diff options
Diffstat (limited to 'ChangeLog')
| -rw-r--r-- | ChangeLog | 246 |
1 files changed, 246 insertions, 0 deletions
diff --git a/ChangeLog b/ChangeLog new file mode 100644 index 0000000..644fbdb --- /dev/null +++ b/ChangeLog @@ -0,0 +1,246 @@ +2009-09-25 Seth Vidal <skvidal@fedoraproject.org> + + * urlgrabber/__init__.py: bump version to 3.9.1 + +2009-09-25 Seth Vidal <skvidal@fedoraproject.org> + + * makefile: clean up everything in make clean + +2009-09-25 Seth Vidal <skvidal@fedoraproject.org> + + * test/runtests.py, test/test_grabber.py, test/test_keepalive.py, + urlgrabber/__init__.py, urlgrabber/byterange.py, + urlgrabber/grabber.py, urlgrabber/keepalive.py, + urlgrabber/mirror.py, urlgrabber/progress.py, + urlgrabber/sslfactory.py: cleanup all the old urlgrabber urllib code + that's not being used delete sslfactory and keepalive fix up the + unittests to match the existing code + +2009-09-24 Seth Vidal <skvidal@fedoraproject.org> + + * urlgrabber/grabber.py: update documentation for ssl options and + size/max_header_size options + +2009-09-23 Seth Vidal <skvidal@fedoraproject.org> + + * urlgrabber/grabber.py: - fix the reget testcases (and regets in general) with the max size + check - make the errorcode more obvious when we go over the range - + obviously don't do the check if all of our max values are None (or + even 0 since that is a silly number for a Max) + +2009-09-22 Seth Vidal <skvidal@fedoraproject.org> + + * urlgrabber/grabber.py: handle endless-data problems safely: "A + malicious server could cause libcurl to download an infinite amount + of data, potentially causing all of memory or disk to be filled. + Setting the CURLOPT_MAXFILESIZE_LARGE option is not sufficient to + guard against this. Instead, the app should monitor the amount of + data received within the write or progress callback and abort once + the limit is reached." had to restructure a good bit of the error + handling to do this but it works for both endless headers and + endless content. + +2009-09-21 Seth Vidal <skvidal@fedoraproject.org> + + * urlgrabber/grabber.py: make sure the value we get back from the + parse150 and other calls is converted to an int before we make it + 'size' rhbug: #524705 + +2009-09-02 Seth Vidal <skvidal@fedoraproject.org> + + * urlgrabber/grabber.py: make file:// url not found msgs clearer and + hopefully fix a couple of ctrl-c issues. + +2009-08-27 Seth Vidal <skvidal@fedoraproject.org> + + * urlgrabber/grabber.py: make proxy=_none_ properly disable all + proxies as per the docs + +2009-08-14 Seth Vidal <skvidal@fedoraproject.org> + + * urlgrabber/grabber.py: - add full contingent of ssl options: - client keys - client + certs - capath/cainfo - client key passwords - client key and + cert types - verifypeer/verifyhost - add a number of common errors + to do_perform() - when an error is unknown, and doesn't make sense + report complete pycurl error code - when the filename is '' and not + None and we're doing a urlgrab() try to open the file anyway + rather than silently swallowing the data into a StringIO and + discarding it. + +2009-08-13 Seth Vidal <skvidal@fedoraproject.org> + + * urlgrabber/grabber.py: add _to_utf8() method to pycurlfileobject + make sure postfield data is to_utf8'd before setting the option + otherwise pycurl is unhappy if the postfield data is a unicode + object instead of a string object. closes rh bug + https://bugzilla.redhat.com/show_bug.cgi?id=515797 + +2009-08-12 Seth Vidal <skvidal@fedoraproject.org> + + * urlgrabber/grabber.py: initial pass at setting more advanced ssl + options. verify peer and verify host work as expected. + +2009-08-07 Seth Vidal <skvidal@fedoraproject.org> + + * urlgrabber/grabber.py: keep from making tmpfiles all over /tmp on + any local file:// urlopen() by doing it in StringIO instead of + mkstemp(). Sort of fixes + https://bugzilla.redhat.com/show_bug.cgi?id=516178 + +2009-08-06 Seth Vidal <skvidal@fedoraproject.org> + + * urlgrabber/grabber.py: - fix intrrupt handler and document why keyboardinterrupt is going + to be so weird in pycurl - disable signals and make sure we don't + handle/intercept any in the pycurl code. - set 'check_timestamp' + regets as NotImplemented. The work around is multiple connections. + it is possible but not immediately useful since, afaict, NOTHING + uses the check_timestamp regets. + +2009-08-05 Seth Vidal <skvidal@fedoraproject.org> + + * urlgrabber/grabber.py: - make sure regets work when our filename is unicode - make sure we + are not resetting self.append = False when we don't need to + +2009-08-05 Seth Vidal <skvidal@fedoraproject.org> + + * urlgrabber/grabber.py: - make sure we tell pycurl to get the filetime when downloading - + set a couple of options as 'True/False' instead of 1,0 - for + readability - make sure the option passed to timeout is an int - not + a string + +2009-08-04 Seth Vidal <skvidal@fedoraproject.org> + + * urlgrabber/grabber.py: missed setting the value from opts.timeout + - doesn't really HURT what will happen b/c if your connect takes + longer than 5minutes then you're SCREWED + +2009-08-04 Seth Vidal <skvidal@fedoraproject.org> + + * urlgrabber/grabber.py: handle timeouts more correctly (with the + exception) and set timeouts to be connect timeouts since libcurl + seems to actually honor timeouts - as opposed to urllib. closes rh + bug # 515497 + +2009-07-31 Seth Vidal <skvidal@fedoraproject.org> + + * ChangeLog, makefile, urlgrabber/__init__.py: changelog + release + date touchup + +2009-07-31 Seth Vidal <skvidal@fedoraproject.org> + + * makefile: add a few more things to be cleaned out + +2009-07-31 Seth Vidal <skvidal@fedoraproject.org> + + * ChangeLog: update changelog + +2009-07-31 Seth Vidal <skvidal@fedoraproject.org> + + * urlgrabber/grabber.py: - make readlines() work for mirrorlists in yum (which probably + shouldn't be using it anyway) - do a do_grab() in _do_open() which + may or may not be a good idea - I could also make the _do_grab() + happen when someone attempts to hit a method beyond the file object + open + +2009-07-30 Seth Vidal <skvidal@fedoraproject.org> + + * urlgrabber/grabber.py: - make basic posts work + +2009-07-30 Seth Vidal <skvidal@fedoraproject.org> + + * maint/git2cl: add git2cl + +2009-07-30 Seth Vidal <skvidal@fedoraproject.org> + + * urlgrabber/grabber.py: when I first started this I hacked + something into URLGrabberFileObject - this reverts that hack + +2009-07-30 Seth Vidal <skvidal@fedoraproject.org> + + * ChangeLog, maint/cvs2cl.pl, maint/usermap, test/runtests.py, + urlgrabber/__init__.py: - clean up some unused files - update the changelog - bump the + version - update the copyright in a couple of places + +2009-07-30 Seth Vidal <skvidal@fedoraproject.org> + + * MANIFEST.in, makefile: - make makefile work again without using cvs - add makefile to + MANIFEST.in + +2009-07-30 Seth Vidal <skvidal@fedoraproject.org> + + * urlgrabber/grabber.py: - make simple/most proxies work - remove unnnecessary 'have_range' + check for pycyurl obj + +2009-07-29 Seth Vidal <skvidal@fedoraproject.org> + + * urlgrabber/grabber.py: - add range support - get rid of the .part file thing - it makes + range-regets harder than they need to be - make sure regets behave + +2009-07-29 Seth Vidal <skvidal@fedoraproject.org> + + * urlgrabber/grabber.py: implement throttle/bandwidth controls in + pycurl tested with the progress call back - seems to work very well + + +2009-07-29 Seth Vidal <skvidal@fedoraproject.org> + + * urlgrabber/grabber.py: get the content-length/size for ftp pkgs + too - steals parse150 from ftplib. Should work for A LOT of ftp + servers, but not all of them - add self.scheme for which protocol + we're using here. + +2009-07-29 James Antill <james@and.org> + + * urlgrabber/byterange.py: Import fix for ftp ports in old urilib + code (probably worthless now, but meh) + +2009-07-29 James Antill <james@and.org> + + * urlgrabber/progress.py: Import progress patches from Fedora. + These were done over a couple of years: . cleanup UI. . dynamic + terminal widths. . deal with serial console. . total download + stuff. + +2009-07-28 Seth Vidal <skvidal@fedoraproject.org> + + * test/runtests.py, urlgrabber/grabber.py: implement + PyCurlFileObject. This makes the default and forklifts all the code + to pycurl. This is not finished but is functional for a significant + number of the tests. things known to be broken: - proxies - http + POST - non-header-based byte-ranges - certain types of read + operations when downloading a file to memory instead of to a + filename + +2009-05-15 Seth Vidal <skvidal@fedoraproject.org> + + * urlgrabber/grabber.py: make it use *args instead of silly if + statements + +2009-05-15 Seth Vidal <skvidal@fedoraproject.org> + + * urlgrabber/grabber.py: modify urlgraberror so it has a url + attribute and includes the url in all error messages. + +2006-12-12 mstenner <mstenner> + + * urlgrabber/grabber.py: more debugging code to expose options + +2006-12-08 mstenner <mstenner> + + * scripts/urlgrabber, test/test_grabber.py, urlgrabber/grabber.py, + urlgrabber/keepalive.py: lots of changes... improved clarity of + cached objects, improved debugging and logging, more options to the + urlgrabber script. + +2006-12-07 mstenner <mstenner> + + * scripts/urlgrabber, urlgrabber/grabber.py: Minor doc updates and + error handling in grabber.py. Complete rewrite of the urlgrabber + script. + +2006-12-05 mstenner <mstenner> + + * Minor fix to make byteranges work with some servers. _do_grab now + only reads as much as it needs to, rather than reading until the + server sends EOF. + |