doggdad commited on
Commit
7c33da8
·
verified ·
1 Parent(s): deac747

Update src/app.py

Browse files
Files changed (1) hide show
  1. src/app.py +11 -6
src/app.py CHANGED
@@ -37,7 +37,7 @@ load_dotenv()
37
  SPACE_ID = os.getenv("SPACE_ID")
38
  IS_SPACES = SPACE_ID is not None
39
 
40
-
41
  if IS_SPACES:
42
  LANCEDB_HOST_FILE = "/tmp/.lancedb"
43
  VIDEO_DIR = "/tmp/videos/video1"
@@ -45,9 +45,9 @@ if IS_SPACES:
45
  else:
46
  LANCEDB_HOST_FILE = "./shared_data/.lancedb"
47
  VIDEO_DIR = "./shared_data/videos/video1"
48
-
49
- #LANCEDB_HOST_FILE = "./shared_data_2/.lancedb"
50
- #VIDEO_DIR = "./shared_data_2/videos/video1"
51
  TBL_NAME = "vectorstore"
52
 
53
  # Initialize components
@@ -119,17 +119,22 @@ def preprocess_and_store(youtube_url: str):
119
  """Download video, extract frames+metadata, embed & store in LanceDB"""
120
  try:
121
  video_url = youtube_url
 
122
  if os.getenv("SPACE_ID"):
123
  video_dir = "/tmp/videos/video1"
124
  else:
125
  video_dir = "./shared_data/videos/video1"
 
 
126
 
127
  # Use the robust download function
128
- video_filepath = download_video_robust(video_url, video_dir)
 
129
 
130
  # Try to download subtitle, but don't fail if it doesn't work
131
  try:
132
- video_transcript_filepath = download_youtube_subtitle(video_url, video_dir)
 
133
  except Exception as e:
134
  print(f"Warning: Could not download subtitles: {e}")
135
  # Create empty transcript file
 
37
  SPACE_ID = os.getenv("SPACE_ID")
38
  IS_SPACES = SPACE_ID is not None
39
 
40
+ '''
41
  if IS_SPACES:
42
  LANCEDB_HOST_FILE = "/tmp/.lancedb"
43
  VIDEO_DIR = "/tmp/videos/video1"
 
45
  else:
46
  LANCEDB_HOST_FILE = "./shared_data/.lancedb"
47
  VIDEO_DIR = "./shared_data/videos/video1"
48
+ '''
49
+ LANCEDB_HOST_FILE = "./shared_data/.lancedb"
50
+ VIDEO_DIR = "./shared_data/videos/video1"
51
  TBL_NAME = "vectorstore"
52
 
53
  # Initialize components
 
119
  """Download video, extract frames+metadata, embed & store in LanceDB"""
120
  try:
121
  video_url = youtube_url
122
+ '''
123
  if os.getenv("SPACE_ID"):
124
  video_dir = "/tmp/videos/video1"
125
  else:
126
  video_dir = "./shared_data/videos/video1"
127
+ '''
128
+ video_dir = "./shared_data/videos/video1"
129
 
130
  # Use the robust download function
131
+ #video_filepath = download_video_robust(video_url, video_dir)
132
+ video_filepath = "./shared_data/videos/video1/Welcome back to Planet Earth.mp4"
133
 
134
  # Try to download subtitle, but don't fail if it doesn't work
135
  try:
136
+ #video_transcript_filepath = download_youtube_subtitle(video_url, video_dir)
137
+ video_transcript_filepath = "./shared_data/videos/video1/generated_captions.vtt"
138
  except Exception as e:
139
  print(f"Warning: Could not download subtitles: {e}")
140
  # Create empty transcript file