xHP6RUth

1
2
3# Usage: Run python captcha.py
4# Output: 
5# Requires Python 3
6
7
8# anti_captcha_secret_key.txt should hold the 32 hexadecimal character secret key from anti-captcha.com
9# NOTE: Will use 1 captcha (about 0.22 cents) per run
10
11CAPTCHA_CLIENT_KEY = None
12with open('anti_captcha_secret_key.txt', 'r', encoding='utf-8') as f:
13    CAPTCHA_CLIENT_KEY = f.read().strip()
14    if len(CAPTCHA_CLIENT_KEY) != 32:
15        raise RuntimeError('CAPTCHA_CLIENT_KEY has wrong length: %d!!!' % len(CAPTCHA_CLIENT_KEY))
16        
17
18import http.client
19import json
20import time
21import re
22from urllib.parse import urlparse, parse_qs, urlencode
23
24
25def CreateTaskForCaptcha(url, siteKey):
26    conn = http.client.HTTPSConnection('api.anti-captcha.com', timeout=30.0)
27
28    postData = {}
29    postData['clientKey'] = CAPTCHA_CLIENT_KEY
30    
31    taskObj = {}
32    taskObj['type'] = "NoCaptchaTaskProxyless"
33    taskObj['websiteURL'] = url
34    taskObj['websiteKey'] = siteKey
35    postData['task'] = taskObj
36    
37    
38    body = json.dumps(postData).encode('utf-8')
39    
40    print(body)
41    conn.request('POST', '/createTask', body=body)
42
43    res = conn.getresponse()
44    
45    response = json.loads(res.read().decode('utf-8', 'ignore'))
46
47    conn.close()
48
49    print(response)
50
51    if 'error' in response:
52        raise RuntimeError('Anti-captcha API error: "%s"' % response["error"])
53
54    taskID = int(response["taskId"])
55    
56    return taskID
57    
58    
59def GetCompletedCaptcha(taskID):
60    counter = 0
61
62    while True:
63        time.sleep(6.0)
64    
65        conn = http.client.HTTPSConnection('api.anti-captcha.com', timeout=30.0)
66
67        postData = {}
68        postData['clientKey'] = CAPTCHA_CLIENT_KEY
69        postData['taskId'] = taskID
70
71        body = json.dumps(postData).encode('utf-8')
72        
73        print(body)
74        conn.request('POST', '/getTaskResult', body=body)
75        
76        res = conn.getresponse()
77    
78        response = json.loads(res.read().decode('utf-8', 'ignore'))
79
80        print(response)
81
82        if 'status' in response and response['status'] == 'ready':
83            conn.close()
84            return response["solution"]["gRecaptchaResponse"]
85        elif 'errorId' in response and response['errorId'] != 0:
86            raise RuntimeError('Getting completed captcha resulted in error %d, "%s"' % (response['errorId'], response["errorDescription"]))
87        else:
88            print('Not ready yet....')
89
90        conn.close()
91        
92        counter += 1
93        
94        if counter > 60:
95            raise RuntimeError('We\'ve probably messed something up, task %d cannot be completed it seems' % taskID)
96
97
98
99formActionReg = re.compile(r'<form action="([^"]*)" method="POST">')
100formInputReg = re.compile(r'<input type="[^"]*" name="([^"]*)" value="([^"]*)">')
101siteKeyReg = re.compile(r'<div class="g-recaptcha" data-sitekey="([a-zA-Z0-9\-_]*)">')
102
103def GrabDataOffCaptchaHTML(pageTxt):
104    formActionMatch = formActionReg.search(pageTxt)
105    
106    if formActionMatch is None:
107        raise RuntimeError('Could not find form action!!')
108
109    formAction = formActionMatch.group(1)
110
111    formTxt = pageTxt.split('<form action="',1)[-1].split('</form>',1)[0]
112    siteKeyMatch = siteKeyReg.search(formTxt)
113    if siteKeyMatch is None:
114        raise RuntimeError('Could not find site key!!')
115        
116    siteKey = siteKeyMatch.group(1)
117
118    retVal = {}
119    retVal['action'] = formAction
120    retVal['site_key'] = siteKey
121
122    inputs = {}
123    for match in formInputReg.finditer(formTxt):
124        inputName = match.group(1)
125        inputValue = match.group(2)
126        inputs[inputName] = inputValue
127
128
129    retVal['inputs'] = inputs
130
131    return retVal
132
133
134def FinishCaptcha(url, formAction, headers, formInputs):
135    conn = http.client.HTTPSConnection('www.youtube.com', timeout=30.0)
136
137    body = urlencode(formInputs).encode('utf-8')
138    headers['Content-Type'] = 'application/x-www-form-urlencoded'
139    print(body)
140    conn.request("POST", formAction, body=body, headers=headers)
141    
142    res = conn.getresponse()
143    
144    print('Captcha result headers:')
145    print(res.getheaders())
146    
147    pageTxt = res.read().decode('utf-8', 'ignore')
148    
149    cookieVals = {}
150        
151    for headerName,headerValue in res.getheaders():
152        if headerName.lower() == 'set-cookie':
153            cookieParts = headerValue.split(';',1)[0].split('=')
154            cookieVals[cookieParts[0]] = cookieParts[1]
155        
156    conn.close()
157    
158    return cookieVals
159
160def CookieDictToString(cookieVals):
161    CookieStr = ''
162    for key,val in cookieVals.items():
163        if CookieStr != '':
164            CookieStr = CookieStr + '; '
165            
166        CookieStr = CookieStr + '%s=%s' % (key, val)
167
168    return CookieStr
169
170def Test429():
171
172    # Random video, could be any
173    path = '/watch?v=n9KjA471Y3k'
174
175    print('------------Fetching watch page----------')
176    conn = http.client.HTTPSConnection('www.youtube.com', timeout=30.0)
177    conn.request("GET", path)
178    
179    res = conn.getresponse()
180    
181    if res.status != 429:
182        print('Not 429\'d, dont bother with captcha')
183        return
184        
185    cookieVals = {}
186        
187    for headerName,headerValue in res.getheaders():
188        if headerName.lower() == 'set-cookie':
189            cookieParts = headerValue.split(';',1)[0].split('=')
190            cookieVals[cookieParts[0]] = cookieParts[1]
191            
192        
193    CookieStr = CookieDictToString(cookieVals)
194    
195    print('Initial watch page cookies: "%s"' % CookieStr)
196        
197    pageTxt = res.read().decode('utf-8', 'ignore')
198    
199    conn.close()
200
201    captchaData = GrabDataOffCaptchaHTML(pageTxt)
202
203    print('------------Creating task for captcha----------')
204    taskID = CreateTaskForCaptcha('https://www.youtube.com' + path, captchaData['site_key'])
205    
206    print('------------Waiting for task to finish----------')
207    captchaResult = GetCompletedCaptcha(taskID)
208
209    inputs = captchaData['inputs']
210    
211    inputs["g-recaptcha-response"] = captchaResult
212    
213    print(inputs)
214    
215    headers = {}
216    headers['Cookie'] = CookieStr
217    
218    print('---------------Submitting form request to Captcha page----------')
219    newCookieVals = FinishCaptcha(path, captchaData['action'], headers, inputs)
220    
221    if len(newCookieVals) == 0:
222        print('ERROR: Could not get any new cookies from the captcha form response, possibly ran consecutively too fast')
223    
224    for key,val in newCookieVals.items():
225        cookieVals[key] = val
226    
227    print('-----------------Output----------------')
228    print('Cookie vals: %s' % str(cookieVals))
229    print('Cookie string: %s' % CookieDictToString(cookieVals))  
230
231
232# Uncomment this to actually run the test
233# Test429()