Comarer 2 files

Compare.php

<?php

namespace Compare;
/**
 * нужно написать программу которая сравнивает два текстовых файла (терабайтных)
 * и выводит в третий файл строки, которые есть в первом, но нет во втором.
 *
 * To run this utility just run  `php ./Compare.php input_file1.txt input_file2.txt output_file.txt`
 */
class Compare
{

    /**
     * @var int Input file reading in bytes
     */
    private $readLength = 1048576; // 1024 * 1024

    /**
     * @var resource first input file handler
     */
    private $inputFile1;
    /**
     * @var resource second input file handler
     */
    private $inputFile2;
    /**
     * @var resource output file handler
     */
    private $outputFile;

    /**
     * Simple construct and open
     *
     * @param $inputFile1
     * @param $inputFile2
     * @param $outputFile
     * @throws \Exception
     */
    public function __construct($inputFile1, $inputFile2, $outputFile) // here you can add configs for redefine params
    {
        if (!$this->inputFile1 = @fopen($inputFile1, 'r')) {
            throw new \Exception('Can\'t open first input file');
        }
        if (!$this->inputFile2 = @fopen($inputFile2, 'r')) {
            throw new \Exception('Can\'t open second input file');
        }
        if (!$this->outputFile = @fopen($outputFile, 'a')) {
            throw new \Exception('Can\'t open output file');
        }
        ftruncate($this->outputFile, 0);
        rewind($this->outputFile);
    }

    /**
     * Close open resources
     */
    public function __destruct()
    {
        fclose($this->inputFile1);
        fclose($this->inputFile2);
        fclose($this->outputFile);
    }

    /**
     * Main runner
     */
    public function run()
    {
        $fpos = 0;
        while (!feof($this->inputFile1)) {
            $this->existString($fpos) ?
                $this->skipString($fpos) :
                $this->writeString($fpos);
            $fpos = ftell($this->inputFile1);
        }
    }

    /**
     * Function to find string (start in $pos) from first file in second
     *
     * @param $pos
     * @return bool
     */
    public function existString($pos)
    {
        $equalPart = true;
        fseek($this->inputFile1, $pos);
        rewind($this->inputFile2);

        while (!feof($this->inputFile2)) { // read every line of file 2
            $rightBlock = fgets($this->inputFile2, $this->readLength);

            //if string already not equal we need to finish string in file 2
            if (!$equalPart) {
                if ($this->eol($rightBlock) || feof($this->inputFile2)) {
                    fseek($this->inputFile1, $pos);
                    $equalPart = true; // we want to believe that next iteration make true
                }
                continue;
            }

            $leftBlock = fgets($this->inputFile1, $this->readLength);

            if ($leftBlock !== $rightBlock) {
                if ($this->eol($rightBlock) || feof($this->inputFile2)) { //second file string end too early
                    fseek($this->inputFile1, $pos);
                    $equalPart = true;
                } else {
                    $equalPart = false;
                }
            } elseif ($this->eol($leftBlock) || feof($this->inputFile1)) {
                return true;
            }
        }

        return false; //So we have no string in second file at all

    }

    /**
     * Write string from input file
     *
     * @param $pos String position for write
     */
    public function writeString($pos)
    {
        fseek($this->inputFile1, $pos);
        while (!feof($this->inputFile1)) {

            $block = fgets($this->inputFile1, $this->readLength);
            fwrite($this->outputFile, $block);

            if ($this->eol($block) || feof($this->inputFile1)) {
                break;
            }
        }
    }

    /**
     * Skip string from input file
     *
     * @param $pos String position for write
     */
    public function skipString($pos)
    {
        fseek($this->inputFile1, $pos);
        while (!feof($this->inputFile1)) {
            $block = fgets($this->inputFile1, $this->readLength);
            if ($this->eol($block) || feof($this->inputFile1)) {
                break;
            }
        }
    }


    /**
     * Check if block has end of line
     *
     * @param $block
     * @return bool
     */
    public function eol($block)
    {
        if (preg_match('/(\n|\r|\r\n)$/', $block)) { //End Of Line - last block in string
            return true;
        } else {
            return false;
        }
    }


}

//-------------------------------- <<< PROGRAM ------------------------------------
if (empty($argv[1])) {
    die("Please show me first input file\n");
} else {
    $input1 = $argv[1];
}
if (empty($argv[2])) {
    die("Please show me second input file\n");
} else {
    $input2 = $argv[2];
}

if (empty($argv[3])) {
    die("Please show me output file\n");
} else {
    $output = $argv[3];
}


try {
    $comparer = new Compare($input1, $input2, $output);
    $comparer->run();
} catch (\Exception $e) { //use native Exceptions
    echo "Error: {$e->getMessage()}\n";
}
//-------------------------------- >>> PROGRAM ------------------------------------