This PR implements `expandSymlink` on Windows with POSIX readlink
semantics: it expands exactly one hop and returns the stored link target
without resolving the full chain.
The main design question was whether Windows symlink expansion should be
built on path-finalization APIs such as `GetFinalPathNameByHandleW`, or
on direct reparse-point inspection. Current `expandSymlink` is a
single-hop "what target is stored in this link object?" operation and
most of other ways to resolve symlinks on Windows actually try to answer
the "final true file location" question in various slightly-incompatible
ways.
The full final-path resolution on Windows is substantially more complex
than readlink and is planned as a follow-up.
## Implementation choice
Implements Windows `expandSymlink` by:
- opening the path with `FILE_FLAG_OPEN_REPARSE_POINT`
- calling `DeviceIoControl(FSCTL_GET_REPARSE_POINT)`
- parsing the reparse payload for `IO_REPARSE_TAG_SYMLINK` and
`IO_REPARSE_TAG_MOUNT_POINT`
- decoding the UTF-16 slice referenced by the payload
- returning the stored target
This is the right primitive for the API:
- does not depend on whole-path finalization
- works for both symlinks and junctions
- matches the existing Linux behaviour
`widestrs` changes allow using WideCString views without temporary
allocations.
Windows prohibits symlink creation without admin rights, so,
unfortunately, the tests are conditionally skipped by default. Manually
running `testament` in an admin console is required.
## Behaviour:
- One hop only
- Relative symlink targets are returned unchanged
- Absolute Windows targets are converted from stored NT-style prefixes
to usable Win32 forms when applicable
- Non-links, malformed payloads, and unsupported reparse tags raise
`OSError`
## Future work
Path canonicalization, i.e. "final true file location". Which is, BTW,
different from `absolutePath`, which works on paths only and doesn't hit
the underlying FS. So this needs to be an API extension.
I'd like to follow-up with this when I sort through the docs, for now
you can resolve symlinks in a loop.
---------
Co-authored-by: Andreas Rumpf <rumpf_a@web.de>
* docgen: implement cross-document links
Fully implements https://github.com/nim-lang/RFCs/issues/125
Follow-up of: https://github.com/nim-lang/Nim/pull/18642 (for internal links)
and https://github.com/nim-lang/Nim/issues/20127.
Overview
--------
Explicit import-like directive is required, called `.. importdoc::`.
(the syntax is % RST, Markdown will use it for a while).
Then one can reference any symbols/headings/anchors, as if they
were in the local file (but they will be prefixed with a module name
or markup document in link text).
It's possible to reference anything from anywhere (any direction
in `.nim`/`.md`/`.rst` files).
See `doc/docgen.md` for full description.
Working is based on `.idx` files, hence one needs to generate
all `.idx` beforehand. A dedicated option `--index:only` is introduced
(and a separate stage for `--index:only` is added to `kochdocs.nim`).
Performance note
----------------
Full run for `./koch docs` now takes 185% of the time before this PR.
(After: 315 s, before: 170 s on my PC).
All the time seems to be spent on `--index:only` run, which takes
almost as much (85%) of normal doc run -- it seems that most time
is spent on file parsing, turning off HTML generation phase has not
helped much.
(One could avoid it by specifying list of files that can be referenced
and pre-processing only them. But it can become error-prone and I assume
that these linke will be **everywhere** in the repository anyway,
especially considering https://github.com/nim-lang/RFCs/issues/478.
So every `.nim`/`.md` file is processed for `.idx` first).
But that's all without significant part of repository converted to
cross-module auto links. To estimate impact I checked the time for
`doc`ing a few files (after all indexes have been generated), and
everywhere difference was **negligible**.
E.g. for `lib/std/private/osfiles.nim` that `importdoc`s large
`os.idx` and hence should have been a case with relatively large
performance impact, but:
* After: 0.59 s.
* Before: 0.59 s.
So Nim compiler works so slow that doc part basically does not matter :-)
Testing
-------
1) added `extlinks` test to `nimdoc/`
2) checked that `theindex.html` is still correct
2) fixed broken auto-links for modules that were derived from `os.nim`
by adding appropriate ``importdoc``
Implementation note
-------------------
Parsing and formating of `.idx` entries is moved into a dedicated
`rstidx.nim` module from `rstgen.nim`.
`.idx` file format changed:
* fields are not escaped in most cases because we need original
strings for referencing, not HTML ones
(the exception is linkTitle for titles and headings).
Escaping happens later -- on the stage of `rstgen` buildIndex, etc.
* all lines have fixed number of columns 6
* added discriminator tag as a first column,
it always allows distinguish Nim/markup entries, titles/headings, etc.
`rstgen` does not rely any more (in most cases) on ad-hoc logic
to determine what type each entry is.
* there is now always a title entry added at the first line.
* add a line number as 6th column
* linkTitle (4th) column has a different format: before it was like
`module: funcName()`, now it's `proc funcName()`.
(This format is also propagated to `theindex.html` and search results,
I kept it that way since I like it more though it's discussible.)
This column is what used for Nim symbols resolution.
* also changed details on column format for headings and titles:
"keyword" is original, "linkTitle" is HTML one
* fix paths on Windows + more clear code
* Update compiler/docgen.nim
Co-authored-by: Andreas Rumpf <rumpf_a@web.de>
* Handle .md and .nim paths uniformly in findRefFile
* handle titles better + more comments
* don't allow markup overwrite index title for .nim files
Co-authored-by: Andreas Rumpf <rumpf_a@web.de>
