feat: add acoustic dedupe pipeline and docker runtime

am_deduper/scanner.py

@@ -1,19 +1,17 @@
 from __future__ import annotations
 
-import os
 from pathlib import Path
 from typing import Iterator
 
 from .constants import MUSIC_EXTENSIONS
 from .db import connect
 from .hashing import fast_hash
+from .metadata import read_metadata
 
 
 def iter_music(root: Path) -> Iterator[Path]:
     for path in root.rglob("*"):
-        if not path.is_file() or path.is_symlink():
-            continue
-        if path.name.startswith(".") or ".dedupe" in path.parts:
+        if not path.is_file() or path.is_symlink() or path.name.startswith(".") or ".dedupe" in path.parts:
             continue
         if path.suffix.lower() in MUSIC_EXTENSIONS:
             yield path
@@ -22,26 +20,21 @@ def iter_music(root: Path) -> Iterator[Path]:
 def index_file(conn, path: Path) -> None:
     st = path.stat()
     fh = fast_hash(path)
-    conn.execute(
-        """
-        INSERT INTO audio_tracks(file_path, file_size, fast_hash, inode, device, status)
-        VALUES (?, ?, ?, ?, ?, 'active')
-        ON CONFLICT(file_path) DO UPDATE SET
-          file_size=excluded.file_size,
-          fast_hash=excluded.fast_hash,
-          inode=excluded.inode,
-          device=excluded.device,
-          updated_at=CURRENT_TIMESTAMP,
-          status='active'
-        """,
-        (str(path), st.st_size, fh, st.st_ino, st.st_dev),
-    )
+    meta = read_metadata(path)
+    flags = ",".join(meta.get("version_flags", []))
+    conn.execute("""
+        INSERT INTO audio_tracks(file_path,file_size,fast_hash,duration,artist,title,album,format,bitrate,sample_rate,bit_depth,version_flags,inode,device,status)
+        VALUES (?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,'active')
+        ON CONFLICT(file_path) DO UPDATE SET file_size=excluded.file_size, fast_hash=excluded.fast_hash,
+          duration=excluded.duration,artist=excluded.artist,title=excluded.title,album=excluded.album,format=excluded.format,
+          bitrate=excluded.bitrate,sample_rate=excluded.sample_rate,bit_depth=excluded.bit_depth,version_flags=excluded.version_flags,
+          inode=excluded.inode,device=excluded.device,updated_at=CURRENT_TIMESTAMP,status='active'
+    """, (str(path),st.st_size,fh,meta.get("duration"),meta.get("artist"),meta.get("title"),meta.get("album"),meta.get("format"),meta.get("bitrate"),meta.get("sample_rate"),meta.get("bit_depth"),flags,st.st_ino,st.st_dev))
 
 
 def scan(root: Path, db_path: Path) -> int:
     count = 0
     with connect(db_path) as conn:
         for path in iter_music(root):
-            index_file(conn, path)
-            count += 1
+            index_file(conn, path); count += 1
     return count