As of the memchr 2.6 release, its Iterator::count method is specialized
to only count the number of occurrences instead of finding the offset of
each occurrence. This replaces ripgrep's use of the bytecount crate.
While micro-benchmarks suggest that memchr's method has better
throughput than bytecount, it turned out to be an illusion. Namely, on a
~13GB haystack prior to this change:
$ time rg-bytecount 'You killed my friend, my best friend, my lifelong friend!' OpenSubtitles2018.raw.en --line-number
441450441:- You killed my friend, my best friend, my lifelong friend!
real 1.473
user 1.186
sys 0.286
maxmem 12512 MB
faults 0
And then after:
$ time rg 'You killed my friend, my best friend, my lifelong friend!' OpenSubtitles2018.raw.en --line-number
441450441:- You killed my friend, my best friend, my lifelong friend!
real 1.532
user 1.280
sys 0.250
maxmem 12512 MB
faults 0
But perf is just about in the same ballpark. That's good enough for me
at the moment in order to drop the extra dependency.
I did this because the marginal cost of adding the Iterator::count()
specialization to memchr was extremely small.
We currently implement globs by converting them to regexes, and in doing
so, sometimes use grouping. In all but one case, we used non-capturing
groups. But for alternations, we used capturing groups, which was likely
just an oversight. We don't make use of capture groups at all, and while
they usually don't have any overhead, they lead to weird cases like this
one: https://github.com/rust-lang/regex/issues/1059
That particular issue is also a bug in the regex crate itself, which is
fixed in https://github.com/rust-lang/regex/pull/1062. Note though that
the bug fix in the regex crate is required. Even with this patch to
globset, memory usage is reduced (by about half in rust-lang/regex#1059)
but is not returned to where it was prior to the regex 1.9 release.
It turns out our fast path for -w/--word-regexp wasn't quite correct in
some cases. Namely, we use `(?m:^|\W)(<original-regex>)(?m:\W|$)` as the
implementation of -w/--word-regexp since `\b(<original-regex>)\b` has
some unintuitive results in certain cases, specifically when
<original-regex> matches non-word characters at match boundaries.
The problem is that using this formulation means that you need to
extract the capture group around <original-regex> to find the "real"
match, since the surrounding (^|\W) and (\W|$) aren't part of the match.
This is fine, but the capture group engine is usually slow, so we have a
fast path where we try to deduce the correct match boundary after an
initial match (before running capture groups). The problem is that doing
this is rather tricky because it's hard to know, in general, whether the
`^` or the `\W` matched.
This still doesn't seem quite right overall, but we at least fix one
more case.
Fixes#2574
This PR adds `*.bat` and `*.cmd` file types.
In doing so, it makes a distinction between batch files (old standard
from the MS-DOS era) and command scripts (new flavor - can operate on
batch files, although `*.cmd` is preferred for various reasons, the
main one being batch files will set `ERRORLEVEL` following inconsistent
MS-DOS style rules[1]).
PR #2556
[1]: https://groups.google.com/g/microsoft.public.win2000.cmdprompt.admin/c/XHeUq8oe2wk/m/LIEViGNmkK0J#i106
Previously, sorting worked by sorting the parents and then sorting the
children within each parent. This was done during traversal, but it only
works when sorting parents preserves the overall order. This generally
only works for '--sort path' in ascending order.
This commit fixes the rest of the sorting behavior by collecting all of
the paths to search and then sorting them before searching. We only
collect all of the paths when sorting was requested.
Fixes#2243, Closes#2361
This causes ripgrep to stop searching an individual file after it has
found a non-matching line. But this only occurs after it has found a
matching line.
Fixes#1790, Closes#1930
Adds a new eprintln_locked macro which locks STDOUT before logging
to STDERR. This patch also replaces instances of eprintln with
eprintln_locked to avoid interleaving lines.
Fixes#1941, Closes#1968
Previously, we were only doing a binary existence check on Windows. And
in fact, the main point there wasn't binary existence, but ensuring we
didn't accidentally resolve a binary name relative to the CWD, which
could result in executing a program one didn't mean to run.
However, it is useful to be able to check whether a binary exists on any
platform when associating a glob with a binary. If the binary doesn't
exist, then the association can fail eagerly and let some other glob
apply.
Closes#1946
We also make py/python, md/markdown and ts/typescript aliases of one
another.
Note that this only introduces aliases at the point where default types
are defined. This just makes them a bit easier to read/write, and also
makes it easier to expose more names that describe the same thing.
Fixes#1857, Closes#1895
*.adb and *.ads are the usual extensions for Ada source code,
and *.gpr indicates a GPRbuild project file used for Ada, and
these days often being combined with alire for package dependency
resolution. Alire stores a bunch of files named alire.toml in
different directories in your (gitignored) cache/dependencies/...
Closes#2013
Eclasses are "ebuild libraries" and generally if you're filtering
for/filtering out an ebuild/eclass, you don't want the other either.
Followup to 4dfea016b915bb1e88679361de83a91e60447835
Closes#2437
This matches the behavior of GNU grep which does not ignore
before-context and after-context completely if the context flag is also
provided.
Note that this change wasn't done just to match GNU grep. In this case,
GNU grep has the more sensible behavior.
Fixes#2288, Closes#2451
I have reservations about this, but it looks useful and doesn't seem
terribly onerous to support. The `ignore` crate will really always need
to have some kind of logic supporting this in some form I think.
Closes#2482
This removes most of the Unicode features as they aren't currently
used. We can always add them back later if necessary.
We can avoid the unicode-perl feature by changing `\s` to `[[:space:]]`,
which uses the ASCII-only definition of `\s`. Since we don't expect
non-ASCII whitespace in git config files, this seems okay.
Closes#2502
GraphQL file extensions: .graphql and .graphqls (schema)
We could also add `.gql`, but perhaps it's less correct to do so. We'll
start conservatively here, and we can always add `.gql` later.
Closes#2439, Closes#2508
When `resolve_binary()` attempts to resolve a path to a program on
Windows while searching for a program in `PATH` without an extension,
`ripgrep` will assume the extension of the file to be `.exe` as it's
the *de facto* standard, which will work most (99.99%) of the time...
...unless the binary is a COM executable (we're on Windows, duh).
Closes#2523
This is mostly a copy of the prefix literal extractor in regex-syntax,
but with a tweaked notion of Seq that keeps track of whether it's a
prefix of an expression or not. If it isn't, then we can't cross it as a
suffix to another Seq.
This new extractor should be a lot more robust than the old one. We
actually will keep going through the regex to try and find the "best"
literals to search for (according to some heuristic).
This makes it easier to enable the `logging` feature for regex-automata.
I wish I could just enable it unconditionally, but it winds up producing
a lot of output because ripgrep uses regexes for things other than the
primary search (like every glob). Sigh.
