Registration framework maintained on bitbucket

The django-registration framework is now maintained on bitbucket, which
requires some changes to the in-tree symlink and the documentation.

Signed-off-by: martin f. krafft <madduck@madduck.net>
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

tools: Update license for post-receive hook.

Martin has allowed distribution under the GPL.
 http://lists.ozlabs.org/pipermail/patchwork/2010-June/000263.html

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

tools: Add a simple sample Git post-receive hook

This patch adds a post-receive hook to lib/git, which can be used to
update Patchwork following a push to the Git repository.

Surely, it can be improved. One thing to do would be to export the state
map to git-config somehow.

Signed-off-by: martin f. krafft <madduck@madduck.net>
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

tools: De-hyphenate Git commands in update-patchwork-status.py

Even though update-patchwork-status.py doesn't really do anything yet,
it uses the ancient way to call Git commands, so this patch fixes that.

Signed-off-by: martin f. krafft <madduck@madduck.net>
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

templates/patch-list: add default text for patches with no name

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

views/bundle/public: Lookup on user

Currently, the public bundle view is only looking up bundles by their
name. This may return multiple bundles if two users have bundles with
the same name.

This change adds the user to the bundle lookup.

Reported-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Add TODO item: changing primary email addresses

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

views: Use request scheme in sample pwclientrc file.

Adapt the pwclientrc file to suit the current scheme in use.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

pwclient: allow authenticated https connections

If the xmlrpc server is using HTTPS, we need to adapt our custom
transport to use SafeTransport.make_connection

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Tested-by: Paul Mundt <lethal@linux-sh.org>

pwclient get: fix for handling messages encoded in UTF-8

Trying `pwclient get` with a message containing UTF-8 characters you
would get the following error.

    $ python -v pwclient get ID
    [...]
    # trying /usr/lib/python2.5/encodings/ascii.so
    # trying /usr/lib/python2.5/encodings/asciimodule.so
    # trying /usr/lib/python2.5/encodings/ascii.py
    # /usr/lib/python2.5/encodings/ascii.pyc matches /usr/lib/python2.5/encodings/ascii.py
    import encodings.ascii # precompiled from /usr/lib/python2.5/encodings/ascii.pyc
    Failed to write to subject.patch
    [...]

You can reproduce it with

    $ pwclient get http://patchwork.openembedded.org/patch/1440/

and the fix is the same as in 4fc6a16910e595401e919abde28cca436193e0dd
by Mauro Carvalho Chehab.

Signed-off-by: Paul Menzel <paulepanter@users.sourceforge.net>

pwclient: fallback to patch_get_by_hash

patch_get_by_project_hash isn't in older xmlrpc interfaces, so
fall back to patch_get_by_hash if we see an xmlrpc fault when
calling this function.

Reported-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006@gmx.net>
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Adjust postgres "grant all" script to new name of bundle table

Since c105cb0 (Initial bundle reordering support, 2009-01-29), the
patchwork_bundle_patches table and its associated id sequence does not
exist.  The script to grant appropriate privileges was not updated when
this change was made, and causes the set-up procedure described in
docs/INSTALL to fail.

This change adjust the script to grant appropriate access rights to
match the change in the DB schema.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

models: Add Reported-By to parsed comment fields

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

docs: Fix typo in INSTALL

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

pwclient,xmlrpc: Specify project for patch hash lookups

We may see the same hash for two patches if one is sent to two lists
tracked by patchwork.

This change adds a new xmlrpc method, patch_get_by_project_hash, which
adds the linkname parameter to the lookup. We need to add a new method
instead of replacing patch_get_by_hash, to avoid breaking existing
pwclient apps.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

pwclient: Add newline to 'no hash' message

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Copy headers from original mail into patch mbox output

Preserve the Cc and To headers in the mbox output by parsing the saved
original headers, and adding them to the mail object.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Use psycopg2 as default DB engine

As the 'postgres' engine seems to have been deprecated, use the
postgres_psycopg2 engine instead

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Add pwclient items to TODO

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Add Message-Id to xmlrpc patch_get_mbox headers

While trying to utilize patchwork for my daily workflow I noticed that
the xmlrpc patch_get_mbox function does not add the original message
id to the headers. That forces me to look it up with the browser which
is a major PITA.

Add Message-Id to the headers.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

pwclient fix for handling messages with UTF-8

pwclient works fine viewing utf-8 messages, if you don't try to redirect stdout.

However, if you do things like pwclient view 5063 >/tmp/patch, it fails
due to the weird character handling logic in python:

Traceback (most recent call last):
  File "/usr/local/bin/pwclient", line 446, in ?
    main()
  File "/usr/local/bin/pwclient", line 411, in main
    print s
UnicodeEncodeError: 'ascii' codec can't encode character u'\xe1' in position 249: ordinal not in range(128)

Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

xmlrpc: Remove duplicated "name" field

Remove duplicated "name" field from the list of allowed
fields for a patch list filter in the XML RPC interface.

Signed-off-by: Abhishek Kulkarni <adkulkar@umail.iu.edu>
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

tests: check some edge cases in bundle reordering tests

While testing BundleReorderTest, I felt like adding more test cases to
check if reordering is touching only the right patches on the bundle.

This increases the number of patches on the test bundle to 5, and
changes the code to do the following tests:

- Reorder all patches
- Reorder only the first 3 patches
- Reorder only the last 3 patches
- Reorder only the middle 3 patches
- On all above tests, check if the 'order' field on the BundlePatch
  objects is still consistent

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

tests: Fix 'order_start' on BundleReorderTest unit test

The 'order_start' parameter on the reorder form should be a patch ID,
not the 'order' field of BundlePatch.

This fixes the following test failure:

> ======================================================================
> FAIL: testBundleReorder (patchwork.tests.bundles.BundleReorderTest)
> ----------------------------------------------------------------------
> Traceback (most recent call last):
>   File "/mnt/common/code/admin/rh-patchwork/apps/patchwork/tests/bundles.py", line 358, in testBundleReorder
>     self.failUnlessEqual(response.status_code, 200)
> AssertionError: 404 != 200
>
> ----------------------------------------------------------------------
> Ran 123 tests in 4.527s
>
> FAILED (failures=1)

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

parser: allow short-form List-Id headers

Some lists (eg, nongnu.org) have short-form List-Id headers. For
example:

List-Id: qemu-devel.nongnu.org

Although RFC 2919 prescribes that:

 list-id-header = "List-ID:" [phrase] "<" list-id ">" CRLF

we should allow for these headers anyway.

Based on an original patch from "J.H." <warthog9@kernel.org>.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

docs: add note about -pass-header Authorization' to INSTALL

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Fix unarchive action

We want to un-archive, not re-archive.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Add nacked-by tag

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

views: order bundle correctly

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

templates: close table cell in project maintainer list

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

views: separate project views into separate file

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

docs: remove completed todo items

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

docs: add items to todo

Additions to pwclient.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

tests: test creating bundles with existing names

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

migration: fix whitespace

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

migration: reset bundlepatch sequence on setup

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Add extra patch meta tags (Tested/Reviewed/Nacked-by)

Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Fix non-ascii character encodings on xmlrpc interface

The current xmlrpc interface dies when trying to retrieve patches
that contain non-ascii characters.

This change uses unicode() rather than str().

Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[views] Check count() for duplicate bundle patches, rather than try/except

If the exception is raised, the transaction will still be in a bad
state, due to the foreign key constraint.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[views] Don't create bundles with no name

Check to see that we have a valid name before creating.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[docs] Remove done things from TODO

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[mbox] Add extra newline after patch comment

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[mbox] Put ACKs before '---' updates

Update the Patch's mbox() function to split into '--- update'
sections, allowing us to add Acks and Signoffs in the right place

Add tests too.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[sql] Fix up bundle-reorder migration script

We need to add grants for the new table, and drop the old to
prevent reference constraint errors.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[tests] Add initial bundle tests

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Bundle reordering support

Bundles can now be reordered and saved.

Add dependency on jquery in INSTALL.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[views] Allow Order class to be user-editable

This can be enabled with the editable_order parameter to
generic_list

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[sql] Perserve bundle ordering in bundle reorder migration script

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Normalise order numbers in bundle-ordering migration SQL

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Complete bundle ordering migration script

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Initial bundle reordering support

todo:
 * implement ajax reorder api
 * finish migration script
 * order patches in bundle view

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Add per-user default filters to TODO

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[models] use case-insensitve lookup when creating confirmation

We were getting multiple entries when the case-sensitive lookup
failed.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Add TODO

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Use 'from' import for urlresolvers

Some python installs can't seem to find the fully-qualified module,
so use the 'from'-style import instead.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[models] Make patches unique on (msgid, project), not just (msgid)

On patchwork.ozlabs.org, we may see multiple patches for different
projects, but with the same message-id.

We want these patches to show up on both projects, so we need to change
the current UNIQUE contstraint on msgid.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[views] Use standard UserAdmin for admin site

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[parser] Handle special case of "New newline at end of file"

If a file doesn't have a newline at the end, diff will insert a "\ No
newline at end of file" line in the middle of the last hunk. parser.py
was counting this line as part of the hunk line count. This had the side
effect of putting the last line of the hunk in the patch comment, and
truncating out the last line of the diff.

Handle this special case by not including this line in the internal line
counter. This appears to be the only case where diff will insert a
string like this within a hunk.

Signed-off-by: Nate Case <ncase@xes-inc.com>
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[parser] Handle patches with no content charset defined

If we don't have an incoming charset defined, assume utf-8.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[parser] Convert patch content to unicode before parsing

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[packages] Fix bug with django 1.0

We needed an extra to_field_name

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[parser] Accept x-diff patches

We should accept x-diff attachments as well as x-patch.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[packages] update to django 1.0

With a minor code change, and a small workaround.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[docs] Create python and packages directories

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[models] Fix exception on mbox view with non-ascii submitter name

We need to unicode the name instead of str-ing it.

Add test.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[sql] use separate grant-all scripts for postgresql and mysql

Mysql doesn't support granting to multiple tables, and requires a
different username format.

Would be nice to code the permissions somewhere, then generate the
grant statements as required.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[encoding] Don't output patch mbox as quoted-printable

git-am doesn't like quoted-printable, so output mbox files as raw 7- or
8-bit mbox files. This means we have to create a new MIMEText class, so
that the content isn't automatically QP-encoded on __init__().

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[models] Create patch mbox with utf-8 encoding

We were getting exceptions on the mbox view when looking at a
non-ascii patch.

Add test to suit.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[sql] Default data should be in fixtures, not custom SQL

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[tests] Remove print from SenderCorrelationTest

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[tests] Add tests for utf-8 patches

.. which expose a bug in the patch parser, so fix that.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[views] Don't use bundle owner for current user

This made the top auth area show the bundle owner's details, not
the bundle owner's.

Bundle owner can still be accessed through bundle.owner in the template.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[html] Fix apparently-unclosed script tag

Firefox doesn't seem to grok the <script/> notation.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[html] Add select-all checkbox for patch lists

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[html] Reference common.js in base template

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[html] Consolidate js files

We don't need most of the stuff in htdocs/js/. Put the only function
required into a common.js, which we can add other stuff to.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[views] Change default ordering to latest-first

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[css] Separate patchforms from list a little

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[html] Reinstate 'about' link

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[css] Remove horizontal scrollbar

We don't need to specify width on #title.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[views] Restructure profile view, simplify bundle access

Make bundles more like todo lists - the list itself has its own page,
accessible from the top user links.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[views] Fix empty bundle showing entire project's patches

If a bundle is empty, we end up triggering the 'if not patches' check
in generic_list(), which repopulates the patch list with the current
project's patches.

So, check for None instead, and leave the patch list empty.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Add comment signoffs and acks to mbox text

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[test] Move helper functions into tests/utils.py

So that we can use them for other tests

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[models] Improve comment filter

We should be able to search on the patch_id faster.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[parser] Fix spacing for encoded headers

We need to join the fragments with a space.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[parser] Merge senders with different case

... and add tests

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[test] 2.4-ise test cases

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Add bin/__init__.py

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[parser] Decode From: headers

We're getting a few utf-8 encoded From: addresses, so decode before
saving to the DB.

Also, add tests.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[parser] Don't remove --- update lines

We'd like to keep update lines in the patch comments, so change the
signature-removal code to allow them to pass through.

Also, add appropriate tests.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[test] Add test for signatures in comments

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

[test] Add patchwork testing infrastructure

Add a small set of initial tests for the patch parsing code.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Remove superfluous print from parsemail.py

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Fix handling of newlines in subjects

By doing two passes of whitespace normalisation

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Drop project.linkname from patch subject lines

.. otherwise every cell patch will be prefixed with [Cbe-oss-dev]

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Improve [PATCH] subject parsing

We'd like to leave some fragments of [PATCH]-style headers in the
patch subject, so add functionality to do this in clean_subject

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Add 404 template

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Fix 'activate' mispelling

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Fix mishandling of existing querystring parameters

Currently, we only get the first character of order= parameters
in the 'remove this filter' querystring.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>