Recent commits to derivepassphrase.git (b65b618949ca77aa0ad481ae2a517d0b469a4d9f)

Fix formatting and linting issues

2025-01-15T20:09:20+01:00

</pre>

Merge topic branch 'l10n-machinery' into master

2025-01-15T19:31:19+01:00

* t/l10n-machinery:
  Convert translatable string construction in enums to positional args only
  Normalize and unwrap translatable strings (messages and comments)
  Add hypothesis-based tests for l10n machinery
  Fix obvious coverage failures among the translatable strings machinery
  Split off comments in translatable string enum value constructor
  Sort arguments to translatable string enum value constructor
  Issue new context IDs for translatable messages
  Fix debug output of messages with trimmed filenames
  Support SOURCE_DATE_EPOCH for .po file generation
  Move translation string preparation into the TranslatableString class
  Properly support trimmed filenames in translation strings
  Generate debug translations automatically, in code and on the command-line
  Hide translation template enums behind a type alias
</pre>

Convert translatable string construction in enums to positional args only

2025-01-15T19:23:47+01:00

This too makes the rendered code more readable by elliding repetitive
information.
</pre>

Normalize and unwrap translatable strings (messages and comments)

2025-01-15T16:01:44+01:00

Rewrite one-paragraph comments and strings as one-line strings, with
implicit line continuation if necessary.  This is how they would appear
anyway if printed via `repr`, and in the `mkdocstrings`-extracted API
documentation.  Further trim the "TRANSLATORS:" prefix from comments; it
is added automatically during `.po` file writing as necessary.

Some minor changes to the docstrings have also been made:

  * `Label`
      - `VAULT_METAVAR_SERVICE`: Trim `as "service_metavar"` from the`
        comments for consistency.
  * DebugMsgTemplate
      - `VAULT_NATIVE_EVP_BYTESTOKEY_RESULT`: Use the correct message.
      - `VAULT_NATIVE_CHECKING_MAC_DETAILS`,
        `VAULT_NATIVE_V02_PAYLOAD_MAC_POSTPROCESSING`: Fix reference to
        `InfoMsgTemplate.VAULT_NATIVE_PARSING_IV_PAYLOAD_AND_MAC`
  * InfoMsgTemplate
      - `VAULT_NATIVE_PARSING_IV_PAYLOAD_MAC`: Use "these terms" instead
        of "this term".

The diff is hard to read in normal "line" mode, but much clearer if read
in "word-diff" mode (with quotes also treated as whitespace).
</pre>

Add hypothesis-based tests for l10n machinery

2025-01-14T03:06:58+01:00

This specifically tests the `DebugTranslations` object, and various
asserted properties of `TranslatedStrings` and `TranslatableStrings`
such as hashability and equality under various modifications (e.g.
trimming the filename).

To further support this, `TranslatedStrings` now know their own format
string replacement fields (except for `%`-formatted strings), and the
`DebugTranslations` class now uses that knowledge instead of duplicating
it inline.  Additionally, because `TranslatedString`s are always
interpolated, there is a new `TranslatedString.constant` constructor
for `str` templates which ensures that nothing is interpolated.

Some code branches previously excluded from coverage are now considered
again.
</pre>

Fix obvious coverage failures among the translatable strings machinery

2025-01-13T23:28:32+01:00

Use examples in doctests where it makes didactical sense.
</pre>

Split off comments in translatable string enum value constructor

2025-01-13T22:44:58+01:00

Now that the constructor for translatable string enum values is
homogeneous, split the constructor into a curried two-step function
call.  This should hopefully look much better visually separated when
rendered in the API documentation than the current solution, because the
rendering normalizes the multiline strings into single-line strings with
escape sequences.

(This commit only splits the arguments into the two-step function call;
the arguments have not yet been trimmed or normalized.)
</pre>

Sort arguments to translatable string enum value constructor

2025-01-13T18:05:33+01:00

</pre>

Issue new context IDs for translatable messages

2025-01-13T17:36:18+01:00

Use an ID system similar to the trove classifiers used the Python
packaging authority.

(Enum members were not reordered according to this new ID system.)
</pre>

Fix debug output of messages with trimmed filenames

2025-01-13T17:28:25+01:00

We do need to keep track of the trimming status when adding those
variants to the cache, because the debug output formatter will discover
the trimmed fields in the template anyway and otherwise attempt to emit
them.  So store the set of trimmed fields in the cache for each message,
and pass it on to the formatter.  We will assume that if a field gets
trimmed, then it had a `None` value, which is true for all the fields we
currently support trimming.
</pre>