Aggiornato cloudscraper

lib/cloudscraper/interpreters/__init__.py

@@ -49,7 +49,7 @@ class JavaScriptInterpreter(ABC):
 
     def solveChallenge(self, body, domain):
         try:
-            return float(self.eval(body, domain))
+            return '{0:.10f}'.format(float(self.eval(body, domain)))
         except Exception:
             raise CloudflareSolveError(
                 'Error trying to solve Cloudflare IUAM Javascript, they may have changed their technique.'

lib/cloudscraper/interpreters/encapsulated.py

@@ -9,32 +9,38 @@ def template(body, domain):
 
     try:
         js = re.search(
-            r'setTimeout\(function\(\){\s+(.*?a\.value = \S+)',
+            r'setTimeout\(function\(\){\s+(.*?a\.value\s*=\s*\S+toFixed\(10\);)',
             body,
             re.M | re.S
         ).group(1)
     except Exception:
         raise ValueError('Unable to identify Cloudflare IUAM Javascript on website. {}'.format(BUG_REPORT))
 
-    jsEnv = '''
-        String.prototype.italics=function(str) {{return "<i>" + this + "</i>";}};
+    jsEnv = '''String.prototype.italics=function(str) {{return "<i>" + this + "</i>";}};
+        var subVars= {{{subVars}}};
         var document = {{
             createElement: function () {{
                 return {{ firstChild: {{ href: "https://{domain}/" }} }}
             }},
-            getElementById: function () {{
-                return {{"innerHTML": "{innerHTML}"}};
+            getElementById: function (str) {{
+                return {{"innerHTML": subVars[str]}};
             }}
         }};
     '''
 
     try:
-        innerHTML = re.search(
-            r'<div(?: [^<>]*)? id="([^<>]*?)">([^<>]*?)</div>',
-            body,
-            re.MULTILINE | re.DOTALL
+        js = js.replace(
+            r"(setInterval(function(){}, 100),t.match(/https?:\/\//)[0]);",
+            r"t.match(/https?:\/\//)[0];"
         )
-        innerHTML = innerHTML.group(2) if innerHTML else ''
+
+        k = re.search(r" k\s*=\s*'(?P<k>\S+)';", body).group('k')
+        r = re.compile(r'<div id="{}(?P<id>\d+)">\s*(?P<jsfuck>[^<>]*)</div>'.format(k))
+
+        subVars = ''
+        for m in r.finditer(body):
+            subVars = '{}\n\t\t{}{}: {},\n'.format(subVars, k, m.group('id'), m.group('jsfuck'))
+        subVars = subVars[:-2]
 
     except:  # noqa
         logging.error('Error extracting Cloudflare IUAM Javascript. {}'.format(BUG_REPORT))
@@ -46,7 +52,7 @@ def template(body, domain):
             ' ',
             jsEnv.format(
                 domain=domain,
-                innerHTML=innerHTML
+                subVars=subVars
             ),
             re.MULTILINE | re.DOTALL
         ),

lib/cloudscraper/interpreters/native.py

@@ -100,8 +100,8 @@ class ChallengeInterpreter(JavaScriptInterpreter):
 
         # ------------------------------------------------------------------------------- #
 
-        def flatten(l):
-            return sum(map(flatten, l), []) if isinstance(l, list) else [l]
+        def flatten(lists):
+            return sum(map(flatten, lists), []) if isinstance(lists, list) else [lists]
 
         # ------------------------------------------------------------------------------- #
 
@@ -114,6 +114,7 @@ class ChallengeInterpreter(JavaScriptInterpreter):
             # Hackery Parser for Math
             stack = []
             bstack = []
+
             for i in flatten(pyparsing.nestedExpr().parseString(jsFuck).asList()):
                 if i == '+':
                     stack.append(bstack)
@@ -152,13 +153,35 @@ class ChallengeInterpreter(JavaScriptInterpreter):
             try:
                 jsfuckChallenge = re.search(
                     r"setTimeout\(function\(\){\s+var.*?f,\s*(?P<variable>\w+).*?:(?P<init>\S+)};"
-                    r".*?\('challenge-form'\);\s+;(?P<challenge>.*?a\.value)"
-                    r"(?:.*id=\"cf-dn-.*?>(?P<k>\S+)<)?",
+                    r".*?\('challenge-form'\);.*?;(?P<challenge>.*?a\.value)\s*=\s*\S+\.toFixed\(10\);",
                     body,
                     re.DOTALL | re.MULTILINE
                 ).groupdict()
             except AttributeError:
-                raise CloudflareSolveError('There was an issue extracting the Cloudflare challenge.')
+                raise CloudflareSolveError('There was an issue extracting "jsfuckChallenge" from the Cloudflare challenge.')
+
+            kJSFUCK = re.search(r'(;|)\s*k.=(?P<kJSFUCK>\S+);', jsfuckChallenge['challenge'], re.S | re.M)
+            if kJSFUCK:
+                try:
+                    kJSFUCK = jsfuckToNumber(kJSFUCK.group('kJSFUCK'))
+                except IndexError:
+                    raise CloudflareSolveError('There was an issue extracting "kJSFUCK" from the Cloudflare challenge.')
+
+                try:
+                    kID = re.search(r"\s*k\s*=\s*'(?P<kID>\S+)';", body).group('kID')
+                except IndexError:
+                    raise CloudflareSolveError('There was an issue extracting "kID" from the Cloudflare challenge.')
+
+                try:
+                    r = re.compile(r'<div id="{}(?P<id>\d+)">\s*(?P<jsfuck>[^<>]*)</div>'.format(kID))
+
+                    kValues = {}
+                    for m in r.finditer(body):
+                        kValues[int(m.group('id'))] = m.group('jsfuck')
+
+                    jsfuckChallenge['k'] = kValues[kJSFUCK]
+                except (AttributeError, IndexError):
+                    raise CloudflareSolveError('There was an issue extracting "kValues" from the Cloudflare challenge.')
 
             jsfuckChallenge['challenge'] = re.finditer(
                 r'{}.*?([+\-*/])=(.*?);(?=a\.value|{})'.format(
@@ -193,8 +216,8 @@ class ChallengeInterpreter(JavaScriptInterpreter):
 
             # ------------------------------------------------------------------------------- #
 
-            if not jsfuckChallenge['k'] and '+ t.length' in body:
-                jschl_answer += len(domain)
+            # if not jsfuckChallenge['k'] and '+ t.length' in body:
+            #    jschl_answer += len(domain)
 
             # ------------------------------------------------------------------------------- #