asahi417 commited on
Commit
fdf60eb
·
1 Parent(s): 389a271
Files changed (1) hide show
  1. download_audio.py +15 -7
download_audio.py CHANGED
@@ -3,12 +3,13 @@ import tarfile
3
  import zipfile
4
  import gzip
5
  import traceback
6
- import requests
7
  from os.path import join as p_join
8
  from tqdm import tqdm
9
  from multiprocessing import Pool
10
  from typing import Optional
11
  from urllib3.connection import ConnectionError
 
12
 
13
  import pandas as pd
14
 
@@ -16,7 +17,7 @@ import pandas as pd
16
  url_metadata_s2s = "https://dl.fbaipublicfiles.com/seamless/data/seamless_align_nov2023_extension/seamless.dataset.metadata.public.enA-jaA.tsv.gz"
17
  url_metadata_s2t = "https://dl.fbaipublicfiles.com/seamless/data/seamless.dataset.metadata.public.enA-jpn.withduration.tsv.gz"
18
  cache_dir_root = "./download"
19
- n_pool = 8
20
 
21
 
22
  def wget(url: str, cache_dir: str, filename: Optional[str] = None):
@@ -24,9 +25,10 @@ def wget(url: str, cache_dir: str, filename: Optional[str] = None):
24
  filename = os.path.basename(url) if not filename else filename
25
  output_file = p_join(cache_dir, filename)
26
  try:
27
- with open(output_file, "wb") as f:
28
- r = requests.get(url)
29
- f.write(r.content)
 
30
  except ConnectionError or KeyboardInterrupt:
31
  traceback.print_exc()
32
  os.remove(output_file)
@@ -77,8 +79,14 @@ def process_dataset(url_metadata):
77
  ) for _, r in df_metadata.iterrows()]
78
  inputs = [x for x in inputs if not os.path.exists(p_join(cache_dir_root, "audio", x[1]))]
79
  print(f"{len(inputs)} urls to download")
80
- with Pool(n_pool) as pool:
81
- pool.starmap(get_audio, tqdm(inputs, total=len(inputs)))
 
 
 
 
 
 
82
 
83
 
84
  if __name__ == '__main__':
 
3
  import zipfile
4
  import gzip
5
  import traceback
6
+ # import requests
7
  from os.path import join as p_join
8
  from tqdm import tqdm
9
  from multiprocessing import Pool
10
  from typing import Optional
11
  from urllib3.connection import ConnectionError
12
+ import urllib.request
13
 
14
  import pandas as pd
15
 
 
17
  url_metadata_s2s = "https://dl.fbaipublicfiles.com/seamless/data/seamless_align_nov2023_extension/seamless.dataset.metadata.public.enA-jaA.tsv.gz"
18
  url_metadata_s2t = "https://dl.fbaipublicfiles.com/seamless/data/seamless.dataset.metadata.public.enA-jpn.withduration.tsv.gz"
19
  cache_dir_root = "./download"
20
+ n_pool = int(os.getenv("N_POOL", 8))
21
 
22
 
23
  def wget(url: str, cache_dir: str, filename: Optional[str] = None):
 
25
  filename = os.path.basename(url) if not filename else filename
26
  output_file = p_join(cache_dir, filename)
27
  try:
28
+ urllib.request.urlretrieve(url, output_file)
29
+ # with open(output_file, "wb") as f:
30
+ # r = requests.get(url)
31
+ # f.write(r.content)
32
  except ConnectionError or KeyboardInterrupt:
33
  traceback.print_exc()
34
  os.remove(output_file)
 
79
  ) for _, r in df_metadata.iterrows()]
80
  inputs = [x for x in inputs if not os.path.exists(p_join(cache_dir_root, "audio", x[1]))]
81
  print(f"{len(inputs)} urls to download")
82
+ if n_pool == 1:
83
+ for url, filename in tqdm(inputs, total=len(inputs)):
84
+ flag = get_audio(url, filename)
85
+ if flag:
86
+ print(f"failed:\n{url}")
87
+ else:
88
+ with Pool(n_pool) as pool:
89
+ pool.starmap(get_audio, tqdm(inputs, total=len(inputs)))
90
 
91
 
92
  if __name__ == '__main__':