MichaelR207 commited on
Commit
b34b79d
1 Parent(s): 349ca84

Fix MultiSim.py to download large files properly from huggingface

Browse files
Files changed (1) hide show
  1. MultiSim.py +16 -24
MultiSim.py CHANGED
@@ -18,6 +18,7 @@ import pandas as pd
18
  import os
19
  from collections import defaultdict
20
  import urllib.parse
 
21
 
22
 
23
  import datasets
@@ -300,34 +301,27 @@ class MultiSim(datasets.GeneratorBasedBuilder):
300
  # dl_manager is a datasets.download.DownloadManager that can be used to download and extract URLS
301
  # It can accept any type or nested list/dict and will give back the same structure with the url replaced with path to local files.
302
  # By default the archives will be extracted and a path to a cached folder where they are extracted is returned instead of the archive
303
- download_urls = {}
 
 
304
  if (self.config.name == 'all'):
305
  for subcorpus in _SUBCORPORA:
306
- download_urls[subcorpus + "-train"] = _URLS[subcorpus+"-train"]
307
- download_urls[subcorpus + "-test"] = _URLS[subcorpus+"-test"]
308
- download_urls[subcorpus + "-val"] = _URLS[subcorpus+"-val"]
309
  elif (self.config.name in _LANGUAGES):
310
  lang_code = _LANGUAGES[self.config.name]
311
  for subcorpus in _SUBCORPORA:
312
  if _SUBCORPORA[subcorpus]['language'] == lang_code:
313
- download_urls[subcorpus + "-train"] = _URLS[subcorpus+"-train"]
314
- download_urls[subcorpus + "-test"] = _URLS[subcorpus+"-test"]
315
- download_urls[subcorpus + "-val"] = _URLS[subcorpus+"-val"]
316
  elif (self.config.name in _SUBCORPORA):
317
- download_urls[self.config.name + "-train"] = _URLS[self.config.name+"-train"]
318
- download_urls[self.config.name + "-test"] = _URLS[self.config.name+"-test"]
319
- download_urls[self.config.name + "-val"] = _URLS[self.config.name+"-val"]
320
  else:
321
  print("Invalid configuration name: " + self.config.name + ". Try 'all', 'English', 'ASSET', etc.")
322
-
323
- downloaded_files = dl_manager.download_and_extract(download_urls)
324
-
325
  return [
326
  datasets.SplitGenerator(
327
  name=datasets.Split.TRAIN,
328
  # These kwargs will be passed to _generate_examples
329
  gen_kwargs={
330
- "filepaths": downloaded_files,
331
  "split": "train",
332
  },
333
  ),
@@ -335,7 +329,7 @@ class MultiSim(datasets.GeneratorBasedBuilder):
335
  name=datasets.Split.VALIDATION,
336
  # These kwargs will be passed to _generate_examples
337
  gen_kwargs={
338
- "filepaths": downloaded_files,
339
  "split": "val",
340
  },
341
  ),
@@ -343,7 +337,7 @@ class MultiSim(datasets.GeneratorBasedBuilder):
343
  name=datasets.Split.TEST,
344
  # These kwargs will be passed to _generate_examples
345
  gen_kwargs={
346
- "filepaths": downloaded_files,
347
  "split": "test"
348
  },
349
  ),
@@ -355,18 +349,16 @@ class MultiSim(datasets.GeneratorBasedBuilder):
355
  # The `key` is for legacy reasons (tfds) and is not important in itself, but must be unique for each example.
356
  df = pd.DataFrame()
357
 
358
- if (len(filepaths.keys()) > 1):
359
- for dataset, path in filepaths.items():
360
- if os.path.exists(path):
361
- if dataset.endswith("-"+split):
362
- df = pd.concat([df, pd.read_csv(path)])
363
 
364
  # shuffle the combined dataset
365
  df = df.sample(frac=1, random_state=3600).reset_index(drop=True)
366
  else:
367
- dataset = list(filepaths.keys())[0]
368
- if os.path.exists(filepaths[dataset]):
369
- df = pd.read_csv(filepaths[dataset])
370
 
371
  if len(df) > 0:
372
  for key, row in df.iterrows():
 
18
  import os
19
  from collections import defaultdict
20
  import urllib.parse
21
+ from huggingface_hub import snapshot_download
22
 
23
 
24
  import datasets
 
301
  # dl_manager is a datasets.download.DownloadManager that can be used to download and extract URLS
302
  # It can accept any type or nested list/dict and will give back the same structure with the url replaced with path to local files.
303
  # By default the archives will be extracted and a path to a cached folder where they are extracted is returned instead of the archive
304
+ dataset_path = snapshot_download(repo_id="MichaelR207/MultiSim", repo_type="dataset")
305
+
306
+ filepaths = []
307
  if (self.config.name == 'all'):
308
  for subcorpus in _SUBCORPORA:
309
+ filepaths.append(os.path.join(dataset_path,_SUBCORPORA[subcorpus]['path']))
 
 
310
  elif (self.config.name in _LANGUAGES):
311
  lang_code = _LANGUAGES[self.config.name]
312
  for subcorpus in _SUBCORPORA:
313
  if _SUBCORPORA[subcorpus]['language'] == lang_code:
314
+ filepaths.append(os.path.join(dataset_path,_SUBCORPORA[subcorpus]['path']))
 
 
315
  elif (self.config.name in _SUBCORPORA):
316
+ filepaths = [os.path.join(dataset_path,_SUBCORPORA[self.config.name]['path'])]
 
 
317
  else:
318
  print("Invalid configuration name: " + self.config.name + ". Try 'all', 'English', 'ASSET', etc.")
 
 
 
319
  return [
320
  datasets.SplitGenerator(
321
  name=datasets.Split.TRAIN,
322
  # These kwargs will be passed to _generate_examples
323
  gen_kwargs={
324
+ "filepaths": filepaths,
325
  "split": "train",
326
  },
327
  ),
 
329
  name=datasets.Split.VALIDATION,
330
  # These kwargs will be passed to _generate_examples
331
  gen_kwargs={
332
+ "filepaths": filepaths,
333
  "split": "val",
334
  },
335
  ),
 
337
  name=datasets.Split.TEST,
338
  # These kwargs will be passed to _generate_examples
339
  gen_kwargs={
340
+ "filepaths": filepaths,
341
  "split": "test"
342
  },
343
  ),
 
349
  # The `key` is for legacy reasons (tfds) and is not important in itself, but must be unique for each example.
350
  df = pd.DataFrame()
351
 
352
+ if (len(filepaths) > 1):
353
+ for filepath in filepaths:
354
+ if os.path.exists(filepath + "_" + split + ".csv"):
355
+ df = pd.concat([df, pd.read_csv(filepath + "_" + split + ".csv")])
 
356
 
357
  # shuffle the combined dataset
358
  df = df.sample(frac=1, random_state=3600).reset_index(drop=True)
359
  else:
360
+ if os.path.exists(filepaths[0] + "_" + split + ".csv"):
361
+ df = pd.read_csv(filepaths[0] + "_" + split + ".csv")
 
362
 
363
  if len(df) > 0:
364
  for key, row in df.iterrows():