/**
* Configure and make a request and return its results.
*
* @param string $file
* @param string $type
*
* @return string
*
* @throws \Exception
*/
protected function request($file, $type)
{
// check if is cached
if (isset($this->cache[sha1($file)][$type])) {
return $this->cache[sha1($file)][$type];
}
// parameters for cURL request
$headers = [];
switch ($type) {
case 'html':
$resource = 'tika';
$headers[] = 'Accept: text/html';
break;
case 'mime':
$name = basename($file);
$resource = 'detect/stream';
$headers[] = "Content-Disposition: attachment, filename={$name}";
break;
case 'lang':
$resource = 'language/stream';
break;
case 'meta':
$resource = 'meta';
$headers[] = 'Accept: application/json';
break;
case 'text':
$resource = 'tika';
$headers[] = 'Accept: text/plain';
break;
case 'version':
$resource = 'version';
break;
default:
throw new Exception("Unknown type {$type}");
}
// cURL base options
$options = [CURLOPT_PUT => true];
// remote file options
if ($file && preg_match('/^http/', $file)) {
$options[CURLOPT_INFILE] = fopen($file, 'r');
} elseif ($file && file_exists($file) && is_readable($file)) {
$options[CURLOPT_INFILE] = fopen($file, 'r');
$options[CURLOPT_INFILESIZE] = filesize($file);
} elseif ($type == 'version') {
$options = [CURLOPT_PUT => false];
} else {
throw new Exception("File {$file} can't be opened");
}
// sets headers
$options[CURLOPT_HTTPHEADER] = $headers;
// cURL init and options
$options[CURLOPT_URL] = "http://{$this->host}:{$this->port}" . "/{$resource}";
// get the response and the HTTP status code
list($response, $status) = $this->exec($options);
switch ($status) {
// request completed successfully
case 200:
if ($type == 'meta') {
$response = Metadata::make($response, $file);
}
break;
// request completed sucessfully but result is empty
// request completed sucessfully but result is empty
case 204:
$response = null;
break;
// unsupported media type
// unsupported media type
case 415:
throw new Exception('Unsupported media type');
break;
// unprocessable entity
// unprocessable entity
case 422:
throw new Exception('Unprocessable document');
break;
// server error
// server error
case 500:
throw new Exception('Error while processing document');
break;
// unexpected
// unexpected
default:
throw new Exception("Unexpected response for /{$resource} ({$status})");
}
// cache certain responses
if (in_array($type, ['lang', 'meta'])) {
$this->cache[sha1($file)][$type] = $response;
}
return $response;
}
/**
* Configure and make a request and return its results.
*
* @param string $file
* @param string $type
*
* @return string
*
* @throws \Exception
*/
protected function request($file, $type)
{
// check if is cached
if (isset($this->cache[sha1($file)][$type])) {
return $this->cache[sha1($file)][$type];
}
// parameters for cURL request
$arguments = [];
switch ($type) {
case 'html':
$arguments[] = '--html';
break;
case 'mime':
$arguments[] = '--detect';
break;
case 'lang':
$arguments[] = '--language';
break;
case 'meta':
$arguments[] = '--metadata --json';
break;
case 'text':
$arguments[] = '--text';
break;
case 'version':
$arguments[] = '--version';
break;
default:
throw new Exception("Unknown type {$type}");
}
// invalid file
if ($file && !preg_match('/^http/', $file) && !file_exists($file)) {
throw new Exception("File {$file} can't be opened");
}
// add last argument
if ($file) {
$arguments[] = "'{$file}'";
}
// build command
$command = "java -jar '{$this->path}' " . implode(' ', $arguments);
// run command and process output
$response = trim(shell_exec($command));
// metadata response
if ($type == 'meta') {
// fix for invalid? json returned only with images
$response = str_replace(basename($file) . '"}{', '", ', $response);
$response = Metadata::make($response, $file);
}
// cache certain responses
if (in_array($type, ['lang', 'meta'])) {
$this->cache[sha1($file)][$type] = $response;
}
return $response;
}
/**
* Configure and make a request and return its results.
*
* @param string $type
* @param string $file
*
* @return string
*
* @throws \Exception
*/
public function request($type, $file = null)
{
// check if is cached
if (isset($this->cache[sha1($file)][$type])) {
return $this->cache[sha1($file)][$type];
}
// parameters for command
$arguments = [];
switch ($type) {
case 'html':
$arguments[] = '--html';
break;
case 'lang':
$arguments[] = '--language';
break;
case 'mime':
$arguments[] = '--detect';
break;
case 'meta':
$arguments[] = '--metadata --json';
break;
case 'text':
$arguments[] = '--text';
break;
case 'version':
$arguments[] = '--version';
break;
default:
throw new Exception("Unknown type {$type}");
}
// invalid local file
if ($file && !preg_match('/^http/', $file) && !file_exists($file)) {
throw new Exception("File {$file} can't be opened");
} elseif ($file && preg_match('/^http/', $file) && !preg_match('/200/', @get_headers($file)[0])) {
throw new Exception("File {$file} can't be opened", 2);
}
// add last argument
if ($file) {
$arguments[] = "'{$file}'";
}
// build command
$command = ($this->java ?: 'java') . " -jar '{$this->path}' " . implode(' ', $arguments);
// run command
$exit = -1;
$response = null;
$descriptors = [['pipe', 'r'], ['pipe', 'w'], ['file', '/tmp/tika-error.log', 'a']];
$process = proc_open($command, $descriptors, $pipes);
// get output if command runs ok
if (is_resource($process)) {
fclose($pipes[0]);
$response = trim(stream_get_contents($pipes[1]));
fclose($pipes[1]);
$exit = proc_close($process);
} else {
throw new Exception("Error running command {$command}");
}
// exception if exit value is not zero
if ($exit > 0) {
throw new Exception("Unexpected exit value ({$exit}) for command {$command}");
}
// metadata response
if ($type == 'meta') {
// fix for invalid? json returned only with images
$response = str_replace(basename($file) . '"}{', '", ', $response);
$response = Metadata::make($response, $file);
}
// cache certain responses
if (in_array($type, ['lang', 'meta'])) {
$this->cache[sha1($file)][$type] = $response;
}
return $response;
}
/**
* 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 $retries = [];
// check if is cached
if (isset($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
$headers = [];
switch ($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 'version':
$resource = 'version';
break;
default:
throw new Exception("Unknown type {$type}");
}
// base options
$options = $this->options;
// remote file options
if ($file && preg_match('/^http/', $file)) {
$options[CURLOPT_INFILE] = fopen($file, 'r');
} elseif ($file && file_exists($file) && is_readable($file)) {
$options[CURLOPT_INFILE] = fopen($file, 'r');
$options[CURLOPT_INFILESIZE] = filesize($file);
} elseif ($type == 'version') {
$options[CURLOPT_PUT] = false;
} else {
throw new Exception("File {$file} can't be opened");
}
// sets headers
$options[CURLOPT_HTTPHEADER] = $headers;
// cURL init and options
$options[CURLOPT_URL] = "http://{$this->host}:{$this->port}" . "/{$resource}";
// get the response and the HTTP status code
list($response, $status) = $this->exec($options);
// request completed successfully
if ($status == 200) {
if ($type == 'meta') {
$response = Metadata::make($response, $file);
}
} elseif ($status == 204) {
$response = null;
} elseif ($status == 500 && $retries[sha1($file)]--) {
usleep(100000);
$response = $this->request($type, $file);
} else {
$this->error($status, $resource);
}
// cache certain responses
if (in_array($type, ['lang', 'meta'])) {
$this->cache[sha1($file)][$type] = $response;
}
return $response;
}