... it turns out that rustembedded/cross:armv7-unknown-linux-musleabi
doesn't exist. And looking more closely, it looks like the Cross project
has decided to shake things up and publish images to ghcr instead. So we
migrate everything over to that.
As the FIXME comment says, ripgrep is not yet using the new line
terminator option in regex-automata exposed for exactly this purpose.
Because of that, line anchors like `(?m:^)` and `(?m:$)` will only match
`\n` as a line terminator. This means that when --null-data is used in
combination with --line-regexp, the anchors inserted by --line-regexp
will not match correctly. This is only a big deal in the "fast" path,
which requires the regex engine to deal with line terminators itself
correctly. The slow path strips line terminators regardless of what they
are, and so the line anchors can match (begin/end of haystack).
Fixes#2658
Basically, unless the -a/--text flag is given, it is generally always an
error to search for an explicit NUL byte because the binary detection
will prevent it from matching.
Fixes#1838
ripgrep began it's life with docopt for argument parsing. Then it moved
to Clap and stayed there for a number of years. Clap has served ripgrep
well, and it probably could continue to serve ripgrep well, but I ended
up deciding to move off of it.
Why?
The first time I had the thought of moving off of Clap was during the
2->3->4 transition. I thought the 3.x and 4.x releases were great, but
for me, it ended up moving a little too quickly. Since the release of
4.x was telegraphed around when 3.x came out, I decided to just hold off
and wait to migrate to 4.x instead of doing a 3.x migration followed
shortly by another 4.x migration. Of course, I just never ended up doing
the migration at all. I never got around to it and there just wasn't a
compelling reason for me to upgrade. While I never investigated it, I
saw an upgrade as a non-trivial amount of work in part because I didn't
encapsulate the usage of Clap enough.
The above is just what got me started thinking about it. It wasn't
enough to get me to move off of it on its own. What ended up pushing me
over the edge was a combination of factors:
* As mentioned above, I didn't want to run on the migration treadmill.
This has proven to not be much of an issue, but at the time of the
2->3->4 releases, I didn't know how long Clap 4.x would be out before a
5.x would come out.
* The release of lexopt[1] caught my eye. IMO, that crate demonstrates
exactly how something new can arrive on the scene and just thoroughly
solve a problem minimalistically. It has the docs, the reasoning, the
simple API, the tests and good judgment. It gets all the weird corner
cases right that Clap also gets right (and is part of why I was
originally attracted to Clap).
* I have an overall desire to reduce the size of my dependency tree. In
part because a smaller dependency tree tends to correlate with better
compile times, but also in part because it reduces my reliance and trust
on others. It lets me be the "master" of ripgrep's destiny by reducing
the amount of behavior that is the result of someone else's decision
(whether good or bad).
* I perceived that Clap solves a more general problem than what I
actually need solved. Despite the vast number of flags that ripgrep has,
its requirements are actually pretty simple. We just need simple
switches and flags that support one value. No multi-value flags. No
sub-commands. And probably a lot of other functionality that Clap has
that makes it so flexible for so many different use cases. (I'm being
hand wavy on the last point.)
With all that said, perhaps most importantly, the future of ripgrep
possibly demands a more flexible CLI argument parser. In today's world,
I would really like, for example, flags like `--type` and `--type-not`
to be able to accumulate their repeated values into a single sequence
while respecting the order they appear on the CLI. For example, prior
to this migration, `rg regex-automata -Tlock -ttoml` would not return
results in `Cargo.lock` in this repository because the `-Tlock` always
took priority even though `-ttoml` appeared after it. But with this
migration, `-ttoml` now correctly overrides `-Tlock`. We would like to
do similar things for `-g/--glob` and `--iglob` and potentially even
now introduce a `-G/--glob-not` flag instead of requiring users to use
`!` to negate a glob. (Which I had done originally to work-around this
problem.) And some day, I'd like to add some kind of boolean matching to
ripgrep perhaps similar to how `git grep` does it. (Although I haven't
thought too carefully on a design yet.) In order to do that, I perceive
it would be difficult to implement correctly in Clap.
I believe that this last point is possible to implement correctly in
Clap 2.x, although it is awkward to do so. I have not looked closely
enough at the Clap 4.x API to know whether it's still possible there. In
any case, these were enough reasons to move off of Clap and own more of
the argument parsing process myself.
This did require a few things:
* I had to write my own logic for how arguments are combined into one
single state object. Of course, I wanted this. This was part of the
upside. But it's still code I didn't have to write for Clap.
* I had to write my own shell completion generator.
* I had to write my own `-h/--help` output generator.
* I also had to write my own man page generator. Well, I had to do this
with Clap 2.x too, although my understanding is that Clap 4.x supports
this. With that said, without having tried it, my guess is that I
probably wouldn't have liked the output it generated because I
ultimately had to write most of the roff by hand myself to get the man
page I wanted. (This also had the benefit of dropping the build
dependency on asciidoc/asciidoctor.)
While this is definitely a fair bit of extra work, it overall only cost
me a couple days. IMO, that's a good trade off given that this code is
unlikely to change again in any substantial way. And it should also
allow for more flexible semantics going forward.
Fixes#884, Fixes#1648, Fixes#1701, Fixes#1814, Fixes#1966
[1]: https://docs.rs/lexopt/0.3.0/lexopt/index.html
This commit represents the initial work to get hyperlinks working and
was submitted as part of PR #2483. Subsequent commits largely retain the
functionality and structure of the hyperlink support added here, but
rejigger some things around.
When searching subdirectories the path was not correctly built and
included duplicate parts. This fix will remove the duplicate part if
possible.
Fixes#1757, Closes#2295
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 was originally fixed by using non-capturing groups when joining
patterns in crates/core/args.rs, but before that landed, it ended up
getting fixed via a refactor in the course of migrating to regex 1.9.
Namely, it's now fixed by pushing pattern joining down into the regex
layer, so that patterns can be joined in the most effective way
possible.
Still, #2488 contains a useful test, so we bring that in here. The
test actually failed for `rg -e ')('`, since it expected the command to
fail with a syntax error. But my refactor actually causes this command
to succeed. And indeed, #2488 worked around this by special casing a
single pattern. That work-around fixes it for the single pattern case,
but doesn't fix it for the -w or -X or multi-pattern case. So for now,
we're content to leave well enough alone. The only real way to fix this
for real is to parse each regexp individual and verify that each is
valid on its own. It's not clear that doing so is worth it.
Fixes#2480, Closes#2488
When a glob pattern ended with a \/, and since we permit backslash
escapes, the glob parser gave a "dangling escape" error. Which is weird,
because the \ is clearly not dangling.
The issue is that the layer above the glob parser, the gitignore parser,
was stripping the trailing / so that it wouldn't be part of the matching
logic. Of course, stripping the trailing / while it is escaped without
removing the backslash escape is wrong. So we do that here.
Fixes#2236
This furthers our kludge of dealing with PCRE2's look-around in the
printer. Because of our bad abstraction boundaries, we added a kludge to
deal with PCRE2 look-around by extending the bytes we search by a fixed
amount to hopefully permit any look-around to operate. But because of
that kludge, we wind up over extending ourselves in some cases and
dragging along those extra bytes.
We had fixed this for simple searching by simply rejecting any matches
past the end point. But we didn't do the same for replacements. So this
commit extends our kludge to replacements.
Thanks to @sonohgong for diagnosing the problem and proposing a fix. I
mostly went with their solution, but adding the new replacement routine
as an internal helper rather than a new APIn in the 'grep-matcher'
crate.
Fixes#2095, Fixes#2208
Sadly, PCRE2 has different behavior (but doesn't panic). We should look
into that, but for now, this is good enough.
Also, update the CHANGELOG.
Ref #1891
This is a weird bug where our optimization for handling -w more quickly
than we would otherwise failed. In particular, if the original regex can
match the empty string, then our word boundary detection would produce
invalid indices to the start the next search at. We "fix" it by simply
bailing when the indices are known to be incorrect.
This wasn't a problem in a previous release since ripgrep 13 tweaked how
word boundaries are detected in commit efd9cfb2.
Fixes#1891
This fixes a bug where PCRE2 look-around could change the result of a
match if it observed a line terminator in the printer. And in
particular, this is precisely how the searcher operates: the line is
considered unto itself *without* the line terminator.
Fixes#1401
This commit hacks in a bug fix for handling look-around across multiple
lines. The main problem is that by the time the matching lines are sent
to the printer, the surrounding context---which some look-behind or
look-ahead might have matched---could have been dropped if it wasn't
part of the set of matching lines. Therefore, when the printer re-runs
the regex engine in some cases (to do replacements, color matches, etc
etc), it won't be guaranteed to see the same matches that the searcher
found.
Overall, this is a giant clusterfuck and suggests that the way I divided
the abstraction boundary between the printer and the searcher is just
wrong. It's likely that the searcher needs to handle more of the work of
matching and pass that info on to the printer. The tricky part is that
this additional work isn't always needed. Ultimately, this means a
serious re-design of the interface between searching and printing. Sigh.
The way this fix works is to smuggle the underlying buffer used by the
searcher through into the printer. Since these bugs only impact
multi-line search (otherwise, searches are only limited to matches
across a single line), and since multi-line search always requires
having the entire file contents in a single contiguous slice (memory
mapped or on the heap), it follows that the buffer we pass through when
we need it is, in fact, the entire haystack. So this commit refactors
the printer's regex searching to use that buffer instead of the intended
bundle of bytes containing just the relevant matching portions of that
same buffer.
There is one last little hiccup: PCRE2 doesn't seem to have a way to
specify an ending position for a search. So when we re-run the search to
find matches, we can't say, "but don't search past here." Since the
buffer is likely to contain the entire file, we really cannot do
anything here other than specify a fixed upper bound on the number of
bytes to search. So if look-ahead goes more than N bytes beyond the
match, this code will break by simply being unable to find the match. In
practice, this is probably pretty rare. I believe that if we did a
better fix for this bug by fixing the interfaces, then we'd probably try
to have PCRE2 find the pertinent matches up front so that it never needs
to re-discover them.
Fixes#1412
This commit fixes a subtle bug in multi-line replacement of line
terminators.
The problem is that even though ripgrep supports multi-line searches, it
is *still* line oriented. It still needs to print line numbers, for
example. For this reason, there are various parts in the printer that
iterate over lines in order to format them into the desired output.
This turns out to be problematic in some cases. #1311 documents one of
those cases (with line numbers enabled to highlight a point later):
$ printf "hello\nworld\n" | rg -n -U "\n" -r "?"
1:hello?
2:world?
But the desired output is this:
$ printf "hello\nworld\n" | rg -n -U "\n" -r "?"
1:hello?world?
At first I had thought that the main problem was that the printer was
taking ownership of writing line terminators, even if the input already
had them. But it's more subtle than that. If we fix that issue, we get
output like this instead:
$ printf "hello\nworld\n" | rg -n -U "\n" -r "?"
1:hello?2:world?
Notice how '2:' is printed before 'world?'. The reason it works this way
is because matches are reported to the printer in a line oriented way.
That is, the printer gets a block of lines. The searcher guarantees that
all matches that start or end in any of those lines also end or start in
another line in that same block. As a result, the printer uses this
assumption: once it has processed a block of lines, the next match will
begin on a new and distinct line. Thus, things like '2:' are printed.
This is generally all fine and good, but an impedance mismatch arises
when replacements are used. Because now, the replacement can be used to
change the "block of lines" approach. Now, in terms of the output, the
subsequent match might actually continue the current line since the
replacement might get rid of the concept of lines altogether.
We can sometimes work around this. For example:
$ printf "hello\nworld\n" | rg -U "\n(.)?" -r '?$1'
hello?world?
Why does this work? It's because the '(.)' after the '\n' causes the
match to overlap between lines. Thus, the searcher guarantees that the
block sent to the printer contains every line.
And there in lay the solution: all we need to do is tweak the multi-line
searcher so that it combines lines with matches that directly adjacent,
instead of requiring at least one byte of overlap. Fixing that solves
the issue above. It does cause some tests to fail:
* The binary3 test in the searcher crate fails because adjacent line
matches are now one part of block, and that block is scanned for
binary data. To preserve the essence of the test, we insert a couple
dummy lines to split up the blocks.
* The JSON CRLF test. It was testing that we didn't output any messages
with an empty 'submatches' array. That is indeed still the case. The
difference is that the messages got combined because of the adjacent
line merging behavior. This is a slight change to the output, but is
still correct.
Fixes#1311
It turns out that the vimgrep format really only wants one line per
match, even when that match spans multiple lines.
We continue to support the previous behavior (print all lines in a
match) in the `grep-printer` crate. We add a new option to enable the
"only print the first line" behavior, and unconditionally enable it in
ripgrep. We can do that because the option has no effect in single-line
mode, since, well, in that case matches are guaranteed to span one line
anyway.
Fixes#1866
In the case where after-context is requested with a match count limit,
we need to be careful not to reset the state tracking the remaining
context lines.
Fixes#1380, Closes#1642
Previously, we were only looking for the UTF-16 BOM for determining
whether to do transcoding or not. But we should also look for the UTF-8
BOM as well.
Fixes#1638, Closes#1697
This fixes a bug where it was assumed that 'is_suffix' when CRLF
handling was enabled mean that '\r\n' was present. But that's not the
case, and it is intentional that 'is_suffix' only looks for '\n'. (Which
is why #1803 wasn't taken, which tries to fix this by changing
'is_suffix'.)
Fixes#1765, Closes#1803
This was once part of ripgrep, but at some point, was unintentionally
removed. The value of this warning is that since ripgrep tries to be
"smart" by default, it can be surprising if it doesn't search certain
things. This warning covers the case when ripgrep searches *nothing*,
which happens somewhat more frequently than you might expect. e.g., If
you're searching within an ignore directory.
Note that for now, we only print this message when the user has not
supplied any explicit paths. It's not clear that we want to print this
otherwise, and in particular, it seems that the message shows up too
eagerly. e.g., 'rg foo does-not-exist' will both print an error about
'does-not-exist' not existing, *and* the message about no files being
searched, which seems annoying in this case. We can always refine this
logic later.
Fixes#1404, Closes#1762
This fixes a bug where using \A or (?-m)^ in combination with
-U/--multiline would permit matches that aren't anchored to the
beginning of the file. The underlying cause was an optimization that
occurred when mmaps couldn't be used. Namely, ripgrep tries to still
read the input incrementally if it knows the pattern can't match through
a new line. But the detection logic was flawed, since it didn't account
for line anchors. This commit fixes that.
Fixes#1878, Fixes#1879
It turned out that --vimgrep wasn't quite getting the column of each
match correctly. Instead of printing column numbers relative to the
current line, it was printing column numbers as byte offsets relative to
where the match began. To fix this, we simply subtract the offset of the
line number from the beginning of the match. If the beginning of the
match came before the start of the current line, then there's really
nothing sensible we can do other than to use a column number of 1, which
we now document.
Interestingly, existing tests were checking that the previous behavior
was intended. My only defense is that I somehow tricked myself into
thinking it was a byte offset instead of a column number.
Kudos to @bfrg for calling this out in #1866:
https://github.com/BurntSushi/ripgrep/issues/1866#issuecomment-841635553
In order to implement --count-matches, we simply re-execute the regex on
the spans reported by the searcher. The spans always correspond to the
lines that participated in the match. This is the correct thing to do,
except when the regex contains look-ahead (or look-behind).
In particular, the look-around permits the regex's match success to
depends on an arbitrary point before or after the lines actually
reported as participating in the match. Since only the matched lines are
reported to the printer, it is possible for subsequent searching on
those lines to fail.
A true fix for this would somehow make the total span available to the
printer. But that seems tricky since it isn't always available. For
PCRE2's case in multiline mode, it is available because we force it to
be so for correctness.
For now, we simply detect this corner case heuristically. If the match
count is zero, then it necessarily means there is some kind of
look-around that isn't matching. So we set the match count to 1. This is
probably incorrect in some cases, although my brain can't quite come up
with a concrete example. Nevertheless, this is strictly better than the
status quo.
Fixes#1573
This is why I was so intent on clearing the PR queue. This will
effectively invalidate all existing patches, so I wanted to start from a
clean slate.
We do make one little tweak: we put the default type definitions in
their own file and tell rustfmt to keep its grubby mits off of it. We
also sort it lexicographically and hopefully will enforce that from here
on.
Due to how walkdir works if symlinks are not followed, symlinks to
directories are seen as simple files by ripgrep. This caused a panic
in some cases due to receiving a WalkEvent::Exit event without a
corresponding WalkEvent::Dir event.
This is fixed by looking at the metadata of the file in the case of a
symlink to determine if it's a directory. We are careful to only do
this stat check when the depth of the entry is 0, as this bug only
impacts us when 1) we aren't following symlinks generally and 2) the
user provides a symlinked directory that we do follow as a top-level
path to search.
Fixes#1389, Closes#1397
This appears to be another transcription bug from copying this code from
the prefix literal detection from inside the regex crate. Namely, when
it comes to inner literals, we only want to treat counted repetition as
two separate cases: the case when the minimum match is 0 and the case
when the minimum match is more than 0. In the former case, we treat
`e{0,n}` as `e*` and in the latter we treat `e{m,n}` where `m >= 1` as
just `e`.
We could definitely do better here. e.g., This means regexes like
`(foo){10}` will only have `foo` extracted as a literal, where searching
for the full literal would likely be faster.
The actual bug here was that we were not implementing this logic
correctly. Namely, we weren't always "cutting" the literals in the
second case to prevent them from being expanded.
Fixes#1319, Closes#1367
Git looks for this file in GIT_COMMON_DIR, which is usually the same
as GIT_DIR (.git). However, when searching inside a linked worktree,
.git is usually a file that contains the path of the actual git dir,
which in turn contains a file "commondir" which references the directory
where info/exclude may reside, alongside other configuration shared across
all worktrees. This directory is usually the git dir of the main worktree.
Unlike git this does *not* read environment variables GIT_DIR and
GIT_COMMON_DIR, because it is not clear how to interpret them when
searching multiple repositories.
Fixes#1445, Closes#1446
Specifically, when searching stdin, if the current directory has a
directory named `-`, then the `--with-filename` flag would automatically
be turned on. This is because `--with-filename` is automatically enabled
when ripgrep is given a single path that is a directory. When ripgrep is
given empty arguments, and if it is searching stdin, then its default
path list is just simple `["-"]`. The `is_dir` check passes, and
`--with-filename` gets enabled.
This commit fixes the problem by checking whether the path is `-` first.
If so, then we assume it isn't a directory. This is fine, since if it is
a directory and one asks to search it explicitly, then ripgrep will
interpret `-` as stdin anyway (which is arguably a bug on its own, but
probably not one worth fixing).
Fixes#1223, Closes#1292
It turns out that when the -F flag was used, if any of the patterns
contained a regex meta character (such as `.`), then we winded up
escaping the pattern first before handing it off to Aho-Corasick, which
treats all patterns literally.
We continue to apply band-aides here and just avoid Aho-Corasick if
there is an escape in any of the literal patterns. This is unfortunate,
but making this work better requires more refactoring, and the right
solution is to get this optimization pushed down into the regex engine.
Fixes#1334
In an effort to strip line terminators, we assumed their existence. But
a pattern file may not end with a line terminator, so we shouldn't
unconditionally strip them.
We fix this by moving to bstr's line handling, which does this for us
automatically.
This fixes what appears to be a pretty egregious regression where the
`-F/--fixed-strings` flag wasn't be applied to patterns supplied via
the `-f/--file` flag. The same bug existed for the `-x/--line-regexp`
flag as well, which we fix here.
Fixes#1176
This changes how ripgrep emit exit status codes. In particular, any error
that occurs while searching will now cause ripgrep to emit a `2` exit
code, where as it previously would emit either a `0` or a `1` code based
on whether it matched or not. That is, ripgrep would only emit a `2` exit
code for a catastrophic error.
This tweak includes additional logic that GNU grep adheres to, which seems
like good sense. Namely, if -q/--quiet is given, and an error occurs and
a match occurs, then ripgrep will emit a `0` exit code.
Closes#1159
Previously, we relied on clap to handle printing either an error
message, or --help/--version output, in addition to setting the exit
status code. Unfortunately, for --help/--version output, clap was
panicking if the write failed, which can happen in fairly common
scenarios via a broken pipe error. e.g., `rg -h | head`.
We fix this by using clap's "safe" API and doing the printing ourselves.
We also set the exit code to `2` when an invalid command has been given.
Fixes#1125 and partially addresses #1159
This fixes a bug where a BOM prefix was included. While this was somewhat
intentional in order to have a faithful "UTF8 passthru" option, in
practice, this causes problems such as breaking patterns like `^` in a
really non-obvious way.
The actual fix was to add a new API to encoding_rs_io, which this commit
brings in.
Fixes#1163
Previously, `man gitignore` specified that `**` was invalid unless it
was used in one of a few specific circumstances, i.e., `**`, `a/**`,
`**/b` or `a/**/b`. That is, `**` always had to be surrounded by either
a path separator or the beginning/end of the pattern.
It turns out that git itself has treated `**` outside the above contexts
as valid for quite a while, so there was an inconsistency between the
spec `man gitignore` and the implementation, and it wasn't clear which
was actually correct.
@okdana filed a bug against git[1] and got this fixed. The spec was wrong,
which has now been fixed [2] and updated[2].
This commit brings ripgrep in line with git and treats `**` outside of
the above contexts as two consecutive `*` patterns. We deprecate the
`InvalidRecursive` error since it is no longer used.
Fixes#373, Fixes#1098
[1] - https://public-inbox.org/git/C16A9F17-0375-42F9-90A9-A92C9F3D8BBA@dana.is
[2] - 627186d020
[3] - https://git-scm.com/docs/gitignore
This fixes a bug where repeated use of ** didn't behave as it should. In
particular, each use of `**` added a new requirement directory depth
requirement. For example, something like `**/**/b` would match
`foo/bar/b`, but it wouldn't match `foo/b` even though it should. In
particular, `**` semantics demand "infinite" depth, so repeated uses of
`**` should just coalesce as if only one was given.
We do this coalescing in the parser. It's a little tricky because we
treat `**/a`, `a/**` and `a/**/b` as distinct tokens with their own
regex conversions. We also test the crap out of it.
Fixes#1174
When deciding whether to add the `**/` prefix or not, we should choose
not to add it if the pattern is simply a bare `**`. Previously, we were
only not adding it if it was `**/`, which is correct, but we also need
to do it for `**` since `**` can already match anywhere.
There's likely a more principled solution to this, but this works for
now.
Fixes#1173
This commit fixes a bug where both of the following commands always
reported an error:
rg --files-with-matches foo file
rg --files-without-match foo file
In particular, the printer was erroneously respecting the `path` option
even the the summary kind was `PathWithMatch` or `PathWithoutMatch`. The
documented behavior is that those summary kinds always require a path,
and thus, the `path` option has no effect. We fix this by correcting the
case analysis.
This also fixes a bug where the exit code for `--files-without-match`
was not set correctly. We update the printer's `has_match` method to
report the correct value.
Fixes#1106, Closes#1130
The --ignore-file-case-insensitive flag causes all
.gitignore/.rgignore/.ignore files to have their globs matched without
regard for case. Because this introduces a potentially significant
performance regression, this is always disabled by default. Users that
need case insensitive matching can enable it on a case by case basis.
Closes#1164, Closes#1170
It seems the inner literal detector fails spectacularly in cases of
concatenations that involve groups. The issue here is that if the prefix
of a group inside a concatenation can match the empty string, then any
literals generated to that point in the concatenation need to be cut
such that they are never extended. The detector isn't really built to
handle this case, so we just act conservative cut literals whenever we
see a sub-group. This may make some regexes slower, but the inner
literal detector already misses plenty of cases.
Literal detection (including in the regex engine) is a key component
that needs to be completely rethought at some point.
Fixes#1064
This basically rewrites every integration test. We reduce the amount of
magic involved here in terms of which arguments are being passed to
ripgrep processes. To make up for the boiler plate saved by the magic,
we make the Dir (formerly WorkDir) type a bit nicer to use, along with a
new TestCommand that wraps a std::process::Command. In exchange, we get
tests that are easier to read and write.
We also run every test with the `--pcre2` flag to make sure that works,
when PCRE2 is available.
