CanRunCommand.php

<?php
/**
 * Created by PhpStorm.
 * User: hocvt
 * Date: 6/6/17
 * Time: 15:01
 */

namespace App\Colombo;


use Symfony\Component\Process\Process;

abstract class CanRunCommand {
	
	protected $process;
	protected $process_options = [];
	private $command;
	protected $bin;
	protected $timeout = 300;
	
	/**
	 * CanRunCommand constructor.
	 *
	 * @param string $bin
	 */
	public function __construct($bin = '') {
		$this->bin($bin);
		$this->process = new Process('');
		$this->process->setTimeout($this->timeout);
	}
	
	
	protected function validateRun()
	{
		$status = $this->process->getExitCode();
		$error  = $this->process->getErrorOutput();
		
		if ($status !== 0 and $error !== '') {
			throw new \RuntimeException(
				sprintf(
					"The exit status code %s says something went wrong:\n stderr: %s\n stdout: %s\ncommand: %s.",
					$status,
					$error,
					$this->process->getOutput(),
					$this->command
				)
			);
		}
	}
	protected function buildCommand($append = ''){
		$command = $this->bin;
		$command .= " " . $this->buildOptions();
		return $command . $append;
	}
	
	protected function buildOptions(){
		$options = ' ';
		foreach($this->process_options as $k => $v){
			if($v !== false){
				$options .= $k . " ";
			}else{
				continue;
			}
			if($v !== true){
				$options .= $v . " ";
			}
		}
		
		return $options;
	}
	
	public function run($command)
	{
		$this->command = escapeshellcmd($command);
		$this->process->setCommandLine($this->command);
		$this->process->run();
		$this->validateRun();
		
		return $this;
	}
	public function bin($bin = ''){
		if(!empty($bin)){
			$this->bin = $bin;
		}
		return $this->bin;
	}
	public function timeout($timeout = ''){
		if(!empty($timeout)){
			$this->timeout = $timeout;
		}
		return $this->timeout;
	}
	public function options($key = null, $value = null){
		if(is_array($key)){
			if($value === true){
				$this->process_options = $key;
			}else{
				$this->process_options = array_merge($this->process_options, $key);
			}
		}elseif ($key != null){
			if($value !== null){
				$this->process_options[$key] = $value;
			}
			return $this->process_options;
		}
		return $this->process_options;
	}
	public function output()
	{
		return $this->process->getOutput();
	}
	
	
}

PdfInfo.php

<?php
/**
 * Created by PhpStorm.
 * User: hocvt
 * Date: 6/7/17
 * Time: 09:40
 */

namespace App\Colombo\Converter;


use App\Colombo\CanRunCommand;
use Carbon\Carbon;

class PdfInfo extends CanRunCommand {
	
	/**
	 * PdfInfo constructor.
	 *
	 * @param string $bin
	 */
	public function __construct($bin = 'pdfinfo') {
		parent::__construct($bin);
	}
	
	public function read($path){
		$command = $this->buildCommand($path);
		$this->run($command);
		return $this->parseResult();
	}
	
	private function parseResult(){
		$output = $this->output();
		$lines = preg_split("/\n/", $output);
		$output = [];
		foreach ($lines as $line){
			$exploded = preg_split("/\:\s+/", $line);
			if(count($exploded) == 2){
				if(stripos($exploded[0], 'date')){
					$value = Carbon::parse($exploded[1]);
				}else{
					$value = $exploded[1];
				}
				$key = mb_strtolower(str_replace(" ", "_", $exploded[0]));
				$output[$key] = $value;
			}
		}
		
		return $output;
	}
	
}

PdftotextConverter.php

<?php
/**
 * Created by PhpStorm.
 * User: hocvt
 * Date: 6/6/17
 * Time: 14:45
 */

namespace App\Colombo\Converter;


use App\Colombo\CanRunCommand;

class PdftotextConverter extends CanRunCommand {
	
	private $tmp_folder;
	protected $process;
	protected $process_options = [
		'-nodrm' => true,
		'-zoom' => 1,
		'-i' => true,
		'-c' => true,
		'-l' => 2,
	];
	
	public function __construct($bin = 'pdfinfo', $cache = '') {
		$this->tmp_folder = storage_path('framework/pdftotext');
		parent::__construct($bin);
	}
	
	private function checkWritable(){
		if(!file_exists($this->tmp_folder)){
			@mkdir($this->tmp_folder);
		}
		$this->tmp_folder = tempnam($this->tmp_folder, 'pdftohtml_');
		@unlink($this->tmp_folder);
		@mkdir($this->tmp_folder);
		if(!file_exists($this->tmp_folder)){
			throw new \Exception("Can not read/write tmp folder");
		}
	}
	
	/**
	 * Main feature
	 *
	 * @param $path
	 * @param int $page default convert 2 pages
	 *
	 * @return array
	 */
	public function getHtmlPages($path, $page = 3){
		$this->checkWritable();
		$this->options('-l', $page);
		$tmp_path = $this->tmp_folder . DIRECTORY_SEPARATOR . "source.pdf";
		copy($path, $tmp_path);
		$command = $this->buildCommand($tmp_path);
		$this->run($command);
		$pages = [];
		for($i = 1; $i <= $page; $i++){
			$page_path = $this->tmp_folder . DIRECTORY_SEPARATOR . "source-" . $i . ".html";
			if(file_exists($page_path)){
				$pages[$i] = file_get_contents($page_path);
			}else{
				break;
			}
		}
		return $pages;
	}
	
	/** Helper functions */
	
	/**
	 * Delete cache file before this time
	 *
	 * @param PdftotextConverter $instant
	 * @param $time
	 */
	public static function clean(PdftotextConverter $instant = null, $time = null){
		$instant = $instant == null ? new self() : $instant;
		$folder = glob($instant->tmp_folder . DIRECTORY_SEPARATOR . "*");
		foreach ($folder as $file){
			@unlink($file);
		}
		@rmdir($instant->tmp_folder);
	}
	
	function __destruct() {
		self::clean($this);
	}
	
	
}