Upgrade regex and regex-syntax

Supersedes: #1334 Supersedes: #1343
sharkdp · Jul 18, 2023 · e2b8514 · e2b8514
1 parent 686d194
commit e2b8514
Show file tree

Hide file tree

Showing 3 changed files with 30 additions and 19 deletions.
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/Cargo.toml b/Cargo.toml
@@ -39,8 +39,8 @@ argmax = "0.3.1"
 atty = "0.2"
 ignore = "0.4.20"
 num_cpus = "1.15"
-regex = "1.8.3"
-regex-syntax = "0.6"
+regex = "1.9.1"
+regex-syntax = "0.7"
 ctrlc = "3.2"
 humantime = "2.1"
 globset = "0.4"

diff --git a/src/regex_helper.rs b/src/regex_helper.rs
@@ -3,7 +3,7 @@ use regex_syntax::ParserBuilder;
 
 /// Determine if a regex pattern contains a literal uppercase character.
 pub fn pattern_has_uppercase_char(pattern: &str) -> bool {
-    let mut parser = ParserBuilder::new().allow_invalid_utf8(true).build();
+    let mut parser = ParserBuilder::new().utf8(false).build();
 
     parser
         .parse(pattern)
@@ -16,16 +16,18 @@ fn hir_has_uppercase_char(hir: &Hir) -> bool {
     use regex_syntax::hir::*;
 
     match hir.kind() {
-        HirKind::Literal(Literal::Unicode(c)) => c.is_uppercase(),
-        HirKind::Literal(Literal::Byte(b)) => char::from(*b).is_uppercase(),
+        HirKind::Literal(Literal(bytes)) => match std::str::from_utf8(&bytes) {
+            Ok(s) => s.chars().any(|c| c.is_uppercase()),
+            Err(_) => bytes.iter().any(|b| char::from(*b).is_uppercase()),
+        },
         HirKind::Class(Class::Unicode(ranges)) => ranges
             .iter()
             .any(|r| r.start().is_uppercase() || r.end().is_uppercase()),
         HirKind::Class(Class::Bytes(ranges)) => ranges
             .iter()
             .any(|r| char::from(r.start()).is_uppercase() || char::from(r.end()).is_uppercase()),
-        HirKind::Group(Group { hir, .. }) | HirKind::Repetition(Repetition { hir, .. }) => {
-            hir_has_uppercase_char(hir)
+        HirKind::Capture(Capture { sub, .. }) | HirKind::Repetition(Repetition { sub, .. }) => {
+            hir_has_uppercase_char(sub)
         }
         HirKind::Concat(hirs) | HirKind::Alternation(hirs) => {
             hirs.iter().any(hir_has_uppercase_char)
@@ -36,7 +38,7 @@ fn hir_has_uppercase_char(hir: &Hir) -> bool {
 
 /// Determine if a regex pattern only matches strings starting with a literal dot (hidden files)
 pub fn pattern_matches_strings_with_leading_dot(pattern: &str) -> bool {
-    let mut parser = ParserBuilder::new().allow_invalid_utf8(true).build();
+    let mut parser = ParserBuilder::new().utf8(false).build();
 
     parser
         .parse(pattern)
@@ -56,15 +58,18 @@ fn hir_matches_strings_with_leading_dot(hir: &Hir) -> bool {
         HirKind::Concat(hirs) => {
             let mut hirs = hirs.iter();
             if let Some(hir) = hirs.next() {
-                if hir.kind() != &HirKind::Anchor(Anchor::StartText) {
+                if hir.kind() != &HirKind::Look(Look::Start) {
                     return false;
                 }
             } else {
                 return false;
             }
 
             if let Some(hir) = hirs.next() {
-                hir.kind() == &HirKind::Literal(Literal::Unicode('.'))
+                match hir.kind() {
+                    HirKind::Literal(Literal(bytes)) => bytes.starts_with(&[b'.']),
+                    _ => false,
+                }
             } else {
                 false
             }