From 916e6fcabbac5e79526be1cef7d855b30187ff72 Mon Sep 17 00:00:00 2001
From: serversdown <brian@serversdown.net>
Date: Thu, 2 Jul 2026 06:30:11 +0000
Subject: [PATCH] feat(db): zip recent events' waveform files

---
 sfm/db_snapshot.py        | 26 ++++++++++++++++++++++++++
 tests/test_db_snapshot.py | 34 ++++++++++++++++++++++++++++++++++
 2 files changed, 60 insertions(+)

diff --git a/sfm/db_snapshot.py b/sfm/db_snapshot.py
index 0bd2d68..45bd3d8 100644
--- a/sfm/db_snapshot.py
+++ b/sfm/db_snapshot.py
@@ -6,6 +6,7 @@ unit-testable without HTTP.
 """
 from __future__ import annotations
 import sqlite3
+import zipfile
 from pathlib import Path
 
 
@@ -22,3 +23,28 @@ def snapshot_db(src_path: Path, dest_path: Path) -> Path:
         src.close()
         dest.close()
     return Path(dest_path)
+
+
+def build_recent_waveforms_zip(db, store, n: int, dest_path: Path) -> dict:
+    """Zip every on-disk file (event file + .h5/.a5.pkl/.sfm.json siblings) for
+    the n most-recent events, arcname '<serial>/<filename>'. Events with no
+    resolvable files on disk are skipped. Returns {'events': int, 'files': int}."""
+    rows = db.query_events(limit=n, offset=0)
+    events = 0
+    files = 0
+    with zipfile.ZipFile(dest_path, "w", compression=zipfile.ZIP_DEFLATED) as zf:
+        for row in rows:
+            serial = row.get("serial")
+            fn = row.get("blastware_filename")
+            if not serial or not fn:
+                continue
+            serial_dir = store.root / serial
+            matched = sorted(serial_dir.glob(f"{fn}*")) if serial_dir.is_dir() else []
+            matched = [p for p in matched if p.is_file()]
+            if not matched:
+                continue
+            events += 1
+            for p in matched:
+                zf.write(p, arcname=f"{serial}/{p.name}")
+                files += 1
+    return {"events": events, "files": files}
diff --git a/tests/test_db_snapshot.py b/tests/test_db_snapshot.py
index 32cacaa..0a5f39b 100644
--- a/tests/test_db_snapshot.py
+++ b/tests/test_db_snapshot.py
@@ -45,3 +45,37 @@ def test_snapshot_db_produces_queryable_copy(tmp_path):
     n = conn.execute("SELECT COUNT(*) FROM t").fetchone()[0]
     conn.close()
     assert n == 5
+
+
+def test_recent_waveforms_zip_bundles_all_per_event_files(tmp_path):
+    import zipfile
+    from sfm.database import SeismoDb
+    from sfm.waveform_store import WaveformStore
+    from sfm.db_snapshot import build_recent_waveforms_zip
+
+    db = SeismoDb(tmp_path / "seismo_relay.db")   # creates schema on init
+    store = WaveformStore(tmp_path / "waveforms")
+
+    def add(serial, fn, ts):
+        d = store.root / serial
+        d.mkdir(parents=True, exist_ok=True)
+        for suffix in ("", ".h5", ".a5.pkl", ".sfm.json"):
+            (d / f"{fn}{suffix}").write_bytes(b"data")
+        _seed_event(db.db_path, serial, fn, ts)
+
+    add("BE111", "AAAA1111.OE0H", "2026-06-01T00:00:00")
+    add("BE111", "BBBB2222.OE0H", "2026-06-02T00:00:00")   # newest
+
+    dest = tmp_path / "wf.zip"
+    stats = build_recent_waveforms_zip(db, store, n=1, dest_path=dest)
+
+    assert stats["events"] == 1
+    assert stats["files"] == 4
+    with zipfile.ZipFile(dest) as zf:
+        names = set(zf.namelist())
+    assert names == {
+        "BE111/BBBB2222.OE0H",
+        "BE111/BBBB2222.OE0H.h5",
+        "BE111/BBBB2222.OE0H.a5.pkl",
+        "BE111/BBBB2222.OE0H.sfm.json",
+    }