I have these hooks enabled on many of my projects, but for some reason
forgot them here which led to the issue described in
https://github.com/zoni/obsidian-export/pull/181
This should help prevent that from happening again.
This lets the compiler reason about the lifetimes of objects used by the
postprocessor, if the callback captures variables.
See zoni/obsidian-export#175
This license achieves everything that dual-licensing under MIT + Apache
aims for, but without the weirdness of being under two licenses.
Having checked external contributions, I feel pretty confident that I
can unilaterally make this license change, as people have only
contributed a handful of one-line changes of no significance towards
copyrighted work up to this point.
The unicode standard allows for certain (visually) identical characters to
be represented in different ways.
For example the character ä may be represented as a single combined
codepoint "Latin Small Letter A with Diaeresis" (U+00E4) or by the
combination of "Latin Small Letter A" (U+0061) followed by "Combining
Diaeresis" (U+0308).
When encoded with UTF-8, these are represented as respectively the two
bytes 0xC3 0xA4, and the three bytes 0x61 0xCC 0x88.
A user linking to notes with these characters in their titles would
expect these two variants to link to the same file, given they are
visually identical and have the exact same semantic meaning.
The unicode standard defines a method to deconstruct and normalize these
forms, so that a byte comparison on the normalized forms of these
variants ends up comparing the same thing. This is called Unicode
Normalization, defined in Unicode® Standard Annex #15
(http://www.unicode.org/reports/tr15/).
The W3C Working Group has written an excellent explanation of the
problems regarding string matching, and how unicode normalization helps
with this process: https://www.w3.org/TR/charmod-norm/#unicodeNormalization
With this change, obsidian-export will perform unicode normalization
(specifically the C (or NFC) normalization form) on all note titles
while looking up link references, ensuring visually identical links are
treated as being similar, even if they were encoded as different
variants.
A special thanks to Hans Raaf (@oderwat) for reporting and helping track
down this issue.
---
Closes#126
No need to depend on a third-party hook repository when each of these
checks is easily defined and run through system commands.
This also allows us to actually run tests, which is current unsupported
(https://github.com/doublify/pre-commit-rust/pull/19)
Instead of passing clones of context and the markdown tree to
postprocessors, pass them a mutable reference which may be modified
in-place.
This is a breaking change to the postprocessor implementation, changing
both the input arguments as well as the return value:
```diff
- dyn Fn(Context, MarkdownEvents) -> (Context, MarkdownEvents, PostprocessorResult) + Send + Sync;
+ dyn Fn(&mut Context, &mut MarkdownEvents) -> PostprocessorResult + Send + Sync;
```
With this change the postprocessor API becomes a little more ergonomic
to use however, especially making the intent around return statements more clear.
This may make it more clear to users that these are precompiled, binary
files. This is especially relevant on Windows, where the convention is
that executable files have a .exe` extension, as seen in #49.
This change introduces a new `--hard-linebreaks` CLI argument. When
used, this converts soft line breaks to hard line breaks, mimicking
Obsidian's "Strict line breaks" setting.
Implementation detail: I considered naming this flag
`--strict-line-breaks` to be consistent with Obsidian itself, however I
feel the name is somewhat misleading and ill-chosen.
