add generation script for package stats

code/files.py

@@ -19,29 +19,6 @@ def filesize_csv(input_dir: Path, output_dir: Path) -> None:
             writer.writerow([p_date, p_fname, p_mtime, p_size])
 
 
-def newer_than_last_clean(
-    input_glob: Generator[Path], output_glob: Generator[Path], desc: str | None = None
-) -> bool:
-    last_raw_update: datetime = datetime(1, 1, 1)
-    oldest_clean: datetime = datetime.now()
-    for file in input_glob:
-        mdate = datetime.fromtimestamp(file.stat().st_mtime)
-        if mdate > last_raw_update:
-            last_raw_update = mdate
-    for file in output_glob:
-        mdate = datetime.fromtimestamp(file.stat().st_mtime)
-        if mdate < oldest_clean:
-            oldest_clean = mdate
-    if oldest_clean > last_raw_update:
-        print(
-            "INFO: Skip creating cleaned data"
-            + f"{f' for {desc}' if desc else ''}"
-            + f", last processing {oldest_clean} newer than {last_raw_update}"
-        )
-        return False
-    return True
-
-
 def ensure_dirs(input_dir: Path, output_dir: Path):
     if not input_dir.is_dir():
         raise ValueError