From 341a19e0d05bc6f0cabeb2dc756b37cfc047f8f0 Mon Sep 17 00:00:00 2001
From: Andrew Gallant <jamslam@gmail.com>
Date: Mon, 31 Jul 2023 08:51:09 -0400
Subject: [PATCH] regex: fix fast path for -w/--word-regexp flag (#2576)

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
---
 CHANGELOG.md             |  2 ++
 crates/regex/src/word.rs | 16 ++++++++++++++--
 tests/regression.rs      | 15 +++++++++++++++
 3 files changed, 31 insertions(+), 2 deletions(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 1b435273..c637aeae 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -38,6 +38,8 @@ Bug fixes:
   Fix bug when using inline regex flags with `-e/--regexp`.
 * [BUG #2523](https://github.com/BurntSushi/ripgrep/issues/2523):
   Make executable searching take `.com` into account on Windows.
+* [BUG #2574](https://github.com/BurntSushi/ripgrep/issues/2574):
+  Fix bug in `-w/--word-regexp` that would result in incorrect match offsets.
 
 
 13.0.0 (2021-06-12)
diff --git a/crates/regex/src/word.rs b/crates/regex/src/word.rs
index af4480ab..52fb61ce 100644
--- a/crates/regex/src/word.rs
+++ b/crates/regex/src/word.rs
@@ -128,6 +128,9 @@ impl WordMatcher {
         // The reason why we cannot handle the ^/$ cases here is because we
         // can't assume anything about the original pattern. (Try commenting
         // out the checks for ^/$ below and run the tests to see examples.)
+        //
+        // NOTE(2023-07-31): After fixing #2574, this logic honestly still
+        // doesn't seem correct. Regex composition is hard.
         let input = Input::new(haystack).span(at..haystack.len());
         let mut cand = match self.regex.find(input) {
             None => return Ok(None),
@@ -136,8 +139,17 @@ impl WordMatcher {
         if cand.start() == 0 || cand.end() == haystack.len() {
             return Err(());
         }
-        let (_, slen) = bstr::decode_utf8(&haystack[cand]);
-        let (_, elen) = bstr::decode_last_utf8(&haystack[cand]);
+        // We decode the chars on either side of the match. If either char is
+        // a word character, then that means the ^/$ matched and not \W. In
+        // that case, we defer to the slower engine.
+        let (ch, slen) = bstr::decode_utf8(&haystack[cand]);
+        if ch.map_or(true, regex_syntax::is_word_character) {
+            return Err(());
+        }
+        let (ch, elen) = bstr::decode_last_utf8(&haystack[cand]);
+        if ch.map_or(true, regex_syntax::is_word_character) {
+            return Err(());
+        }
         let new_start = cand.start() + slen;
         let new_end = cand.end() - elen;
         // This occurs the original regex can match the empty string. In this
diff --git a/tests/regression.rs b/tests/regression.rs
index b9076803..5ef741cf 100644
--- a/tests/regression.rs
+++ b/tests/regression.rs
@@ -1173,3 +1173,18 @@ rgtest!(r2480, |dir: Dir, mut cmd: TestCommand| {
     cmd.args(&["--only-matching", "-e", "(?i)notfoo", "-e", "bar", "file"]);
     cmd.assert_err();
 });
+
+// See: https://github.com/BurntSushi/ripgrep/issues/2574
+rgtest!(r2574, |dir: Dir, mut cmd: TestCommand| {
+    dir.create("haystack", "some.domain.com\nsome.domain.com/x\n");
+    let got = cmd
+        .args(&[
+            "--no-filename",
+            "--no-unicode",
+            "-w",
+            "-o",
+            r"(\w+\.)*domain\.(\w+)",
+        ])
+        .stdout();
+    eqnice!("some.domain.com\nsome.domain.com\n", got);
+});