Ignore unchanged nodes at the start/end of the file

This is the first part of #84.

CHANGELOG.md

@@ -1,5 +1,9 @@
 ## 0.17 (unreleased)
 
+### Diffing
+
+Improved performance when all file changes are close together.
+
 ## 0.16 (released 22 January 2022)
 
 ### Parsing

src/main.rs

@@ -24,6 +24,7 @@ mod style;
 mod summary;
 mod syntax;
 mod tree_sitter_parser;
+mod unchanged;
 
 #[macro_use]
 extern crate log;
@@ -48,6 +49,7 @@ use sliders::fix_all_sliders;
 use std::{env, path::Path};
 use summary::DiffResult;
 use typed_arena::Arena;
+use unchanged::skip_unchanged_at_ends;
 use walkdir::WalkDir;
 
 use crate::{
@@ -305,10 +307,15 @@ fn diff_file_content(display_path: &str, lhs_bytes: &[u8], rhs_bytes: &[u8]) ->
             let rhs = tsp::parse(&arena, &rhs_src, &ts_lang);
 
             init_info(&lhs, &rhs);
-            mark_syntax(lhs.get(0).copied(), rhs.get(0).copied());
 
-            fix_all_sliders(&lhs);
+            let (possibly_changed_lhs, possibly_changed_rhs) = skip_unchanged_at_ends(&lhs, &rhs);
-            fix_all_sliders(&rhs);
+            mark_syntax(
+                possibly_changed_lhs.get(0).copied(),
+                possibly_changed_rhs.get(0).copied(),
+            );
+
+            fix_all_sliders(&possibly_changed_lhs);
+            fix_all_sliders(&possibly_changed_rhs);
 
             let lhs_positions = change_positions(&lhs_src, &rhs_src, &lhs);
             let rhs_positions = change_positions(&rhs_src, &lhs_src, &rhs);

src/syntax.rs

@@ -44,10 +44,10 @@ impl<'a> fmt::Debug for ChangeKind<'a> {
 pub struct SyntaxInfo<'a> {
     /// The syntax node that occurs after this one, in a depth-first
     /// tree traversal.
-    next: Cell<Option<&'a Syntax<'a>>>,
+    pub next: Cell<Option<&'a Syntax<'a>>>,
     /// The syntax node that occurs before this one, in a depth-first
     /// tree traversal.
-    prev: Cell<Option<&'a Syntax<'a>>>,
+    pub prev: Cell<Option<&'a Syntax<'a>>>,
     /// The parent syntax node, if present.
     parent: Cell<Option<&'a Syntax<'a>>>,
     /// Does the previous syntax node occur on the same line as the

src/unchanged.rs

@@ -0,0 +1,95 @@
+use crate::syntax::{ChangeKind, Syntax};
+
+pub fn skip_unchanged_at_ends<'a>(
+    lhs_nodes: &[&'a Syntax<'a>],
+    rhs_nodes: &[&'a Syntax<'a>],
+) -> (Vec<&'a Syntax<'a>>, Vec<&'a Syntax<'a>>) {
+    let mut lhs_nodes = lhs_nodes;
+    let mut rhs_nodes = rhs_nodes;
+
+    while let (Some(lhs_node), Some(rhs_node)) = (lhs_nodes.first(), rhs_nodes.first()) {
+        if lhs_node.content_id() == rhs_node.content_id() {
+            skip_pair(lhs_node, rhs_node);
+
+            lhs_nodes = &lhs_nodes[1..];
+            rhs_nodes = &rhs_nodes[1..];
+        } else {
+            break;
+        }
+    }
+
+    while let (Some(lhs_node), Some(rhs_node)) = (lhs_nodes.last(), rhs_nodes.last()) {
+        if lhs_node.content_id() == rhs_node.content_id() {
+            skip_pair(lhs_node, rhs_node);
+
+            lhs_nodes = &lhs_nodes[..lhs_nodes.len() - 1];
+            rhs_nodes = &rhs_nodes[..rhs_nodes.len() - 1];
+        } else {
+            break;
+        }
+    }
+
+    (Vec::from(lhs_nodes), Vec::from(rhs_nodes))
+}
+
+fn skip_pair<'a>(lhs_node: &'a Syntax<'a>, rhs_node: &'a Syntax<'a>) {
+    lhs_node.set_change_deep(ChangeKind::Unchanged(rhs_node));
+    rhs_node.set_change_deep(ChangeKind::Unchanged(lhs_node));
+
+    detach(lhs_node);
+    detach(rhs_node);
+}
+
+fn detach(node: &Syntax) {
+    if let Some(next_node) = node.next() {
+        node.info().next.set(None);
+        next_node.info().prev.set(None);
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::{
+        guess_language,
+        syntax::init_info,
+        tree_sitter_parser::{from_language, parse},
+    };
+    use typed_arena::Arena;
+
+    #[test]
+    fn test_unchanged_at_start() {
+        let arena = Arena::new();
+        let config = from_language(guess_language::Language::EmacsLisp);
+
+        let lhs_nodes = parse(&arena, "unchanged A B", &config);
+        let rhs_nodes = parse(&arena, "unchanged X", &config);
+        init_info(&lhs_nodes, &rhs_nodes);
+
+        let (lhs_after_skip, rhs_after_skip) = skip_unchanged_at_ends(&lhs_nodes, &rhs_nodes);
+
+        assert_eq!(lhs_nodes[0].change(), Some(ChangeKind::Unchanged(rhs_nodes[0])));
+        assert_eq!(rhs_nodes[0].change(), Some(ChangeKind::Unchanged(lhs_nodes[0])));
+
+        assert_eq!(lhs_after_skip.len(), 2);
+        assert_eq!(rhs_after_skip.len(), 1);
+    }
+
+    #[test]
+    fn test_unchanged_at_end() {
+        let arena = Arena::new();
+        let config = from_language(guess_language::Language::EmacsLisp);
+
+        let lhs_nodes = parse(&arena, "A B unchanged", &config);
+        let rhs_nodes = parse(&arena, "X unchanged", &config);
+        init_info(&lhs_nodes, &rhs_nodes);
+
+        let (lhs_after_skip, rhs_after_skip) = skip_unchanged_at_ends(&lhs_nodes, &rhs_nodes);
+
+        assert_eq!(lhs_nodes[2].change(), Some(ChangeKind::Unchanged(rhs_nodes[1])));
+        assert_eq!(rhs_nodes[1].change(), Some(ChangeKind::Unchanged(lhs_nodes[2])));
+
+        assert_eq!(lhs_after_skip.len(), 2);
+        assert_eq!(rhs_after_skip.len(), 1);
+    }
+}