Allow skipping missing files in AMI download (#1318)

lhotse/recipes/ami.py

@@ -198,7 +198,10 @@ def download_audio(
             wav_dir.mkdir(parents=True, exist_ok=True)
             wav_path = wav_dir / wav_name
             resumable_download(
-                wav_url, filename=wav_path, force_download=force_download
+                wav_url,
+                filename=wav_path,
+                force_download=force_download,
+                missing_ok=True,
             )
         elif mic == "mdm":
             for array in MDM_ARRAYS:
@@ -208,19 +211,12 @@ def download_audio(
                     wav_dir = target_dir / "wav_db" / item / "audio"
                     wav_dir.mkdir(parents=True, exist_ok=True)
                     wav_path = wav_dir / wav_name
-                    try:
-                        resumable_download(
-                            wav_url, filename=wav_path, force_download=force_download
-                        )
-                    except urllib.error.HTTPError as err:
-                        if err.code == 404:
-                            logging.warning(
-                                f"{wav_url} does not exist. Skipping this file."
-                            )
-                            if os.path.exists(wav_path) and os.path.isfile(wav_path):
-                                os.remove(wav_path)
-                        else:
-                            raise err
+                    resumable_download(
+                        wav_url,
+                        filename=wav_path,
+                        force_download=force_download,
+                        missing_ok=True,
+                    )
         elif mic == "mdm8-bf":
             wav_name = f"{item}_MDM8.wav"
             wav_url = f"{url}/AMICorpusMirror/amicorpus/beamformed/{item}/{wav_name}"

lhotse/utils.py

@@ -454,6 +454,7 @@ def resumable_download(
     filename: Pathlike,
     force_download: bool = False,
     completed_file_size: Optional[int] = None,
+    missing_ok: bool = False,
 ) -> None:
     # Check if the file exists and get its size
     file_exists = os.path.exists(filename)
@@ -518,7 +519,13 @@ def _download(rq, size):
         except urllib.error.HTTPError as e:
             # "Request Range Not Satisfiable" means the requested range
             # starts after the file ends OR that the server does not support range requests.
-            if e.code == 416:
+            if e.code == 404 and missing_ok:
+                logging.warning(
+                    f"{url} does not exist (error 404). Skipping this file."
+                )
+                if Path(filename).is_file():
+                    os.remove(filename)
+            elif e.code == 416:
                 content_range = e.headers.get("Content-Range", None)
                 if content_range is None:
                     # sometimes, the server actually supports range requests