Add caller/reference analysis, C# imports, and de-noised search

Agent-workflow tooling and search quality improvements:

- find_callers / find_references (by id or exact name), with caller
  attribution to the innermost enclosing symbol and import-graph narrowing
- index file (single-file re-index) and workspace glob member resolution
- search symbols: exact > prefix > fulltext > substring fallback, inline
  source and line numbers, --limit on all search paths
- search text: reads live from disk (no stale FTS snapshot), collapses
  duplicate snippets, correct multi-word phrase matching
- find dead-symbols: excludes framework-wired entry points (pytest test_*,
  #[test] tests, FastAPI/Flask routes, Celery tasks, click/typer commands)
- find references: filters the definition line and string/comment
  occurrences (incl. Rust raw strings)
- C# import edges via new extract_csharp (using directives)
- richer MCP tool descriptions; brace-in-string tokenizer fix

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

src/index/schema.rs

@@ -1,19 +1,21 @@
-//! SQLite schema + minimal migration (schema_version in metadata).
+//! SQLite schema + migration (schema_version in metadata).
 
-use anyhow::Result;
+use anyhow::{Context, Result};
 use rusqlite::Connection;
 
-const SCHEMA_VERSION: i64 = 1;
+const SCHEMA_VERSION: i64 = 2;
 
+/// Fresh-install schema at the current version. Everything is `IF NOT EXISTS`
+/// so it can be re-run safely (the old version re-ran it on every `version <
+/// SCHEMA_VERSION`, which broke the moment any table already existed).
 const SCHEMA: &str = r#"
-CREATE TABLE files (
+CREATE TABLE IF NOT EXISTS files (
   id INTEGER PRIMARY KEY,
   path TEXT NOT NULL UNIQUE,
   content_hash TEXT NOT NULL,
-  language TEXT NOT NULL,
-  mtime INTEGER NOT NULL
+  language TEXT NOT NULL
 );
-CREATE TABLE symbols (
+CREATE TABLE IF NOT EXISTS symbols (
   id INTEGER PRIMARY KEY,
   file_id INTEGER NOT NULL REFERENCES files(id) ON DELETE CASCADE,
   name TEXT NOT NULL,
@@ -23,31 +25,40 @@ CREATE TABLE symbols (
   line_start INTEGER NOT NULL,
   line_end INTEGER NOT NULL
 );
-CREATE INDEX idx_symbols_file ON symbols(file_id);
-CREATE INDEX idx_symbols_name ON symbols(name);
-CREATE TABLE imports (
+CREATE INDEX IF NOT EXISTS idx_symbols_file ON symbols(file_id);
+CREATE INDEX IF NOT EXISTS idx_symbols_name ON symbols(name);
+CREATE TABLE IF NOT EXISTS imports (
   id INTEGER PRIMARY KEY,
   from_file_id INTEGER NOT NULL REFERENCES files(id) ON DELETE CASCADE,
   to_file_id INTEGER NOT NULL REFERENCES files(id) ON DELETE CASCADE,
   import_type TEXT NOT NULL
 );
-CREATE INDEX idx_imports_from ON imports(from_file_id);
-CREATE INDEX idx_imports_to ON imports(to_file_id);
-CREATE VIRTUAL TABLE symbols_fts USING fts5(
+CREATE INDEX IF NOT EXISTS idx_imports_from ON imports(from_file_id);
+CREATE INDEX IF NOT EXISTS idx_imports_to ON imports(to_file_id);
+CREATE VIRTUAL TABLE IF NOT EXISTS symbols_fts USING fts5(
   name, qualified_name, kind, signature,
   content='symbols', content_rowid='id'
 );
-CREATE TRIGGER symbols_ai AFTER INSERT ON symbols BEGIN
+CREATE TRIGGER IF NOT EXISTS symbols_ai AFTER INSERT ON symbols BEGIN
   INSERT INTO symbols_fts(rowid, name, qualified_name, kind, signature)
   VALUES (new.id, new.name, new.qualified_name, new.kind, new.signature);
 END;
-CREATE TRIGGER symbols_ad AFTER DELETE ON symbols BEGIN
+CREATE TRIGGER IF NOT EXISTS symbols_ad AFTER DELETE ON symbols BEGIN
   INSERT INTO symbols_fts(symbols_fts, rowid, name, qualified_name, kind, signature)
   VALUES ('delete', old.id, old.name, old.qualified_name, old.kind, old.signature);
 END;
-CREATE VIRTUAL TABLE content_fts USING fts5(path UNINDEXED, content);
+CREATE VIRTUAL TABLE IF NOT EXISTS content_fts USING fts5(path UNINDEXED, content);
 "#;
 
+/// Incremental migrations. `MIGRATIONS[i]` upgrades from schema version `i+1`
+/// to `i+2`; version 0 is a fresh database and is created via [`SCHEMA`]
+/// directly.
+const MIGRATIONS: &[&str] = &[
+    // v1 -> v2: the `mtime` column was written but never read (change detection
+    // is content-hash based). Drop the dead column.
+    "ALTER TABLE files DROP COLUMN mtime;",
+];
+
 /// Create schema if missing, or migrate to the current version.
 pub fn init(conn: &mut Connection) -> Result<()> {
     conn.execute_batch(
@@ -60,12 +71,22 @@ pub fn init(conn: &mut Connection) -> Result<()> {
             |r| r.get(0),
         )
         .unwrap_or(0);
-    if version < SCHEMA_VERSION {
+
+    if version == 0 {
         conn.execute_batch(SCHEMA)?;
-        conn.execute(
-            "INSERT OR REPLACE INTO metadata (key, value) VALUES ('schema_version', ?1)",
-            [SCHEMA_VERSION.to_string()],
-        )?;
+    } else {
+        for (i, migration) in MIGRATIONS.iter().enumerate() {
+            // i-th migration upgrades version (i+1) -> (i+2).
+            if version <= (i as i64) + 1 {
+                conn.execute_batch(migration)
+                    .with_context(|| format!("migration to schema v{} failed", i + 2))?;
+            }
+        }
     }
+
+    conn.execute(
+        "INSERT OR REPLACE INTO metadata (key, value) VALUES ('schema_version', ?1)",
+        [SCHEMA_VERSION.to_string()],
+    )?;
     Ok(())
 }