aboutsummaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeLines
* Token remover: ignore DMsGravatar MarkKoz2020-06-15-0/+13
| | | | | It's a private channel so there's no risk of a token "leaking". Furthermore, messages cannot be deleted in DMs.
* Add the C# guild to the whitelistGravatar Dennis Pham2020-06-15-0/+1
|
* Merge pull request #937 from python-discord/bug/filters/928/non-ascii-tokenGravatar Mark2020-06-12-154/+357
|\ | | | | Make token detection more robust and completely rewrite its tests
| * Merge remote-tracking branch 'origin/master' into ↵Gravatar MarkKoz2020-06-02-1044/+2456
| |\ | | | | | | | | | bug/filters/928/non-ascii-token
| * | Token remover: remove the `delete_message` functionGravatar MarkKoz2020-06-02-19/+9
| | | | | | | | | | | | | | | It's redundant; there's no benefit here in abstracting two lines of code into a function.
| * | Test find_token_in_message returns None for invalid matchesGravatar MarkKoz2020-05-28-0/+15
| | | | | | | | | | | | | | | This covers the case when a token is matched, but its user ID and timestamp turn out to be invalid.
| * | Adjust find_token_in_message tests for the recent cog changesGravatar MarkKoz2020-05-28-19/+20
| | | | | | | | | | | | | | | | | | | | | | | | | | | It now supports the changes that switched to finditer, added match groups, and added the Token NamedTuple. It also accounts for the is_maybe_token function being removed. For the sake of simplicity, call assertions on is_valid_user_id and is_valid_timestamp were not made.
| * | Remove is_maybe_token testsGravatar MarkKoz2020-05-28-33/+0
| | | | | | | | | | | | | | | The function was removed due to redundancy. Therefore, its tests are obsolete.
| * | Add more thorough and realistic inputs for token ID and timestamp testsGravatar MarkKoz2020-05-27-18/+52
| | | | | | | | | | | | | | | The tests for valid inputs and invalid inputs were split to make them more readable.
| * | Adjust token remover tests to use the Token NamedTupleGravatar MarkKoz2020-05-27-6/+8
| | |
| * | Switch findall to finditer in assertionsGravatar MarkKoz2020-05-27-4/+4
| | | | | | | | | | | | | | | `find_token_in_message` now uses the latter so the tests should adjust accordingly.
| * | Use real token values for testing multiple matches in regexGravatar MarkKoz2020-05-25-3/+4
| | |
| * | Fix multiple match text for token regexGravatar MarkKoz2020-05-25-2/+3
| | | | | | | | | | | | | | | | | | | | | It has to account for the addition of groups. It's easiest to compare the entire string so `finditer` is used to return re.Match objects; the tuples of `findall` would be cumbersome. Also threw in a change to use `assertCountEqual` cause the order doesn't really matter.
| * | Fix valid token regex testGravatar MarkKoz2020-05-25-2/+2
| | | | | | | | | | | | | | | | | | | | | | | | It was broken due to the addition of groups. Rather than returning the full match, `findall` returns groups if any exist. The test was comparing a tuple of groups to the token string, which was of course failing. Now `fullmatch` is used cause it's simpler - just check for `None` and don't worry about iterating matches to search.
| * | Add more valid tokens to test the regex withGravatar MarkKoz2020-05-25-3/+5
| | |
| * | Test token regex won't match non-base64 charactersGravatar MarkKoz2020-05-25-4/+3
| | |
| * | Token remover: specify Discord epoch in secondsGravatar MarkKoz2020-05-25-1/+1
| | | | | | | | | | | | | | | | | | The timestamp in the token is in seconds and is being compared against the epoch. To make life easier, they should use the same unit. Previously, the epoch was in milliseconds.
| * | Token remover: use finditer instead of findallGravatar MarkKoz2020-05-25-4/+3
| | | | | | | | | | | | | | | It makes more sense to use the lazy function when the loop is already short-circuiting on the first valid token it finds.
| * | Token remover: use regex groups and pass the token as a NamedTupleGravatar MarkKoz2020-05-25-27/+20
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | It felt redundant to be splitting the token in two different functions when regex could take care of this from the outset. ' A NamedTuple was created to house the token. This is nicer than passing an re.Match object, because it's clearer which attributes are available. Even if the regex used named groups, it wouldn't be as obvious which group names exist. Without the split, `is_maybe_token` is dwindled down to a redundant function. Therefore, it's been removed.
| * | Token remover: escape dashes in regexGravatar MarkKoz2020-05-23-1/+1
| | | | | | | | | | | | | | | They need to be escaped when they're in a character set. By default, they are interpreted as part of the character range syntax.
| * | Token remover: match only base64 in regexGravatar MarkKoz2020-05-21-7/+6
| | | | | | | | | | | | | | | | | | | | | Making the regex more accurate reduces false positives at an earlier stage. There's no benefit to matching non-base64 as that would just be weeded out as invalid at a later stage anyway when it tries to decode it.
| * | Token remover: decode ID using URL-safe base64Gravatar MarkKoz2020-05-21-1/+1
| | | | | | | | | | | | | | | Though I've not seen an ID with neither + and \ nor - and _, given that the timestamp uses URL-safe encoding, the ID probably does too.
| * | Add a utility function to pad base64 dataGravatar MarkKoz2020-05-16-2/+8
| | |
| * | Token remover: use strict check for digits in token IDGravatar MarkKoz2020-05-15-2/+5
| | | | | | | | | | | | | | | | | | | | | `isnumeric` would be true for a wide range of characters in Unicode, but the ID must only consist of the characters 0-9 (ASCII digits). In fact, `isdigit` on its own would also match other Unicode characters too.
| * | Token remover: fix timestamp checkGravatar MarkKoz2020-05-14-10/+16
| | | | | | | | | | | | | | | | | | The timestamp calculation was incorrect. The bytes need to be interpreted as big-endian and the result is just a timestamp rather than a snowflake.
| * | Token remover: add logs to clarify why token is invalidGravatar MarkKoz2020-05-13-2/+4
| | |
| * | Add missing comma to token remover log messageGravatar MarkKoz2020-05-13-1/+1
| | |
| * | Fix a test needlessly being a coroutineGravatar MarkKoz2020-05-13-1/+1
| | |
| * | Use subtests for valid ID/timestamp tests and test non-ASCII inputsGravatar MarkKoz2020-05-11-18/+25
| | |
| * | Clean up token remover test importsGravatar MarkKoz2020-05-11-16/+12
| | |
| * | Replace deprecated assertion methodsGravatar MarkKoz2020-05-11-2/+2
| | |
| * | Refactor `TokenRemoverSetupTests` and add a more thorough testGravatar MarkKoz2020-05-11-4/+8
| | | | | | | | | | | | | | | The test now ensures the cog is instantiated and that the instance is passed as an argument to `add_cog`.
| * | Test TokenRemover.take_actionGravatar MarkKoz2020-05-11-43/+30
| | | | | | | | | | | | | | | | | | | | | | | | * Remove `bot.get_cog` mocks in `setUp` * Mock the logger cause it's easier to assert logs * Remove subtests * Assert helper functions were called * Create an autospec for ModLog
| * | Test TokenRemover.format_log_messageGravatar MarkKoz2020-05-11-0/+16
| | |
| * | Test token remover's message deletionGravatar MarkKoz2020-05-11-0/+9
| | |
| * | Simplify token remover's message mockGravatar MarkKoz2020-05-11-9/+3
| | | | | | | | | | | | | | | * Rely on default values for the author * Set the content to a non-empty string
| * | Avoid instantiating the cog when testing static/class methodsGravatar MarkKoz2020-05-11-10/+5
| | |
| * | Token remover: use a string template for the log messageGravatar MarkKoz2020-05-11-4/+11
| | |
| * | Token remover: split some of `take_action` into separate functionsGravatar MarkKoz2020-05-11-11/+21
| | |
| * | Correct the return type annotation for the autospec decoratorGravatar MarkKoz2020-05-11-2/+2
| | |
| * | Test token regex matches valid tokensGravatar MarkKoz2020-05-11-0/+21
| | |
| * | Test is_maybe_tokenGravatar MarkKoz2020-05-11-7/+24
| | |
| * | Token remover: fix `is_maybe_token` returning None instead of FalseGravatar MarkKoz2020-05-11-0/+2
| | | | | | | | | | | | | | | It's annotated as returning a bool and when the split fails it already returns False. To be consistent, it should always return a bool.
| * | Test `is_maybe_token` returns False for missing partsGravatar MarkKoz2020-05-11-0/+10
| | | | | | | | | | | | | | | | | | | | | In practice, this won't ever happen since the regex wouldn't match strings with missing parts. However, the function does check it so may as well test it. It's not necessarily bound to always use inputs from the regex either I suppose.
| * | Fix autospec decorator when used with multiple attributesGravatar MarkKoz2020-05-11-16/+12
| | | | | | | | | | | | | | | | | | | | | | | | The original approach of messing with the `attribute_name` didn't work for reasons I won't discuss here (would require knowledge of patcher internals). The new approach doesn't use patch.multiple but mimics it by applying multiple patch decorators to the function. As a consequence, this can no longer be used as a context manager.
| * | Test token regex doesn't match invalid tokensGravatar MarkKoz2020-05-11-7/+25
| | |
| * | Test `find_token_in_message` returns the found tokenGravatar MarkKoz2020-05-11-0/+24
| | |
| * | Test `find_token_in_message` returns None if no matches foundGravatar MarkKoz2020-05-11-0/+14
| | |
| * | Fix test for token remover ignoring bot messagesGravatar MarkKoz2020-05-11-4/+9
| | | | | | | | | | | | | | | | | | It's not possible to test this via asserting the return value of `on_message` since it never returns anything. Instead, the actual relevant unit, `find_token_in_message,` should be tested.
| * | Allow using arbitrary parameter names with the autospec decoratorGravatar MarkKoz2020-05-11-2/+15
| | | | | | | | | | | | | | | This gives the caller more flexibility. Sometimes attribute names are too long or they don't follow a naming scheme accepted by the linter.
generated by cgit v1.2.3 (git 2.39.1) at 2025-10-27 03:01:46 +0000