switch to regex-cursor (#9422)

2 files changed, 44 insertions, 2 deletions

diff --git a/helix-stdx/Cargo.toml b/helix-stdx/Cargo.toml
index 540a1b99..5ac7c011 100644
--- a/helix-stdx/Cargo.toml
+++ b/helix-stdx/Cargo.toml
@@ -16,6 +16,7 @@ dunce = "1.0"
 etcetera = "0.8"
 ropey = { version = "1.6.1", default-features = false }
 which = "6.0"
+regex-cursor = "0.1.3" 
 
 [dev-dependencies]
 tempfile = "3.10"
diff --git a/helix-stdx/src/rope.rs b/helix-stdx/src/rope.rs
index 4ee39d4a..7b4edda4 100644
--- a/helix-stdx/src/rope.rs
+++ b/helix-stdx/src/rope.rs
@@ -1,11 +1,22 @@
+use std::ops::{Bound, RangeBounds};
+
+pub use regex_cursor::engines::meta::{Builder as RegexBuilder, Regex};
+pub use regex_cursor::regex_automata::util::syntax::Config;
+use regex_cursor::{Input as RegexInput, RopeyCursor};
 use ropey::RopeSlice;
 
-pub trait RopeSliceExt: Sized {
+pub trait RopeSliceExt<'a>: Sized {
     fn ends_with(self, text: &str) -> bool;
     fn starts_with(self, text: &str) -> bool;
+    fn regex_input(self) -> RegexInput<RopeyCursor<'a>>;
+    fn regex_input_at_bytes<R: RangeBounds<usize>>(
+        self,
+        byte_range: R,
+    ) -> RegexInput<RopeyCursor<'a>>;
+    fn regex_input_at<R: RangeBounds<usize>>(self, char_range: R) -> RegexInput<RopeyCursor<'a>>;
 }
 
-impl RopeSliceExt for RopeSlice<'_> {
+impl<'a> RopeSliceExt<'a> for RopeSlice<'a> {
     fn ends_with(self, text: &str) -> bool {
         let len = self.len_bytes();
         if len < text.len() {
@@ -23,4 +34,34 @@ impl RopeSliceExt for RopeSlice<'_> {
         self.get_byte_slice(..len - text.len())
             .map_or(false, |start| start == text)
     }
+
+    fn regex_input(self) -> RegexInput<RopeyCursor<'a>> {
+        RegexInput::new(self)
+    }
+
+    fn regex_input_at<R: RangeBounds<usize>>(self, char_range: R) -> RegexInput<RopeyCursor<'a>> {
+        let start_bound = match char_range.start_bound() {
+            Bound::Included(&val) => Bound::Included(self.char_to_byte(val)),
+            Bound::Excluded(&val) => Bound::Excluded(self.char_to_byte(val)),
+            Bound::Unbounded => Bound::Unbounded,
+        };
+        let end_bound = match char_range.end_bound() {
+            Bound::Included(&val) => Bound::Included(self.char_to_byte(val)),
+            Bound::Excluded(&val) => Bound::Excluded(self.char_to_byte(val)),
+            Bound::Unbounded => Bound::Unbounded,
+        };
+        self.regex_input_at_bytes((start_bound, end_bound))
+    }
+    fn regex_input_at_bytes<R: RangeBounds<usize>>(
+        self,
+        byte_range: R,
+    ) -> RegexInput<RopeyCursor<'a>> {
+        let input = match byte_range.start_bound() {
+            Bound::Included(&pos) | Bound::Excluded(&pos) => {
+                RegexInput::new(RopeyCursor::at(self, pos))
+            }
+            Bound::Unbounded => RegexInput::new(self),
+        };
+        input.range(byte_range)
+    }
 }