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.zst")),
35    ("2022", include_bytes!("../../mcp_store_v2022.db.zst")),
36    ("2019", include_bytes!("../../mcp_store_v2019.db.zst")),
37    ("2017", include_bytes!("../../mcp_store_v2017.db.zst")),
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    //
91    // `bytes` is the zstd-compressed `.db.zst` payload (see
92    // `VERSION_STORE_BYTES`), not a valid SQLite file itself — it must be
93    // decompressed before `rusqlite::Connection::open` can read it.
94    let decompressed = zstd::stream::decode_all(bytes).with_context(|| {
95        format!("failed to decompress embedded store data for api_version '{api_version}'")
96    })?;
97    static UNIQUE: AtomicU64 = AtomicU64::new(0);
98    let tmp_path = dir.join(format!(
99        "{file}.{}.{}.tmp",
100        std::process::id(),
101        UNIQUE.fetch_add(1, Ordering::Relaxed)
102    ));
103    std::fs::write(&tmp_path, decompressed).with_context(|| {
104        format!(
105            "failed to extract embedded store data to '{}'",
106            tmp_path.display()
107        )
108    })?;
109    std::fs::rename(&tmp_path, &path).with_context(|| {
110        format!(
111            "failed to move extracted store data into place at '{}'",
112            path.display()
113        )
114    })?;
115
116    Ok(path)
117}
118
119const ENDPOINT_COLUMNS: &str = "operation_id, path, method, summary, description, input_schema, output_schema, auth_scheme_ref";
120
121/// Registers the `sqlite-vec` extension once per process, via
122/// `sqlite3_auto_extension` — matching the pattern the `sqlite-vec` crate
123/// itself documents for `rusqlite`.
124fn register_vec_extension() {
125    REGISTER_VEC_EXTENSION.call_once(|| unsafe {
126        #[allow(clippy::missing_transmute_annotations)]
127        rusqlite::ffi::sqlite3_auto_extension(Some(std::mem::transmute(
128            sqlite_vec::sqlite3_vec_init as *const (),
129        )));
130    });
131}
132
133/// Opens `mcp_store.db` read-only: this crate's binaries only ever read
134/// it, except `sqlserver-mcp-populate-embeddings` (a separate `[[bin]]`), which uses
135/// `open_store_read_write` instead.
136pub fn open_store(path: &Path) -> Result<Connection> {
137    register_vec_extension();
138    Connection::open_with_flags(path, OpenFlags::SQLITE_OPEN_READ_ONLY)
139        .with_context(|| format!("failed to open '{}'", path.display()))
140}
141
142/// Read-write counterpart to `open_store` — for `bin/populate_embeddings.rs`,
143/// the one binary that actually writes to `mcp_store.db` (backfilling
144/// `semantic_endpoints`, whose table every other caller only ever reads
145/// from).
146pub fn open_store_read_write(path: &Path) -> Result<Connection> {
147    register_vec_extension();
148    Connection::open_with_flags(path, OpenFlags::SQLITE_OPEN_READ_WRITE)
149        .with_context(|| format!("failed to open '{}'", path.display()))
150}
151
152/// Returns a process-wide, in-memory-backed connection for `api_version`:
153/// on first access, opens the on-disk file read-only via `open_store`,
154/// copies its entire contents into a fresh `:memory:` connection (via
155/// SQLite's backup API), then drops the on-disk connection immediately —
156/// so its file handle/lock is held only for that brief copy, not for the
157/// process's lifetime. This means an external process (a fresh
158/// `sqlserver-mcp-populate-embeddings` run, a deployment replacing the file) can update
159/// `mcp_store.db` without hitting "database is locked" against a live
160/// server, at the cost of the running process only picking up such an
161/// update on its next restart.
162///
163/// Returns the connection behind a `Mutex` rather than handing it out
164/// directly: callers must finish with the guard (and drop it) *before*
165/// any `.await` — `rusqlite::Connection` isn't `Sync`, so holding the
166/// guard across an await point would make the enclosing future
167/// non-`Send`, the same constraint `core/mcp_server.rs`'s tool handlers
168/// already respect for a plain `Connection`.
169pub fn cached_store_connection(api_version: &str) -> Result<&'static Mutex<Connection>> {
170    let path = resolve_store_path(api_version)?;
171    cached_in_memory_connection(api_version, &path)
172}
173
174/// Does the actual work for `cached_store_connection`, taking an explicit
175/// path (rather than resolving one from `api_version` itself) so it's
176/// testable against an arbitrary tempdir path — `cache_key` and `path`
177/// are separate parameters because two different `api_version`s should
178/// never collide in the process-wide cache even if (hypothetically) they
179/// resolved to the same file.
180fn cached_in_memory_connection(cache_key: &str, path: &Path) -> Result<&'static Mutex<Connection>> {
181    static CACHE: OnceLock<Mutex<HashMap<String, &'static Mutex<Connection>>>> = OnceLock::new();
182    let cache = CACHE.get_or_init(|| Mutex::new(HashMap::new()));
183
184    if let Some(conn) = cache.lock().unwrap().get(cache_key) {
185        return Ok(conn);
186    }
187
188    let disk_conn = open_store(path)?;
189    let mut mem_conn =
190        Connection::open_in_memory().context("failed to open an in-memory SQLite connection")?;
191    Backup::new(&disk_conn, &mut mem_conn)
192        .context("failed to start SQLite backup into memory")?
193        .run_to_completion(i32::MAX, Duration::from_millis(0), None)
194        .with_context(|| format!("failed to back up '{}' into memory", path.display()))?;
195    drop(disk_conn);
196
197    let leaked: &'static Mutex<Connection> = Box::leak(Box::new(Mutex::new(mem_conn)));
198    let mut cache = cache.lock().unwrap();
199    Ok(*cache.entry(cache_key.to_string()).or_insert(leaked))
200}
201
202#[derive(Debug, Clone, Serialize)]
203pub struct EndpointRecord {
204    pub operation_id: String,
205    pub path: String,
206    pub method: String,
207    pub summary: Option<String>,
208    pub description: Option<String>,
209    pub input_schema: serde_json::Value,
210    pub output_schema: serde_json::Value,
211    pub auth_scheme_ref: Option<String>,
212}
213
214fn row_to_endpoint(row: &Row) -> rusqlite::Result<EndpointRecord> {
215    let input_schema: String = row.get(5)?;
216    let output_schema: String = row.get(6)?;
217    Ok(EndpointRecord {
218        operation_id: row.get(0)?,
219        path: row.get(1)?,
220        method: row.get(2)?,
221        summary: row.get(3)?,
222        description: row.get(4)?,
223        input_schema: serde_json::from_str(&input_schema).unwrap_or(serde_json::Value::Null),
224        output_schema: serde_json::from_str(&output_schema).unwrap_or(serde_json::Value::Null),
225        auth_scheme_ref: row.get(7)?,
226    })
227}
228
229pub fn get_endpoint(conn: &Connection, operation_id: &str) -> Result<Option<EndpointRecord>> {
230    let mut stmt = conn.prepare(&format!(
231        "SELECT {ENDPOINT_COLUMNS} FROM endpoints WHERE operation_id = ?1"
232    ))?;
233    match stmt.query_row([operation_id], row_to_endpoint) {
234        Ok(record) => Ok(Some(record)),
235        Err(rusqlite::Error::QueryReturnedNoRows) => Ok(None),
236        Err(err) => Err(err.into()),
237    }
238}
239
240pub fn list_endpoints(conn: &Connection) -> Result<Vec<EndpointRecord>> {
241    let mut stmt = conn.prepare(&format!("SELECT {ENDPOINT_COLUMNS} FROM endpoints"))?;
242    let rows = stmt.query_map([], row_to_endpoint)?;
243    Ok(rows.collect::<rusqlite::Result<Vec<_>>>()?)
244}
245
246#[derive(Debug, Clone, Serialize)]
247pub struct SearchResult {
248    pub operation_id: String,
249    pub summary: Option<String>,
250    pub similarity: f64,
251}
252
253/// k-nearest-neighbor search over `semantic_endpoints` (sqlite-vec's
254/// documented `MATCH ... AND k = ...` query form). `query_embedding` must
255/// come from the same model as the vectors it's compared against — see
256/// `services::embedding_service`. Bound as a raw little-endian `f32` blob,
257/// the same wire format `bin/populate_embeddings.rs` writes (sqlite-vec
258/// accepts either that or a JSON array per-call, independent of how any
259/// given row was originally inserted, so this is a consistency choice, not
260/// a correctness requirement).
261///
262/// `similarity` is a simple `1 - distance` transform of sqlite-vec's L2
263/// distance, not a true cosine-similarity computation; since embeddings
264/// are normalized (unit vectors), L2-distance ordering already matches
265/// cosine-similarity ordering, so result *ranking* is correct even though
266/// the displayed score is an approximation.
267pub fn search_endpoints(
268    conn: &Connection,
269    query_embedding: &[f32],
270    limit: usize,
271) -> Result<Vec<SearchResult>> {
272    let blob: Vec<u8> = query_embedding
273        .iter()
274        .flat_map(|value| value.to_le_bytes())
275        .collect();
276
277    let mut stmt = conn.prepare(
278        "SELECT e.operation_id, e.summary, s.distance
279         FROM semantic_endpoints s
280         JOIN endpoints e ON e.operation_id = s.operation_id
281         WHERE s.embedding MATCH ?1 AND k = ?2
282         ORDER BY s.distance",
283    )?;
284    let rows = stmt.query_map(rusqlite::params![blob, limit], |row| {
285        let distance: f64 = row.get(2)?;
286        Ok(SearchResult {
287            operation_id: row.get(0)?,
288            summary: row.get(1)?,
289            similarity: 1.0 - distance,
290        })
291    })?;
292    Ok(rows.collect::<rusqlite::Result<Vec<_>>>()?)
293}
294
295#[cfg(test)]
296mod tests {
297    use super::*;
298
299    /// Guards against `VERSION_STORE_FILES` and `VERSION_STORE_BYTES`
300    /// silently drifting apart — every `api_version` this crate lists must
301    /// resolve to embedded bytes, or `resolve_store_path` fails at runtime
302    /// for exactly that version and nothing else, which is easy to miss in
303    /// review since the two arrays are edited in different places.
304    #[test]
305    fn every_version_store_file_has_embedded_bytes() {
306        let file_labels: std::collections::HashSet<_> = VERSION_STORE_FILES
307            .iter()
308            .map(|(label, _)| *label)
309            .collect();
310        let byte_labels: std::collections::HashSet<_> = VERSION_STORE_BYTES
311            .iter()
312            .map(|(label, _)| *label)
313            .collect();
314        assert_eq!(file_labels, byte_labels);
315    }
316
317    /// Regression test: `resolve_store_path` used to `std::fs::write` the
318    /// shared extraction path directly, which let one thread's `open_store`
319    /// race another thread's in-progress truncate — exactly what happens
320    /// when multiple MCP tool calls run concurrently against the same
321    /// `api_version` (each hits this path via `cached_store_connection`).
322    /// The rename-into-place fix must make every one of these opens see a
323    /// complete file rather than an intermittently empty one.
324    #[test]
325    fn resolve_store_path_survives_concurrent_calls() {
326        let api_version = VERSION_STORE_FILES[0].0.to_string();
327        let handles: Vec<_> = (0..16)
328            .map(|_| {
329                let api_version = api_version.clone();
330                std::thread::spawn(move || {
331                    let path = resolve_store_path(&api_version).unwrap();
332                    open_store(&path).unwrap();
333                })
334            })
335            .collect();
336        for handle in handles {
337            handle.join().unwrap();
338        }
339    }
340
341    /// Builds a read-write connection with the same schema mcpify's shared
342    /// pipeline writes, and seeds it with one row — real usage never
343    /// creates this schema (mcpify already wrote it before any generated
344    /// code runs), so this setup is test-only fixture, not production
345    /// code these tests exercise.
346    fn seeded_store(path: &Path) -> Connection {
347        unsafe {
348            #[allow(clippy::missing_transmute_annotations)]
349            rusqlite::ffi::sqlite3_auto_extension(Some(std::mem::transmute(
350                sqlite_vec::sqlite3_vec_init as *const (),
351            )));
352        }
353        let conn = Connection::open(path).unwrap();
354        conn.execute(
355            "CREATE TABLE endpoints (
356                operation_id TEXT PRIMARY KEY,
357                path TEXT NOT NULL,
358                method TEXT NOT NULL,
359                summary TEXT,
360                description TEXT,
361                input_schema TEXT NOT NULL,
362                output_schema TEXT NOT NULL,
363                auth_scheme_ref TEXT
364            )",
365            [],
366        )
367        .unwrap();
368        conn.execute(
369            "CREATE VIRTUAL TABLE semantic_endpoints USING vec0(
370                operation_id TEXT PRIMARY KEY,
371                embedding FLOAT[4]
372            )",
373            [],
374        )
375        .unwrap();
376        conn.execute(
377            "INSERT INTO endpoints (operation_id, path, method, summary, description, input_schema, output_schema, auth_scheme_ref)
378             VALUES ('listWidgets', '/widgets', 'GET', 'List widgets', NULL, '{}', '[]', NULL)",
379            [],
380        )
381        .unwrap();
382        let embedding: Vec<u8> = [1.0f32, 0.0, 0.0, 0.0]
383            .iter()
384            .flat_map(|v| v.to_le_bytes())
385            .collect();
386        conn.execute(
387            "INSERT INTO semantic_endpoints (operation_id, embedding) VALUES ('listWidgets', ?1)",
388            rusqlite::params![embedding],
389        )
390        .unwrap();
391        conn
392    }
393
394    #[test]
395    fn get_endpoint_returns_a_seeded_row() {
396        let dir = tempfile::tempdir().unwrap();
397        let path = dir.path().join("mcp_store.db");
398        let _conn = seeded_store(&path);
399
400        let store = open_store(&path).unwrap();
401        let endpoint = get_endpoint(&store, "listWidgets").unwrap().unwrap();
402        assert_eq!(endpoint.path, "/widgets");
403        assert_eq!(endpoint.method, "GET");
404        assert_eq!(endpoint.summary.as_deref(), Some("List widgets"));
405    }
406
407    #[test]
408    fn get_endpoint_returns_none_for_an_unknown_operation() {
409        let dir = tempfile::tempdir().unwrap();
410        let path = dir.path().join("mcp_store.db");
411        let _conn = seeded_store(&path);
412
413        let store = open_store(&path).unwrap();
414        assert!(get_endpoint(&store, "unknownOp").unwrap().is_none());
415    }
416
417    #[test]
418    fn list_endpoints_returns_every_row() {
419        let dir = tempfile::tempdir().unwrap();
420        let path = dir.path().join("mcp_store.db");
421        let _conn = seeded_store(&path);
422
423        let store = open_store(&path).unwrap();
424        let endpoints = list_endpoints(&store).unwrap();
425        assert_eq!(endpoints.len(), 1);
426        assert_eq!(endpoints[0].operation_id, "listWidgets");
427    }
428
429    #[test]
430    fn search_endpoints_finds_the_nearest_neighbor() {
431        let dir = tempfile::tempdir().unwrap();
432        let path = dir.path().join("mcp_store.db");
433        let _conn = seeded_store(&path);
434
435        let store = open_store(&path).unwrap();
436        let results = search_endpoints(&store, &[1.0, 0.0, 0.0, 0.0], 5).unwrap();
437        assert_eq!(results.len(), 1);
438        assert_eq!(results[0].operation_id, "listWidgets");
439        assert!(results[0].similarity > 0.99);
440    }
441
442    #[test]
443    fn cached_in_memory_connection_serves_seeded_data() {
444        let dir = tempfile::tempdir().unwrap();
445        let path = dir.path().join("mcp_store.db");
446        let _conn = seeded_store(&path);
447
448        let cached = cached_in_memory_connection("cached-serves-seeded-data", &path).unwrap();
449        let endpoint = get_endpoint(&cached.lock().unwrap(), "listWidgets")
450            .unwrap()
451            .unwrap();
452        assert_eq!(endpoint.path, "/widgets");
453
454        // vec0 search must also work against the in-memory copy —
455        // `register_vec_extension`'s `sqlite3_auto_extension` applies
456        // process-wide, so it isn't specific to file-backed connections.
457        let results = search_endpoints(&cached.lock().unwrap(), &[1.0, 0.0, 0.0, 0.0], 5).unwrap();
458        assert_eq!(results.len(), 1);
459        assert_eq!(results[0].operation_id, "listWidgets");
460    }
461
462    #[test]
463    fn cached_in_memory_connection_holds_no_lingering_lock_on_the_disk_file() {
464        let dir = tempfile::tempdir().unwrap();
465        let path = dir.path().join("mcp_store.db");
466        let _conn = seeded_store(&path);
467
468        let cached = cached_in_memory_connection("cached-releases-disk-handle", &path).unwrap();
469
470        // If the disk connection were still open, removing the file out
471        // from under it would be the exact failure mode this fix exists
472        // to avoid.
473        std::fs::remove_file(&path).unwrap();
474
475        let endpoint = get_endpoint(&cached.lock().unwrap(), "listWidgets")
476            .unwrap()
477            .unwrap();
478        assert_eq!(endpoint.path, "/widgets");
479    }
480
481    #[test]
482    fn cached_in_memory_connection_reuses_the_same_connection_for_the_same_key() {
483        let dir = tempfile::tempdir().unwrap();
484        let path = dir.path().join("mcp_store.db");
485        let _conn = seeded_store(&path);
486
487        let first = cached_in_memory_connection("cached-reuses-same-key", &path).unwrap()
488            as *const Mutex<Connection>;
489        let second = cached_in_memory_connection("cached-reuses-same-key", &path).unwrap()
490            as *const Mutex<Connection>;
491        assert_eq!(
492            first, second,
493            "expected the same cached connection, not a fresh backup"
494        );
495    }
496}
sqlserver_mcp_catalog/data/store.rs

sqlserver_mcp_catalog/data/
store.rs
