2 files changed, 393 insertions, 28 deletions

diff --git a/crates/syntax/Cargo.toml b/crates/syntax/Cargo.toml
index c343f2f70..aa39ce554 100644
--- a/crates/syntax/Cargo.toml
+++ b/crates/syntax/Cargo.toml
@@ -17,6 +17,7 @@ rustc_lexer = { version = "683.0.0", package = "rustc-ap-rustc_lexer" }
 rustc-hash = "1.1.0"
 arrayvec = "0.5.1"
 once_cell = "1.3.1"
+indexmap = "1.4.0"
 # This crate transitively depends on `smol_str` via `rowan`.
 # ideally, `serde` should be enabled by `rust-analyzer`, but we enable it here
 # to reduce number of compilations
@@ -26,10 +27,9 @@ serde = { version = "1.0.106", features = ["derive"] }
 stdx = { path = "../stdx", version = "0.0.0" }
 text_edit = { path = "../text_edit", version = "0.0.0" }
 parser = { path = "../parser", version = "0.0.0" }
+test_utils = { path = "../test_utils" }
 
 [dev-dependencies]
 walkdir = "2.3.1"
 rayon = "1"
 expect-test = "1.0"
-
-test_utils = { path = "../test_utils" }
diff --git a/crates/syntax/src/algo.rs b/crates/syntax/src/algo.rs
index ea199f9b8..4f9a7a6e8 100644
--- a/crates/syntax/src/algo.rs
+++ b/crates/syntax/src/algo.rs
@@ -2,11 +2,14 @@
 
 use std::{
     fmt,
+    hash::BuildHasherDefault,
     ops::{self, RangeInclusive},
 };
 
+use indexmap::IndexMap;
 use itertools::Itertools;
 use rustc_hash::FxHashMap;
+use test_utils::mark;
 use text_edit::TextEditBuilder;
 
 use crate::{
@@ -106,42 +109,56 @@ pub enum InsertPosition<T> {
     After(T),
 }
 
+type FxIndexMap<K, V> = IndexMap<K, V, BuildHasherDefault<rustc_hash::FxHasher>>;
+
+#[derive(Debug)]
 pub struct TreeDiff {
     replacements: FxHashMap<SyntaxElement, SyntaxElement>,
+    deletions: Vec<SyntaxElement>,
+    // the vec as well as the indexmap are both here to preserve order
+    insertions: FxIndexMap<SyntaxElement, Vec<SyntaxElement>>,
 }
 
 impl TreeDiff {
     pub fn into_text_edit(&self, builder: &mut TextEditBuilder) {
+        for (anchor, to) in self.insertions.iter() {
+            to.iter().for_each(|to| builder.insert(anchor.text_range().end(), to.to_string()));
+        }
         for (from, to) in self.replacements.iter() {
             builder.replace(from.text_range(), to.to_string())
         }
+        for text_range in self.deletions.iter().map(SyntaxElement::text_range) {
+            builder.delete(text_range);
+        }
     }
 
     pub fn is_empty(&self) -> bool {
-        self.replacements.is_empty()
+        self.replacements.is_empty() && self.deletions.is_empty() && self.insertions.is_empty()
     }
 }
 
 /// Finds minimal the diff, which, applied to `from`, will result in `to`.
 ///
-/// Specifically, returns a map whose keys are descendants of `from` and values
-/// are descendants of `to`, such that  `replace_descendants(from, map) == to`.
+/// Specifically, returns a structure that consists of a replacements, insertions and deletions
+/// such that applying this map on `from` will result in `to`.
 ///
-/// A trivial solution is a singleton map `{ from: to }`, but this function
-/// tries to find a more fine-grained diff.
+/// This function tries to find a fine-grained diff.
 pub fn diff(from: &SyntaxNode, to: &SyntaxNode) -> TreeDiff {
-    let mut buf = FxHashMap::default();
-    // FIXME: this is both horrible inefficient and gives larger than
-    // necessary diff. I bet there's a cool algorithm to diff trees properly.
-    go(&mut buf, from.clone().into(), to.clone().into());
-    return TreeDiff { replacements: buf };
-
-    fn go(
-        buf: &mut FxHashMap<SyntaxElement, SyntaxElement>,
-        lhs: SyntaxElement,
-        rhs: SyntaxElement,
-    ) {
-        if lhs.kind() == rhs.kind()
+    let mut diff = TreeDiff {
+        replacements: FxHashMap::default(),
+        insertions: FxIndexMap::default(),
+        deletions: Vec::new(),
+    };
+    let (from, to) = (from.clone().into(), to.clone().into());
+
+    // FIXME: this is horrible inefficient. I bet there's a cool algorithm to diff trees properly.
+    if !syntax_element_eq(&from, &to) {
+        go(&mut diff, from, to);
+    }
+    return diff;
+
+    fn syntax_element_eq(lhs: &SyntaxElement, rhs: &SyntaxElement) -> bool {
+        lhs.kind() == rhs.kind()
             && lhs.text_range().len() == rhs.text_range().len()
             && match (&lhs, &rhs) {
                 (NodeOrToken::Node(lhs), NodeOrToken::Node(rhs)) => {
@@ -150,18 +167,47 @@ pub fn diff(from: &SyntaxNode, to: &SyntaxNode) -> TreeDiff {
                 (NodeOrToken::Token(lhs), NodeOrToken::Token(rhs)) => lhs.text() == rhs.text(),
                 _ => false,
             }
-        {
-            return;
-        }
-        if let (Some(lhs), Some(rhs)) = (lhs.as_node(), rhs.as_node()) {
-            if lhs.children_with_tokens().count() == rhs.children_with_tokens().count() {
-                for (lhs, rhs) in lhs.children_with_tokens().zip(rhs.children_with_tokens()) {
-                    go(buf, lhs, rhs)
-                }
+    }
+
+    fn go(diff: &mut TreeDiff, lhs: SyntaxElement, rhs: SyntaxElement) {
+        let (lhs, rhs) = match lhs.as_node().zip(rhs.as_node()) {
+            Some((lhs, rhs)) => (lhs, rhs),
+            _ => {
+                mark::hit!(diff_node_token_replace);
+                diff.replacements.insert(lhs, rhs);
                 return;
             }
+        };
+
+        let mut rhs_children = rhs.children_with_tokens();
+        let mut lhs_children = lhs.children_with_tokens();
+        let mut last_lhs = None;
+        loop {
+            let lhs_child = lhs_children.next();
+            match (lhs_child.clone(), rhs_children.next()) {
+                (None, None) => break,
+                (None, Some(element)) => match last_lhs.clone() {
+                    Some(prev) => {
+                        mark::hit!(diff_insert);
+                        diff.insertions.entry(prev).or_insert_with(Vec::new).push(element);
+                    }
+                    // first iteration, this means we got no anchor element to insert after
+                    // therefor replace the parent node instead
+                    None => {
+                        mark::hit!(diff_replace_parent);
+                        diff.replacements.insert(lhs.clone().into(), rhs.clone().into());
+                        break;
+                    }
+                },
+                (Some(element), None) => {
+                    mark::hit!(diff_delete);
+                    diff.deletions.push(element);
+                }
+                (Some(ref lhs_ele), Some(ref rhs_ele)) if syntax_element_eq(lhs_ele, rhs_ele) => {}
+                (Some(lhs_ele), Some(rhs_ele)) => go(diff, lhs_ele, rhs_ele),
+            }
+            last_lhs = lhs_child.or(last_lhs);
         }
-        buf.insert(lhs, rhs);
     }
 }
 
@@ -404,3 +450,322 @@ fn to_green_element(element: SyntaxElement) -> NodeOrToken<rowan::GreenNode, row
         NodeOrToken::Token(it) => it.green().clone().into(),
     }
 }
+
+#[cfg(test)]
+mod tests {
+    use expect_test::{expect, Expect};
+    use itertools::Itertools;
+    use parser::SyntaxKind;
+    use test_utils::mark;
+    use text_edit::TextEdit;
+
+    use crate::{AstNode, SyntaxElement};
+
+    #[test]
+    fn replace_node_token() {
+        mark::check!(diff_node_token_replace);
+        check_diff(
+            r#"use node;"#,
+            r#"ident"#,
+            expect![[r#"
+                insertions:
+
+
+
+                replacements:
+
+                Line 0: Token([email protected] "use") -> ident
+
+                deletions:
+
+                Line 1: " "
+                Line 1: node
+                Line 1: ;
+            "#]],
+        );
+    }
+
+    #[test]
+    fn insert() {
+        mark::check!(diff_insert);
+        check_diff(
+            r#"use foo;"#,
+            r#"use foo;
+use bar;"#,
+            expect![[r#"
+                insertions:
+
+                Line 0: Node([email protected])
+                -> "\n"
+                -> use bar;
+
+                replacements:
+
+
+
+                deletions:
+
+
+            "#]],
+        );
+    }
+
+    #[test]
+    fn replace_parent() {
+        mark::check!(diff_replace_parent);
+        check_diff(
+            r#""#,
+            r#"use foo::bar;"#,
+            expect![[r#"
+                insertions:
+
+
+
+                replacements:
+
+                Line 0: Node([email protected]) -> use foo::bar;
+
+                deletions:
+
+
+            "#]],
+        );
+    }
+
+    #[test]
+    fn delete() {
+        mark::check!(diff_delete);
+        check_diff(
+            r#"use foo;
+            use bar;"#,
+            r#"use foo;"#,
+            expect![[r#"
+                insertions:
+
+
+
+                replacements:
+
+
+
+                deletions:
+
+                Line 1: "\n            "
+                Line 2: use bar;
+            "#]],
+        );
+    }
+
+    #[test]
+    fn insert_use() {
+        check_diff(
+            r#"
+use expect_test::{expect, Expect};
+
+use crate::AstNode;
+"#,
+            r#"
+use expect_test::{expect, Expect};
+use text_edit::TextEdit;
+
+use crate::AstNode;
+"#,
+            expect![[r#"
+                insertions:
+
+                Line 4: Token([email protected] "\n")
+                -> use crate::AstNode;
+                -> "\n"
+
+                replacements:
+
+                Line 2: Token([email protected] "\n\n") -> "\n"
+                Line 4: Token([email protected] "crate") -> text_edit
+                Line 4: Token([email protected] "AstNode") -> TextEdit
+                Line 4: Token([email protected] "\n") -> "\n\n"
+
+                deletions:
+
+
+            "#]],
+        )
+    }
+
+    #[test]
+    fn remove_use() {
+        check_diff(
+            r#"
+use expect_test::{expect, Expect};
+use text_edit::TextEdit;
+
+use crate::AstNode;
+"#,
+            r#"
+use expect_test::{expect, Expect};
+
+use crate::AstNode;
+"#,
+            expect![[r#"
+                insertions:
+
+
+
+                replacements:
+
+                Line 2: Token([email protected] "\n") -> "\n\n"
+                Line 3: Node([email protected]) -> crate
+                Line 3: Token([email protected] "TextEdit") -> AstNode
+                Line 3: Token([email protected] "\n\n") -> "\n"
+
+                deletions:
+
+                Line 4: use crate::AstNode;
+                Line 5: "\n"
+            "#]],
+        )
+    }
+
+    #[test]
+    fn merge_use() {
+        check_diff(
+            r#"
+use std::{
+    fmt,
+    hash::BuildHasherDefault,
+    ops::{self, RangeInclusive},
+};
+"#,
+            r#"
+use std::fmt;
+use std::hash::BuildHasherDefault;
+use std::ops::{self, RangeInclusive};
+"#,
+            expect![[r#"
+                insertions:
+
+                Line 2: Node([email protected])
+                -> ::
+                -> fmt
+                Line 6: Token([email protected] "\n")
+                -> use std::hash::BuildHasherDefault;
+                -> "\n"
+                -> use std::ops::{self, RangeInclusive};
+                -> "\n"
+
+                replacements:
+
+                Line 2: Token([email protected] "std") -> std
+
+                deletions:
+
+                Line 2: ::
+                Line 2: {
+                    fmt,
+                    hash::BuildHasherDefault,
+                    ops::{self, RangeInclusive},
+                }
+            "#]],
+        )
+    }
+
+    #[test]
+    fn early_return_assist() {
+        check_diff(
+            r#"
+fn main() {
+    if let Ok(x) = Err(92) {
+        foo(x);
+    }
+}
+            "#,
+            r#"
+fn main() {
+    let x = match Err(92) {
+        Ok(it) => it,
+        _ => return,
+    };
+    foo(x);
+}
+            "#,
+            expect![[r#"
+                insertions:
+
+                Line 3: Node([email protected])
+                -> " "
+                -> match Err(92) {
+                        Ok(it) => it,
+                        _ => return,
+                    }
+                -> ;
+                Line 5: Token([email protected] "}")
+                -> "\n"
+                -> }
+
+                replacements:
+
+                Line 3: Token([email protected] "if") -> let
+                Line 3: Token([email protected] "let") -> x
+                Line 3: Node([email protected]) -> =
+                Line 5: Token([email protected] "\n") -> "\n    "
+                Line 5: Token([email protected] "}") -> foo(x);
+
+                deletions:
+
+                Line 3: " "
+                Line 3: Ok(x)
+                Line 3: " "
+                Line 3: =
+                Line 3: " "
+                Line 3: Err(92)
+            "#]],
+        )
+    }
+
+    fn check_diff(from: &str, to: &str, expected_diff: Expect) {
+        let from_node = crate::SourceFile::parse(from).tree().syntax().clone();
+        let to_node = crate::SourceFile::parse(to).tree().syntax().clone();
+        let diff = super::diff(&from_node, &to_node);
+
+        let line_number =
+            |syn: &SyntaxElement| from[..syn.text_range().start().into()].lines().count();
+
+        let fmt_syntax = |syn: &SyntaxElement| match syn.kind() {
+            SyntaxKind::WHITESPACE => format!("{:?}", syn.to_string()),
+            _ => format!("{}", syn),
+        };
+
+        let insertions = diff.insertions.iter().format_with("\n", |(k, v), f| {
+            f(&format!(
+                "Line {}: {:?}\n-> {}",
+                line_number(k),
+                k,
+                v.iter().format_with("\n-> ", |v, f| f(&fmt_syntax(v)))
+            ))
+        });
+
+        let replacements = diff
+            .replacements
+            .iter()
+            .sorted_by_key(|(syntax, _)| syntax.text_range().start())
+            .format_with("\n", |(k, v), f| {
+                f(&format!("Line {}: {:?} -> {}", line_number(k), k, fmt_syntax(v)))
+            });
+
+        let deletions = diff
+            .deletions
+            .iter()
+            .format_with("\n", |v, f| f(&format!("Line {}: {}", line_number(v), &fmt_syntax(v))));
+
+        let actual = format!(
+            "insertions:\n\n{}\n\nreplacements:\n\n{}\n\ndeletions:\n\n{}\n",
+            insertions, replacements, deletions
+        );
+        expected_diff.assert_eq(&actual);
+
+        let mut from = from.to_owned();
+        let mut text_edit = TextEdit::builder();
+        diff.into_text_edit(&mut text_edit);
+        text_edit.finish().apply(&mut from);
+        assert_eq!(&*from, to, "diff did not turn `from` to `to`");
+    }
+}