kaitai py

maple-py/compare_pdb.py

@@ -0,0 +1,57 @@
+"""Parse two .pdb files with the kaitai schema and summarise differences."""
+import sys
+from pathlib import Path
+
+sys.path.insert(0, str(Path(__file__).parent / "src"))
+from io import BytesIO
+from kaitaistruct import KaitaiStream
+from kaitai.rekordbox_pdb import RekordboxPdb
+
+
+PAGE_TYPE_NAMES = {t.value: t.name for t in RekordboxPdb.PageType}
+
+
+def summarise(path: Path):
+    data = path.read_bytes()
+    pdb = RekordboxPdb(is_ext=False, _io=KaitaiStream(BytesIO(data)))
+    print(f"\n=== {path} ({path.stat().st_size} bytes) ===")
+    print(f"page_size={pdb.len_page} num_tables={pdb.num_tables} next_unused_page={pdb.next_unused_page} sequence={pdb.sequence}")
+    for tbl in pdb.tables:
+        ptype = PAGE_TYPE_NAMES.get(tbl.type.value, str(tbl.type))
+        page_count = 0
+        row_count = 0
+        err = None
+        try:
+            ref = tbl.first_page
+            while True:
+                page = ref.body
+                page_count += 1
+                if page.is_data_page:
+                    for rg in page.row_groups:
+                        for rpw in rg.rows:
+                            if rpw.present:
+                                row_count += 1
+                if ref.index == tbl.last_page.index:
+                    break
+                ref = page.next_page
+                if page_count > 5000:
+                    break
+        except Exception as e:
+            err = f"{type(e).__name__}: {e}"
+        line = f"  {ptype:24s} first={tbl.first_page.index:>4} last={tbl.last_page.index:>4}  pages={page_count:>3} rows={row_count}"
+        if err:
+            line += f"  ERR={err}"
+        print(line)
+    return pdb
+
+
+def main():
+    base = Path(__file__).parent
+    ref = base / "data/MAPLE/PIONEER/rekordbox/export.pdb"
+    ours = base / "data/usb/PIONEER/rekordbox/export.pdb"
+    summarise(ref)
+    summarise(ours)
+
+
+if __name__ == "__main__":
+    main()