system HF staff commited on
Commit
20b5612
1 Parent(s): 501dddf

Update files from the datasets library (from 1.16.0)

Browse files

Release notes: https://github.com/huggingface/datasets/releases/tag/1.16.0

Files changed (2) hide show
  1. README.md +1 -0
  2. reuters21578.py +52 -33
README.md CHANGED
@@ -1,4 +1,5 @@
1
  ---
 
2
  languages:
3
  - en
4
  paperswithcode_id: reuters-21578
1
  ---
2
+ pretty_name: Reuters-21578 Text Categorization Collection
3
  languages:
4
  - en
5
  paperswithcode_id: reuters-21578
reuters21578.py CHANGED
@@ -17,7 +17,6 @@
17
  """Reuters 21578"""
18
 
19
 
20
- import os
21
  from textwrap import dedent
22
 
23
  import datasets
@@ -253,33 +252,53 @@ class Reuters21578(datasets.GeneratorBasedBuilder):
253
  )
254
 
255
  def _split_generators(self, dl_manager):
256
- dl_dir = dl_manager.download_and_extract(_DATA_URL)
257
- files = [os.path.join(dl_dir, "reut2-" + "%03d" % i + ".sgm") for i in range(22)]
258
  if self.config.name == "ModHayes":
259
  return [
260
  datasets.SplitGenerator(
261
  name=datasets.Split.TEST,
262
- gen_kwargs={"filepath": files, "split": "PUBLISHED-TESTSET"},
 
 
 
 
263
  ),
264
  datasets.SplitGenerator(
265
  name=datasets.Split.TRAIN,
266
- gen_kwargs={"filepath": files, "split": "TRAINING-SET"},
 
 
 
 
267
  ),
268
  ]
269
  else:
270
  return [
271
- datasets.SplitGenerator(name=datasets.Split.TEST, gen_kwargs={"filepath": files, "split": "TEST"}),
272
- datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepath": files, "split": "TRAIN"}),
273
- datasets.SplitGenerator(name="unused", gen_kwargs={"filepath": files, "split": "NOT-USED"}),
 
 
 
 
 
 
 
 
 
 
 
 
 
274
  ]
275
 
276
- def _generate_examples(self, filepath, split):
277
  """This function returns the examples in the raw (text) form."""
278
- for file in filepath:
279
- with open(
280
- file, encoding="utf-8", errors="ignore"
281
- ) as f: # only the file reut2-017 has one line non UTF-8 encoded so we can ignore it
282
- line = f.readline()
283
  while line:
284
  if line.startswith("<REUTERS"):
285
  lewis_split = ""
@@ -301,7 +320,7 @@ class Reuters21578(datasets.GeneratorBasedBuilder):
301
  old_id = line[4].split("=")[1]
302
  new_id = line[5].split("=")[1][:-1]
303
  has_topic = line[1].split("=")[1]
304
- line = f.readline()
305
  if (
306
  (self.config.name == "ModHayes" and split not in cgis_split)
307
  or (
@@ -325,7 +344,7 @@ class Reuters21578(datasets.GeneratorBasedBuilder):
325
  ): # skip example that are not in the current split
326
  li = line
327
  while li and not li.startswith("<REUTERS"):
328
- li = f.readline()
329
  if li:
330
  line = li
331
  elif line.startswith("<TOPICS>"):
@@ -333,65 +352,65 @@ class Reuters21578(datasets.GeneratorBasedBuilder):
333
  line = line.split("<D>")
334
  topics = [topic.replace("</D>", "") for topic in line[1:]]
335
  topics = [topic.replace("</TOPICS>\n", "") for topic in topics]
336
- line = f.readline()
337
  elif line.startswith("<PLACES>"):
338
  if line.replace("\n", "") != "<PLACES></PLACES>":
339
  line = line.split("<D>")
340
  places = [place.replace("</D>", "") for place in line[1:]]
341
  places = [place.replace("</PLACES>\n", "") for place in places]
342
- line = f.readline()
343
  elif line.startswith("<PEOPLE>"):
344
  if line.replace("\n", "") != "<PEOPLE></PEOPLE>":
345
  line = line.split("<D>")
346
  people = [p.replace("</D>", "") for p in line[1:]]
347
  people = [p.replace("</PEOPLE>\n", "") for p in people]
348
- line = f.readline()
349
  elif line.startswith("<ORGS>"):
350
  if line.replace("\n", "") != "<ORGS></ORGS>":
351
  line = line.split("<D>")
352
  orgs = [org.replace("</D>", "") for org in line[1:]]
353
  orgs = [org.replace("</ORGS>\n", "") for org in orgs]
354
- line = f.readline()
355
  elif line.startswith("<EXCHANGES>"):
356
  if line.replace("\n", "") != "<EXCHANGES></EXCHANGES>":
357
  line = line.split("<D>")
358
  exchanges = [ex.replace("</D>", "") for ex in line[1:]]
359
  exchanges = [ex.replace("</EXCHANGES>\n", "") for ex in exchanges]
360
- line = f.readline()
361
  elif line.startswith("<DATE>"):
362
  date = line.replace("\n", "")
363
  date = line[6:-8]
364
- line = f.readline()
365
  elif line.startswith("<TITLE>"):
366
  title = line[7:-9]
367
- line = f.readline()
368
  elif "*<TITLE>" in line:
369
  # These lines start with a variable number of * chars
370
  title = line.split("*<TITLE>")[1][:-1]
371
- line = f.readline()
372
  while "</TITLE>" not in line:
373
  # Convert any \n in TYPE="BRIEF" text to spaces to match other titles
374
  title += " " + line[:-1]
375
- line = f.readline()
376
  elif "<BODY>" in line:
377
  text = line.split("<BODY>")[1]
378
- line = f.readline()
379
  while "</BODY>" not in line:
380
  text += line
381
- line = f.readline()
382
  elif line.startswith('<TEXT TYPE="UNPROC">'):
383
  text_type = '"UNPROC"'
384
  text = line[20:]
385
- line = f.readline()
386
  while "</TEXT>" not in line:
387
  text += line
388
- line = f.readline()
389
  elif line.startswith('<TEXT TYPE="BRIEF">'):
390
  text_type = '"BRIEF"'
391
- line = f.readline()
392
  elif line.startswith("<TEXT>"):
393
  text_type = '"NORM"'
394
- line = f.readline()
395
  elif line.startswith("</REUTERS>"):
396
  yield new_id, {
397
  "lewis_split": lewis_split,
@@ -408,6 +427,6 @@ class Reuters21578(datasets.GeneratorBasedBuilder):
408
  "text": text,
409
  "text_type": text_type,
410
  }
411
- line = f.readline()
412
  else:
413
- line = f.readline()
17
  """Reuters 21578"""
18
 
19
 
 
20
  from textwrap import dedent
21
 
22
  import datasets
252
  )
253
 
254
  def _split_generators(self, dl_manager):
255
+ archive = dl_manager.download(_DATA_URL)
256
+ filepaths = ["reut2-" + "%03d" % i + ".sgm" for i in range(22)]
257
  if self.config.name == "ModHayes":
258
  return [
259
  datasets.SplitGenerator(
260
  name=datasets.Split.TEST,
261
+ gen_kwargs={
262
+ "filepaths": filepaths,
263
+ "split": "PUBLISHED-TESTSET",
264
+ "files": dl_manager.iter_archive(archive),
265
+ },
266
  ),
267
  datasets.SplitGenerator(
268
  name=datasets.Split.TRAIN,
269
+ gen_kwargs={
270
+ "filepaths": filepaths,
271
+ "split": "TRAINING-SET",
272
+ "files": dl_manager.iter_archive(archive),
273
+ },
274
  ),
275
  ]
276
  else:
277
  return [
278
+ datasets.SplitGenerator(
279
+ name=datasets.Split.TEST,
280
+ gen_kwargs={"filepaths": filepaths, "split": "TEST", "files": dl_manager.iter_archive(archive)},
281
+ ),
282
+ datasets.SplitGenerator(
283
+ name=datasets.Split.TRAIN,
284
+ gen_kwargs={"filepaths": filepaths, "split": "TRAIN", "files": dl_manager.iter_archive(archive)},
285
+ ),
286
+ datasets.SplitGenerator(
287
+ name="unused",
288
+ gen_kwargs={
289
+ "filepaths": filepaths,
290
+ "split": "NOT-USED",
291
+ "files": dl_manager.iter_archive(archive),
292
+ },
293
+ ),
294
  ]
295
 
296
+ def _generate_examples(self, filepaths, split, files):
297
  """This function returns the examples in the raw (text) form."""
298
+ for path, f in files:
299
+ if path in filepaths:
300
+ # only the file reut2-017 has one line non UTF-8 encoded so we can ignore it
301
+ line = f.readline().decode("utf-8", errors="ignore")
 
302
  while line:
303
  if line.startswith("<REUTERS"):
304
  lewis_split = ""
320
  old_id = line[4].split("=")[1]
321
  new_id = line[5].split("=")[1][:-1]
322
  has_topic = line[1].split("=")[1]
323
+ line = f.readline().decode("utf-8", errors="ignore")
324
  if (
325
  (self.config.name == "ModHayes" and split not in cgis_split)
326
  or (
344
  ): # skip example that are not in the current split
345
  li = line
346
  while li and not li.startswith("<REUTERS"):
347
+ li = f.readline().decode("utf-8", errors="ignore")
348
  if li:
349
  line = li
350
  elif line.startswith("<TOPICS>"):
352
  line = line.split("<D>")
353
  topics = [topic.replace("</D>", "") for topic in line[1:]]
354
  topics = [topic.replace("</TOPICS>\n", "") for topic in topics]
355
+ line = f.readline().decode("utf-8", errors="ignore")
356
  elif line.startswith("<PLACES>"):
357
  if line.replace("\n", "") != "<PLACES></PLACES>":
358
  line = line.split("<D>")
359
  places = [place.replace("</D>", "") for place in line[1:]]
360
  places = [place.replace("</PLACES>\n", "") for place in places]
361
+ line = f.readline().decode("utf-8", errors="ignore")
362
  elif line.startswith("<PEOPLE>"):
363
  if line.replace("\n", "") != "<PEOPLE></PEOPLE>":
364
  line = line.split("<D>")
365
  people = [p.replace("</D>", "") for p in line[1:]]
366
  people = [p.replace("</PEOPLE>\n", "") for p in people]
367
+ line = f.readline().decode("utf-8", errors="ignore")
368
  elif line.startswith("<ORGS>"):
369
  if line.replace("\n", "") != "<ORGS></ORGS>":
370
  line = line.split("<D>")
371
  orgs = [org.replace("</D>", "") for org in line[1:]]
372
  orgs = [org.replace("</ORGS>\n", "") for org in orgs]
373
+ line = f.readline().decode("utf-8", errors="ignore")
374
  elif line.startswith("<EXCHANGES>"):
375
  if line.replace("\n", "") != "<EXCHANGES></EXCHANGES>":
376
  line = line.split("<D>")
377
  exchanges = [ex.replace("</D>", "") for ex in line[1:]]
378
  exchanges = [ex.replace("</EXCHANGES>\n", "") for ex in exchanges]
379
+ line = f.readline().decode("utf-8", errors="ignore")
380
  elif line.startswith("<DATE>"):
381
  date = line.replace("\n", "")
382
  date = line[6:-8]
383
+ line = f.readline().decode("utf-8", errors="ignore")
384
  elif line.startswith("<TITLE>"):
385
  title = line[7:-9]
386
+ line = f.readline().decode("utf-8", errors="ignore")
387
  elif "*<TITLE>" in line:
388
  # These lines start with a variable number of * chars
389
  title = line.split("*<TITLE>")[1][:-1]
390
+ line = f.readline().decode("utf-8", errors="ignore")
391
  while "</TITLE>" not in line:
392
  # Convert any \n in TYPE="BRIEF" text to spaces to match other titles
393
  title += " " + line[:-1]
394
+ line = f.readline().decode("utf-8", errors="ignore")
395
  elif "<BODY>" in line:
396
  text = line.split("<BODY>")[1]
397
+ line = f.readline().decode("utf-8", errors="ignore")
398
  while "</BODY>" not in line:
399
  text += line
400
+ line = f.readline().decode("utf-8", errors="ignore")
401
  elif line.startswith('<TEXT TYPE="UNPROC">'):
402
  text_type = '"UNPROC"'
403
  text = line[20:]
404
+ line = f.readline().decode("utf-8", errors="ignore")
405
  while "</TEXT>" not in line:
406
  text += line
407
+ line = f.readline().decode("utf-8", errors="ignore")
408
  elif line.startswith('<TEXT TYPE="BRIEF">'):
409
  text_type = '"BRIEF"'
410
+ line = f.readline().decode("utf-8", errors="ignore")
411
  elif line.startswith("<TEXT>"):
412
  text_type = '"NORM"'
413
+ line = f.readline().decode("utf-8", errors="ignore")
414
  elif line.startswith("</REUTERS>"):
415
  yield new_id, {
416
  "lewis_split": lewis_split,
427
  "text": text,
428
  "text_type": text_type,
429
  }
430
+ line = f.readline().decode("utf-8", errors="ignore")
431
  else:
432
+ line = f.readline().decode("utf-8", errors="ignore")