github_mcp/data/
store.rs

1// GitHub v3 REST API 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    ("gh-2026-03-10", "mcp_store.db"),
28    ("ghec-2026-03-10", "mcp_store_vghec-2026-03-10.db"),
29    ("ghes-3.21", "mcp_store_vghes-3.21.db"),
30    ("ghes-3.20", "mcp_store_vghes-3.20.db"),
31    ("ghes-3.19", "mcp_store_vghes-3.19.db"),
32];
33
34const VERSION_STORE_BYTES: &[(&str, &[u8])] = &[
35    ("gh-2026-03-10", include_bytes!("../../mcp_store.db.zst")),
36    (
37        "ghec-2026-03-10",
38        include_bytes!("../../mcp_store_vghec-2026-03-10.db.zst"),
39    ),
40    (
41        "ghes-3.21",
42        include_bytes!("../../mcp_store_vghes-3.21.db.zst"),
43    ),
44    (
45        "ghes-3.20",
46        include_bytes!("../../mcp_store_vghes-3.20.db.zst"),
47    ),
48    (
49        "ghes-3.19",
50        include_bytes!("../../mcp_store_vghes-3.19.db.zst"),
51    ),
52];
53// mcpify:versions:end
54
55/// Resolves the active `api_version` (from the config cascade) to its
56/// store file. Every `.db` this crate supports is embedded into the
57/// compiled binary via `include_bytes!` (`VERSION_STORE_BYTES`) exactly
58/// like `validator.rs` embeds each version's schema — there is no
59/// filesystem fallback chain to reason about, and this crate never
60/// depends on any particular `.db` file existing anywhere on disk after
61/// `cargo install`. The one difference from a schema lookup: SQLite
62/// needs a real file to open a `Connection` against (unlike a `&[u8]`
63/// JSON schema, read directly from memory), so this extracts the
64/// embedded bytes to a fixed path in the OS temp dir the first time a
65/// given `api_version` is requested in this process, then reuses that
66/// same path on every later call (see `EXTRACTED` below) — a rebuilt
67/// binary with different embedded bytes (a `populate_embeddings` re-run,
68/// an `add-version` update) still can never be shadowed by a stale
69/// leftover from a previous install, since a fresh process always
70/// extracts fresh.
71pub fn resolve_store_path(api_version: &str) -> Result<PathBuf> {
72    // `cached_store_connection` calls this on every tool invocation, and
73    // tool calls run concurrently (each MCP request is its own tokio
74    // task) — caching the resolved path per `api_version`, guarded by
75    // this same mutex, means the actual extract-and-rename-into-place
76    // below only ever runs once per `api_version` per process: every
77    // concurrent first caller blocks here rather than racing each other
78    // to write the same destination file. That race used to be real: on
79    // Windows, `rename`-ing over a destination another thread already has
80    // open via `open_store` can fail outright ("failed to move extracted
81    // store data into place") rather than just being non-atomic, unlike
82    // POSIX where the same rename silently succeeds.
83    static EXTRACTED: OnceLock<Mutex<HashMap<String, PathBuf>>> = OnceLock::new();
84    let extracted = EXTRACTED.get_or_init(|| Mutex::new(HashMap::new()));
85    let mut extracted = extracted.lock().unwrap();
86    if let Some(path) = extracted.get(api_version) {
87        return Ok(path.clone());
88    }
89
90    let file = VERSION_STORE_FILES
91        .iter()
92        .find(|(label, _)| *label == api_version)
93        .map(|(_, file)| *file)
94        .with_context(|| format!("unknown api_version '{api_version}' — run the 'versions' command to see what's available"))?;
95    let bytes = VERSION_STORE_BYTES
96        .iter()
97        .find(|(label, _)| *label == api_version)
98        .map(|(_, bytes)| *bytes)
99        .with_context(|| format!("no embedded store data for api_version '{api_version}'"))?;
100
101    let mut dir = std::env::temp_dir();
102    dir.push(concat!(env!("CARGO_PKG_NAME"), "-store"));
103    std::fs::create_dir_all(&dir)
104        .with_context(|| format!("failed to create temp dir '{}'", dir.display()))?;
105
106    let path = dir.join(file);
107    // Writes to a uniquely-named sibling file first, then `rename`s it
108    // into place, rather than writing `path` directly, so a *different*
109    // process sharing this same temp dir (a fresh `populate_embeddings`
110    // run, a second server instance starting up concurrently) never
111    // observes a momentarily-empty or partially-written file and fails
112    // with "no such table: endpoints". `rename` within the same
113    // directory is atomic on both POSIX and Windows with respect to
114    // content — a reader always sees either the complete previous copy
115    // or the complete new one, never a partial write — even though, as
116    // noted above, Windows can still refuse the rename outright if
117    // something in *this* process already has the destination open;
118    // the per-`api_version` cache above is what actually prevents that.
119    //
120    // `bytes` is the zstd-compressed `.db.zst` payload (see
121    // `VERSION_STORE_BYTES`), not a valid SQLite file itself — it must be
122    // decompressed before `rusqlite::Connection::open` can read it.
123    let decompressed = zstd::stream::decode_all(bytes).with_context(|| {
124        format!("failed to decompress embedded store data for api_version '{api_version}'")
125    })?;
126    static UNIQUE: AtomicU64 = AtomicU64::new(0);
127    let tmp_path = dir.join(format!(
128        "{file}.{}.{}.tmp",
129        std::process::id(),
130        UNIQUE.fetch_add(1, Ordering::Relaxed)
131    ));
132    std::fs::write(&tmp_path, decompressed).with_context(|| {
133        format!(
134            "failed to extract embedded store data to '{}'",
135            tmp_path.display()
136        )
137    })?;
138    std::fs::rename(&tmp_path, &path).with_context(|| {
139        format!(
140            "failed to move extracted store data into place at '{}'",
141            path.display()
142        )
143    })?;
144
145    extracted.insert(api_version.to_string(), path.clone());
146    Ok(path)
147}
148
149const ENDPOINT_COLUMNS: &str = "operation_id, path, method, summary, description, input_schema, output_schema, auth_scheme_ref";
150
151/// Registers the `sqlite-vec` extension once per process, via
152/// `sqlite3_auto_extension` — matching the pattern the `sqlite-vec` crate
153/// itself documents for `rusqlite`.
154fn register_vec_extension() {
155    REGISTER_VEC_EXTENSION.call_once(|| unsafe {
156        #[allow(clippy::missing_transmute_annotations)]
157        rusqlite::ffi::sqlite3_auto_extension(Some(std::mem::transmute(
158            sqlite_vec::sqlite3_vec_init as *const (),
159        )));
160    });
161}
162
163/// Opens `mcp_store.db` read-only: this crate's binaries only ever read
164/// it, except `github-mcp-populate-embeddings` (a separate `[[bin]]`), which uses
165/// `open_store_read_write` instead.
166pub fn open_store(path: &Path) -> Result<Connection> {
167    register_vec_extension();
168    Connection::open_with_flags(path, OpenFlags::SQLITE_OPEN_READ_ONLY)
169        .with_context(|| format!("failed to open '{}'", path.display()))
170}
171
172/// Read-write counterpart to `open_store` — for `bin/populate_embeddings.rs`,
173/// the one binary that actually writes to `mcp_store.db` (backfilling
174/// `semantic_endpoints`, whose table every other caller only ever reads
175/// from).
176pub fn open_store_read_write(path: &Path) -> Result<Connection> {
177    register_vec_extension();
178    Connection::open_with_flags(path, OpenFlags::SQLITE_OPEN_READ_WRITE)
179        .with_context(|| format!("failed to open '{}'", path.display()))
180}
181
182/// Returns a process-wide, in-memory-backed connection for `api_version`:
183/// on first access, opens the on-disk file read-only via `open_store`,
184/// copies its entire contents into a fresh `:memory:` connection (via
185/// SQLite's backup API), then drops the on-disk connection immediately —
186/// so its file handle/lock is held only for that brief copy, not for the
187/// process's lifetime. This means an external process (a fresh
188/// `github-mcp-populate-embeddings` run, a deployment replacing the file) can update
189/// `mcp_store.db` without hitting "database is locked" against a live
190/// server, at the cost of the running process only picking up such an
191/// update on its next restart.
192///
193/// Returns the connection behind a `Mutex` rather than handing it out
194/// directly: callers must finish with the guard (and drop it) *before*
195/// any `.await` — `rusqlite::Connection` isn't `Sync`, so holding the
196/// guard across an await point would make the enclosing future
197/// non-`Send`, the same constraint `core/mcp_server.rs`'s tool handlers
198/// already respect for a plain `Connection`.
199pub fn cached_store_connection(api_version: &str) -> Result<&'static Mutex<Connection>> {
200    let path = resolve_store_path(api_version)?;
201    cached_in_memory_connection(api_version, &path)
202}
203
204/// Does the actual work for `cached_store_connection`, taking an explicit
205/// path (rather than resolving one from `api_version` itself) so it's
206/// testable against an arbitrary tempdir path — `cache_key` and `path`
207/// are separate parameters because two different `api_version`s should
208/// never collide in the process-wide cache even if (hypothetically) they
209/// resolved to the same file.
210fn cached_in_memory_connection(cache_key: &str, path: &Path) -> Result<&'static Mutex<Connection>> {
211    static CACHE: OnceLock<Mutex<HashMap<String, &'static Mutex<Connection>>>> = OnceLock::new();
212    let cache = CACHE.get_or_init(|| Mutex::new(HashMap::new()));
213
214    if let Some(conn) = cache.lock().unwrap().get(cache_key) {
215        return Ok(conn);
216    }
217
218    let disk_conn = open_store(path)?;
219    let mut mem_conn =
220        Connection::open_in_memory().context("failed to open an in-memory SQLite connection")?;
221    Backup::new(&disk_conn, &mut mem_conn)
222        .context("failed to start SQLite backup into memory")?
223        .run_to_completion(i32::MAX, Duration::from_millis(0), None)
224        .with_context(|| format!("failed to back up '{}' into memory", path.display()))?;
225    drop(disk_conn);
226
227    let leaked: &'static Mutex<Connection> = Box::leak(Box::new(Mutex::new(mem_conn)));
228    let mut cache = cache.lock().unwrap();
229    Ok(*cache.entry(cache_key.to_string()).or_insert(leaked))
230}
231
232#[derive(Debug, Clone, Serialize)]
233pub struct EndpointRecord {
234    pub operation_id: String,
235    pub path: String,
236    pub method: String,
237    pub summary: Option<String>,
238    pub description: Option<String>,
239    pub input_schema: serde_json::Value,
240    pub output_schema: serde_json::Value,
241    pub auth_scheme_ref: Option<String>,
242}
243
244fn row_to_endpoint(row: &Row) -> rusqlite::Result<EndpointRecord> {
245    let input_schema: String = row.get(5)?;
246    let output_schema: String = row.get(6)?;
247    Ok(EndpointRecord {
248        operation_id: row.get(0)?,
249        path: row.get(1)?,
250        method: row.get(2)?,
251        summary: row.get(3)?,
252        description: row.get(4)?,
253        input_schema: serde_json::from_str(&input_schema).unwrap_or(serde_json::Value::Null),
254        output_schema: serde_json::from_str(&output_schema).unwrap_or(serde_json::Value::Null),
255        auth_scheme_ref: row.get(7)?,
256    })
257}
258
259pub fn get_endpoint(conn: &Connection, operation_id: &str) -> Result<Option<EndpointRecord>> {
260    let mut stmt = conn.prepare(&format!(
261        "SELECT {ENDPOINT_COLUMNS} FROM endpoints WHERE operation_id = ?1"
262    ))?;
263    match stmt.query_row([operation_id], row_to_endpoint) {
264        Ok(record) => Ok(Some(record)),
265        Err(rusqlite::Error::QueryReturnedNoRows) => Ok(None),
266        Err(err) => Err(err.into()),
267    }
268}
269
270pub fn list_endpoints(conn: &Connection) -> Result<Vec<EndpointRecord>> {
271    let mut stmt = conn.prepare(&format!("SELECT {ENDPOINT_COLUMNS} FROM endpoints"))?;
272    let rows = stmt.query_map([], row_to_endpoint)?;
273    Ok(rows.collect::<rusqlite::Result<Vec<_>>>()?)
274}
275
276#[derive(Debug, Clone, Serialize)]
277pub struct SearchResult {
278    pub operation_id: String,
279    pub summary: Option<String>,
280    pub similarity: f64,
281}
282
283/// k-nearest-neighbor search over `semantic_endpoints` (sqlite-vec's
284/// documented `MATCH ... AND k = ...` query form). `query_embedding` must
285/// come from the same model as the vectors it's compared against — see
286/// `services::embedding_service`. Bound as a raw little-endian `f32` blob,
287/// the same wire format `bin/populate_embeddings.rs` writes (sqlite-vec
288/// accepts either that or a JSON array per-call, independent of how any
289/// given row was originally inserted, so this is a consistency choice, not
290/// a correctness requirement).
291///
292/// `similarity` is a simple `1 - distance` transform of sqlite-vec's L2
293/// distance, not a true cosine-similarity computation; since embeddings
294/// are normalized (unit vectors), L2-distance ordering already matches
295/// cosine-similarity ordering, so result *ranking* is correct even though
296/// the displayed score is an approximation.
297pub fn search_endpoints(
298    conn: &Connection,
299    query_embedding: &[f32],
300    limit: usize,
301) -> Result<Vec<SearchResult>> {
302    let blob: Vec<u8> = query_embedding
303        .iter()
304        .flat_map(|value| value.to_le_bytes())
305        .collect();
306
307    let mut stmt = conn.prepare(
308        "SELECT e.operation_id, e.summary, s.distance
309         FROM semantic_endpoints s
310         JOIN endpoints e ON e.operation_id = s.operation_id
311         WHERE s.embedding MATCH ?1 AND k = ?2
312         ORDER BY s.distance",
313    )?;
314    let rows = stmt.query_map(rusqlite::params![blob, limit], |row| {
315        let distance: f64 = row.get(2)?;
316        Ok(SearchResult {
317            operation_id: row.get(0)?,
318            summary: row.get(1)?,
319            similarity: 1.0 - distance,
320        })
321    })?;
322    Ok(rows.collect::<rusqlite::Result<Vec<_>>>()?)
323}
324
325#[cfg(test)]
326mod tests {
327    use super::*;
328
329    /// Guards against `VERSION_STORE_FILES` and `VERSION_STORE_BYTES`
330    /// silently drifting apart — every `api_version` this crate lists must
331    /// resolve to embedded bytes, or `resolve_store_path` fails at runtime
332    /// for exactly that version and nothing else, which is easy to miss in
333    /// review since the two arrays are edited in different places.
334    #[test]
335    fn every_version_store_file_has_embedded_bytes() {
336        let file_labels: std::collections::HashSet<_> = VERSION_STORE_FILES
337            .iter()
338            .map(|(label, _)| *label)
339            .collect();
340        let byte_labels: std::collections::HashSet<_> = VERSION_STORE_BYTES
341            .iter()
342            .map(|(label, _)| *label)
343            .collect();
344        assert_eq!(file_labels, byte_labels);
345    }
346
347    /// Regression test: `resolve_store_path` used to `std::fs::write` the
348    /// shared extraction path directly, which let one thread's `open_store`
349    /// race another thread's in-progress truncate — exactly what happens
350    /// when multiple MCP tool calls run concurrently against the same
351    /// `api_version` (each hits this path via `cached_store_connection`).
352    /// The rename-into-place fix must make every one of these opens see a
353    /// complete file rather than an intermittently empty one.
354    #[test]
355    fn resolve_store_path_survives_concurrent_calls() {
356        let api_version = VERSION_STORE_FILES[0].0.to_string();
357        let handles: Vec<_> = (0..16)
358            .map(|_| {
359                let api_version = api_version.clone();
360                std::thread::spawn(move || {
361                    let path = resolve_store_path(&api_version).unwrap();
362                    open_store(&path).unwrap();
363                })
364            })
365            .collect();
366        for handle in handles {
367            handle.join().unwrap();
368        }
369    }
370
371    /// Builds a read-write connection with the same schema mcpify's shared
372    /// pipeline writes, and seeds it with one row — real usage never
373    /// creates this schema (mcpify already wrote it before any generated
374    /// code runs), so this setup is test-only fixture, not production
375    /// code these tests exercise.
376    fn seeded_store(path: &Path) -> Connection {
377        unsafe {
378            #[allow(clippy::missing_transmute_annotations)]
379            rusqlite::ffi::sqlite3_auto_extension(Some(std::mem::transmute(
380                sqlite_vec::sqlite3_vec_init as *const (),
381            )));
382        }
383        let conn = Connection::open(path).unwrap();
384        conn.execute(
385            "CREATE TABLE endpoints (
386                operation_id TEXT PRIMARY KEY,
387                path TEXT NOT NULL,
388                method TEXT NOT NULL,
389                summary TEXT,
390                description TEXT,
391                input_schema TEXT NOT NULL,
392                output_schema TEXT NOT NULL,
393                auth_scheme_ref TEXT
394            )",
395            [],
396        )
397        .unwrap();
398        conn.execute(
399            "CREATE VIRTUAL TABLE semantic_endpoints USING vec0(
400                operation_id TEXT PRIMARY KEY,
401                embedding FLOAT[4]
402            )",
403            [],
404        )
405        .unwrap();
406        conn.execute(
407            "INSERT INTO endpoints (operation_id, path, method, summary, description, input_schema, output_schema, auth_scheme_ref)
408             VALUES ('listWidgets', '/widgets', 'GET', 'List widgets', NULL, '{}', '[]', NULL)",
409            [],
410        )
411        .unwrap();
412        let embedding: Vec<u8> = [1.0f32, 0.0, 0.0, 0.0]
413            .iter()
414            .flat_map(|v| v.to_le_bytes())
415            .collect();
416        conn.execute(
417            "INSERT INTO semantic_endpoints (operation_id, embedding) VALUES ('listWidgets', ?1)",
418            rusqlite::params![embedding],
419        )
420        .unwrap();
421        conn
422    }
423
424    #[test]
425    fn get_endpoint_returns_a_seeded_row() {
426        let dir = tempfile::tempdir().unwrap();
427        let path = dir.path().join("mcp_store.db");
428        let _conn = seeded_store(&path);
429
430        let store = open_store(&path).unwrap();
431        let endpoint = get_endpoint(&store, "listWidgets").unwrap().unwrap();
432        assert_eq!(endpoint.path, "/widgets");
433        assert_eq!(endpoint.method, "GET");
434        assert_eq!(endpoint.summary.as_deref(), Some("List widgets"));
435    }
436
437    #[test]
438    fn get_endpoint_returns_none_for_an_unknown_operation() {
439        let dir = tempfile::tempdir().unwrap();
440        let path = dir.path().join("mcp_store.db");
441        let _conn = seeded_store(&path);
442
443        let store = open_store(&path).unwrap();
444        assert!(get_endpoint(&store, "unknownOp").unwrap().is_none());
445    }
446
447    #[test]
448    fn list_endpoints_returns_every_row() {
449        let dir = tempfile::tempdir().unwrap();
450        let path = dir.path().join("mcp_store.db");
451        let _conn = seeded_store(&path);
452
453        let store = open_store(&path).unwrap();
454        let endpoints = list_endpoints(&store).unwrap();
455        assert_eq!(endpoints.len(), 1);
456        assert_eq!(endpoints[0].operation_id, "listWidgets");
457    }
458
459    #[test]
460    fn search_endpoints_finds_the_nearest_neighbor() {
461        let dir = tempfile::tempdir().unwrap();
462        let path = dir.path().join("mcp_store.db");
463        let _conn = seeded_store(&path);
464
465        let store = open_store(&path).unwrap();
466        let results = search_endpoints(&store, &[1.0, 0.0, 0.0, 0.0], 5).unwrap();
467        assert_eq!(results.len(), 1);
468        assert_eq!(results[0].operation_id, "listWidgets");
469        assert!(results[0].similarity > 0.99);
470    }
471
472    #[test]
473    fn cached_in_memory_connection_serves_seeded_data() {
474        let dir = tempfile::tempdir().unwrap();
475        let path = dir.path().join("mcp_store.db");
476        let _conn = seeded_store(&path);
477
478        let cached = cached_in_memory_connection("cached-serves-seeded-data", &path).unwrap();
479        let endpoint = get_endpoint(&cached.lock().unwrap(), "listWidgets")
480            .unwrap()
481            .unwrap();
482        assert_eq!(endpoint.path, "/widgets");
483
484        // vec0 search must also work against the in-memory copy —
485        // `register_vec_extension`'s `sqlite3_auto_extension` applies
486        // process-wide, so it isn't specific to file-backed connections.
487        let results = search_endpoints(&cached.lock().unwrap(), &[1.0, 0.0, 0.0, 0.0], 5).unwrap();
488        assert_eq!(results.len(), 1);
489        assert_eq!(results[0].operation_id, "listWidgets");
490    }
491
492    #[test]
493    fn cached_in_memory_connection_holds_no_lingering_lock_on_the_disk_file() {
494        let dir = tempfile::tempdir().unwrap();
495        let path = dir.path().join("mcp_store.db");
496        // Dropped immediately after seeding: this is the *fixture's* own
497        // handle, not the one under test, and an open handle here would
498        // itself block the `remove_file` below on Windows (which — unlike
499        // POSIX — refuses to delete a file that's still open anywhere),
500        // masking whether `cached_in_memory_connection` released its own.
501        drop(seeded_store(&path));
502
503        let cached = cached_in_memory_connection("cached-releases-disk-handle", &path).unwrap();
504
505        // If the disk connection were still open, removing the file out
506        // from under it would be the exact failure mode this fix exists
507        // to avoid.
508        std::fs::remove_file(&path).unwrap();
509
510        let endpoint = get_endpoint(&cached.lock().unwrap(), "listWidgets")
511            .unwrap()
512            .unwrap();
513        assert_eq!(endpoint.path, "/widgets");
514    }
515
516    #[test]
517    fn cached_in_memory_connection_reuses_the_same_connection_for_the_same_key() {
518        let dir = tempfile::tempdir().unwrap();
519        let path = dir.path().join("mcp_store.db");
520        let _conn = seeded_store(&path);
521
522        let first = cached_in_memory_connection("cached-reuses-same-key", &path).unwrap()
523            as *const Mutex<Connection>;
524        let second = cached_in_memory_connection("cached-reuses-same-key", &path).unwrap()
525            as *const Mutex<Connection>;
526        assert_eq!(
527            first, second,
528            "expected the same cached connection, not a fresh backup"
529        );
530    }
531}
github_mcp/data/store.rs

github_mcp/data/
store.rs
