├── README.md
├── functions.php
├── get_keys.php
├── keys.txt
├── scraper.php
└── validate_keys.php


/README.md:
--------------------------------------------------------------------------------
 1 | Works both on Windows and Linux. Tested on PHP 8.1. <br />
 2 | To run:
 3 | ```properties
 4 | php scraper.php
 5 | ```
 6 | If the script is throwing errors, which it does if you are using a VPN, then use slow mode.
 7 | 
 8 | ```properties
 9 | php scraper.php -mode slow
10 | ```
11 | 
12 | The working keys are written in text files with the model name (gpt-4.txt, gpt-3.5-turbo.txt).
13 | 
14 | 


--------------------------------------------------------------------------------
/functions.php:
--------------------------------------------------------------------------------
 1 | <?php
 2 | $models = ['gpt-3.5-turbo', 'gpt-4', 'gpt-4o-mini', 'gpt-4o'];
 3 | 
 4 | function getOs(){
 5 |     if (strpos(php_uname(),'Windows')!==false || strpos(php_uname(),'WINNT')!==false)
 6 |         return true;
 7 |     else
 8 |         return false;
 9 | }
10 | function isScriptsRunning($scriptName){
11 |     if (getOs())
12 |         $output = shell_exec('wmic process where "name like \'php%\'" get processid,commandline');
13 |     else
14 |         $output = shell_exec('ps -C php -f');
15 |     if (strpos($output, $scriptName) !== false)
16 |     {
17 |         sleep(1);
18 |         return true;
19 |     }
20 |     else
21 |         return false;
22 | }
23 | 
24 | function isValidKey($ch){
25 |     if (!curl_errno($ch)) 
26 |     {
27 |         $httpCode = curl_getinfo($ch, CURLINFO_HTTP_CODE);
28 |         switch ($httpCode) {
29 |           case 200:
30 |               return true;
31 |           default: 
32 |               return false;
33 |         }
34 |       }
35 | }
36 | 
37 | function cleanupFile($file){
38 |     $fd = fopen($file, 'w');
39 |     fclose($fd);
40 | }
41 | 
42 | function pleaseWaitAnimation($message){
43 |     $animation = ['|','/','-','\\','\\'];
44 |     foreach ($animation as $char)
45 |     {
46 |         echo $message . "[".$char."]"."\r";
47 |         usleep(20000);
48 |     }
49 | }
50 | 
51 | function isFileEmpty($file){
52 |     if (filesize($file)===0)
53 |         throw new Exception("File is empty");
54 | }
55 | 
56 | function ifSlowModeDelay($mode){
57 |     if ($mode==='slow')
58 |         return usleep(50000);
59 | }
60 | 


--------------------------------------------------------------------------------
/get_keys.php:
--------------------------------------------------------------------------------
 1 | <?php
 2 | $keys = [];
 3 | $pattern = '/(sk-\w)<\/span>([a-zA-Z0-9]{47})/';
 4 | $char = explode('--char=', $argv[1])[1];
 5 | $index = explode('--index=', $argv[2])[1];
 6 | $ch = curl_init();
 7 | curl_setopt($ch, CURLOPT_URL, "https://huggingface.co/search/full-text?q=s'k-$char&limit=100&skip=$index");
 8 | curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
 9 | curl_setopt($ch, CURLOPT_TIMEOUT, 90);
10 | $output = curl_exec($ch);
11 | 
12 | if ($output === false)
13 |     throw new Exception(curl_error($ch));
14 | preg_match_all($pattern, $output, $keys);
15 | $fd = fopen('keys.txt', 'a+');
16 | foreach ($keys[0] as $key) 
17 | {
18 |     $key = str_replace('</span>', '', $key);
19 |     fwrite($fd, $key . PHP_EOL);
20 | }
21 | fclose($fd);
22 | 


--------------------------------------------------------------------------------
/keys.txt:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/gasievt/huggingface-openai-key-scraper/56c30cf0e381c86884f1e2022f7a850f8b616cd9/keys.txt


--------------------------------------------------------------------------------
/scraper.php:
--------------------------------------------------------------------------------
 1 | <?php
 2 | include "./functions.php";
 3 | $mode = 'default';
 4 | $OS = getOs(); //true is windows
 5 | while (count($argv) > 0)
 6 | {
 7 |     $flag = array_shift($argv);
 8 |     switch ($flag)
 9 |     {
10 |         case '-mode':
11 |         {
12 |             if (count($argv)===0)
13 |             {
14 |                 echo "ERROR: No value was provided for flag $flag" . PHP_EOL;
15 |                 die();
16 |             }
17 |             global $mode;
18 |             $arg = array_shift($argv);
19 |             switch ($arg) 
20 |             {
21 |                 case 'slow':
22 |                 {
23 |                    $mode = 'slow';
24 |                    break;
25 |                 }
26 |                 default:
27 |                     echo "ERROR: No such mode: $arg" . PHP_EOL;
28 |                     die();
29 |             }
30 |         }
31 |     }
32 | }
33 | $chars = array_merge(range('a', 'z'), range('0', '9'));
34 | $path = __DIR__;
35 | foreach ($chars as $char)
36 | {
37 |     for ($i=0; $i<=1000; $i+=100)
38 |     {
39 |         echo "Scraping: $char char" . ' ' . "Page № $i" . PHP_EOL;
40 |         $command = '';
41 |         if ($OS)
42 |         {
43 |             $command = "php $path/get_keys.php --char=$char --index=$i > NUL:";
44 |             proc_open($command, [], $pipe);
45 |         }
46 |         else
47 |         {
48 |             $command = "php $path/get_keys.php --char=$char --index=$i > /dev/null &";
49 |             exec($command);
50 |         }
51 |         ifSlowModeDelay($mode);
52 |     }
53 | }
54 | 
55 | do {
56 |     $flag = false;
57 |     if (isScriptsRunning('get_keys.php'))
58 |         pleaseWaitAnimation('Please wait');
59 |     else
60 |     {
61 |         echo PHP_EOL .'Done!' . PHP_EOL;
62 |         $flag = true;
63 |     }
64 | }
65 | while (!$flag);
66 | 
67 | isFileEmpty('keys.txt');
68 | $uniqueKeys = array_unique(explode(PHP_EOL, file_get_contents('keys.txt')));
69 | cleanupFile('keys.txt');
70 | 
71 | foreach ($uniqueKeys as $key)
72 | {
73 |     echo "Validating $key" . PHP_EOL;
74 |     $command = '';
75 |     if ($OS)
76 |     {
77 |         $command = "php $path/validate_keys.php --key=$key > NUL:";
78 |         proc_open($command, [], $pipe);
79 |     }
80 |     else
81 |     {
82 |         $command = "php $path/validate_keys.php --key=$key > /dev/null &";
83 |         exec($command);
84 |     }
85 |     ifSlowModeDelay($mode);
86 | }
87 | 
88 | do {
89 |     $flag = false;
90 |     if (isScriptsRunning('validate_keys.php'))
91 |         pleaseWaitAnimation("Validating keys");
92 |     else
93 |     {
94 |         echo PHP_EOL . 'Done!' . PHP_EOL;
95 |         $flag = true;
96 |     }
97 | }
98 | while (!$flag);
99 | 


--------------------------------------------------------------------------------
/validate_keys.php:
--------------------------------------------------------------------------------
 1 | <?php
 2 |     include "./functions.php";
 3 |     $key = explode('--key=', $argv[1])[1];
 4 |     $ch = curl_init();
 5 |     curl_setopt($ch, CURLOPT_URL, "https://api.openai.com/v1/chat/completions");
 6 |     curl_setopt($ch, CURLOPT_POST, 1);
 7 |     curl_setopt($ch, CURLOPT_TIMEOUT, 90);
 8 |     $headers = [
 9 |         "Content-Type: application/json",
10 |         "Authorization: Bearer $key"
11 |     ];
12 |     curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);
13 |     foreach ($models as $model)
14 |     {
15 |         curl_setopt($ch, CURLOPT_POSTFIELDS,
16 |             json_encode(['model'=>$model,
17 |              'messages'=>[['role'=>'user', 'content'=>'Hi']],
18 |              'temperature'=>0.7]));
19 |         if (curl_exec($ch)===false)
20 |             throw new Exception(curl_error($ch));   
21 |         if (isValidKey($ch))
22 |         {
23 |             $fd = fopen("$model.txt", 'a+');
24 |             fwrite($fd, $key . PHP_EOL);
25 |             fclose($fd);
26 |         }   
27 |     }


--------------------------------------------------------------------------------