A simple php cli tool to traverse movie directories and download the missing subtitles from subscene.com (suited for movies from yify torrents)

sub.php

#!/usr/bin/env php
<?php

/*
 * A simple php cli tool to traverse movie directories
 * and download the missing subtitles from subscene.com
 * (Especially suited for movies from yify torrents)
 *
 * Usage: php sub.php /path/to/movies/
 *
 * @author Jitendra Adhikari <jiten.adhikary@gmail.com>
 */

/* START user config */
// The path to cache metadata and  subtitles for later use
$cach = 'D:/share/subtitles/';
/* END user config */

$isCli = 'cli' == php_sapi_name() or defined('STDIN');
if (!$isCli) {
    exit('To be run from Command Line Terminal only.');
}
if ($argc < 2) {
    $syntax = '<fullpath:string> <force:bool> <caching:bool>';
    exit(
"Usage 1:
    php sub.php {$syntax}
Usage 2: (with sub.bat in path)
    sub {$syntax}
");
}

// Pad
$argv += ['', '', 0, 0];

$path = $argv[1];
$subt = 'http://subscene.com';
$temp = $cach.'temp.zip';
$opts = compact('subt', 'cach', 'temp');
$vids = [
    'mp4', 'avi', 'mkv', 'm4v', 'flv',
    'webm', 'mov', 'wmv', 'mpg', 'mpeg',
];
$proc = [];
if (is_file($cach.'meta.json')) {
    $proc = json_decode(file_get_contents($cach.'meta.json'), true);
}

function saveMeta($proc, $cach)
{
    static $saved = false;
    if ($saved) {
        return;
    }
    file_put_contents(
        $cach.'meta.json',
        json_encode(array_filter($proc), JSON_PRETTY_PRINT)
    );
    $saved = true;
}

function saveSrt($url, $opts)
{
    $saved = false;
    extract($opts, EXTR_OVERWRITE);
    if (preg_match(
        '~<a href="(/subtitle/download[^"]+)"~s',
        @file_get_contents($subt.$url),
        $link
    )) {
        file_put_contents(
            $temp,
            @file_get_contents($subt.$link[1])
        );
        $zip = new ZipArchive();
        if ($zip->open($temp) === true) {
            $fn = $zip->getNameIndex(0);
            if ($fn and
                copy("zip://{$temp}#{$fn}", $cacheFile) and
                copy($cacheFile, $srtFile)
            ) {
                $saved = true;
                echo ' :: SAVED';
            }
            $zip->close();
        }
    }

    return $saved;
}

echo PHP_EOL;
$count = $eCount = $sCount = $oCount = $cCount = 0;
foreach (new RecursiveIteratorIterator(
    new RecursiveDirectoryIterator($path),
    RecursiveIteratorIterator::SELF_FIRST
) as $key => $file
) {
    $srtFile = str_replace('.'.$file->getExtension(), '.srt', $file);
    if ($file->isFile() and
        in_array($file->getExtension(), $vids)
    ) {
        $film = basename(dirname($file));
        if (strlen($film) < 5) {
            continue;
        }
        ++$count;
        $cacheFile = $cach.basename($srtFile);
        echo str_pad(substr($film, 0, 69).' ', 70, '-');

        if (is_file($srtFile)) {
            ++$oCount;
            if (!is_file($cacheFile) and !is_file($cach.$film)) {
                copy($srtFile, $cacheFile);
                echo ' :: CACHE', PHP_EOL;
            } else {
                echo ' :: OK', PHP_EOL;
            }
            continue;
        }
        if ($argv[3] == 1 or (isset($proc[$film]) and $argv[2] != 1)) {
            ++$sCount;
            echo ' :: SKIP', PHP_EOL;
            continue;
        }

        $proc[$film] = 0;
        preg_match('`\(\d{4}\)`', $film, $year);
        $year = trim(reset($year), '()');
        $name = implode('-', array_map(function ($p) {
            return preg_replace('`[\W]+`', '', trim(strtolower($p)));
        }, preg_split('/\s+/', strstr($film, ' (', true))));

        if (($a = is_file($cacheFile)) or is_file($cach.$film)) {
            if (!$a) {
                $cacheFile = $cach.$film;
            }
            if (copy($cacheFile, $srtFile)) {
                ++$oCount;
                echo ' :: SAVED';
            }
            echo PHP_EOL;
            continue;
        }
        $html2 = '';
        $saved = false;
        $html = @file_get_contents("http://subscene.com/subtitles/{$name}/english");
        $yify = '~<a href="([^"]+)">\s+<span.*?English.*?</span>\s+<span>\s+.*?YIFY\s+</span>\s+</a>~ms';
        $othr = '~<a href="([^"]+)">\s+<span.*?English.*?</span>\s+<span>\s+.*?\s+</span>\s+</a>~ms';
        $opts = array_merge($opts, compact('srtFile', 'cacheFile'));
        // Four priority order:
        // #1. name/yifi
        if (preg_match($yify, $html, $ancr)) {
            $saved = saveSrt($subt.$ancr[1], $opts);
        }
        // #2. name-year/yifi
        if (!$saved and $year) {
            $html2 = @file_get_contents("http://subscene.com/subtitles/{$name}-{$year}/english");
            if (preg_match($yify, $html2, $ancr)) {
                $saved = saveSrt($ancr[1], $opts);
            }
        }
        // #3. name/othr
        if (!$saved and preg_match($othr, $html, $ancr)) {
            $saved = saveSrt($ancr[1], $opts);
        }
        // #4. name-year/othr
        if (!$saved and $year) {
            if (!$html2) {
                $html2 = @file_get_contents("http://subscene.com/subtitles/{$name}-{$year}/english");
            }
            if (preg_match($othr, $html2, $ancr)) {
                $saved = saveSrt($ancr[1], $opts);
            }
        }

        if (!$saved) {
            ++$eCount;
            $proc[$film] = 1;
            echo ' :: ERROR';
        } else {
            ++$cCount;
        }

        echo PHP_EOL;
    }
}

ksort($proc);
saveMeta($proc, $cach);
echo "\nOperation Completed Normally.
    Total    : {$count}
    Okay     : {$oCount}
    Download : {$cCount}
    Skip     : {$sCount}
    Error    : {$eCount}
";