- added legal_file::sanitize_dirname(), analogous to sanitize_filename.
- revised item model to use new function when adding or updating an album.
- added some legal_file unit tests.
- revised some item model unit tests.
- disallowed backslashes in item validation.
- refactored the validation logic in the item model a bit.
- added no_backslash error messages in edit album/photo/movie forms.
- fixed error messages in add album forum (some missing, some text different from edit)
- added unit tests
- updated to v58 to correct any existing backslashes in item names
the state of the world. These were uncovered when we ran the tests on
Travis which runs the tests in a different order than what I (and
apparently others) use on their dev boxes.
- legal_file - added sanitize_filname() to sanitize photo/movie filenames.
- admin_watermarks - revised add() to use new function.
- item model - added _process_data_file_info() to validate the data file, get its metadata, and sanitize the item name.
- item model - revised save() for new items to use _process_data_file_info *before* the slug is checked.
- item model - revised save() for updated items to use _process_data_file_info.
- item model - revised save() for updated items to sanitize name if changed.
- uploader - removed call to smash_extensions (item model does this when it calls sanitize_filename).
- Legal_File_Helper_Test - added unit tests for sanitize_filename.
- Item_Model_Test - revised existing unit tests based on changes.
- Item_Model_Test - added new unit tests for names with legal but incorrect extensions.
- Averted take over by HAL with fix #2001...
- photo & movie helpers: modified to throw exceptions when file is known to be unidentifiable/illegal.
- item model: revised to work with exceptions and be more explicit when the data file is invalid.
- item model: removed duplicate get_file_metadata call for updated items.
- admin_watermarks controller: revised to work with exceptions (really cleans up logic here).
- graphics helper: revised to handle invalid placeholders (a nearly-impossible corner case, but still...).
- photo & movie helper tests: revised to work with exceptions, added new tests for illegal files with valid extensions.
- item model tests: revised to work with exceptions, added new tests for illegal files with valid extensions.
- Add extra condition to Item_Model_Test::urls_test to test cache busters of missing files.
- Previous commit fixes unit test for empty album url, but now no test checks missing files.
- Reduced from four places that alter names/slugs to two (one to populate empty slugs, one to handle conflicting names/slugs).
- For empty slugs, fill with generic human-readable name (e.g. "photo") instead of random value.
- For conflicting names/slugs, add suffix that's sequential (e.g. "-01"), only using random after 99 conflicts.
- Made conflict-finding code check filenames with no extensions.
- Renamed _randomize_name_or_slug_on_conflict to _check_and_fix_conflicts.
- Added unit tests. Also cleaned up existing unit tests to reflect new logic and removed a redundant test.
- Added installer logic to correct existing items now considered in conflict.
- Revised gallery_task to look for duplicate names based on new criteria.
- Added test::random_unique_photo and test::random_unique_photo_unsaved to uniquify test photos.
- Uniquified the black dot of test.jpg by coloring it with the six-digit hex code already used to name the random photos (e.g. "name_a48801.jpg").
- Modified four tests in Item_Model_Test that check photo file contents to use new functions to guarantee uniqueness.
Squashed commit of the following:
commit 945316a8c220b12adb687c896bcc5e86f99f46a4
Author: Chad Parry <github@chad.parry.org>
Date: Thu Jul 21 01:11:13 2011 -0600
Add a test for the sunny-day scenario where a rule changes a data file's MIME type.
commit 4ee1ee000c8f4d8ebaae66f637bc71080486fd73
Author: Chad Parry <github@chad.parry.org>
Date: Thu Jul 21 00:49:47 2011 -0600
Ensure that a third-party cannot swap out a legitimate photo with an unsafe file type.
commit 7dd0105bfc59c150e5640e693778f51bbaa44eab
Author: Chad Parry <github@chad.parry.org>
Date: Thu Jul 21 00:48:19 2011 -0600
Update the MIME type and other meta-data when a new data file is provided.
commit 5a8844c7947b21cf658f22cc61f20ffa9e8f07f2
Author: Chad Parry <github@chad.parry.org>
Date: Thu Jul 21 00:30:01 2011 -0600
Remove a unit test that no longer applies. Replacement data files are allowed to have different MIME types.
commit 0de9c6283ce4f5773cad8e92b6785d6a1f7b5e46
Author: Chad Parry <github@chad.parry.org>
Date: Thu Jul 21 00:27:45 2011 -0600
If one rule fails, then abort processing, rather than trying to proceed to subsequent rules.
commit 41d379c2b777ae7b3a11f528971228e234f8976f
Author: Chad Parry <github@chad.parry.org>
Date: Thu Jul 21 00:10:10 2011 -0600
Replace an overly-complicated regular expression with a simple in_array, at Bharat's suggestion.
commit 1b3f7111d4c2607baaa2da0aab3b501f2d9a1426
Merge: 8f7904a 403f64b
Author: Chad Parry <github@chad.parry.org>
Date: Wed Jul 20 21:02:56 2011 -0600
Merge branch 'master' into rawphoto
commit 8f7904ab62c71a7e4ee68762f936030b4dcb4ea1
Merge: e950573 771de0a
Author: Chad Parry <github@chad.parry.org>
Date: Sat Jun 25 14:12:39 2011 -0600
Merge branches 'master' and 'rawphoto' into rawphoto
commit e95057337996351e49915d9f85d007d50103a4be
Author: Chad Parry <github@chad.parry.org>
Date: Wed Jun 15 20:24:18 2011 -0600
Merge branches 'rawphoto-squash' and 'rawphoto' into rawphoto
random::hash()
random::string()
random::percent()
random::int()
So that we don't have lots of different ways to get random values all
over the code. Follow-on to #1527.
Refactor the rotate code in Quick_Controller to replace the data file,
and then have gallery_event::item_updated_data_file() pick up after
the change is saved, rebuild the image and handle album covers. This
is much more portable than before and it will allow any mechanism (eg:
REST) to replace the source image.
not capturing all cases for setting and resetting $original, which
leads to some weird and hard to reproduce behavior.
Instead, if we need the original just reload it from the database.
This may result in a somewhat excessive load in places, but we'll have
to fix that in a later optimization pass.
