From c1be59f40ae7d1268961b95c5734297b5ca3d872 Mon Sep 17 00:00:00 2001 From: Jeremy Hylton Date: Sun, 14 Dec 2003 05:27:34 +0000 Subject: SF patch 852995: add processors feature to urllib2 John J. Lee writes: "the patch makes it possible to implement functionality like HTTP cookie handling, Refresh handling, etc. etc. using handler objects. At the moment urllib2's handler objects aren't quite up to the job, which results in a lot of cut-n-paste and subclassing. I believe the changes are backwards-compatible, with the exception of people who've reimplemented build_opener()'s functionality -- those people would need to call opener.add_handler(HTTPErrorProcessor). The main change is allowing handlers to implement methods like: http_request(request) http_response(request, response) In addition to the usual http_open(request) http_error{_*}(...) " Note that the change isn't well documented at least in part because handlers aren't well documented at all. Need to fix this. Add a bunch of new tests. It appears that none of these tests actually use the network, so they don't need to be guarded by a resource flag. --- Doc/lib/liburllib2.tex | 32 ++- Lib/test/test_urllib2.py | 643 ++++++++++++++++++++++++++++++++++++++++++++--- Lib/urllib2.py | 194 +++++++++----- 3 files changed, 783 insertions(+), 86 deletions(-) diff --git a/Doc/lib/liburllib2.tex b/Doc/lib/liburllib2.tex index 3f8ff3a..0020b6c 100644 --- a/Doc/lib/liburllib2.tex +++ b/Doc/lib/liburllib2.tex @@ -52,7 +52,7 @@ front of the \var{handler}s, unless the \var{handler}s contain them, instances of them or subclasses of them: \class{ProxyHandler}, \class{UnknownHandler}, \class{HTTPHandler}, \class{HTTPDefaultErrorHandler}, \class{HTTPRedirectHandler}, -\class{FTPHandler}, \class{FileHandler} +\class{FTPHandler}, \class{FileHandler}, \class{HTTPErrorProcessor}. If the Python installation has SSL support (\function{socket.ssl()} exists), \class{HTTPSHandler} will also be added. @@ -248,6 +248,15 @@ when used more than once have a (header-specific) way of gaining the same functionality using only one header. \end{methoddesc} +\begin{methoddesc}[Request]{add_unredirected_header}{key, header} +Add a header that will not be added to a redirected request. +\end{methoddesc} + +\begin{methoddesc}[Request]{has_header}{header} +Return whether the instance has the named header (checks both regular +and unredirected). +\end{methoddesc} + \begin{methoddesc}[Request]{get_full_url}{} Return the URL given in the constructor. \end{methoddesc} @@ -286,6 +295,12 @@ following methods are searched, and added to the possible chains. \item \method{\var{protocol}_error_\var{type}()} --- signal that the handler knows how to handle \var{type} errors from \var{protocol}. + \item \method{\var{protocol}_request()} --- + signal that the handler knows how to pre-process \var{protocol} + requests. + \item \method{\var{protocol}_response()} --- + signal that the handler knows how to post-process \var{protocol} + responses. \end{itemize} \end{methoddesc} @@ -620,6 +635,21 @@ Raise a \exception{URLError} exception. \end{methoddesc} +\subsection{HTTPErrorProcessor Objects \label{http-error-processor-objects}} + +\begin{methoddesc}[HTTPErrorProcessor]{unknown_open}{} +Process HTTP error responses. + +For 200 error codes, the response object is returned immediately. + +For non-200 error codes, this simply passes the job on to the +\method{\var{protocol}_error_\var{code}()} handler methods, via +\method{OpenerDirector.error()}. Eventually, +\class{urllib2.HTTPDefaultErrorHandler} will raise an +\exception{HTTPError} if no other handler handles the error. +\end{methoddesc} + + \subsection{Examples \label{urllib2-examples}} This example gets the python.org main page and displays the first 100 diff --git a/Lib/test/test_urllib2.py b/Lib/test/test_urllib2.py index 8238bca..d8bf83e 100644 --- a/Lib/test/test_urllib2.py +++ b/Lib/test/test_urllib2.py @@ -1,31 +1,616 @@ -from test.test_support import verify -import urllib2 +import unittest +from test import test_support + import os +import StringIO + +import urllib2 +from urllib2 import Request, OpenerDirector + +# XXX +# Request +# CacheFTPHandler (hard to write) +# parse_keqv_list, parse_http_list (I'm leaving this for Anthony Baxter +# and Greg Stein, since they're doing Digest Authentication) +# Authentication stuff (ditto) +# ProxyHandler, CustomProxy, CustomProxyHandler (I don't use a proxy) +# GopherHandler (haven't used gopher for a decade or so...) + +class TrivialTests(unittest.TestCase): + def test_trivial(self): + # A couple trivial tests + + self.assertRaises(ValueError, urllib2.urlopen, 'bogus url') + + # XXX Name hacking to get this to work on Windows. + fname = os.path.abspath(urllib2.__file__).replace('\\', '/') + if fname[1:2] == ":": + fname = fname[2:] + # And more hacking to get it to work on MacOS. This assumes + # urllib.pathname2url works, unfortunately... + if os.name == 'mac': + fname = '/' + fname.replace(':', '/') + elif os.name == 'riscos': + import string + fname = os.expand(fname) + fname = fname.translate(string.maketrans("/.", "./")) + + file_url = "file://%s" % fname + f = urllib2.urlopen(file_url) + + buf = f.read() + f.close() + + +class MockOpener: + addheaders = [] + def open(self, req, data=None): + self.req, self.data = req, data + def error(self, proto, *args): + self.proto, self.args = proto, args + +class MockFile: + def read(self, count=None): pass + def readline(self, count=None): pass + def close(self): pass + +class MockResponse(StringIO.StringIO): + def __init__(self, code, msg, headers, data, url=None): + StringIO.StringIO.__init__(self, data) + self.code, self.msg, self.headers, self.url = code, msg, headers, url + def info(self): + return self.headers + def geturl(self): + return self.url + +class FakeMethod: + def __init__(self, meth_name, action, handle): + self.meth_name = meth_name + self.handle = handle + self.action = action + def __call__(self, *args): + return self.handle(self.meth_name, self.action, *args) + +class MockHandler: + def __init__(self, methods): + self._define_methods(methods) + def _define_methods(self, methods): + for spec in methods: + if len(spec) == 2: name, action = spec + else: name, action = spec, None + meth = FakeMethod(name, action, self.handle) + setattr(self.__class__, name, meth) + def handle(self, fn_name, action, *args, **kwds): + self.parent.calls.append((self, fn_name, args, kwds)) + if action is None: + return None + elif action == "return self": + return self + elif action == "return response": + res = MockResponse(200, "OK", {}, "") + return res + elif action == "return request": + return Request("http://blah/") + elif action.startswith("error"): + code = action[action.rfind(" ")+1:] + try: + code = int(code) + except ValueError: + pass + res = MockResponse(200, "OK", {}, "") + return self.parent.error("http", args[0], res, code, "", {}) + elif action == "raise": + raise urllib2.URLError("blah") + assert False + def close(self): pass + def add_parent(self, parent): + self.parent = parent + self.parent.calls = [] + def __lt__(self, other): + if not hasattr(other, "handler_order"): + # No handler_order, leave in original order. Yuck. + return True + return self.handler_order < other.handler_order + +def add_ordered_mock_handlers(opener, meth_spec): + """Create MockHandlers and add them to an OpenerDirector. + + meth_spec: list of lists of tuples and strings defining methods to define + on handlers. eg: + + [["http_error", "ftp_open"], ["http_open"]] + + defines methods .http_error() and .ftp_open() on one handler, and + .http_open() on another. These methods just record their arguments and + return None. Using a tuple instead of a string causes the method to + perform some action (see MockHandler.handle()), eg: + + [["http_error"], [("http_open", "return request")]] + + defines .http_error() on one handler (which simply returns None), and + .http_open() on another handler, which returns a Request object. + + """ + handlers = [] + count = 0 + for meths in meth_spec: + class MockHandlerSubclass(MockHandler): pass + h = MockHandlerSubclass(meths) + h.handler_order = count + h.add_parent(opener) + count = count + 1 + handlers.append(h) + opener.add_handler(h) + return handlers + +class OpenerDirectorTests(unittest.TestCase): + + def test_handled(self): + # handler returning non-None means no more handlers will be called + o = OpenerDirector() + meth_spec = [ + ["http_open", "ftp_open", "http_error_302"], + ["ftp_open"], + [("http_open", "return self")], + [("http_open", "return self")], + ] + handlers = add_ordered_mock_handlers(o, meth_spec) + + req = Request("http://example.com/") + r = o.open(req) + # Second .http_open() gets called, third doesn't, since second returned + # non-None. Handlers without .http_open() never get any methods called + # on them. + # In fact, second mock handler defining .http_open() returns self + # (instead of response), which becomes the OpenerDirector's return + # value. + self.assert_(r == handlers[2]) + calls = [(handlers[0], "http_open"), (handlers[2], "http_open")] + for expected, got in zip(calls, o.calls): + handler, name, args, kwds = got + self.assert_((handler, name) == expected) + self.assert_(args == (req,)) + + def test_handler_order(self): + o = OpenerDirector() + handlers = [] + for meths, handler_order in [ + ([("http_open", "return self")], 500), + (["http_open"], 0), + ]: + class MockHandlerSubclass(MockHandler): pass + h = MockHandlerSubclass(meths) + h.handler_order = handler_order + handlers.append(h) + o.add_handler(h) + + r = o.open("http://example.com/") + # handlers called in reverse order, thanks to their sort order + self.assert_(o.calls[0][0] == handlers[1]) + self.assert_(o.calls[1][0] == handlers[0]) + + def test_raise(self): + # raising URLError stops processing of request + o = OpenerDirector() + meth_spec = [ + [("http_open", "raise")], + [("http_open", "return self")], + ] + handlers = add_ordered_mock_handlers(o, meth_spec) + + req = Request("http://example.com/") + self.assertRaises(urllib2.URLError, o.open, req) + self.assert_(o.calls == [(handlers[0], "http_open", (req,), {})]) + +## def test_error(self): +## # XXX this doesn't actually seem to be used in standard library, +## # but should really be tested anyway... + + def test_http_error(self): + # XXX http_error_default + # http errors are a special case + o = OpenerDirector() + meth_spec = [ + [("http_open", "error 302")], + [("http_error_400", "raise"), "http_open"], + [("http_error_302", "return response"), "http_error_303", + "http_error"], + [("http_error_302")], + ] + handlers = add_ordered_mock_handlers(o, meth_spec) + + class Unknown: + def __eq__(self, other): return True + + req = Request("http://example.com/") + r = o.open(req) + assert len(o.calls) == 2 + calls = [(handlers[0], "http_open", (req,)), + (handlers[2], "http_error_302", (req, Unknown(), 302, "", {}))] + for expected, got in zip(calls, o.calls): + handler, method_name, args = expected + self.assert_((handler, method_name) == got[:2]) + assert args == got[2] + + def test_processors(self): + # *_request / *_response methods get called appropriately + o = OpenerDirector() + meth_spec = [ + [("http_request", "return request"), + ("http_response", "return response")], + [("http_request", "return request"), + ("http_response", "return response")], + ] + handlers = add_ordered_mock_handlers(o, meth_spec) + + req = Request("http://example.com/") + r = o.open(req) + # processor methods are called on *all* handlers that define them, + # not just the first handler that handles the request + calls = [(handlers[0], "http_request"), (handlers[1], "http_request"), + (handlers[0], "http_response"), (handlers[1], "http_response")] + + for i, (handler, name, args, kwds) in enumerate(o.calls): + if i < 2: + # *_request + self.assert_((handler, name) == calls[i]) + self.assert_(len(args) == 1) + self.assert_(isinstance(args[0], Request)) + else: + # *_response + self.assert_((handler, name) == calls[i]) + self.assert_(len(args) == 2) + self.assert_(isinstance(args[0], Request)) + # response from opener.open is None, because there's no + # handler that defines http_open to handle it + self.assert_(args[1] is None or + isinstance(args[1], MockResponse)) + + +class HandlerTests(unittest.TestCase): + + def test_ftp(self): + class MockFTPWrapper: + def __init__(self, data): self.data = data + def retrfile(self, filename, filetype): + self.filename, self.filetype = filename, filetype + return StringIO.StringIO(self.data), len(self.data) + + class NullFTPHandler(urllib2.FTPHandler): + def __init__(self, data): self.data = data + def connect_ftp(self, user, passwd, host, port, dirs): + self.user, self.passwd = user, passwd + self.host, self.port = host, port + self.dirs = dirs + self.ftpwrapper = MockFTPWrapper(self.data) + return self.ftpwrapper + + import ftplib, socket + data = "rheum rhaponicum" + h = NullFTPHandler(data) + o = h.parent = MockOpener() + + for url, host, port, type_, dirs, filename, mimetype in [ + ("ftp://localhost/foo/bar/baz.html", + "localhost", ftplib.FTP_PORT, "I", + ["foo", "bar"], "baz.html", "text/html"), +# XXXX Bug: FTPHandler tries to gethostbyname "localhost:80", with the +# port still there. +## ("ftp://localhost:80/foo/bar/", +## "localhost", 80, "D", +## ["foo", "bar"], "", None), +# XXXX bug: second use of splitattr() in FTPHandler should be splitvalue() +## ("ftp://localhost/baz.gif;type=a", +## "localhost", ftplib.FTP_PORT, "A", +## [], "baz.gif", "image/gif"), + ]: + r = h.ftp_open(Request(url)) + # ftp authentication not yet implemented by FTPHandler + self.assert_(h.user == h.passwd == "") + self.assert_(h.host == socket.gethostbyname(host)) + self.assert_(h.port == port) + self.assert_(h.dirs == dirs) + self.assert_(h.ftpwrapper.filename == filename) + self.assert_(h.ftpwrapper.filetype == type_) + headers = r.info() + self.assert_(headers["Content-type"] == mimetype) + self.assert_(int(headers["Content-length"]) == len(data)) + + def test_file(self): + import time, rfc822, socket + h = urllib2.FileHandler() + o = h.parent = MockOpener() + + #from test_support import TESTFN + TESTFN = "test.txt" + towrite = "hello, world\n" + for url in [ + "file://localhost%s/%s" % (os.getcwd(), TESTFN), + "file://%s/%s" % (os.getcwd(), TESTFN), + "file://%s%s/%s" % (socket.gethostbyname('localhost'), + os.getcwd(), TESTFN), + "file://%s%s/%s" % (socket.gethostbyname(socket.gethostname()), + os.getcwd(), TESTFN), + # XXX Windows / Mac format(s), ... ? + ]: + f = open(TESTFN, "w") + try: + try: + f.write(towrite) + finally: + f.close() + + r = h.file_open(Request(url)) + try: + data = r.read() + read_time = time.time() + headers = r.info() + newurl = r.geturl() + finally: + r.close() + finally: + os.remove(TESTFN) + self.assert_(data == towrite) + self.assert_(headers["Content-type"] == "text/plain") + self.assert_(headers["Content-length"] == "13") + # Fudge Last-modified string comparison by one second to + # prevent spurious failure on crossing a second boundary while + # executing this test. + unfudged = rfc822.formatdate(read_time) + fudged = rfc822.formatdate(read_time-1) + self.assert_(headers["Last-modified"] in [unfudged, fudged]) + + for url in [ + "file://localhost:80%s/%s" % (os.getcwd(), TESTFN), +# XXXX bug: these fail with socket.gaierror, should be URLError +## "file://%s:80%s/%s" % (socket.gethostbyname('localhost'), +## os.getcwd(), TESTFN), +## "file://somerandomhost.ontheinternet.com%s/%s" % +## (os.getcwd(), TESTFN), + ]: + try: + f = open(TESTFN, "w") + try: + f.write(towrite) + finally: + f.close() + + self.assertRaises(urllib2.URLError, + h.file_open, Request(url)) + finally: + os.remove(TESTFN) + + h = urllib2.FileHandler() + o = h.parent = MockOpener() + # XXXX why does // mean ftp (and /// mean not ftp!), and where + # is file: scheme specified? I think this is really a bug, and + # what was intended was to distinguish between URLs like: + # file:/blah.txt (a file) + # file://localhost/blah.txt (a file) + # file:///blah.txt (a file) + # file://ftp.example.com/blah.txt (an ftp URL) + for url, ftp in [ + ("file://ftp.example.com//foo.txt", True), + ("file://ftp.example.com///foo.txt", False), +# XXXX bug: fails with OSError, should be URLError + ("file://ftp.example.com/foo.txt", False), + ]: + req = Request(url) + try: + h.file_open(req) + # XXXX remove OSError when bug fixed + except (urllib2.URLError, OSError): + self.assert_(not ftp) + else: + self.assert_(o.req is req) + self.assert_(req.type == "ftp") + + def test_http(self): + class MockHTTPClass: + def __init__(self): + self.req_headers = [] + self.data = None + self.raise_on_endheaders = False + def __call__(self, host): + self.host = host + return self + def set_debuglevel(self, level): self.level = level + def putrequest(self, method, selector): + self.method, self.selector = method, selector + def putheader(self, key, value): + self.req_headers.append((key, value)) + def endheaders(self): + if self.raise_on_endheaders: + import socket + raise socket.error() + def send(self, data): self.data = data + def getreply(self): return 200, "OK", {} + def getfile(self): return MockFile() + + h = urllib2.AbstractHTTPHandler() + o = h.parent = MockOpener() + + url = "http://example.com/" + for method, data in [("GET", None), ("POST", "blah")]: + req = Request(url, data, {"Foo": "bar"}) + req.add_unredirected_header("Spam", "eggs") + http = MockHTTPClass() + r = h.do_open(http, req) + + # result attributes + r.read; r.readline # wrapped MockFile methods + r.info; r.geturl # addinfourl methods + r.code, r.msg == 200, "OK" # added from MockHTTPClass.getreply() + hdrs = r.info() + hdrs.get; hdrs.has_key # r.info() gives dict from .getreply() + self.assert_(r.geturl() == url) + + self.assert_(http.host == "example.com") + self.assert_(http.level == 0) + self.assert_(http.method == method) + self.assert_(http.selector == "/") + self.assert_(http.req_headers == [("Foo", "bar"), ("Spam", "eggs")]) + self.assert_(http.data == data) + + # check socket.error converted to URLError + http.raise_on_endheaders = True + self.assertRaises(urllib2.URLError, h.do_open, http, req) + + # check adding of standard headers + o.addheaders = [("Spam", "eggs")] + for data in "", None: # POST, GET + req = Request("http://example.com/", data) + r = MockResponse(200, "OK", {}, "") + newreq = h.do_request(req) + if data is None: # GET + self.assert_("Content-length" not in req.unredirected_hdrs) + self.assert_("Content-type" not in req.unredirected_hdrs) + else: # POST + self.assert_(req.unredirected_hdrs["Content-length"] == "0") + self.assert_(req.unredirected_hdrs["Content-type"] == + "application/x-www-form-urlencoded") + # XXX the details of Host could be better tested + self.assert_(req.unredirected_hdrs["Host"] == "example.com") + self.assert_(req.unredirected_hdrs["Spam"] == "eggs") + + # don't clobber existing headers + req.add_unredirected_header("Content-length", "foo") + req.add_unredirected_header("Content-type", "bar") + req.add_unredirected_header("Host", "baz") + req.add_unredirected_header("Spam", "foo") + newreq = h.do_request(req) + self.assert_(req.unredirected_hdrs["Content-length"] == "foo") + self.assert_(req.unredirected_hdrs["Content-type"] == "bar") + self.assert_(req.unredirected_hdrs["Host"] == "baz") + self.assert_(req.unredirected_hdrs["Spam"] == "foo") + + def test_errors(self): + h = urllib2.HTTPErrorProcessor() + o = h.parent = MockOpener() + + url = "http://example.com/" + req = Request(url) + # 200 OK is passed through + r = MockResponse(200, "OK", {}, "", url) + newr = h.http_response(req, r) + self.assert_(r is newr) + self.assert_(not hasattr(o, "proto")) # o.error not called + # anything else calls o.error (and MockOpener returns None, here) + r = MockResponse(201, "Created", {}, "", url) + self.assert_(h.http_response(req, r) is None) + self.assert_(o.proto == "http") # o.error called + self.assert_(o.args == (req, r, 201, "Created", {})) + + def test_redirect(self): + from_url = "http://example.com/a.html" + to_url = "http://example.com/b.html" + h = urllib2.HTTPRedirectHandler() + o = h.parent = MockOpener() + + # ordinary redirect behaviour + for code in 301, 302, 303, 307: + for data in None, "blah\nblah\n": + method = getattr(h, "http_error_%s" % code) + req = Request(from_url, data) + req.add_header("Nonsense", "viking=withhold") + req.add_unredirected_header("Spam", "spam") + try: + method(req, MockFile(), code, "Blah", {"location": to_url}) + except urllib2.HTTPError: + # 307 in response to POST requires user OK + self.assert_(code == 307 and data is not None) + self.assert_(o.req.get_full_url() == to_url) + try: + self.assert_(o.req.get_method() == "GET") + except AttributeError: + self.assert_(not o.req.has_data()) + self.assert_(o.req.headers["Nonsense"] == "viking=withhold") + self.assert_("Spam" not in o.req.headers) + self.assert_("Spam" not in o.req.unredirected_hdrs) + + # loop detection + req = Request(from_url) + req.origin_req_host = "example.com" + def redirect(h, req, code, url=to_url): + method = getattr(h, "http_error_%s" % code) + method(req, MockFile(), code, "Blah", {"location": url}) + # Note that the *original* request shares the same record of + # redirections with the sub-requests caused by the redirections. + # once + redirect(h, req, 302) + # twice: loop detected + self.assertRaises(urllib2.HTTPError, redirect, h, req, 302) + # and again + self.assertRaises(urllib2.HTTPError, redirect, h, req, 302) + # but this is a different redirect code, so OK... + redirect(h, req, 301) + self.assertRaises(urllib2.HTTPError, redirect, h, req, 301) + # order doesn't matter + redirect(h, req, 303) + redirect(h, req, 307) + self.assertRaises(urllib2.HTTPError, redirect, h, req, 303) + + # detect endless non-repeating chain of redirects + req = Request(from_url) + req.origin_req_host = "example.com" + count = 0 + try: + while 1: + redirect(h, req, 302, "http://example.com/%d" % count) + count = count + 1 + except urllib2.HTTPError: + self.assert_(count == urllib2.HTTPRedirectHandler.max_redirections) + + +class MiscTests(unittest.TestCase): + + def test_build_opener(self): + class MyHTTPHandler(urllib2.HTTPHandler): pass + class FooHandler(urllib2.BaseHandler): + def foo_open(self): pass + class BarHandler(urllib2.BaseHandler): + def bar_open(self): pass + + build_opener = urllib2.build_opener + + o = build_opener(FooHandler, BarHandler) + self.opener_has_handler(o, FooHandler) + self.opener_has_handler(o, BarHandler) + + # can take a mix of classes and instances + o = build_opener(FooHandler, BarHandler()) + self.opener_has_handler(o, FooHandler) + self.opener_has_handler(o, BarHandler) + + # subclasses of default handlers override default handlers + o = build_opener(MyHTTPHandler) + self.opener_has_handler(o, MyHTTPHandler) + + # a particular case of overriding: default handlers can be passed + # in explicitly + o = build_opener() + self.opener_has_handler(o, urllib2.HTTPHandler) + o = build_opener(urllib2.HTTPHandler) + self.opener_has_handler(o, urllib2.HTTPHandler) + o = build_opener(urllib2.HTTPHandler()) + self.opener_has_handler(o, urllib2.HTTPHandler) + + def opener_has_handler(self, opener, handler_class): + for h in opener.handlers: + if h.__class__ == handler_class: + break + else: + self.assert_(False) + + +def test_main(verbose=None): + from test import test_sets + test_support.run_unittest( + TrivialTests, + OpenerDirectorTests, + HandlerTests, + MiscTests, + ) -# A couple trivial tests - -try: - urllib2.urlopen('bogus url') -except ValueError: - pass -else: - verify(0) - -# XXX Name hacking to get this to work on Windows. -fname = os.path.abspath(urllib2.__file__).replace('\\', '/') -if fname[1:2] == ":": - fname = fname[2:] -# And more hacking to get it to work on MacOS. This assumes -# urllib.pathname2url works, unfortunately... -if os.name == 'mac': - fname = '/' + fname.replace(':', '/') -elif os.name == 'riscos': - import string - fname = os.expand(fname) - fname = fname.translate(string.maketrans("/.", "./")) - -file_url = "file://%s" % fname -f = urllib2.urlopen(file_url) - -buf = f.read() -f.close() +if __name__ == "__main__": + test_main(verbose=True) diff --git a/Lib/urllib2.py b/Lib/urllib2.py index 5c90aea..3fbb5e3 100644 --- a/Lib/urllib2.py +++ b/Lib/urllib2.py @@ -105,6 +105,7 @@ import socket import sys import time import urlparse +import bisect try: from cStringIO import StringIO @@ -192,6 +193,7 @@ class Request: self.headers = {} for key, value in headers.items(): self.add_header(key, value) + self.unredirected_hdrs = {} def __getattr__(self, attr): # XXX this is a fallback mechanism to guard against these @@ -248,6 +250,15 @@ class Request: # useful for something like authentication self.headers[key.capitalize()] = val + def add_unredirected_header(self, key, val): + # will not be added to a redirected request + self.unredirected_hdrs[key.capitalize()] = val + + def has_header(self, header_name): + return bool(header_name in self.headers or + header_name in self.unredirected_hdrs) + + class OpenerDirector: def __init__(self): server_version = "Python-urllib/%s" % __version__ @@ -256,40 +267,44 @@ class OpenerDirector: self.handlers = [] self.handle_open = {} self.handle_error = {} + self.process_response = {} + self.process_request = {} def add_handler(self, handler): - added = 0 + added = False for meth in dir(handler): - if meth[-5:] == '_open': - protocol = meth[:-5] - if protocol in self.handle_open: - self.handle_open[protocol].append(handler) - self.handle_open[protocol].sort() - else: - self.handle_open[protocol] = [handler] - added = 1 - continue - i = meth.find('_') - j = meth[i+1:].find('_') + i + 1 - if j != -1 and meth[i+1:j] == 'error': - proto = meth[:i] + i = meth.find("_") + protocol = meth[:i] + condition = meth[i+1:] + + if condition.startswith("error"): + j = meth[i+1:].find("_") + i + 1 kind = meth[j+1:] try: kind = int(kind) except ValueError: pass - dict = self.handle_error.get(proto, {}) - if kind in dict: - dict[kind].append(handler) - dict[kind].sort() - else: - dict[kind] = [handler] - self.handle_error[proto] = dict - added = 1 + lookup = self.handle_error.get(protocol, {}) + self.handle_error[protocol] = lookup + elif condition == "open": + kind = protocol + lookup = getattr(self, "handle_"+condition) + elif condition in ["response", "request"]: + kind = protocol + lookup = getattr(self, "process_"+condition) + else: continue + + handlers = lookup.setdefault(kind, []) + if handlers: + bisect.insort(handlers, handler) + else: + handlers.append(handler) + added = True + if added: - self.handlers.append(handler) - self.handlers.sort() + # XXX why does self.handlers need to be sorted? + bisect.insort(self.handlers, handler) handler.add_parent(self) def __del__(self): @@ -320,13 +335,32 @@ class OpenerDirector: if data is not None: req.add_data(data) + protocol = req.get_type() + + # pre-process request + meth_name = protocol+"_request" + for processor in self.process_request.get(protocol, []): + meth = getattr(processor, meth_name) + req = meth(req) + + response = self._open(req, data) + + # post-process response + meth_name = protocol+"_response" + for processor in self.process_response.get(protocol, []): + meth = getattr(processor, meth_name) + response = meth(req, response) + + return response + + def _open(self, req, data=None): result = self._call_chain(self.handle_open, 'default', 'default_open', req) if result: return result - type_ = req.get_type() - result = self._call_chain(self.handle_open, type_, type_ + \ + protocol = req.get_type() + result = self._call_chain(self.handle_open, protocol, protocol + '_open', req) if result: return result @@ -339,7 +373,7 @@ class OpenerDirector: # XXX http[s] protocols are special-cased dict = self.handle_error['http'] # https is not different than http proto = args[2] # YUCK! - meth_name = 'http_error_%d' % proto + meth_name = 'http_error_%s' % proto http_err = 1 orig_args = args else: @@ -372,7 +406,7 @@ def build_opener(*handlers): opener = OpenerDirector() default_classes = [ProxyHandler, UnknownHandler, HTTPHandler, HTTPDefaultErrorHandler, HTTPRedirectHandler, - FTPHandler, FileHandler] + FTPHandler, FileHandler, HTTPErrorProcessor] if hasattr(httplib, 'HTTPS'): default_classes.append(HTTPSHandler) skip = [] @@ -400,8 +434,10 @@ class BaseHandler: def add_parent(self, parent): self.parent = parent + def close(self): self.parent = None + def __lt__(self, other): if not hasattr(other, "handler_order"): # Try to preserve the old behavior of having custom classes @@ -411,11 +447,29 @@ class BaseHandler: return self.handler_order < other.handler_order +class HTTPErrorProcessor(BaseHandler): + """Process HTTP error responses.""" + handler_order = 1000 # after all other processing + + def http_response(self, request, response): + code, msg, hdrs = response.code, response.msg, response.info() + + if code != 200: + response = self.parent.error( + 'http', request, response, code, msg, hdrs) + + return response + + https_response = http_response + class HTTPDefaultErrorHandler(BaseHandler): def http_error_default(self, req, fp, code, msg, hdrs): raise HTTPError(req.get_full_url(), code, msg, hdrs, fp) class HTTPRedirectHandler(BaseHandler): + # maximum number of redirections before assuming we're in a loop + max_redirections = 10 + def redirect_request(self, req, fp, code, msg, headers, newurl): """Return a Request or None in response to a redirect. @@ -459,14 +513,20 @@ class HTTPRedirectHandler(BaseHandler): return # loop detection - new.error_302_dict = {} - if hasattr(req, 'error_302_dict'): - if len(req.error_302_dict)>10 or \ - newurl in req.error_302_dict: + # .redirect_dict has a key (url, code) if url was previously + # visited as a result of a redirection with that code. The + # code is needed in addition to the URL because visiting a URL + # twice isn't necessarily a loop: there is more than one way + # to redirect (301, 302, 303, 307, refresh). + key = (newurl, code) + if hasattr(req, 'redirect_dict'): + visited = new.redirect_dict = req.redirect_dict + if key in visited or len(visited) >= self.max_redirections: raise HTTPError(req.get_full_url(), code, self.inf_msg + msg, headers, fp) - new.error_302_dict.update(req.error_302_dict) - new.error_302_dict[newurl] = newurl + else: + visited = new.redirect_dict = req.redirect_dict = {} + visited[key] = None # Don't close the fp until we are sure that we won't use it # with HTTPError. @@ -853,6 +913,38 @@ class ProxyDigestAuthHandler(BaseHandler, AbstractDigestAuthHandler): class AbstractHTTPHandler(BaseHandler): + def __init__(self, debuglevel=0): + self._debuglevel = debuglevel + + def set_http_debuglevel(self, level): + self._debuglevel = level + + def do_request(self, request): + host = request.get_host() + if not host: + raise URLError('no host given') + + if request.has_data(): # POST + data = request.get_data() + if not request.has_header('Content-type'): + request.add_unredirected_header( + 'Content-type', + 'application/x-www-form-urlencoded') + if not request.has_header('Content-length'): + request.add_unredirected_header( + 'Content-length', '%d' % len(data)) + + scheme, sel = splittype(request.get_selector()) + sel_host, sel_path = splithost(sel) + if not request.has_header('Host'): + request.add_unredirected_header('Host', sel_host or host) + for name, value in self.parent.addheaders: + name = name.capitalize() + if not request.has_header(name): + request.add_unredirected_header(name, value) + + return request + # XXX Should rewrite do_open() to use the new httplib interface, # would be a little simpler. @@ -862,26 +954,13 @@ class AbstractHTTPHandler(BaseHandler): raise URLError('no host given') h = http_class(host) # will parse host:port - if req.has_data(): - data = req.get_data() - h.putrequest('POST', req.get_selector()) - if not 'Content-type' in req.headers: - h.putheader('Content-type', - 'application/x-www-form-urlencoded') - if not 'Content-length' in req.headers: - h.putheader('Content-length', '%d' % len(data)) - else: - h.putrequest('GET', req.get_selector()) + h.set_debuglevel(self._debuglevel) - scheme, sel = splittype(req.get_selector()) - sel_host, sel_path = splithost(sel) - h.putheader('Host', sel_host or host) - for name, value in self.parent.addheaders: - name = name.capitalize() - if name not in req.headers: - h.putheader(name, value) + h.putrequest(req.get_method(), req.get_selector()) for k, v in req.headers.items(): h.putheader(k, v) + for k, v in req.unredirected_hdrs.items(): + h.putheader(k, v) # httplib will attempt to connect() here. be prepared # to convert a socket error to a URLError. try: @@ -889,14 +968,15 @@ class AbstractHTTPHandler(BaseHandler): except socket.error, err: raise URLError(err) if req.has_data(): - h.send(data) + h.send(req.get_data()) code, msg, hdrs = h.getreply() fp = h.getfile() - if code == 200: - return addinfourl(fp, hdrs, req.get_full_url()) - else: - return self.parent.error('http', req, fp, code, msg, hdrs) + response = addinfourl(fp, hdrs, req.get_full_url()) + # XXXX should these be methods, for uniformity with rest of interface? + response.code = code + response.msg = msg + return response class HTTPHandler(AbstractHTTPHandler): @@ -904,6 +984,7 @@ class HTTPHandler(AbstractHTTPHandler): def http_open(self, req): return self.do_open(httplib.HTTP, req) + http_request = AbstractHTTPHandler.do_request if hasattr(httplib, 'HTTPS'): class HTTPSHandler(AbstractHTTPHandler): @@ -911,6 +992,7 @@ if hasattr(httplib, 'HTTPS'): def https_open(self, req): return self.do_open(httplib.HTTPS, req) + https_request = AbstractHTTPHandler.do_request class UnknownHandler(BaseHandler): def unknown_open(self, req): -- cgit v0.12