Ticket #6104: get-page-cancel.diff

twisted/web/client.py

@@ -311 +311 @@
     def __init__(self, url, method='GET', postdata=None, headers=None,
                  agent="Twisted PageGetter", timeout=0, cookies=None,
                  followRedirect=True, redirectLimit=20,
-                 afterFoundGet=False):
+                 afterFoundGet=False, canceller=None):
         self.followRedirect = followRedirect
         self.redirectLimit = redirectLimit
         self._redirectCount = 0
@@ -336 +336 @@
 
         self.waiting = 1
         self._disconnectedDeferred = defer.Deferred()
-        self.deferred = defer.Deferred()
+        self.deferred = defer.Deferred(canceller)
         # Make sure the first callback on the result Deferred pauses the
         # callback chain until the request connection is closed.
         self.deferred.addBoth(self._waitForDisconnect)
@@ -408 +408 @@
         result has yet been provided to the result Deferred, provide the
         connection failure reason as an error result.
         """
-        if self.waiting:
-            self.waiting = 0
-            # If the connection attempt failed, there is nothing more to
-            # disconnect, so just fire that Deferred now.
-            self._disconnectedDeferred.callback(None)
-            self.deferred.errback(reason)
+        self.noPage(reason)
+        # If the connection attempt failed, there is nothing more to
+        # disconnect, so just fire that Deferred now.
+        self._disconnectedDeferred.callback(None)
 
 
 
@@ -427 +425 @@
                  method='GET', postdata=None, headers=None,
                  agent="Twisted client", supportPartial=0,
                  timeout=0, cookies=None, followRedirect=1,
-                 redirectLimit=20, afterFoundGet=False):
+                 redirectLimit=20, afterFoundGet=False,
+                 canceller=None):
         self.requestedPartial = 0
         if isinstance(fileOrName, types.StringTypes):
             self.fileName = fileOrName
@@ -445 +444 @@
             self, url, method=method, postdata=postdata, headers=headers,
             agent=agent, timeout=timeout, cookies=cookies,
             followRedirect=followRedirect, redirectLimit=redirectLimit,
-            afterFoundGet=afterFoundGet)
+            afterFoundGet=afterFoundGet, canceller=canceller)
 
 
     def gotHeaders(self, headers):
@@ -596 +595 @@
 
     @return: The factory created by C{factoryFactory}
     """
+    def cancel(d):
+        factory.noPage(defer.CancelledError())
+        connector.disconnect()
     scheme, host, port, path = _parse(url)
-    factory = factoryFactory(url, *args, **kwargs)
+    factory = factoryFactory(url, canceller=cancel, *args, **kwargs)
     if scheme == 'https':
         from twisted.internet import ssl
         if contextFactory is None:
             contextFactory = ssl.ClientContextFactory()
-        reactor.connectSSL(host, port, factory, contextFactory)
+        connector = reactor.connectSSL(host, port, factory, contextFactory)
     else:
-        reactor.connectTCP(host, port, factory)
+        connector = reactor.connectTCP(host, port, factory)
     return factory
 
 
@@ -615 +617 @@
     Download a page. Return a deferred, which will callback with a
     page (as a string) or errback with a description of the error.
 
+    If the deferred is cancelled before the request completes, the
+    connection is closed and the deferred will fire with a
+    L{defer.CancelledError}.
+
     See L{HTTPClientFactory} to see what extra arguments can be passed.
     """
     return _makeGetterFactory(
@@ -628 +634 @@
     """
     Download a web page to a file.
 
+    Download a page. Return a deferred, which will callback with None
+    or errback with a description of the error.
+
+    If the deferred is cancelled before the request completes, the
+    connection is closed and the deferred will fire with a
+    L{defer.CancelledError}.
+
     @param file: path to file on filesystem, or file-like object.
 
-    See HTTPDownloader to see what extra args can be passed.
+    See L{HTTPDownloader} to see what extra args can be passed.
     """
     factoryFactory = lambda url, *a, **kw: HTTPDownloader(url, file, *a, **kw)
     return _makeGetterFactory(

twisted/web/test/test_webclient.py

@@ -507 +507 @@
             defer.TimeoutError)
 
 
+    def test_getPageCancelImmediately(self):
+        """
+        Call L{getPage} and cancel it before it has chance to try to
+        connect to the server. The L{Deferred} returned by L{getPage}
+        fails with L{defer.CancelledError}.
+        """
+        d = client.getPage(self.getURL("wait"), timeout=10)
+        d.cancel()
+        return self.assertFailure(d, defer.CancelledError)
+
+
+    def test_getPageCancelLater(self):
+        """
+        Call L{getPage} and cancel it after it has connected to the server,
+        but before it was able to deliver the results. The L{Deferred}
+        returned by L{getPage} fails with L{defer.CancelledError}.
+        """
+        d = client.getPage(self.getURL("wait"), timeout=10)
+        reactor.callLater(0.01, d.cancel)
+        return self.assertFailure(d, defer.CancelledError)
+
+
+    def test_getPageCancelAfter(self):
+        """
+        Call L{getPage} and attempt to cancel it after it has delivered
+        the results. The cancellation is ignored and the L{Deferred}
+        returned by L{getPage} is called back with the page contents.
+        """
+        d = client.getPage(self.getURL("file"))
+        def cancelRequest(content):
+            d.cancel()
+            return content
+        d.addCallback(cancelRequest)
+        d.addCallback(self.assertEqual, "0123456789")
+        return d
+
+
     def testDownloadPage(self):
         downloads = []
         downloadData = [("file", self.mktemp(), "0123456789"),