| msg77405 - (view) |
Author: Jakub Wilk (jwilk) |
Date: 2008-12-09 12:13 |
| $ cat urltest2.5 #!/usr/bin/python2.5 from urllib2 import urlopen for line in urlopen('http://python.org/'): print line break $ ./urltest2.5 $ cat urltest3.0 #!/usr/bin/python3.0 from urllib.request import urlopen for line in urlopen('http://python.org/'): print(line) break $ ./urltest3.0 Traceback (most recent call last): File "./urltest3.0", line 3, in for line in urlopen('http://python.org/'): TypeError: 'addinfourl' object is not iterable |
|
|
| msg77409 - (view) |
Author: Senthil Kumaran (orsenthil) *  |
Date: 2008-12-09 12:50 |
| I verified this bug in the Py3.0 and Py3.1. Shall come out with a patch for it. |
|
|
| msg77415 - (view) |
Author: Jeremy Hylton (jhylton)  |
Date: 2008-12-09 13:48 |
| Oops. I didn't think it translate the code in addinfobase to the new style of iterators. Jeremy On Tue, Dec 9, 2008 at 7:50 AM, Senthil <report@bugs.python.org> wrote: > > Senthil <orsenthil@gmail.com> added the comment: > > I verified this bug in the Py3.0 and Py3.1. Shall come out with a patch > for it. > > ---------- > nosy: +orsenthil > > _______________________________________ > Python tracker <report@bugs.python.org> > <http://bugs.python.org/issue4608> > _______________________________________ > _______________________________________________ > Python-bugs-list mailing list > Unsubscribe: http://mail.python.org/mailman/options/python-bugs-list/jeremy%40alum.mit.edu > > |
|
|
| msg78139 - (view) |
Author: Senthil Kumaran (orsenthil) * |
Date: 2008-12-21 05:47 |
| Here is a patch to fix the issue. Jeremy, is it approach okay? Or do you have any other suggestion? |
|
|
| msg78416 - (view) |
Author: Jakub Wilk (jwilk) |
Date: 2008-12-28 16:03 |
| Regarding Senthil's patch: __next__() method seems superfluous to me (and the implementation is buggy). |
|
|
| msg78880 - (view) |
Author: Senthil Kumaran (orsenthil) * |
Date: 2009-01-02 21:30 |
| Jakub, I have attached a revision to the patch. You are right, when __iter__ returns self.fp (as in previous patch), the __next__ is superflous. But, I was thinking of __iter__ returning an instance of addbase, instead of self.fp and in that case __next__ was required. But I see that i had not changed self.fp to self. This is implemented in the similar lines of IOBase class, io.py w.r.t to your other comment, why do you think __next__ implementation is incorrect? Thanks, Senthil |
|
|
| msg78944 - (view) |
Author: Jakub Wilk (jwilk) |
Date: 2009-01-03 10:52 |
| Oops, __next__ is OK. Sorry for the confusion. |
|
|
| msg79575 - (view) |
Author: Facundo Batista (facundobatista) * |
Date: 2009-01-10 21:00 |
| Senthil, do you think you could provide a test case for this? Thank you! |
|
|
| msg81426 - (view) |
Author: Daniel Diniz (ajaksu2) * |
Date: 2009-02-08 23:38 |
| Test cases attached. The second one highlights a bug in the current patch, as it fails to return a line longer than 65475 chars. This behavior doesn't match trunk's. |
|
|
| msg86365 - (view) |
Author: Senthil Kumaran (orsenthil) * |
Date: 2009-04-23 12:32 |
| This issue is already fixed by jeremy at Revision 70815, wherein "The response from an HTTP request is now an HTTPResponse instance instead of an addinfourl() wrapper instance." So the issue won't be present in the py3k code ( confirmed). However, the test added by Daniel,which tests for urlopen() for a request which is b"verylong" * 8192 still fails. It is not just with iteration; but test_200 will fail too if the request is a large chunk. This is only in py3k branch, test will pass in the trunk code. I am investigating further. |
|
|
| msg113245 - (view) |
Author: Brian Brazil (bbrazil) * |
Date: 2010-08-08 09:43 |
| This looks as though its a short write: [pid 28343] recvfrom(5, "GET / HTTP/1.1\r\nAccept-Encoding:"..., 8192, 0, NULL, NULL) = 118 [pid 28343] poll([{fd=5, events=POLLOUT, revents=POLLOUT}], 1, 10000) = 1 [pid 28343] sendto(5, "HTTP/1.0 200 OK\r\n", 17, 0, NULL, 0) = 17 [pid 28343] poll([{fd=5, events=POLLOUT, revents=POLLOUT}], 1, 10000) = 1 [pid 28343] sendto(5, "Server: TestHTTP/ Python/3.2a1+\r"..., 33, 0, NULL, 0) = 33 [pid 28343] poll([{fd=5, events=POLLOUT, revents=POLLOUT}], 1, 10000) = 1 [pid 28343] sendto(5, "Date: Sun, 08 Aug 2010 09:41:08 "..., 37, 0, NULL, 0) = 37 [pid 28343] poll([{fd=5, events=POLLOUT, revents=POLLOUT}], 1, 10000) = 1 [pid 28343] sendto(5, "Content-type: text/plain\r\n", 26, 0, NULL, 0) = 26 [pid 28343] poll([{fd=5, events=POLLOUT, revents=POLLOUT}], 1, 10000) = 1 [pid 28343] sendto(5, "\r\n", 2, 0, NULL, 0) = 2 [pid 28343] poll([{fd=5, events=POLLOUT, revents=POLLOUT}], 1, 10000) = 1 [pid 28343] sendto(5, "verylongverylongverylongverylong"..., 56001, 0, NULL, 0) = 49054 [pid 28343] shutdown(5, 1 /* send */) = 0 |
|
|
| msg113260 - (view) |
Author: Brian Brazil (bbrazil) * |
Date: 2010-08-08 13:23 |
| The attached patch handles short writes, and adds ajaksu2's tests. |
|
|
| msg113280 - (view) |
Author: Florent Xicluna (flox) * |
Date: 2010-08-08 16:27 |
| Thanks, Brian. Pushed with revision 83833. |
|
|
| msg134129 - (view) |
Author: Albert Hopkins (marduk) |
Date: 2011-04-20 07:54 |
| This issue appears to persist when the protocol used is FTP: root@tp-db $ cat test.py from urllib.request import urlopen for line in urlopen('ftp://gentoo.osuosl.org/pub/gentoo/releases/'): print(line) break root@tp-db $ python3.2 test.py Traceback (most recent call last): File "test.py", line 2, in for line in urlopen('ftp://gentoo.osuosl.org/pub/gentoo/releases/'): TypeError: 'addinfourl' object is not iterable |
|
|
| msg134130 - (view) |
Author: Albert Hopkins (marduk) |
Date: 2011-04-20 07:56 |
| Oops, previous example was a directory, but it's the same if the url points to a ftp file. |
|
|
| msg134319 - (view) |
Author: Rafael Zanella (zanella) |
Date: 2011-04-24 00:30 |
| The patch that makes addinfourl() iterable was not commited due to the change to HTTP request see: (http://bugs.python.org/issue4608#msg86365). Since urllib is protocol agnostic it should behave the same with FTP, right? So, where to fix? Change the addinfourl() to become itrable or change the FTPHandler return? |
|
|
| msg139100 - (view) |
Author: Stefan Schwarzer (sschwarzer) |
Date: 2011-06-25 16:37 |
| It turned out that although the addinfourl instance had the `__iter__` attribute in `addbase.__init__` correctly assigned, `__iter__` wasn't found by the `iter` builtin. It seems that `iter` always tries to use the `__iter__` method of the _class_ and doesn't look at the instance. Riccardo Attilio Galli and I made the attached patch. The patch also fixes a corresponding `TypeError` for "file://" URLs, not just "ftp://" URLs. |
|
|
| msg139170 - (view) |
Author: Roundup Robot (python-dev) |
Date: 2011-06-26 12:30 |
| New changeset c0a68b948f5d by Raymond Hettinger in branch '3.2': Issue #4608: urllib.request.urlopen does not return an iterable object http://hg.python.org/cpython/rev/c0a68b948f5d New changeset d4aeeddf72e3 by Raymond Hettinger in branch 'default': Issue #4608: urllib.request.urlopen does not return an iterable object http://hg.python.org/cpython/rev/d4aeeddf72e3 |
|
|
| msg139171 - (view) |
Author: Raymond Hettinger (rhettinger) * |
Date: 2011-06-26 12:31 |
| Thanks for the patch. |
|
|