sqlserver_mcp_catalog/data/
store.rs

1// SQL Server 2025 - master/msdb/sandbox combined catalog MCP server — generated by mcpify. Do not hand-edit.
2//
3// Single data-access layer against mcp_store.db — both the relational
4// `endpoints` table and the `semantic_endpoints` vec0 virtual table
5// (REQ-2.4.1's single-store consolidation). Mirrors the shape of
6// mcpify's own src/db/open.rs almost directly, since both the generator
7// and the generated project use the identical rusqlite+sqlite-vec crate
8// pair. Unlike mcpify's own src/db/schema.rs, this module never creates
9// tables — mcpify's shared pipeline (Story 5/6) already wrote
10// mcp_store.db's schema before any generated code runs.
11
12use std::collections::HashMap;
13use std::path::{Path, PathBuf};
14use std::sync::atomic::{AtomicU64, Ordering};
15use std::sync::{Mutex, Once, OnceLock};
16use std::time::Duration;
17
18use anyhow::{Context, Result};
19use rusqlite::backup::Backup;
20use rusqlite::{Connection, OpenFlags, Row};
21use serde::Serialize;
22
23static REGISTER_VEC_EXTENSION: Once = Once::new();
24
25// mcpify:versions:begin
26pub const VERSION_STORE_FILES: &[(&str, &str)] = &[
27    ("2025", "mcp_store.db"),
28    ("2022", "mcp_store_v2022.db"),
29    ("2019", "mcp_store_v2019.db"),
30    ("2017", "mcp_store_v2017.db"),
31];
32
33const VERSION_STORE_BYTES: &[(&str, &[u8])] = &[
34    ("2025", include_bytes!("../../mcp_store.db")),
35    ("2022", include_bytes!("../../mcp_store_v2022.db")),
36    ("2019", include_bytes!("../../mcp_store_v2019.db")),
37    ("2017", include_bytes!("../../mcp_store_v2017.db")),
38];
39// mcpify:versions:end
40
41/// Resolves the active `api_version` (from the config cascade) to its
42/// store file. Every `.db` this crate supports is embedded into the
43/// compiled binary via `include_bytes!` (`VERSION_STORE_BYTES`) exactly
44/// like `validator.rs` embeds each version's schema — there is no
45/// filesystem fallback chain to reason about, and this crate never
46/// depends on any particular `.db` file existing anywhere on disk after
47/// `cargo install`. The one difference from a schema lookup: SQLite
48/// needs a real file to open a `Connection` against (unlike a `&[u8]`
49/// JSON schema, read directly from memory), so this extracts the
50/// embedded bytes to a fixed path in the OS temp dir on every call —
51/// cheap, since it only runs once per process start, and it means a
52/// rebuilt binary with different embedded bytes (a `populate_embeddings`
53/// re-run, an `add-version` update) can never be shadowed by a stale
54/// leftover from a previous install at that same path.
55pub fn resolve_store_path(api_version: &str) -> Result<PathBuf> {
56    let file = VERSION_STORE_FILES
57        .iter()
58        .find(|(label, _)| *label == api_version)
59        .map(|(_, file)| *file)
60        .with_context(|| format!("unknown api_version '{api_version}' — run the 'versions' command to see what's available"))?;
61    let bytes = VERSION_STORE_BYTES
62        .iter()
63        .find(|(label, _)| *label == api_version)
64        .map(|(_, bytes)| *bytes)
65        .with_context(|| format!("no embedded store data for api_version '{api_version}'"))?;
66
67    let mut dir = std::env::temp_dir();
68    dir.push(concat!(env!("CARGO_PKG_NAME"), "-store"));
69    std::fs::create_dir_all(&dir)
70        .with_context(|| format!("failed to create temp dir '{}'", dir.display()))?;
71
72    let path = dir.join(file);
73    // Always (re)writes rather than skipping when the path already
74    // exists: the extraction path doesn't vary by build, so a stale copy
75    // from a previous install (older embedded bytes, e.g. before a
76    // `populate_embeddings` re-run or an `add-version` update) would
77    // otherwise linger forever, silently serving outdated data. The
78    // write is cheap — this runs once per process start, not per query.
79    //
80    // Writes to a uniquely-named sibling file first, then `rename`s it
81    // into place, rather than writing `path` directly: `cached_store_connection`
82    // calls this on every tool invocation (not just once at startup), and
83    // tool calls run concurrently (each MCP request is its own tokio
84    // task) — a direct write would let one task's `open_store` observe
85    // another task's in-progress truncate, reading a momentarily-empty
86    // file and failing with "no such table: endpoints". `rename` within
87    // the same directory is atomic on both POSIX and Windows, so every
88    // reader sees either the complete previous copy or the complete new
89    // one, never a partial write.
90    static UNIQUE: AtomicU64 = AtomicU64::new(0);
91    let tmp_path = dir.join(format!(
92        "{file}.{}.{}.tmp",
93        std::process::id(),
94        UNIQUE.fetch_add(1, Ordering::Relaxed)
95    ));
96    std::fs::write(&tmp_path, bytes).with_context(|| {
97        format!(
98            "failed to extract embedded store data to '{}'",
99            tmp_path.display()
100        )
101    })?;
102    std::fs::rename(&tmp_path, &path).with_context(|| {
103        format!(
104            "failed to move extracted store data into place at '{}'",
105            path.display()
106        )
107    })?;
108
109    Ok(path)
110}
111
112const ENDPOINT_COLUMNS: &str = "operation_id, path, method, summary, description, input_schema, output_schema, auth_scheme_ref";
113
114/// Registers the `sqlite-vec` extension once per process, via
115/// `sqlite3_auto_extension` — matching the pattern the `sqlite-vec` crate
116/// itself documents for `rusqlite`.
117fn register_vec_extension() {
118    REGISTER_VEC_EXTENSION.call_once(|| unsafe {
119        #[allow(clippy::missing_transmute_annotations)]
120        rusqlite::ffi::sqlite3_auto_extension(Some(std::mem::transmute(
121            sqlite_vec::sqlite3_vec_init as *const (),
122        )));
123    });
124}
125
126/// Opens `mcp_store.db` read-only: this crate's binaries only ever read
127/// it, except `sqlserver-mcp-populate-embeddings` (a separate `[[bin]]`), which uses
128/// `open_store_read_write` instead.
129pub fn open_store(path: &Path) -> Result<Connection> {
130    register_vec_extension();
131    Connection::open_with_flags(path, OpenFlags::SQLITE_OPEN_READ_ONLY)
132        .with_context(|| format!("failed to open '{}'", path.display()))
133}
134
135/// Read-write counterpart to `open_store` — for `bin/populate_embeddings.rs`,
136/// the one binary that actually writes to `mcp_store.db` (backfilling
137/// `semantic_endpoints`, whose table every other caller only ever reads
138/// from).
139pub fn open_store_read_write(path: &Path) -> Result<Connection> {
140    register_vec_extension();
141    Connection::open_with_flags(path, OpenFlags::SQLITE_OPEN_READ_WRITE)
142        .with_context(|| format!("failed to open '{}'", path.display()))
143}
144
145/// Returns a process-wide, in-memory-backed connection for `api_version`:
146/// on first access, opens the on-disk file read-only via `open_store`,
147/// copies its entire contents into a fresh `:memory:` connection (via
148/// SQLite's backup API), then drops the on-disk connection immediately —
149/// so its file handle/lock is held only for that brief copy, not for the
150/// process's lifetime. This means an external process (a fresh
151/// `sqlserver-mcp-populate-embeddings` run, a deployment replacing the file) can update
152/// `mcp_store.db` without hitting "database is locked" against a live
153/// server, at the cost of the running process only picking up such an
154/// update on its next restart.
155///
156/// Returns the connection behind a `Mutex` rather than handing it out
157/// directly: callers must finish with the guard (and drop it) *before*
158/// any `.await` — `rusqlite::Connection` isn't `Sync`, so holding the
159/// guard across an await point would make the enclosing future
160/// non-`Send`, the same constraint `core/mcp_server.rs`'s tool handlers
161/// already respect for a plain `Connection`.
162pub fn cached_store_connection(api_version: &str) -> Result<&'static Mutex<Connection>> {
163    let path = resolve_store_path(api_version)?;
164    cached_in_memory_connection(api_version, &path)
165}
166
167/// Does the actual work for `cached_store_connection`, taking an explicit
168/// path (rather than resolving one from `api_version` itself) so it's
169/// testable against an arbitrary tempdir path — `cache_key` and `path`
170/// are separate parameters because two different `api_version`s should
171/// never collide in the process-wide cache even if (hypothetically) they
172/// resolved to the same file.
173fn cached_in_memory_connection(cache_key: &str, path: &Path) -> Result<&'static Mutex<Connection>> {
174    static CACHE: OnceLock<Mutex<HashMap<String, &'static Mutex<Connection>>>> = OnceLock::new();
175    let cache = CACHE.get_or_init(|| Mutex::new(HashMap::new()));
176
177    if let Some(conn) = cache.lock().unwrap().get(cache_key) {
178        return Ok(conn);
179    }
180
181    let disk_conn = open_store(path)?;
182    let mut mem_conn =
183        Connection::open_in_memory().context("failed to open an in-memory SQLite connection")?;
184    Backup::new(&disk_conn, &mut mem_conn)
185        .context("failed to start SQLite backup into memory")?
186        .run_to_completion(i32::MAX, Duration::from_millis(0), None)
187        .with_context(|| format!("failed to back up '{}' into memory", path.display()))?;
188    drop(disk_conn);
189
190    let leaked: &'static Mutex<Connection> = Box::leak(Box::new(Mutex::new(mem_conn)));
191    let mut cache = cache.lock().unwrap();
192    Ok(*cache.entry(cache_key.to_string()).or_insert(leaked))
193}
194
195#[derive(Debug, Clone, Serialize)]
196pub struct EndpointRecord {
197    pub operation_id: String,
198    pub path: String,
199    pub method: String,
200    pub summary: Option<String>,
201    pub description: Option<String>,
202    pub input_schema: serde_json::Value,
203    pub output_schema: serde_json::Value,
204    pub auth_scheme_ref: Option<String>,
205}
206
207fn row_to_endpoint(row: &Row) -> rusqlite::Result<EndpointRecord> {
208    let input_schema: String = row.get(5)?;
209    let output_schema: String = row.get(6)?;
210    Ok(EndpointRecord {
211        operation_id: row.get(0)?,
212        path: row.get(1)?,
213        method: row.get(2)?,
214        summary: row.get(3)?,
215        description: row.get(4)?,
216        input_schema: serde_json::from_str(&input_schema).unwrap_or(serde_json::Value::Null),
217        output_schema: serde_json::from_str(&output_schema).unwrap_or(serde_json::Value::Null),
218        auth_scheme_ref: row.get(7)?,
219    })
220}
221
222pub fn get_endpoint(conn: &Connection, operation_id: &str) -> Result<Option<EndpointRecord>> {
223    let mut stmt = conn.prepare(&format!(
224        "SELECT {ENDPOINT_COLUMNS} FROM endpoints WHERE operation_id = ?1"
225    ))?;
226    match stmt.query_row([operation_id], row_to_endpoint) {
227        Ok(record) => Ok(Some(record)),
228        Err(rusqlite::Error::QueryReturnedNoRows) => Ok(None),
229        Err(err) => Err(err.into()),
230    }
231}
232
233pub fn list_endpoints(conn: &Connection) -> Result<Vec<EndpointRecord>> {
234    let mut stmt = conn.prepare(&format!("SELECT {ENDPOINT_COLUMNS} FROM endpoints"))?;
235    let rows = stmt.query_map([], row_to_endpoint)?;
236    Ok(rows.collect::<rusqlite::Result<Vec<_>>>()?)
237}
238
239#[derive(Debug, Clone, Serialize)]
240pub struct SearchResult {
241    pub operation_id: String,
242    pub summary: Option<String>,
243    pub similarity: f64,
244}
245
246/// k-nearest-neighbor search over `semantic_endpoints` (sqlite-vec's
247/// documented `MATCH ... AND k = ...` query form). `query_embedding` must
248/// come from the same model as the vectors it's compared against — see
249/// `services::embedding_service`. Bound as a raw little-endian `f32` blob,
250/// the same wire format `bin/populate_embeddings.rs` writes (sqlite-vec
251/// accepts either that or a JSON array per-call, independent of how any
252/// given row was originally inserted, so this is a consistency choice, not
253/// a correctness requirement).
254///
255/// `similarity` is a simple `1 - distance` transform of sqlite-vec's L2
256/// distance, not a true cosine-similarity computation; since embeddings
257/// are normalized (unit vectors), L2-distance ordering already matches
258/// cosine-similarity ordering, so result *ranking* is correct even though
259/// the displayed score is an approximation.
260pub fn search_endpoints(
261    conn: &Connection,
262    query_embedding: &[f32],
263    limit: usize,
264) -> Result<Vec<SearchResult>> {
265    let blob: Vec<u8> = query_embedding
266        .iter()
267        .flat_map(|value| value.to_le_bytes())
268        .collect();
269
270    let mut stmt = conn.prepare(
271        "SELECT e.operation_id, e.summary, s.distance
272         FROM semantic_endpoints s
273         JOIN endpoints e ON e.operation_id = s.operation_id
274         WHERE s.embedding MATCH ?1 AND k = ?2
275         ORDER BY s.distance",
276    )?;
277    let rows = stmt.query_map(rusqlite::params![blob, limit], |row| {
278        let distance: f64 = row.get(2)?;
279        Ok(SearchResult {
280            operation_id: row.get(0)?,
281            summary: row.get(1)?,
282            similarity: 1.0 - distance,
283        })
284    })?;
285    Ok(rows.collect::<rusqlite::Result<Vec<_>>>()?)
286}
287
288#[cfg(test)]
289mod tests {
290    use super::*;
291
292    /// Guards against `VERSION_STORE_FILES` and `VERSION_STORE_BYTES`
293    /// silently drifting apart — every `api_version` this crate lists must
294    /// resolve to embedded bytes, or `resolve_store_path` fails at runtime
295    /// for exactly that version and nothing else, which is easy to miss in
296    /// review since the two arrays are edited in different places.
297    #[test]
298    fn every_version_store_file_has_embedded_bytes() {
299        let file_labels: std::collections::HashSet<_> = VERSION_STORE_FILES
300            .iter()
301            .map(|(label, _)| *label)
302            .collect();
303        let byte_labels: std::collections::HashSet<_> = VERSION_STORE_BYTES
304            .iter()
305            .map(|(label, _)| *label)
306            .collect();
307        assert_eq!(file_labels, byte_labels);
308    }
309
310    /// Regression test: `resolve_store_path` used to `std::fs::write` the
311    /// shared extraction path directly, which let one thread's `open_store`
312    /// race another thread's in-progress truncate — exactly what happens
313    /// when multiple MCP tool calls run concurrently against the same
314    /// `api_version` (each hits this path via `cached_store_connection`).
315    /// The rename-into-place fix must make every one of these opens see a
316    /// complete file rather than an intermittently empty one.
317    #[test]
318    fn resolve_store_path_survives_concurrent_calls() {
319        let api_version = VERSION_STORE_FILES[0].0.to_string();
320        let handles: Vec<_> = (0..16)
321            .map(|_| {
322                let api_version = api_version.clone();
323                std::thread::spawn(move || {
324                    let path = resolve_store_path(&api_version).unwrap();
325                    open_store(&path).unwrap();
326                })
327            })
328            .collect();
329        for handle in handles {
330            handle.join().unwrap();
331        }
332    }
333
334    /// Builds a read-write connection with the same schema mcpify's shared
335    /// pipeline writes, and seeds it with one row — real usage never
336    /// creates this schema (mcpify already wrote it before any generated
337    /// code runs), so this setup is test-only fixture, not production
338    /// code these tests exercise.
339    fn seeded_store(path: &Path) -> Connection {
340        unsafe {
341            #[allow(clippy::missing_transmute_annotations)]
342            rusqlite::ffi::sqlite3_auto_extension(Some(std::mem::transmute(
343                sqlite_vec::sqlite3_vec_init as *const (),
344            )));
345        }
346        let conn = Connection::open(path).unwrap();
347        conn.execute(
348            "CREATE TABLE endpoints (
349                operation_id TEXT PRIMARY KEY,
350                path TEXT NOT NULL,
351                method TEXT NOT NULL,
352                summary TEXT,
353                description TEXT,
354                input_schema TEXT NOT NULL,
355                output_schema TEXT NOT NULL,
356                auth_scheme_ref TEXT
357            )",
358            [],
359        )
360        .unwrap();
361        conn.execute(
362            "CREATE VIRTUAL TABLE semantic_endpoints USING vec0(
363                operation_id TEXT PRIMARY KEY,
364                embedding FLOAT[4]
365            )",
366            [],
367        )
368        .unwrap();
369        conn.execute(
370            "INSERT INTO endpoints (operation_id, path, method, summary, description, input_schema, output_schema, auth_scheme_ref)
371             VALUES ('listWidgets', '/widgets', 'GET', 'List widgets', NULL, '{}', '[]', NULL)",
372            [],
373        )
374        .unwrap();
375        let embedding: Vec<u8> = [1.0f32, 0.0, 0.0, 0.0]
376            .iter()
377            .flat_map(|v| v.to_le_bytes())
378            .collect();
379        conn.execute(
380            "INSERT INTO semantic_endpoints (operation_id, embedding) VALUES ('listWidgets', ?1)",
381            rusqlite::params![embedding],
382        )
383        .unwrap();
384        conn
385    }
386
387    #[test]
388    fn get_endpoint_returns_a_seeded_row() {
389        let dir = tempfile::tempdir().unwrap();
390        let path = dir.path().join("mcp_store.db");
391        let _conn = seeded_store(&path);
392
393        let store = open_store(&path).unwrap();
394        let endpoint = get_endpoint(&store, "listWidgets").unwrap().unwrap();
395        assert_eq!(endpoint.path, "/widgets");
396        assert_eq!(endpoint.method, "GET");
397        assert_eq!(endpoint.summary.as_deref(), Some("List widgets"));
398    }
399
400    #[test]
401    fn get_endpoint_returns_none_for_an_unknown_operation() {
402        let dir = tempfile::tempdir().unwrap();
403        let path = dir.path().join("mcp_store.db");
404        let _conn = seeded_store(&path);
405
406        let store = open_store(&path).unwrap();
407        assert!(get_endpoint(&store, "unknownOp").unwrap().is_none());
408    }
409
410    #[test]
411    fn list_endpoints_returns_every_row() {
412        let dir = tempfile::tempdir().unwrap();
413        let path = dir.path().join("mcp_store.db");
414        let _conn = seeded_store(&path);
415
416        let store = open_store(&path).unwrap();
417        let endpoints = list_endpoints(&store).unwrap();
418        assert_eq!(endpoints.len(), 1);
419        assert_eq!(endpoints[0].operation_id, "listWidgets");
420    }
421
422    #[test]
423    fn search_endpoints_finds_the_nearest_neighbor() {
424        let dir = tempfile::tempdir().unwrap();
425        let path = dir.path().join("mcp_store.db");
426        let _conn = seeded_store(&path);
427
428        let store = open_store(&path).unwrap();
429        let results = search_endpoints(&store, &[1.0, 0.0, 0.0, 0.0], 5).unwrap();
430        assert_eq!(results.len(), 1);
431        assert_eq!(results[0].operation_id, "listWidgets");
432        assert!(results[0].similarity > 0.99);
433    }
434
435    #[test]
436    fn cached_in_memory_connection_serves_seeded_data() {
437        let dir = tempfile::tempdir().unwrap();
438        let path = dir.path().join("mcp_store.db");
439        let _conn = seeded_store(&path);
440
441        let cached = cached_in_memory_connection("cached-serves-seeded-data", &path).unwrap();
442        let endpoint = get_endpoint(&cached.lock().unwrap(), "listWidgets")
443            .unwrap()
444            .unwrap();
445        assert_eq!(endpoint.path, "/widgets");
446
447        // vec0 search must also work against the in-memory copy —
448        // `register_vec_extension`'s `sqlite3_auto_extension` applies
449        // process-wide, so it isn't specific to file-backed connections.
450        let results = search_endpoints(&cached.lock().unwrap(), &[1.0, 0.0, 0.0, 0.0], 5).unwrap();
451        assert_eq!(results.len(), 1);
452        assert_eq!(results[0].operation_id, "listWidgets");
453    }
454
455    #[test]
456    fn cached_in_memory_connection_holds_no_lingering_lock_on_the_disk_file() {
457        let dir = tempfile::tempdir().unwrap();
458        let path = dir.path().join("mcp_store.db");
459        let _conn = seeded_store(&path);
460
461        let cached = cached_in_memory_connection("cached-releases-disk-handle", &path).unwrap();
462
463        // If the disk connection were still open, removing the file out
464        // from under it would be the exact failure mode this fix exists
465        // to avoid.
466        std::fs::remove_file(&path).unwrap();
467
468        let endpoint = get_endpoint(&cached.lock().unwrap(), "listWidgets")
469            .unwrap()
470            .unwrap();
471        assert_eq!(endpoint.path, "/widgets");
472    }
473
474    #[test]
475    fn cached_in_memory_connection_reuses_the_same_connection_for_the_same_key() {
476        let dir = tempfile::tempdir().unwrap();
477        let path = dir.path().join("mcp_store.db");
478        let _conn = seeded_store(&path);
479
480        let first = cached_in_memory_connection("cached-reuses-same-key", &path).unwrap()
481            as *const Mutex<Connection>;
482        let second = cached_in_memory_connection("cached-reuses-same-key", &path).unwrap()
483            as *const Mutex<Connection>;
484        assert_eq!(
485            first, second,
486            "expected the same cached connection, not a fresh backup"
487        );
488    }
489}
sqlserver_mcp_catalog/data/store.rs

sqlserver_mcp_catalog/data/
store.rs
