample.zip' => new Metadata()
     *      'sample.zip/example.doc' => new DocumentMetadata()
     *  ]
     *
     * @link https://cwiki.apache.org/confluence/display/TIKA/TikaServer#TikaServer-RecursiveMetadataandContent
     * @throws \Exception
     */
    public function getRecursiveMetadata(string $file, ?string $format = 'ignore'): array
    {
        if(in_array($format, ['text', 'html', 'ignore']) === false)
        {
            throw new Exception("Unknown recursive type (must be text, html, ignore or null)");
        }

        $response = $this->parseJsonResponse($this->request("rmeta/$format", $file) ?: 'ERROR');

        if(is_array($response) === false)
        {
            throw new Exception("Unexpected metadata response for $file");
        }

        $metadata = [];

        foreach($response as $item)
        {
            $name = basename($file);
            if(isset($item->{'X-TIKA:embedded_resource_path'}))
            {
                $name .= $item->{'X-TIKA:embedded_resource_path'};
            }

            $metadata[$name] = Metadata::make($item, $file);
        }

        return $metadata;
    }

    /**
     * Detect language
     *
     * @throws \Exception
     */
    public function getLanguage(string $file): ?string
    {
        return $this->request('lang', $file);
    }

    /**
     * Detect MIME type
     *
     * @throws \Exception
     */
    public function getMIME(string $file): ?string
    {
        return $this->request('mime', $file);
    }

    /**
     * Extracts HTML
     *
     * @throws \Exception
     */
    public function getHTML(string $file, ?callable $callback = null, bool $append = true): ?string
    {
        if(!is_null($callback))
        {
            $this->setCallback($callback, $append);
        }

        return $this->request('html', $file);
    }

    /**
     * Extracts XHTML
     *
     * @throws \Exception
     */
    public function getXHTML(string $file, ?callable $callback = null, bool $append = true): ?string
    {
        if(!is_null($callback))
        {
            $this->setCallback($callback, $append);
        }

        return $this->request('xhtml', $file);
    }

    /**
     * Extracts text
     *
     * @throws \Exception
     */
    public function getText(string $file, ?callable $callback = null, bool $append = true): ?string
    {
        if(!is_null($callback))
        {
            $this->setCallback($callback, $append);
        }

        return $this->request('text', $file);
    }

    /**
     * Extracts main text
     *
     * @throws \Exception
     */
    public function getMainText(string $file, ?callable $callback = null, bool $append = true): ?string
    {
        if(!is_null($callback))
        {
            $this->setCallback($callback, $append);
        }

        return $this->request('text-main', $file);
    }

    /**
     * Returns current Tika version
     *
     * @throws \Exception
     */
    public function getVersion(): ?string
    {
        if(is_null($this->version))

        {
            $this->setVersion($this->request('version'));
        }

        return $this->version;
    }

    /**
     * Set the Tika version
     */
    public function setVersion(string $version): self
    {
        $this->checked = true;
        $this->version = $version;

        return $this;
    }

    /**
     * Return the list of Apache Tika supported versions
     *
     * @throws \Exception
     */
    public function getSupportedVersions(): array
    {
        static $versions = null;

        if(is_null($versions))
        {
            $composer = file_get_contents(dirname(__DIR__) . '/composer.json');

            if($composer === false)
            {
                throw new Exception("An error ocurred trying to read package's composer.json file");
            }

            $versions = json_decode($composer, true)['extra']['supported-versions'] ?? null;

            if(empty($versions))
            {
                throw new Exception("An error ocurred trying to read package's composer.json file");
            }
        }

        return $versions;
    }

    /**
     * Sets the checked flag
     */
    public function setChecked(bool $checked): self
    {
        $this->checked = (bool) $checked;

        return $this;
    }

    /**
     * Checks if instance is checked
     */
    public function isChecked(): bool
    {
        return $this->checked;
    }

    /**
     * Check if a response is cached
     */
    protected function isCached(string $type, string $file): bool
    {
        return isset($this->cache[sha1($file)][$type]);
    }

    /**
     * Get a cached response
     *
     * @return mixed
     */
    protected function getCachedResponse(string $type, string $file)
    {
        return $this->cache[sha1($file)][$type] ?? null;
    }

    /**
     * Check if a request type must be cached
     */
    protected function isCacheable(string $type): bool
    {
        return in_array($type, ['lang', 'meta']);
    }

    /**
     * Caches a response
     *
     * @param mixed $response
     */
    protected function cacheResponse(string $type, $response, string $file): bool
    {
        $this->cache[sha1($file)][$type] = $response;

        return true;
    }

    /**
     * Checks if a specific version is supported
     */
    public function isVersionSupported(string $version): bool
    {
        return in_array($version, $this->getSupportedVersions());
    }

    /**
     * Check if a mime type is supported
     *
     * @param string $mime
     * @return bool
     * @throws \Exception
     */
    public function isMIMETypeSupported(string $mime): bool
    {
        return array_key_exists($mime, $this->getSupportedMIMETypes());
    }

    /**
     * Check the request before executing
     *
     * @throws \Exception
     */
    public function checkRequest(string $type, ?string $file = null): ?string
    {
        // no checks for getters
        if(in_array($type, ['detectors', 'mime-types', 'parsers', 'version']))
        {
            //
        }
        // invalid local file
        elseif($file !== null && !preg_match('/^http/', $file) && !file_exists($file))
        {
            throw new Exception("File $file can't be opened");
        }
        // download remote file if required only for integrated downloader
        elseif($file !== null && preg_match('/^http/', $file) && $this->downloadRemote)
        {
            $file = $this->downloadFile($file);
        }
        // invalid remote file
        elseif($file !== null && preg_match('/^http/', $file))
        {
            $headers = get_headers($file);

            if(empty($headers) || !preg_match('/200/', $headers[0]))
            {
                throw new Exception("File $file can't be opened", 2);
            }
        }

        return $file;
    }

    /**
     * Filter response to fix common issues
     *
     * @param string $response
     * @return string
     */
    protected function filterResponse(string $response): string
    {
        // fix Log4j2 warning
        $response = trim(str_replace
        (
            'WARNING: sun.reflect.Reflection.getCallerClass is not supported. This will impact performance.',
            '',
            $response
        ));

        return trim($response);
    }

    /**
     * Parse the response returned by Apache Tika
     *
     * @return mixed
     * @throws \Exception
     */
    protected function parseJsonResponse(string $response)
    {
        // an empty response throws an error
        if(empty($response) || trim($response) == '')
        {
            throw new Exception('Empty response');
        }

        // decode the JSON response
        $json = json_decode($response);

        // exceptions if metadata is not valid
        if(json_last_error())
        {
            $message = function_exists('json_last_error_msg') ? json_last_error_msg() : 'Error parsing JSON response';

            throw new Exception($message, json_last_error());
        }

        return $json;
    }

    /**
     * Download file to a temporary folder and return its path
     *
     * @link https://wiki.apache.org/tika/TikaJAXRS#Specifying_a_URL_Instead_of_Putting_Bytes
     * @throws \Exception
     */
    protected function downloadFile(string $file): string
    {
        $dest = tempnam(sys_get_temp_dir(), 'TIKA');

        if($dest === false)
        {
            throw new Exception("Can't create a temporary file at " . sys_get_temp_dir());
        }

        $fp = fopen($dest, 'w+');

        if($fp === false)
        {
            throw new Exception("$dest can't be opened");
        }

        $ch = curl_init($file);

        if($ch === false)
        {
            throw new Exception("$file can't be downloaded");
        }

        curl_setopt($ch, CURLOPT_FILE, $fp);
        curl_setopt($ch, CURLOPT_TIMEOUT, 5);
        curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true);
        curl_exec($ch);

        if(curl_errno($ch))
        {
            throw new Exception(curl_error($ch));
        }

        $code = curl_getinfo($ch, CURLINFO_HTTP_CODE);

        curl_close($ch);

        if($code != 200)
        {
            throw new Exception("$file can't be downloaded", $code);
        }

        return $dest;
    }

    /**
     * Must return the supported MIME types
     *
     * @throws \Exception
     */
    abstract public function getSupportedMIMETypes(): array;

    /**
     * Must return the available detectors
     *
     * @throws \Exception
     */
    abstract public function getAvailableDetectors(): array;

    /**
     * Must return the available parsers
     *
     * @throws \Exception
     */
    abstract public function getAvailableParsers(): array;

    /**
     * Check Java binary, JAR path or server connection
     */
    abstract public function check(): void;

    /**
     * Configure and make a request and return its results.
     *
     * @throws \Exception
     */
    abstract public function request(string $type, ?string $file = null): ?string;
}


1			<?php
2
3			namespace Vaites\ApacheTika;
4
5			use Closure;
6			use Exception;
7			use stdClass;
8
9			use Vaites\ApacheTika\Clients\CLIClient;
10			use Vaites\ApacheTika\Clients\WebClient;
11			use Vaites\ApacheTika\Metadata\Metadata;
12			use Vaites\ApacheTika\Metadata\MetadataInterface;
13
14			/**
15			* Apache Tika client interface
16			*
17			* @author David Martínez <[email protected]>
18			* @link https://tika.apache.org/1.24/formats.html
19			*/
20			abstract class Client
21			{
22			protected const MODE = null;
23
24			/**
25			* Checked flag
26			*
27			* @var bool
28			*/
29			protected $checked = false;
30
31			/**
32			* Response using callbacks
33			*
34			* @var string
35			*/
36			protected $response = null;
37
38			/**
39			* Platform (unix or win)
40			*
41			* @var string
42			*/
43			protected $platform = null;
44
45			/**
46			* Apache Tika version
47			*
48			* @var string
49			*/
50			protected $version = null;
51
52			/**
53			* Cached responses to avoid multiple request for the same file.
54			*
55			* @var array
56			*/
57			protected $cache = [];
58
59			/**
60			* Text encoding
61			*
62			* @var string\|null
63			*/
64			protected $encoding = null;
65
66			/**
67			* Callback called on secuential read
68			*
69			* @var callable\|null
70			*/
71			protected $callback = null;
72
73			/**
74			* Enable or disable appending when using callback
75			*
76			* @var bool
77			*/
78			protected $callbackAppend = true;
79
80			/**
81			* Size of chunks for callback
82			*
83			* @var int
84			*/
85			protected $chunkSize = 1048576;
86
87			/**
88			* Remote download flag
89			*
90			* @var bool
91			*/
92			protected $downloadRemote = false;
93
94			/**
95			* Configure client
96			*/
97			public function __construct()
98			{
99			$this->platform = defined('PHP_WINDOWS_VERSION_MAJOR') ? 'win' : 'unix';
100			}
101
102			/**
103			* Get a class instance throwing an exception if check fails
104			*
105			* @param string\|null $param1 path or host
106			* @param string\|int\|null $param2 Java binary path or port for web client
107			* @param array $options options for cURL request
108			* @param bool $check check JAR file or server connection
109			* @return \Vaites\ApacheTika\Clients\CLIClient\|\Vaites\ApacheTika\Clients\WebClient
110			* @throws \Exception
111			*/
112			public static function make(?string $param1 = null, $param2 = null, array $options = [], bool $check = true): Client
113			{
114			if(preg_match('/\.jar$/', func_get_arg(0)))
115			{
116			$path = $param1 ? (string) $param1 : null;
117			$java = $param2 ? (string) $param2 : null;
118
119			return new CLIClient($path, $java, $check);
120			}
121			else
122			{
123			$host = $param1 ? (string) $param1 : null;
124			$port = $param2 ? (int) $param2 : null;
125
126			return new WebClient($host, $port, $options, $check);
127			}
128			}
129
130			/**
131			* Get a class instance delaying the check
132			*
133			* @param string\|null $param1 path or host
134			* @param int\|null $param2 Java binary path or port for web client
135			* @param array $options options for cURL request
136			* @return \Vaites\ApacheTika\Clients\CLIClient\|\Vaites\ApacheTika\Clients\WebClient
137			* @throws \Exception
138			*/
139			public static function prepare($param1 = null, $param2 = null, $options = []): Client
140			{
141			return self::make($param1, $param2, $options, false);
142			}
143
144			/**
145			* Get the encoding
146			*/
147			public function getEncoding(): ?string
148			{
149			return $this->encoding;
150			}
151
152			/**
153			* Set the encoding
154			*
155			* @throws \Exception
156			*/
157			public function setEncoding(string $encoding): self
158			{
159			if(!empty($encoding))
160			{
161			$this->encoding = $encoding;
162			}
163			else
164			{
165			throw new Exception('Invalid encoding');
166			}
167
168			return $this;
169			}
170
171			/**
172			* Get the callback
173			*/
174			public function getCallback(): ?callable
175			{
176			return $this->callback;
177			}
178
179			/**
180			* Set the callback (callable or closure) for call on secuential read
181			*
182			* @throws \Exception
183			*/
184			public function setCallback(callable $callback, bool $append = true): self
185			{
186			if($callback instanceof Closure \|\| is_array($callback))
187			{
188			$this->callbackAppend = (bool) $append;
189			$this->callback = $callback;
190			}
191			elseif(is_string($callback))
192			{
193			$this->callbackAppend = (bool) $append;
194			$this->callback = function($chunk) use ($callback)
195			{
196			return call_user_func_array($callback, [$chunk]);
197			};
198			}
199			else
200			{
201			throw new Exception('Invalid callback');
202			}
203
204			return $this;
205			}
206
207			/**
208			* Get the chunk size
209			*/
210			public function getChunkSize(): int
211			{
212			return $this->chunkSize;
213			}
214
215			/**
216			* Set the chunk size for secuential read
217			*
218			* @throws \Exception
219			*/
220			public function setChunkSize(int $size): self
221			{
222			if(static::MODE == 'cli')
			0 ignored issues – show introduced 2020-08-28 23:18 UTC by Report Bug Copy Issue Report The condition `static::MODE == 'cli'` is always `false`. Loading history...
223			{
224			$this->chunkSize = $size;
225			}
226			else
227			{
228			throw new Exception('Chunk size is not supported on web mode');
229			}
230
231			return $this;
232			}
233
234			/**
235			* Get the remote download flag
236			*/
237			public function getDownloadRemote(): bool
238			{
239			return $this->downloadRemote;
240			}
241
242			/**
243			* Set the remote download flag
244			*/
245			public function setDownloadRemote(bool $download): self
246			{
247			$this->downloadRemote = (bool) $download;
248
249			return $this;
250			}
251
252			/**
253			* Gets file metadata
254			*
255			* @throws \Exception
256			*/
257			public function getMetadata(string $file): MetadataInterface
258			{
259			$response = $this->parseJsonResponse($this->request('meta', $file) ?: 'ERROR');
260
261			if($response instanceof stdClass === false)
262			{
263			throw new Exception("Unexpected metadata response for $file");
264			}
265
266			return Metadata::make($response, $file);
267			}
268
269			/**
270			* Gets recursive file metadata where the returned array indexes are the file name.
271			*
272			* Example: for a sample.zip with an example.doc file, the return array looks like if be defined as:
273			*
274			* [
275			* 'sample.zip' => new Metadata()
276			* 'sample.zip/example.doc' => new DocumentMetadata()
277			* ]
278			*
279			* @link https://cwiki.apache.org/confluence/display/TIKA/TikaServer#TikaServer-RecursiveMetadataandContent
280			* @throws \Exception
281			*/
282			public function getRecursiveMetadata(string $file, ?string $format = 'ignore'): array
283			{
284			if(in_array($format, ['text', 'html', 'ignore']) === false)
285			{
286			throw new Exception("Unknown recursive type (must be text, html, ignore or null)");
287			}
288
289			$response = $this->parseJsonResponse($this->request("rmeta/$format", $file) ?: 'ERROR');
290
291			if(is_array($response) === false)
292			{
293			throw new Exception("Unexpected metadata response for $file");
294			}
295
296			$metadata = [];
297
298			foreach($response as $item)
299			{
300			$name = basename($file);
301			if(isset($item->{'X-TIKA:embedded_resource_path'}))
302			{
303			$name .= $item->{'X-TIKA:embedded_resource_path'};
304			}
305
306			$metadata[$name] = Metadata::make($item, $file);
307			}
308
309			return $metadata;
310			}
311
312			/**
313			* Detect language
314			*
315			* @throws \Exception
316			*/
317			public function getLanguage(string $file): ?string
318			{
319			return $this->request('lang', $file);
320			}
321
322			/**
323			* Detect MIME type
324			*
325			* @throws \Exception
326			*/
327			public function getMIME(string $file): ?string
328			{
329			return $this->request('mime', $file);
330			}
331
332			/**
333			* Extracts HTML
334			*
335			* @throws \Exception
336			*/
337			public function getHTML(string $file, ?callable $callback = null, bool $append = true): ?string
338			{
339			if(!is_null($callback))
340			{
341			$this->setCallback($callback, $append);
342			}
343
344			return $this->request('html', $file);
345			}
346
347			/**
348			* Extracts XHTML
349			*
350			* @throws \Exception
351			*/
352			public function getXHTML(string $file, ?callable $callback = null, bool $append = true): ?string
353			{
354			if(!is_null($callback))
355			{
356			$this->setCallback($callback, $append);
357			}
358
359			return $this->request('xhtml', $file);
360			}
361
362			/**
363			* Extracts text
364			*
365			* @throws \Exception
366			*/
367			public function getText(string $file, ?callable $callback = null, bool $append = true): ?string
368			{
369			if(!is_null($callback))
370			{
371			$this->setCallback($callback, $append);
372			}
373
374			return $this->request('text', $file);
375			}
376
377			/**
378			* Extracts main text
379			*
380			* @throws \Exception
381			*/
382			public function getMainText(string $file, ?callable $callback = null, bool $append = true): ?string
383			{
384			if(!is_null($callback))
385			{
386			$this->setCallback($callback, $append);
387			}
388
389			return $this->request('text-main', $file);
390			}
391
392			/**
393			* Returns current Tika version
394			*
395			* @throws \Exception
396			*/
397			public function getVersion(): ?string
398			{
399			if(is_null($this->version))
			0 ignored issues – show introduced 2022-05-04 20:16 UTC by Report Bug Copy Issue Report The condition `is_null($this->version)` is always `false`. Loading history...
400			{
401			$this->setVersion($this->request('version'));
402			}
403
404			return $this->version;
405			}
406
407			/**
408			* Set the Tika version
409			*/
410			public function setVersion(string $version): self
411			{
412			$this->checked = true;
413			$this->version = $version;
414
415			return $this;
416			}
417
418			/**
419			* Return the list of Apache Tika supported versions
420			*
421			* @throws \Exception
422			*/
423			public function getSupportedVersions(): array
424			{
425			static $versions = null;
426
427			if(is_null($versions))
428			{
429			$composer = file_get_contents(dirname(__DIR__) . '/composer.json');
430
431			if($composer === false)
432			{
433			throw new Exception("An error ocurred trying to read package's composer.json file");
434			}
435
436			$versions = json_decode($composer, true)['extra']['supported-versions'] ?? null;
437
438			if(empty($versions))
439			{
440			throw new Exception("An error ocurred trying to read package's composer.json file");
441			}
442			}
443
444			return $versions;
445			}
446
447			/**
448			* Sets the checked flag
449			*/
450			public function setChecked(bool $checked): self
451			{
452			$this->checked = (bool) $checked;
453
454			return $this;
455			}
456
457			/**
458			* Checks if instance is checked
459			*/
460			public function isChecked(): bool
461			{
462			return $this->checked;
463			}
464
465			/**
466			* Check if a response is cached
467			*/
468			protected function isCached(string $type, string $file): bool
469			{
470			return isset($this->cache[sha1($file)][$type]);
471			}
472
473			/**
474			* Get a cached response
475			*
476			* @return mixed
477			*/
478			protected function getCachedResponse(string $type, string $file)
479			{
480			return $this->cache[sha1($file)][$type] ?? null;
481			}
482
483			/**
484			* Check if a request type must be cached
485			*/
486			protected function isCacheable(string $type): bool
487			{
488			return in_array($type, ['lang', 'meta']);
489			}
490
491			/**
492			* Caches a response
493			*
494			* @param mixed $response
495			*/
496			protected function cacheResponse(string $type, $response, string $file): bool
497			{
498			$this->cache[sha1($file)][$type] = $response;
499
500			return true;
501			}
502
503			/**
504			* Checks if a specific version is supported
505			*/
506			public function isVersionSupported(string $version): bool
507			{
508			return in_array($version, $this->getSupportedVersions());
509			}
510
511			/**
512			* Check if a mime type is supported
513			*
514			* @param string $mime
515			* @return bool
516			* @throws \Exception
517			*/
518			public function isMIMETypeSupported(string $mime): bool
519			{
520			return array_key_exists($mime, $this->getSupportedMIMETypes());
521			}
522
523			/**
524			* Check the request before executing
525			*
526			* @throws \Exception
527			*/
528			public function checkRequest(string $type, ?string $file = null): ?string
529			{
530			// no checks for getters
531			if(in_array($type, ['detectors', 'mime-types', 'parsers', 'version']))
532			{
533			//
534			}
535			// invalid local file
536			elseif($file !== null && !preg_match('/^http/', $file) && !file_exists($file))
537			{
538			throw new Exception("File $file can't be opened");
539			}
540			// download remote file if required only for integrated downloader
541			elseif($file !== null && preg_match('/^http/', $file) && $this->downloadRemote)
542			{
543			$file = $this->downloadFile($file);
544			}
545			// invalid remote file
546			elseif($file !== null && preg_match('/^http/', $file))
547			{
548			$headers = get_headers($file);
549
550			if(empty($headers) \|\| !preg_match('/200/', $headers[0]))
551			{
552			throw new Exception("File $file can't be opened", 2);
553			}
554			}
555
556			return $file;
557			}
558
559			/**
560			* Filter response to fix common issues
561			*
562			* @param string $response
563			* @return string
564			*/
565			protected function filterResponse(string $response): string
566			{
567			// fix Log4j2 warning
568			$response = trim(str_replace
569			(
570			'WARNING: sun.reflect.Reflection.getCallerClass is not supported. This will impact performance.',
571			'',
572			$response
573			));
574
575			return trim($response);
576			}
577
578			/**
579			* Parse the response returned by Apache Tika
580			*
581			* @return mixed
582			* @throws \Exception
583			*/
584			protected function parseJsonResponse(string $response)
585			{
586			// an empty response throws an error
587			if(empty($response) \|\| trim($response) == '')
588			{
589			throw new Exception('Empty response');
590			}
591
592			// decode the JSON response
593			$json = json_decode($response);
594
595			// exceptions if metadata is not valid
596			if(json_last_error())
597			{
598			$message = function_exists('json_last_error_msg') ? json_last_error_msg() : 'Error parsing JSON response';
599
600			throw new Exception($message, json_last_error());
601			}
602
603			return $json;
604			}
605
606			/**
607			* Download file to a temporary folder and return its path
608			*
609			* @link https://wiki.apache.org/tika/TikaJAXRS#Specifying_a_URL_Instead_of_Putting_Bytes
610			* @throws \Exception
611			*/
612			protected function downloadFile(string $file): string
613			{
614			$dest = tempnam(sys_get_temp_dir(), 'TIKA');
615
616			if($dest === false)
617			{
618			throw new Exception("Can't create a temporary file at " . sys_get_temp_dir());
619			}
620
621			$fp = fopen($dest, 'w+');
622
623			if($fp === false)
624			{
625			throw new Exception("$dest can't be opened");
626			}
627
628			$ch = curl_init($file);
629
630			if($ch === false)
631			{
632			throw new Exception("$file can't be downloaded");
633			}
634
635			curl_setopt($ch, CURLOPT_FILE, $fp);
636			curl_setopt($ch, CURLOPT_TIMEOUT, 5);
637			curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true);
638			curl_exec($ch);
639
640			if(curl_errno($ch))
641			{
642			throw new Exception(curl_error($ch));
643			}
644
645			$code = curl_getinfo($ch, CURLINFO_HTTP_CODE);
646
647			curl_close($ch);
648
649			if($code != 200)
650			{
651			throw new Exception("$file can't be downloaded", $code);
652			}
653
654			return $dest;
655			}
656
657			/**
658			* Must return the supported MIME types
659			*
660			* @throws \Exception
661			*/
662			abstract public function getSupportedMIMETypes(): array;
663
664			/**
665			* Must return the available detectors
666			*
667			* @throws \Exception
668			*/
669			abstract public function getAvailableDetectors(): array;
670
671			/**
672			* Must return the available parsers
673			*
674			* @throws \Exception
675			*/
676			abstract public function getAvailableParsers(): array;
677
678			/**
679			* Check Java binary, JAR path or server connection
680			*/
681			abstract public function check(): void;
682
683			/**
684			* Configure and make a request and return its results.
685			*
686			* @throws \Exception
687			*/
688			abstract public function request(string $type, ?string $file = null): ?string;
689			}
690

vaites / php-apache-tika

Client::getMainText() A last analyzed 2025-03-16 10:38 UTC

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like

Client::getMainText() A
last analyzed 2025-03-16 10:38 UTC