From f31923244088009ebccf887431c8dbdf5c0fec32 Mon Sep 17 00:00:00 2001
From: Wondr <wondr.design@gmail.com>
Date: Fri, 5 Jun 2026 11:41:01 +0100
Subject: [PATCH] grep: honor whitespace escapes in BRE

---
 src/matcher.rs     |  6 ++++++
 tests/test_grep.rs | 21 +++++++++++++++++++++
 2 files changed, 27 insertions(+)

diff --git a/src/matcher.rs b/src/matcher.rs
index 6da6d9b..de7f93d 100644
--- a/src/matcher.rs
+++ b/src/matcher.rs
@@ -273,6 +273,12 @@ impl CompiledPattern {
             // GNU grep supports \` and \' as buffer anchors in BRE and ERE.
             syntax.enable_operators(SyntaxOperator::SYNTAX_OPERATOR_ESC_GNU_BUF_ANCHOR);
         }
+        if config.regex_mode == RegexMode::Basic {
+            // GNU grep accepts \s and \S as whitespace shorthands in BRE.
+            // Follow Oniguruma's operator directly here; remaining invalid
+            // UTF-8 differences are engine semantics, not local syntax gaps.
+            syntax.enable_operators(SyntaxOperator::SYNTAX_OPERATOR_ESC_S_WHITE_SPACE);
+        }
         if config.regex_mode == RegexMode::Perl {
             // GNU grep supports `(?P<name>...)`.
             // Unfortunately, the onig crate defines the OP2 flag without the
diff --git a/tests/test_grep.rs b/tests/test_grep.rs
index 6a42175..b310808 100644
--- a/tests/test_grep.rs
+++ b/tests/test_grep.rs
@@ -100,6 +100,27 @@ fn gnu_buffer_anchors() {
         .stdout_only("cat\ntar\n");
 }
 
+#[test]
+fn bre_whitespace_shorthands() {
+    let (_s, mut c) = ucmd();
+    c.args(&[r"\s"])
+        .pipe_in("a b\nxy\n")
+        .succeeds()
+        .stdout_only("a b\n");
+
+    let (_s, mut c) = ucmd();
+    c.args(&["-c", r"\S"])
+        .pipe_in("aS b\n  \nx\n")
+        .succeeds()
+        .stdout_only("2\n");
+
+    let (_s, mut c) = ucmd();
+    c.args(&[r"\s\+"])
+        .pipe_in("a  b\nxy\n")
+        .succeeds()
+        .stdout_only("a  b\n");
+}
+
 #[test]
 fn ere_metacharacters() {
     let cases: &[(&[&str], &str, &str)] = &[