prompt refactor

crates/pdfs-core/src/db/local.rs

@@ -7,6 +7,7 @@ use super::Db;
 use crate::Result;
 use crate::localindex::LocalEntry;
 
+use super::nodes::candidate_trigrams;
 use super::utils::{TRIGRAM_MIN, like_escape};
 
 /// One hit from [`Db::search_local`]: an indexed file on the machine itself, not
@@ -147,6 +148,58 @@ impl Db {
         }
         Ok(out)
     }
+
+    /// Bounded candidates for the fuzzy scorer, including matches in the full
+    /// local path. ORed trigrams admit misspellings while the FTS index prevents
+    /// this from becoming an unbounded filesystem-table scan.
+    pub fn search_local_candidates(&self, query: &str, limit: usize) -> Result<Vec<LocalFileHit>> {
+        let query = query.trim();
+        if query.is_empty() || limit == 0 {
+            return Ok(Vec::new());
+        }
+        let conn = self.conn.lock();
+        let terms = candidate_trigrams(query);
+        let lane_limit = limit.div_ceil(2);
+        let mut rows: Vec<LocalFileHit> = if terms.is_empty() {
+            let pat = format!("%{}%", like_escape(query));
+            let mut stmt = conn.prepare(
+                "SELECT path, name, is_dir, size, mtime FROM local_files
+                 WHERE name LIKE ?1 ESCAPE '\\' OR path LIKE ?1 ESCAPE '\\'
+                 ORDER BY mtime DESC LIMIT ?2",
+            )?;
+            stmt.query_map(params![pat, limit as i64], local_hit)?
+                .collect::<std::result::Result<_, _>>()?
+        } else {
+            let mut stmt = conn.prepare(
+                "SELECT f.path, f.name, f.is_dir, f.size, f.mtime
+                   FROM local_fts x JOIN local_files f ON f.id = x.rowid
+                  WHERE local_fts MATCH ?1
+                  ORDER BY x.rank LIMIT ?2",
+            )?;
+            stmt.query_map(params![terms.join(" OR "), lane_limit as i64], local_hit)?
+                .collect::<std::result::Result<_, _>>()?
+        };
+        if !terms.is_empty()
+            && let Some(first) = query.chars().next()
+        {
+            let prefix = format!("{}%", like_escape(&first.to_string()));
+            let mut stmt = conn.prepare(
+                "SELECT path, name, is_dir, size, mtime FROM local_files
+                 WHERE name COLLATE NOCASE LIKE ?1 ESCAPE '\\'
+                 ORDER BY name COLLATE NOCASE LIMIT ?2",
+            )?;
+            let fallback = stmt
+                .query_map(params![prefix, lane_limit as i64], local_hit)?
+                .collect::<std::result::Result<Vec<_>, _>>()?;
+            for hit in fallback {
+                if !rows.iter().any(|existing| existing.path == hit.path) {
+                    rows.push(hit);
+                }
+            }
+            rows.truncate(limit);
+        }
+        Ok(rows)
+    }
 }
 
 /// Row → [`LocalFileHit`]. Every local-search query selects the same

crates/pdfs-core/src/db/migrations.rs

@@ -9,7 +9,7 @@ use super::Db;
 use crate::Result;
 
 /// Current schema version. Bump on every forward migration added below.
-pub(super) const SCHEMA_VERSION: i64 = 15;
+pub(super) const SCHEMA_VERSION: i64 = 16;
 
 impl Db {
     pub(super) fn migrate(&self) -> Result<()> {
@@ -81,6 +81,9 @@ impl Db {
         if current < 15 {
             tx.execute_batch(MIGRATION_V15)?;
         }
+        if current < 16 {
+            tx.execute_batch(MIGRATION_V16)?;
+        }
         tx.execute(
             "INSERT INTO sync_state (key, value) VALUES ('schema_version', ?1)
              ON CONFLICT(key) DO UPDATE SET value = excluded.value",
@@ -369,3 +372,37 @@ CREATE VIRTUAL TABLE nodes_fts USING fts5(
 INSERT INTO nodes_fts (rowid, name)
   SELECT rowid, name FROM nodes WHERE trashed = 0;
 ";
+
+/// Schema v16: fuzzy search needs a bounded candidate set that includes parent
+/// paths.  Keeping the path in the trigram indexes lets SQLite narrow both name
+/// and path candidates without walking either base table for every keystroke.
+/// Node paths are initially backfilled below and subsequently maintained by the
+/// node write path (including descendants after a folder move/rename).
+const MIGRATION_V16: &str = "
+DROP TABLE IF EXISTS nodes_fts;
+CREATE VIRTUAL TABLE nodes_fts USING fts5(
+  name, path, tokenize='trigram'
+);
+CREATE INDEX IF NOT EXISTS idx_nodes_name_nocase ON nodes(name COLLATE NOCASE);
+INSERT INTO nodes_fts (rowid, name, path)
+WITH RECURSIVE paths(rowid, uid, path) AS (
+  SELECT n.rowid, n.uid, '' FROM nodes n
+   WHERE n.parent_uid IS NULL
+      OR NOT EXISTS (SELECT 1 FROM nodes p WHERE p.uid = n.parent_uid)
+  UNION ALL
+  SELECT n.rowid, n.uid,
+         CASE WHEN paths.path = '' THEN n.name ELSE paths.path || '/' || n.name END
+    FROM nodes n JOIN paths ON n.parent_uid = paths.uid
+)
+SELECT n.rowid, n.name, paths.path
+  FROM nodes n JOIN paths ON paths.rowid = n.rowid
+ WHERE n.trashed = 0;
+
+DROP TABLE IF EXISTS local_fts;
+CREATE VIRTUAL TABLE local_fts USING fts5(
+  name, path, content='local_files', content_rowid='id', tokenize='trigram'
+);
+CREATE INDEX IF NOT EXISTS idx_local_files_name_nocase
+  ON local_files(name COLLATE NOCASE);
+INSERT INTO local_fts(local_fts) VALUES('rebuild');
+";

crates/pdfs-core/src/db/nodes.rs

@@ -81,12 +81,42 @@ impl Db {
                 params![uid],
                 |row| row.get(0),
             )?;
-            tx.execute("DELETE FROM nodes_fts WHERE rowid = ?1", params![rowid])?;
-            if !node.trashed {
-                tx.execute(
-                    "INSERT INTO nodes_fts (rowid, name) VALUES (?1, ?2)",
-                    params![rowid, node.name],
+            // A folder rename/move changes every descendant's searchable path.
+            // Refresh the affected subtree, while retaining efficient rowid
+            // deletes for the overwhelmingly common one-file case.
+            let mut affected = vec![rowid];
+            if node.is_folder() {
+                let mut stmt = tx.prepare(
+                    "WITH RECURSIVE descendants(rowid, uid) AS (
+                       SELECT rowid, uid FROM nodes WHERE parent_uid = ?1
+                       UNION ALL
+                       SELECT n.rowid, n.uid FROM nodes n
+                         JOIN descendants d ON n.parent_uid = d.uid
+                     ) SELECT rowid FROM descendants",
                 )?;
+                for descendant in stmt.query_map(params![uid], |row| row.get(0))? {
+                    affected.push(descendant?);
+                }
+            }
+            for affected_rowid in affected {
+                tx.execute(
+                    "DELETE FROM nodes_fts WHERE rowid = ?1",
+                    params![affected_rowid],
+                )?;
+                let indexed: Option<(String, String)> = tx
+                    .query_row(
+                        "SELECT name, uid FROM nodes WHERE rowid = ?1 AND trashed = 0",
+                        params![affected_rowid],
+                        |row| Ok((row.get(0)?, row.get(1)?)),
+                    )
+                    .optional()?;
+                if let Some((name, indexed_uid)) = indexed {
+                    let path = path_of(&tx, &indexed_uid)?;
+                    tx.execute(
+                        "INSERT INTO nodes_fts (rowid, name, path) VALUES (?1, ?2, ?3)",
+                        params![affected_rowid, name, path],
+                    )?;
+                }
             }
         }
         tx.commit()?;
@@ -175,6 +205,68 @@ impl Db {
         Ok(hits)
     }
 
+    /// Return a bounded, deliberately broad candidate pool for fuzzy ranking.
+    /// Unlike [`search`](Self::search), trigram terms are ORed across both the
+    /// basename and parent path, so a typo can still share enough trigrams to
+    /// enter the pool. Final relevance ordering belongs to the caller.
+    pub fn search_candidates(&self, query: &str, limit: usize) -> Result<Vec<SearchHit>> {
+        let query = query.trim();
+        if query.is_empty() || limit == 0 {
+            return Ok(Vec::new());
+        }
+        let conn = self.conn.lock();
+        let terms = candidate_trigrams(query);
+        let lane_limit = limit.div_ceil(2);
+        let mut rows: Vec<(String, String)> = if terms.is_empty() {
+            let pat = format!("%{}%", like_escape(query));
+            let mut stmt = conn.prepare(
+                "SELECT node_json, uid FROM nodes
+                 WHERE name LIKE ?1 ESCAPE '\\' AND trashed = 0 AND node_json IS NOT NULL
+                 ORDER BY name LIMIT ?2",
+            )?;
+            collect_pairs(stmt.query_map(params![pat, limit as i64], pair)?)?
+        } else {
+            let expression = terms.join(" OR ");
+            let mut stmt = conn.prepare(
+                "SELECT n.node_json, n.uid
+                   FROM nodes_fts f JOIN nodes n ON n.rowid = f.rowid
+                  WHERE nodes_fts MATCH ?1 AND n.trashed = 0 AND n.node_json IS NOT NULL
+                  ORDER BY f.rank LIMIT ?2",
+            )?;
+            collect_pairs(stmt.query_map(params![expression, lane_limit as i64], pair)?)?
+        };
+        // A short substitution can destroy every trigram (`vedio` vs `video`).
+        // Feed the scorer an additional indexed same-initial lane so it can
+        // recover those candidates without scanning the full node table.
+        if !terms.is_empty()
+            && let Some(first) = query.chars().next()
+        {
+            let prefix = format!("{}%", like_escape(&first.to_string()));
+            let mut stmt = conn.prepare(
+                "SELECT node_json, uid FROM nodes
+                 WHERE name COLLATE NOCASE LIKE ?1 ESCAPE '\\'
+                   AND trashed = 0 AND node_json IS NOT NULL
+                 ORDER BY name COLLATE NOCASE LIMIT ?2",
+            )?;
+            let fallback =
+                collect_pairs(stmt.query_map(params![prefix, lane_limit as i64], pair)?)?;
+            for row in fallback {
+                if !rows.iter().any(|(_, uid)| uid == &row.1) {
+                    rows.push(row);
+                }
+            }
+            rows.truncate(limit);
+        }
+        rows.into_iter()
+            .map(|(json, uid)| {
+                Ok(SearchHit {
+                    node: serde_json::from_str(&json)?,
+                    path: path_of(&conn, &uid)?,
+                })
+            })
+            .collect()
+    }
+
     /// Mark (or unmark) a folder's child listing as complete. A listed folder
     /// rehydrates its `children` map on mount even when empty; an unlisted one
     /// re-enumerates from the remote on next access.
@@ -261,6 +353,23 @@ impl Db {
     // in sync on every store/read/evict, so it is authoritative for eviction.
 }
 
+/// Unique quoted character trigrams suitable for an FTS5 OR expression.
+pub(super) fn candidate_trigrams(query: &str) -> Vec<String> {
+    let chars: Vec<char> = query.to_lowercase().chars().collect();
+    let mut terms = Vec::new();
+    for window in chars.windows(TRIGRAM_MIN) {
+        if window.iter().all(|c| c.is_whitespace()) {
+            continue;
+        }
+        let raw: String = window.iter().collect();
+        let quoted = format!("\"{}\"", raw.replace('"', "\"\""));
+        if !terms.contains(&quoted) {
+            terms.push(quoted);
+        }
+    }
+    terms
+}
+
 /// Effective plaintext size of a node for the indexed `size` column: the
 /// claimed size when known, else the on-storage size; folders are 0.
 pub(super) fn node_size(node: &Node) -> i64 {

crates/pdfs-core/src/db/tests.rs

@@ -542,6 +542,95 @@ fn search_short_query_like_fallback() {
     assert_eq!(hits[0].node.name, "ab.txt");
 }
 
+#[test]
+fn fuzzy_candidates_include_typo_and_parent_path_matches() {
+    let db = Db::open_in_memory().unwrap();
+    db.upsert_node(&folder("root", None, "My Files")).unwrap();
+    db.upsert_node(&folder("projects", Some("root"), "Projects"))
+        .unwrap();
+    db.upsert_node(&file("report", "projects", "quarterly-report.pdf", 1))
+        .unwrap();
+    db.upsert_node(&file("video", "projects", "video.mp4", 1))
+        .unwrap();
+
+    let typo = db.search_candidates("quartrly", 20).unwrap();
+    assert!(
+        typo.iter()
+            .any(|hit| hit.node.name == "quarterly-report.pdf")
+    );
+    assert!(
+        db.search_candidates("vedio", 20)
+            .unwrap()
+            .iter()
+            .any(|hit| hit.node.name == "video.mp4")
+    );
+
+    let path = db.search_candidates("projects", 20).unwrap();
+    assert!(
+        path.iter()
+            .any(|hit| hit.node.name == "quarterly-report.pdf")
+    );
+}
+
+#[test]
+fn renaming_parent_refreshes_descendant_search_paths() {
+    let db = Db::open_in_memory().unwrap();
+    db.upsert_node(&folder("root", None, "My Files")).unwrap();
+    db.upsert_node(&folder("folder", Some("root"), "zzbeforezz"))
+        .unwrap();
+    db.upsert_node(&file("child", "folder", "plain.txt", 1))
+        .unwrap();
+    assert!(
+        db.search_candidates("zzbeforezz", 20)
+            .unwrap()
+            .iter()
+            .any(|h| h.node.name == "plain.txt")
+    );
+
+    db.upsert_node(&folder("folder", Some("root"), "yyafteryy"))
+        .unwrap();
+    assert!(
+        !db.search_candidates("zzbeforezz", 20)
+            .unwrap()
+            .iter()
+            .any(|h| h.node.name == "plain.txt")
+    );
+    assert!(
+        db.search_candidates("yyafteryy", 20)
+            .unwrap()
+            .iter()
+            .any(|h| h.node.name == "plain.txt")
+    );
+}
+
+#[test]
+fn local_fuzzy_candidates_include_typo_and_parent_path() {
+    let db = Db::open_in_memory().unwrap();
+    let generation = db.local_begin_scan().unwrap();
+    db.local_upsert_batch(
+        generation,
+        &[
+            local(
+                "/home/u/Projects/quarterly-report.pdf",
+                "quarterly-report.pdf",
+                false,
+            ),
+            local("/home/u/Videos/video.mp4", "video.mp4", false),
+        ],
+    )
+    .unwrap();
+    db.local_finish_scan(generation, 1).unwrap();
+
+    assert_eq!(db.search_local_candidates("quartrly", 20).unwrap().len(), 1);
+    assert_eq!(db.search_local_candidates("projects", 20).unwrap().len(), 1);
+    assert!(
+        db.search_local_candidates("vedio", 20)
+            .unwrap()
+            .iter()
+            .any(|hit| hit.name == "video.mp4")
+    );
+}
+
 #[test]
 fn search_drops_fts_row_on_delete_and_trash() {
     let db = Db::open_in_memory().unwrap();