feature/filter-out-invalid-captions

cdp_backend/pipeline/event_gather_pipeline.py

@@ -160,7 +160,11 @@ def create_event_gather_flow(
                 if session.caption_uri is not None:
                     # If the caption doesn't exist, remove the property
                     # This will result in Speech-to-Text being used instead
-                    if not resource_exists(session.caption_uri):
+                    if not resource_exists(
+                        session.caption_uri
+                    ) or not file_utils.caption_is_valid(
+                        tmp_video_filepath, session.caption_uri
+                    ):
                         log.warning(
                             f"File not found using provided caption URI: "
                             f"'{session.caption_uri}'. "

cdp_backend/tests/utils/test_file_utils.py

@@ -8,6 +8,7 @@
 from typing import Optional
 from unittest import mock
 
+import ffmpeg
 import imageio
 import pytest
 from py._path.local import LocalPath
@@ -16,6 +17,7 @@
 from cdp_backend.utils.file_utils import (
     MAX_THUMBNAIL_HEIGHT,
     MAX_THUMBNAIL_WIDTH,
+    caption_is_valid,
     resource_copy,
 )
 
@@ -312,3 +314,26 @@ def test_clip_and_reformat_video(
     assert outfile.exists()
     assert outfile == expected_outfile
     os.remove(outfile)
+
+
+@pytest.mark.parametrize(
+    "video_uri, caption_uri, end_time, expected",
+    [
+        (EXAMPLE_VIDEO_FILENAME, "boston_captions.vtt", 120, False),
+        (EXAMPLE_VIDEO_FILENAME, "boston_captions.vtt", 60, True),
+    ],
+)
+def test_caption_is_valid(
+    resources_dir: Path, video_uri: str, caption_uri: str, end_time: int, expected: bool
+) -> None:
+    temp_video = "caption-test.mp4"
+    ffmpeg.input(str(bytes(resources_dir / video_uri), encoding="utf-8")).output(
+        temp_video, codec="copy", t=end_time
+    ).run(overwrite_output=True)
+
+    valid = caption_is_valid(
+        temp_video,
+        str(bytes(resources_dir / caption_uri), encoding="utf-8"),
+    )
+    os.remove(temp_video)
+    assert valid == expected

cdp_backend/utils/file_utils.py

@@ -10,9 +10,11 @@
 from typing import Optional, Tuple, Union
 from uuid import uuid4
 
+import ffmpeg
 import fireo
 import fsspec
 import requests
+import webvtt
 from fsspec.core import url_to_fs
 
 from ..database import models as db_models
@@ -626,3 +628,40 @@ def clip_and_reformat_video(
         log.error(ffmpeg_stderr)
 
     return output_path
+
+
+def caption_is_valid(video_uri: str, caption_uri: str) -> bool:
+    """
+    Validate the caption file at the URI provided.
+
+    Parameters
+    ----------
+    video_uri: str
+        The URI for the the target video.
+    uri: str
+        The URI to validate caption file for.
+
+    Returns
+    -------
+    status: bool
+        The validation status.
+
+    Notes
+    -----
+    Duration of the video at video_uri
+    and the duration of the caption file are compared.
+    The caption file is accepted if the durations differ by no more than 20%.
+    """
+    try:
+        ffprobe = ffmpeg.probe(video_uri)
+    except ffmpeg.Error as e:
+        log.warning(f"ffprobe{video_uri}): {e.stderr}")
+        return False
+
+    caption_length = webvtt.read(caption_uri).total_length
+    similar_audio_streams = filter(
+        lambda s: s.get("codec_type", "") == "audio"
+        and math.isclose(float(s.get("duration", "0.0")), caption_length, rel_tol=0.2),
+        ffprobe.get("streams", []),
+    )
+    return any(similar_audio_streams)