Merge ac0249bd13 into b1b60bb7fe

crates/multi_buffer/src/multi_buffer.rs

@@ -7723,12 +7723,24 @@ impl<'a> Iterator for MultiBufferChunks<'a> {
                 let diff_transform_end = diff_transform_end.min(self.range.end);
 
                 if diff_transform_end < chunk_end {
-                    let (before, after) =
-                        chunk.text.split_at(diff_transform_end - self.range.start);
+                    let split_idx = diff_transform_end - self.range.start;
+                    let (before, after) = chunk.text.split_at(split_idx);
                     self.range.start = diff_transform_end;
+                    let mask = (1 << split_idx) - 1;
+                    let chars = chunk.chars & mask;
+                    let tabs = chunk.tabs & mask;
+
                     chunk.text = after;
+                    chunk.chars = chunk.chars >> split_idx;
+                    chunk.tabs = chunk.tabs >> split_idx;
+
+                    // FIXME: We should be handling bitmap for tabs and chars here
+                    // Because we do a split at operation the bitmaps will be off
+
                     Some(Chunk {
                         text: before,
+                        chars,
+                        tabs,
                         ..chunk.clone()
                     })
                 } else {
@@ -7772,6 +7784,7 @@ impl<'a> Iterator for MultiBufferChunks<'a> {
                     self.range.start += "\n".len();
                     Chunk {
                         text: "\n",
+                        chars: 1u128,
                         ..Default::default()
                     }
                 };
@@ -7868,9 +7881,11 @@ impl<'a> Iterator for ExcerptChunks<'a> {
 
         if self.footer_height > 0 {
             let text = unsafe { str::from_utf8_unchecked(&NEWLINES[..self.footer_height]) };
+            let chars = (1 << self.footer_height) - 1;
             self.footer_height = 0;
             return Some(Chunk {
                 text,
+                chars,
                 ..Default::default()
             });
         }

crates/multi_buffer/src/multi_buffer_tests.rs

@@ -7,6 +7,7 @@ use parking_lot::RwLock;
 use rand::prelude::*;
 use settings::SettingsStore;
 use std::env;
+use util::RandomCharIter;
 use util::test::sample_text;
 
 #[ctor::ctor]
@@ -3712,3 +3713,235 @@ fn test_new_empty_buffers_title_can_be_set(cx: &mut App) {
     });
     assert_eq!(multibuffer.read(cx).title(cx), "Hey");
 }
+
+#[gpui::test(iterations = 100)]
+fn test_random_chunk_bitmaps(cx: &mut App, mut rng: StdRng) {
+    let multibuffer = if rng.r#gen() {
+        let len = rng.gen_range(0..10000);
+        let text = RandomCharIter::new(&mut rng).take(len).collect::<String>();
+        let buffer = cx.new(|cx| Buffer::local(text, cx));
+        cx.new(|cx| MultiBuffer::singleton(buffer, cx))
+    } else {
+        MultiBuffer::build_random(&mut rng, cx)
+    };
+
+    let snapshot = multibuffer.read(cx).snapshot(cx);
+
+    let chunks = snapshot.chunks(0..snapshot.len(), false);
+
+    for chunk in chunks {
+        let chunk_text = chunk.text;
+        let chars_bitmap = chunk.chars;
+        let tabs_bitmap = chunk.tabs;
+
+        if chunk_text.is_empty() {
+            assert_eq!(
+                chars_bitmap, 0,
+                "Empty chunk should have empty chars bitmap"
+            );
+            assert_eq!(tabs_bitmap, 0, "Empty chunk should have empty tabs bitmap");
+            continue;
+        }
+
+        assert!(
+            chunk_text.len() <= 128,
+            "Chunk text length {} exceeds 128 bytes",
+            chunk_text.len()
+        );
+
+        // Verify chars bitmap
+        let char_indices = chunk_text
+            .char_indices()
+            .map(|(i, _)| i)
+            .collect::<Vec<_>>();
+
+        for byte_idx in 0..chunk_text.len() {
+            let should_have_bit = char_indices.contains(&byte_idx);
+            let has_bit = chars_bitmap & (1 << byte_idx) != 0;
+
+            if has_bit != should_have_bit {
+                eprintln!("Chunk text bytes: {:?}", chunk_text.as_bytes());
+                eprintln!("Char indices: {:?}", char_indices);
+                eprintln!("Chars bitmap: {:#b}", chars_bitmap);
+            }
+
+            assert_eq!(
+                has_bit, should_have_bit,
+                "Chars bitmap mismatch at byte index {} in chunk {:?}. Expected bit: {}, Got bit: {}",
+                byte_idx, chunk_text, should_have_bit, has_bit
+            );
+        }
+
+        for (byte_idx, byte) in chunk_text.bytes().enumerate() {
+            let is_tab = byte == b'\t';
+            let has_bit = tabs_bitmap & (1 << byte_idx) != 0;
+
+            if has_bit != is_tab {
+                eprintln!("Chunk text bytes: {:?}", chunk_text.as_bytes());
+                eprintln!("Tabs bitmap: {:#b}", tabs_bitmap);
+                assert_eq!(
+                    has_bit, is_tab,
+                    "Tabs bitmap mismatch at byte index {} in chunk {:?}. Byte: {:?}, Expected bit: {}, Got bit: {}",
+                    byte_idx, chunk_text, byte as char, is_tab, has_bit
+                );
+            }
+        }
+    }
+}
+
+#[gpui::test(iterations = 100)]
+fn test_random_chunk_bitmaps_with_diffs(cx: &mut App, mut rng: StdRng) {
+    use buffer_diff::BufferDiff;
+    use util::RandomCharIter;
+
+    let multibuffer = if rng.r#gen() {
+        let len = rng.gen_range(100..10000);
+        let text = RandomCharIter::new(&mut rng).take(len).collect::<String>();
+        let buffer = cx.new(|cx| Buffer::local(text, cx));
+        cx.new(|cx| MultiBuffer::singleton(buffer, cx))
+    } else {
+        MultiBuffer::build_random(&mut rng, cx)
+    };
+
+    let _diff_count = rng.gen_range(1..5);
+    let mut diffs = Vec::new();
+
+    multibuffer.update(cx, |multibuffer, cx| {
+        for buffer_id in multibuffer.excerpt_buffer_ids() {
+            if rng.gen_bool(0.7) {
+                if let Some(buffer_handle) = multibuffer.buffer(buffer_id) {
+                    let buffer_text = buffer_handle.read(cx).text();
+                    let mut base_text = String::new();
+
+                    for line in buffer_text.lines() {
+                        if rng.gen_bool(0.3) {
+                            continue;
+                        } else if rng.gen_bool(0.3) {
+                            let line_len = rng.gen_range(0..50);
+                            let modified_line = RandomCharIter::new(&mut rng)
+                                .take(line_len)
+                                .collect::<String>();
+                            base_text.push_str(&modified_line);
+                            base_text.push('\n');
+                        } else {
+                            base_text.push_str(line);
+                            base_text.push('\n');
+                        }
+                    }
+
+                    if rng.gen_bool(0.5) {
+                        let extra_lines = rng.gen_range(1..5);
+                        for _ in 0..extra_lines {
+                            let line_len = rng.gen_range(0..50);
+                            let extra_line = RandomCharIter::new(&mut rng)
+                                .take(line_len)
+                                .collect::<String>();
+                            base_text.push_str(&extra_line);
+                            base_text.push('\n');
+                        }
+                    }
+
+                    let diff =
+                        cx.new(|cx| BufferDiff::new_with_base_text(&base_text, &buffer_handle, cx));
+                    diffs.push(diff.clone());
+                    multibuffer.add_diff(diff, cx);
+                }
+            }
+        }
+    });
+
+    multibuffer.update(cx, |multibuffer, cx| {
+        if rng.gen_bool(0.5) {
+            multibuffer.set_all_diff_hunks_expanded(cx);
+        } else {
+            let snapshot = multibuffer.snapshot(cx);
+            let text = snapshot.text();
+
+            let mut ranges = Vec::new();
+            for _ in 0..rng.gen_range(1..5) {
+                if snapshot.len() == 0 {
+                    break;
+                }
+
+                let diff_size = rng.gen_range(5..1000);
+                let mut start = rng.gen_range(0..snapshot.len());
+
+                while !text.is_char_boundary(start) {
+                    start = start.saturating_sub(1);
+                }
+
+                let mut end = rng.gen_range(start..snapshot.len().min(start + diff_size));
+
+                while !text.is_char_boundary(end) {
+                    end = end.saturating_add(1);
+                }
+                let start_anchor = snapshot.anchor_after(start);
+                let end_anchor = snapshot.anchor_before(end);
+                ranges.push(start_anchor..end_anchor);
+            }
+            multibuffer.expand_diff_hunks(ranges, cx);
+        }
+    });
+
+    let snapshot = multibuffer.read(cx).snapshot(cx);
+
+    let chunks = snapshot.chunks(0..snapshot.len(), false);
+
+    for chunk in chunks {
+        let chunk_text = chunk.text;
+        let chars_bitmap = chunk.chars;
+        let tabs_bitmap = chunk.tabs;
+
+        if chunk_text.is_empty() {
+            assert_eq!(
+                chars_bitmap, 0,
+                "Empty chunk should have empty chars bitmap"
+            );
+            assert_eq!(tabs_bitmap, 0, "Empty chunk should have empty tabs bitmap");
+            continue;
+        }
+
+        assert!(
+            chunk_text.len() <= 128,
+            "Chunk text length {} exceeds 128 bytes",
+            chunk_text.len()
+        );
+
+        let char_indices = chunk_text
+            .char_indices()
+            .map(|(i, _)| i)
+            .collect::<Vec<_>>();
+
+        for byte_idx in 0..chunk_text.len() {
+            let should_have_bit = char_indices.contains(&byte_idx);
+            let has_bit = chars_bitmap & (1 << byte_idx) != 0;
+
+            if has_bit != should_have_bit {
+                eprintln!("Chunk text bytes: {:?}", chunk_text.as_bytes());
+                eprintln!("Char indices: {:?}", char_indices);
+                eprintln!("Chars bitmap: {:#b}", chars_bitmap);
+            }
+
+            assert_eq!(
+                has_bit, should_have_bit,
+                "Chars bitmap mismatch at byte index {} in chunk {:?}. Expected bit: {}, Got bit: {}",
+                byte_idx, chunk_text, should_have_bit, has_bit
+            );
+        }
+
+        for (byte_idx, byte) in chunk_text.bytes().enumerate() {
+            let is_tab = byte == b'\t';
+            let has_bit = tabs_bitmap & (1 << byte_idx) != 0;
+
+            if has_bit != is_tab {
+                eprintln!("Chunk text bytes: {:?}", chunk_text.as_bytes());
+                eprintln!("Tabs bitmap: {:#b}", tabs_bitmap);
+                assert_eq!(
+                    has_bit, is_tab,
+                    "Tabs bitmap mismatch at byte index {} in chunk {:?}. Byte: {:?}, Expected bit: {}, Got bit: {}",
+                    byte_idx, chunk_text, byte as char, is_tab, has_bit
+                );
+            }
+        }
+    }
+}