Merge #5154 #5157

5154: Structured search debugging r=matklad a=davidlattimore

Adds a "search" mode to the rust-analyzer binary that does structured search (SSR without the replace part). This is intended primarily for debugging why a bit of code isn't matching a pattern.

5157: Use dynamic dispatch in AstDiagnostic r=matklad a=lnicola



Co-authored-by: David Lattimore <[email protected]>
Co-authored-by: Laurențiu Nicola <[email protected]>

Author: 3 people

crates/ra_hir_expand/src/diagnostics.rs

@@ -28,7 +28,7 @@ pub trait Diagnostic: Any + Send + Sync + fmt::Debug + 'static {
 
 pub trait AstDiagnostic {
     type AST;
-    fn ast(&self, db: &impl AstDatabase) -> Self::AST;
+    fn ast(&self, db: &dyn AstDatabase) -> Self::AST;
 }
 
 impl dyn Diagnostic {

crates/ra_hir_ty/src/diagnostics.rs

@@ -32,7 +32,7 @@ impl Diagnostic for NoSuchField {
 impl AstDiagnostic for NoSuchField {
     type AST = ast::RecordField;
 
-    fn ast(&self, db: &impl AstDatabase) -> Self::AST {
+    fn ast(&self, db: &dyn AstDatabase) -> Self::AST {
         let root = db.parse_or_expand(self.source().file_id).unwrap();
         let node = self.source().value.to_node(&root);
         ast::RecordField::cast(node).unwrap()
@@ -65,7 +65,7 @@ impl Diagnostic for MissingFields {
 impl AstDiagnostic for MissingFields {
     type AST = ast::RecordFieldList;
 
-    fn ast(&self, db: &impl AstDatabase) -> Self::AST {
+    fn ast(&self, db: &dyn AstDatabase) -> Self::AST {
         let root = db.parse_or_expand(self.source().file_id).unwrap();
         let node = self.source().value.to_node(&root);
         ast::RecordFieldList::cast(node).unwrap()
@@ -135,7 +135,7 @@ impl Diagnostic for MissingOkInTailExpr {
 impl AstDiagnostic for MissingOkInTailExpr {
     type AST = ast::Expr;
 
-    fn ast(&self, db: &impl AstDatabase) -> Self::AST {
+    fn ast(&self, db: &dyn AstDatabase) -> Self::AST {
         let root = db.parse_or_expand(self.file).unwrap();
         let node = self.source().value.to_node(&root);
         ast::Expr::cast(node).unwrap()
@@ -163,7 +163,7 @@ impl Diagnostic for BreakOutsideOfLoop {
 impl AstDiagnostic for BreakOutsideOfLoop {
     type AST = ast::Expr;
 
-    fn ast(&self, db: &impl AstDatabase) -> Self::AST {
+    fn ast(&self, db: &dyn AstDatabase) -> Self::AST {
         let root = db.parse_or_expand(self.file).unwrap();
         let node = self.source().value.to_node(&root);
         ast::Expr::cast(node).unwrap()
@@ -191,7 +191,7 @@ impl Diagnostic for MissingUnsafe {
 impl AstDiagnostic for MissingUnsafe {
     type AST = ast::Expr;
 
-    fn ast(&self, db: &impl AstDatabase) -> Self::AST {
+    fn ast(&self, db: &dyn AstDatabase) -> Self::AST {
         let root = db.parse_or_expand(self.source().file_id).unwrap();
         let node = self.source().value.to_node(&root);
         ast::Expr::cast(node).unwrap()

crates/ra_ssr/src/lib.rs

@@ -9,10 +9,11 @@ mod replacing;
 #[cfg(test)]
 mod tests;
 
-use crate::matching::Match;
+pub use crate::matching::Match;
+use crate::matching::{record_match_fails_reasons_scope, MatchFailureReason};
 use hir::Semantics;
 use ra_db::{FileId, FileRange};
-use ra_syntax::{ast, AstNode, SmolStr, SyntaxNode};
+use ra_syntax::{ast, AstNode, SmolStr, SyntaxKind, SyntaxNode, TextRange};
 use ra_text_edit::TextEdit;
 use rustc_hash::FxHashMap;
 
@@ -26,7 +27,7 @@ pub struct SsrRule {
 }
 
 #[derive(Debug)]
-struct SsrPattern {
+pub struct SsrPattern {
     raw: parsing::RawSearchPattern,
     /// Placeholders keyed by the stand-in ident that we use in Rust source code.
     placeholders_by_stand_in: FxHashMap<SmolStr, parsing::Placeholder>,
@@ -45,7 +46,7 @@ pub struct SsrError(String);
 
 #[derive(Debug, Default)]
 pub struct SsrMatches {
-    matches: Vec<Match>,
+    pub matches: Vec<Match>,
 }
 
 /// Searches a crate for pattern matches and possibly replaces them with something else.
@@ -64,6 +65,12 @@ impl<'db> MatchFinder<'db> {
         self.rules.push(rule);
     }
 
+    /// Adds a search pattern. For use if you intend to only call `find_matches_in_file`. If you
+    /// intend to do replacement, use `add_rule` instead.
+    pub fn add_search_pattern(&mut self, pattern: SsrPattern) {
+        self.add_rule(SsrRule { pattern, template: "()".parse().unwrap() })
+    }
+
     pub fn edits_for_file(&self, file_id: FileId) -> Option<TextEdit> {
         let matches = self.find_matches_in_file(file_id);
         if matches.matches.is_empty() {
@@ -74,14 +81,40 @@ impl<'db> MatchFinder<'db> {
         }
     }
 
-    fn find_matches_in_file(&self, file_id: FileId) -> SsrMatches {
+    pub fn find_matches_in_file(&self, file_id: FileId) -> SsrMatches {
         let file = self.sema.parse(file_id);
         let code = file.syntax();
         let mut matches = SsrMatches::default();
         self.find_matches(code, &None, &mut matches);
         matches
     }
 
+    /// Finds all nodes in `file_id` whose text is exactly equal to `snippet` and attempts to match
+    /// them, while recording reasons why they don't match. This API is useful for command
+    /// line-based debugging where providing a range is difficult.
+    pub fn debug_where_text_equal(&self, file_id: FileId, snippet: &str) -> Vec<MatchDebugInfo> {
+        use ra_db::SourceDatabaseExt;
+        let file = self.sema.parse(file_id);
+        let mut res = Vec::new();
+        let file_text = self.sema.db.file_text(file_id);
+        let mut remaining_text = file_text.as_str();
+        let mut base = 0;
+        let len = snippet.len() as u32;
+        while let Some(offset) = remaining_text.find(snippet) {
+            let start = base + offset as u32;
+            let end = start + len;
+            self.output_debug_for_nodes_at_range(
+                file.syntax(),
+                FileRange { file_id, range: TextRange::new(start.into(), end.into()) },
+                &None,
+                &mut res,
+            );
+            remaining_text = &remaining_text[offset + snippet.len()..];
+            base = end;
+        }
+        res
+    }
+
     fn find_matches(
         &self,
         code: &SyntaxNode,
@@ -128,6 +161,59 @@ impl<'db> MatchFinder<'db> {
             self.find_matches(&child, restrict_range, matches_out);
         }
     }
+
+    fn output_debug_for_nodes_at_range(
+        &self,
+        node: &SyntaxNode,
+        range: FileRange,
+        restrict_range: &Option<FileRange>,
+        out: &mut Vec<MatchDebugInfo>,
+    ) {
+        for node in node.children() {
+            let node_range = self.sema.original_range(&node);
+            if node_range.file_id != range.file_id || !node_range.range.contains_range(range.range)
+            {
+                continue;
+            }
+            if node_range.range == range.range {
+                for rule in &self.rules {
+                    let pattern =
+                        rule.pattern.tree_for_kind_with_reason(node.kind()).map(|p| p.clone());
+                    out.push(MatchDebugInfo {
+                        matched: matching::get_match(true, rule, &node, restrict_range, &self.sema)
+                            .map_err(|e| MatchFailureReason {
+                                reason: e.reason.unwrap_or_else(|| {
+                                    "Match failed, but no reason was given".to_owned()
+                                }),
+                            }),
+                        pattern,
+                        node: node.clone(),
+                    });
+                }
+            } else if let Some(macro_call) = ast::MacroCall::cast(node.clone()) {
+                if let Some(expanded) = self.sema.expand(&macro_call) {
+                    if let Some(tt) = macro_call.token_tree() {
+                        self.output_debug_for_nodes_at_range(
+                            &expanded,
+                            range,
+                            &Some(self.sema.original_range(tt.syntax())),
+                            out,
+                        );
+                    }
+                }
+            } else {
+                self.output_debug_for_nodes_at_range(&node, range, restrict_range, out);
+            }
+        }
+    }
+}
+
+pub struct MatchDebugInfo {
+    node: SyntaxNode,
+    /// Our search pattern parsed as the same kind of syntax node as `node`. e.g. expression, item,
+    /// etc. Will be absent if the pattern can't be parsed as that kind.
+    pattern: Result<SyntaxNode, MatchFailureReason>,
+    matched: Result<Match, MatchFailureReason>,
 }
 
 impl std::fmt::Display for SsrError {
@@ -136,4 +222,70 @@ impl std::fmt::Display for SsrError {
     }
 }
 
+impl std::fmt::Debug for MatchDebugInfo {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        write!(f, "========= PATTERN ==========\n")?;
+        match &self.pattern {
+            Ok(pattern) => {
+                write!(f, "{:#?}", pattern)?;
+            }
+            Err(err) => {
+                write!(f, "{}", err.reason)?;
+            }
+        }
+        write!(
+            f,
+            "\n============ AST ===========\n\
+            {:#?}\n============================\n",
+            self.node
+        )?;
+        match &self.matched {
+            Ok(_) => write!(f, "Node matched")?,
+            Err(reason) => write!(f, "Node failed to match because: {}", reason.reason)?,
+        }
+        Ok(())
+    }
+}
+
+impl SsrPattern {
+    fn tree_for_kind_with_reason(
+        &self,
+        kind: SyntaxKind,
+    ) -> Result<&SyntaxNode, MatchFailureReason> {
+        record_match_fails_reasons_scope(true, || self.tree_for_kind(kind))
+            .map_err(|e| MatchFailureReason { reason: e.reason.unwrap() })
+    }
+}
+
+impl SsrMatches {
+    /// Returns `self` with any nested matches removed and made into top-level matches.
+    pub fn flattened(self) -> SsrMatches {
+        let mut out = SsrMatches::default();
+        self.flatten_into(&mut out);
+        out
+    }
+
+    fn flatten_into(self, out: &mut SsrMatches) {
+        for mut m in self.matches {
+            for p in m.placeholder_values.values_mut() {
+                std::mem::replace(&mut p.inner_matches, SsrMatches::default()).flatten_into(out);
+            }
+            out.matches.push(m);
+        }
+    }
+}
+
+impl Match {
+    pub fn matched_text(&self) -> String {
+        self.matched_node.text().to_string()
+    }
+}
+
 impl std::error::Error for SsrError {}
+
+#[cfg(test)]
+impl MatchDebugInfo {
+    pub(crate) fn match_failure_reason(&self) -> Option<&str> {
+        self.matched.as_ref().err().map(|r| r.reason.as_str())
+    }
+}

crates/ra_ssr/src/matching.rs

@@ -8,9 +8,7 @@ use crate::{
 use hir::Semantics;
 use ra_db::FileRange;
 use ra_syntax::ast::{AstNode, AstToken};
-use ra_syntax::{
-    ast, SyntaxElement, SyntaxElementChildren, SyntaxKind, SyntaxNode, SyntaxToken, TextRange,
-};
+use ra_syntax::{ast, SyntaxElement, SyntaxElementChildren, SyntaxKind, SyntaxNode, SyntaxToken};
 use rustc_hash::FxHashMap;
 use std::{cell::Cell, iter::Peekable};
 
@@ -44,8 +42,8 @@ macro_rules! fail_match {
 
 /// Information about a match that was found.
 #[derive(Debug)]
-pub(crate) struct Match {
-    pub(crate) range: TextRange,
+pub struct Match {
+    pub(crate) range: FileRange,
     pub(crate) matched_node: SyntaxNode,
     pub(crate) placeholder_values: FxHashMap<Var, PlaceholderMatch>,
     pub(crate) ignored_comments: Vec<ast::Comment>,
@@ -135,7 +133,7 @@ impl<'db, 'sema> MatchState<'db, 'sema> {
         match_state.attempt_match_node(&match_inputs, &pattern_tree, code)?;
         match_state.validate_range(&sema.original_range(code))?;
         match_state.match_out = Some(Match {
-            range: sema.original_range(code).range,
+            range: sema.original_range(code),
             matched_node: code.clone(),
             placeholder_values: FxHashMap::default(),
             ignored_comments: Vec::new(),

crates/ra_ssr/src/replacing.rs

@@ -21,8 +21,10 @@ fn matches_to_edit_at_offset(
 ) -> TextEdit {
     let mut edit_builder = ra_text_edit::TextEditBuilder::default();
     for m in &matches.matches {
-        edit_builder
-            .replace(m.range.checked_sub(relative_start).unwrap(), render_replace(m, file_src));
+        edit_builder.replace(
+            m.range.range.checked_sub(relative_start).unwrap(),
+            render_replace(m, file_src),
+        );
     }
     edit_builder.finish()
 }