Merge "deviate-redirects-3624" branch.

Author: truekonrads
Reviewer: exarkun, jonathanj
Fixes: #3624

Introduce a parameter to HTTPClientFactory that enables redirection behaviour
commonly found in web browsers: following a POST with a 302 response status.



git-svn-id: svn://svn.twistedmatrix.com/svn/Twisted/trunk@27250 bbbe8e31-12d6-0310-92fd-ac37d47ddeeb

twisted/web/client.py

@@ -123,7 +123,11 @@ def handleStatus_301(self):
         self.quietLoss = True
         self.transport.loseConnection()
 
-    handleStatus_302 = lambda self: self.handleStatus_301()
+    def handleStatus_302(self):
+        if self.afterFoundGet:
+            self.handleStatus_303()
+        self.handleStatus_301()
+
 
     def handleStatus_303(self):
         self.factory.method = 'GET'
@@ -226,6 +230,11 @@ class HTTPClientFactory(protocol.ClientFactory):
     @ivar redirectLimit: The maximum number of HTTP redirects that can occur
           before it is assumed that the redirection is endless.
 
+    @type afterFoundGet: C{bool}
+    @ivar afterFoundGet: Deviate from the HTTP 1.1 RFC by handling redirects
+        the same way as most web browsers; if the request method is POST and a
+        302 status is encountered, the redirect is followed with a GET method
+
     @type _redirectCount: int
     @ivar _redirectCount: The current number of HTTP redirects encountered.
     """
@@ -240,13 +249,14 @@ class HTTPClientFactory(protocol.ClientFactory):
 
     def __init__(self, url, method='GET', postdata=None, headers=None,
                  agent="Twisted PageGetter", timeout=0, cookies=None,
-                 followRedirect=True, redirectLimit=20):
+                 followRedirect=True, redirectLimit=20,
+                 afterFoundGet=False):
         self.followRedirect = followRedirect
         self.redirectLimit = redirectLimit
         self._redirectCount = 0
         self.timeout = timeout
         self.agent = agent
-
+        self.afterFoundGet = afterFoundGet
         if cookies is None:
             cookies = {}
         self.cookies = cookies
@@ -282,6 +292,7 @@ def setURL(self, url):
     def buildProtocol(self, addr):
         p = protocol.ClientFactory.buildProtocol(self, addr)
         p.followRedirect = self.followRedirect
+        p.afterFoundGet = self.afterFoundGet
         if self.timeout:
             timeoutCall = reactor.callLater(self.timeout, p.timeout)
             self.deferred.addBoth(self._cancelTimeout, timeoutCall)

twisted/web/test/test_webclient.py

@@ -24,6 +24,46 @@
     ssl = None
 
 
+
+class ExtendedRedirect(resource.Resource):
+    """
+    Redirection resource.
+
+    The HTTP status code is set according to the C{code} query parameter.
+
+    @type lastMethod: C{str}
+    @ivar lastMethod: Last handled HTTP request method
+    """
+    isLeaf = 1
+    lastMethod = None
+
+
+    def __init__(self, url):
+        resource.Resource.__init__(self)
+        self.url = url
+
+
+    def render(self, request):
+        if self.lastMethod:
+            self.lastMethod = request.method
+            return "OK Thnx!"
+        else:
+            self.lastMethod = request.method
+            code = int(request.args['code'][0])
+            return self.redirectTo(self.url, request, code)
+
+
+    def getChild(self, name, request):
+        return self
+
+
+    def redirectTo(self, url, request, code):
+        request.setResponseCode(code)
+        request.setHeader("location", url)
+        return "OK Bye!"
+
+
+
 class ForeverTakingResource(resource.Resource):
     """
     L{ForeverTakingResource} is a resource which never finishes responding
@@ -231,6 +271,8 @@ def setUp(self):
         r.putChild("payload", PayloadResource())
         r.putChild("broken", BrokenDownloadResource())
         r.putChild("cookiemirror", CookieMirrorResource())
+        self.extendedRedirect = ExtendedRedirect('/extendedRedirect')
+        r.putChild("extendedRedirect", self.extendedRedirect)
         self.site = server.Site(r, timeout=None)
         self.wrapper = WrappingFactory(self.site)
         self.port = self._listen(self.wrapper)
@@ -494,6 +536,29 @@ def test_isolatedFollowRedirect(self):
         return d
 
 
+    def test_afterFoundGet(self):
+        """
+        Enabling unsafe redirection behaviour overwrites the method of
+        redirected C{POST} requests with C{GET}.
+        """
+        url = self.getURL('extendedRedirect?code=302')
+        f = client.HTTPClientFactory(url, followRedirect=True, method="POST")
+        self.assertFalse(
+            f.afterFoundGet,
+            "By default, afterFoundGet must be disabled")
+
+        def gotPage(page):
+            self.assertEquals(
+                self.extendedRedirect.lastMethod,
+                "GET",
+                "With afterFoundGet, the HTTP method must change to GET")
+
+        d = client.getPage(
+            url, followRedirect=True, afterFoundGet=True, method="POST")
+        d.addCallback(gotPage)
+        return d
+
+
     def testPartial(self):
         name = self.mktemp()
         f = open(name, "wb")