Script PHP Cek Double Entry Dari Setiap Line Dalam 1 File

  1<?php
  2/**
  3 * This program used to find duplicate entry for each line of text from given
  4 * file. For example to check list of email, proxy, username:password,
  5 * configurations, etc.
  6 * 
  7 * This is not WebBased PHP program, so use it from CLI.
  8 * Depend on your machine, you might need to reset your PHP memory_limit to -1
  9 * for checking large file.
 10 * 
 11 * Still very early release, hasn't been tested on Windows and just for fun
 12 * coding purpose :)
 13 * 
 14 * coded by ditatompel
 15 * Usage : new findDuplicates($argv);
 16 */
 17
 18class FindDuplicates
 19{
 20    var $cmd = array();
 21    var $reqFunc = array('file_get_contents', 'fopen', 'fwrite');
 22    
 23    function __construct ($opts) {
 24        $this->checkReqFunc();
 25        
 26        $this->cmd = $this->args($opts);
 27        if ( !array_key_exists('list', $this->cmd) )
 28            exit( $this->usage($this->cmd['input'][0]) );
 29        $this->enumValues($this->cmd['list']);
 30    }
 31    
 32    function checkReqFunc() {
 33        $ok = 1;
 34        foreach( $this->reqFunc as $func ) {
 35            if ( !function_exists($func) ) {
 36                print 'You need function ' . $func . " to execute this tool!\n";
 37                $ok = 0;
 38            }
 39        }
 40        if(php_sapi_name() != 'cli') {
 41            print "<strong>This is not WebBased program! Use this tool from CLI.</strong>";
 42            $ok = 0;
 43        }
 44        if ( !$ok ) exit;
 45    }
 46    
 47    /**
 48     * Check list file
 49     * @return bool. TRUE if file exists and readable.
 50     */
 51    function cekFile ($file) {
 52        if ( !is_file($file) || !is_readable($file) )
 53            return false;
 54        return true;
 55    }
 56    
 57    private function usage($file) {
 58        $msg = "Usage : ";
 59        $msg .= $file . " --list=[list]\n";
 60        $msg .= "\tOptions : \n";
 61        $msg .= "\t\t-v :\t\t verbose mode.\n";
 62        $msg .= "\t\t--out=[file] :\t Save non duplicate entry to file.\n";
 63        $msg .= "Example :\n";
 64        $msg .= $file . " --list=/path/to/file.txt --out=/path/to/file.txt.new -v\n";
 65        return $msg;
 66    }
 67    
 68    private function enumValues($filename) {
 69        if ( !$this->cekFile($filename) )
 70            exit("File is not readable!\n");
 71        if ( array_key_exists("out", $this->cmd) ) {
 72            $this->checkWrite($this->cmd['out']);
 73            $fh = fopen($this->cmd['out'], 'w');
 74        }
 75        $time_start = microtime(1);
 76        print "File : " . $filename . " (" . $this->format_bytes(filesize($filename)) . ")\n";
 77        $fGetContents = file_get_contents($filename);
 78        $content = explode("\n", $fGetContents);
 79        $totalLines = count($content);
 80        print "Total : " . $this->plural($totalLines,'line') . "\n";
 81        $totalDuplicate = 0;
 82        print "Please be patient, process will be a little longer for large file sizes...\n";
 83        $unique = array_unique($content);
 84        if( $totalLines > count($unique) ) {
 85            for($i = 0; $i < $totalLines; $i++) {
 86                $percentage = (int)number_format($i/$totalLines, 2, '', '');
 87                $prog = "Progress : " . $this->ID_nummeric($i) . "/" . $this->ID_nummeric($totalLines) . " " . $percentage . "%";
 88                echo $prog;
 89                echo "\033[" . strlen($prog) . "D";
 90                if(!array_key_exists($i, $unique)) {
 91                    if ( array_key_exists("v", $this->cmd) )
 92                        print "Duplicate entry on line " . ($i+1) . ": " . $content[$i] . "\n";
 93                    $totalDuplicate++;
 94                }
 95                else {
 96                    if ( array_key_exists("out", $this->cmd) ) {
 97                        if ( !is_writable($this->cmd['out']) )
 98                            exit("[Error] " . $this->cmd['out'] . " is not writable.\n");
 99                        fwrite($fh, $content[$i] . "\n");
100                    }
101                }
102            }
103        }
104        print "Total duplicate entry : " . $this->plural($totalDuplicate)  . "\n";
105        if ( array_key_exists("out", $this->cmd) )
106            print "New list saved to: " . $this->cmd['out'] . " (" . $this->format_bytes(filesize($this->cmd['out'])) . ")\n";
107        $time_end = microtime(1);
108        $execution_time = ($time_end - $time_start);
109        print 'process complete in ' . $execution_time . " sec\n";
110    }
111
112    private function ID_nummeric ($number) {
113        return number_format($number, 0, ',','.');
114    }
115    
116    private function checkWrite($location) {
117        if ( is_dir($location) )
118            exit("[Error] " . $location . " is a directory.\n");
119    }
120    
121    private function plural($number, $txt='item') {
122        if ( $number > 1 ) 
123            return $this->ID_nummeric($number) . " "  . $txt . "s";
124        return $this->ID_nummeric($number) . " "  . $txt;
125    }
126    
127    // yatsynych at gmail dot com
128    function format_bytes($a_bytes) {
129        if ($a_bytes < 1024) {
130            return $a_bytes .' B';
131        } elseif ($a_bytes < 1048576) {
132            return round($a_bytes / 1024, 2) .' KiB';
133        } elseif ($a_bytes < 1073741824) {
134            return round($a_bytes / 1048576, 2) . ' MiB';
135        } elseif ($a_bytes < 1099511627776) {
136            return round($a_bytes / 1073741824, 2) . ' GiB';
137        } elseif ($a_bytes < 1125899906842624) {
138            return round($a_bytes / 1099511627776, 2) .' TiB';
139        } elseif ($a_bytes < 1152921504606846976) {
140            return round($a_bytes / 1125899906842624, 2) .' PiB';
141        } elseif ($a_bytes < 1180591620717411303424) {
142            return round($a_bytes / 1152921504606846976, 2) .' EiB';
143        } elseif ($a_bytes < 1208925819614629174706176) {
144            return round($a_bytes / 1180591620717411303424, 2) .' ZiB';
145        } else {
146            return round($a_bytes / 1208925819614629174706176, 2) .' YiB';
147        }
148    }
149    
150    /**
151     * Make UNIX like parameter command.
152     * This function from losbrutos and modified by earomero. Thankyou. =)
153     * @author losbrutos <[email protected]>
154     * @author earomero <[email protected]>
155     * @param array argv
156     * @return array
157     */
158    private function args($argv) {
159        $_ARG = array();
160        foreach ($argv as $arg) {
161            if (preg_match('#^-{1,2}([a-zA-Z0-9]*)=?(.*)$#', $arg, $matches)) {
162                $key = $matches[1];
163                switch ($matches[2]) {
164                    case '':
165                    case 'true':
166                    $arg = true;
167                    break;
168                    case 'false':
169                    $arg = false;
170                    break;
171                    default:
172                    $arg = $matches[2];
173                }
174                
175                // make unix like -afd == -a -f -d
176                if(preg_match("/^-([a-zA-Z0-9]+)/", $matches[0], $match)) {
177                    $string = $match[1];
178                    for($i=0; strlen($string) > $i; $i++) {
179                        $_ARG[$string[$i]] = true;
180                    }
181                } else {
182                    $_ARG[$key] = $arg;
183                }
184            } else {
185                $_ARG['input'][] = $arg;
186            }
187        }
188        return $_ARG;
189    }
190}
191?>

1$cek = new FindDuplicates ($argv); 

1php [nama_script_phpnya] --list=[nama_file_yang_di_cek]

1php class.FindDuplicates.php --list=openixp.rsc --out=openixp.rsc.new -v