Commit
·
f6569d8
1
Parent(s):
03d8b46
Refactor with context managers for reading files (#9)
Browse files- Refactor dict literals, docstring and download splits (a2a086c3cc01c82a8e42d23442ca06eec344d0f0)
- Use context managers for reading files (8a5bfcbec0586eaf4c23950c6e06becf2d539e02)
- Refactor (52edc1241ee7657ed7521c13ab1d4d1c138e9669)
- Refactor yield per config (8c13accca8a7f73dc3129d093f1a245ef86ce23f)
alt.py
CHANGED
@@ -45,31 +45,27 @@ _WIKI_URL = "https://www2.nict.go.jp/astrec-att/member/mutiyama/ALT/ALT-Parallel
|
|
45 |
|
46 |
|
47 |
class AltParallelConfig(datasets.BuilderConfig):
|
48 |
-
"""BuilderConfig for ALT."""
|
49 |
|
50 |
def __init__(self, languages, **kwargs):
|
51 |
-
"""BuilderConfig for ALT.
|
52 |
|
53 |
Args:
|
54 |
-
|
55 |
-
|
56 |
-
languages: languages that will be used for translation. it should be one of the
|
57 |
**kwargs: keyword arguments forwarded to super.
|
58 |
"""
|
59 |
|
60 |
name = "alt-parallel"
|
61 |
|
62 |
description = "ALT Parallel Corpus"
|
63 |
-
super(
|
64 |
name=name,
|
65 |
description=description,
|
66 |
version=datasets.Version("1.0.0", ""),
|
67 |
**kwargs,
|
68 |
)
|
69 |
|
70 |
-
available_langs =
|
71 |
-
["bg", "en", "en_tok", "fil", "hi", "id", "ja", "khm", "lo", "ms", "my", "th", "vi", "zh"]
|
72 |
-
)
|
73 |
for language in languages:
|
74 |
assert language in available_langs
|
75 |
|
@@ -184,7 +180,7 @@ class Alt(datasets.GeneratorBasedBuilder):
|
|
184 |
else:
|
185 |
data_path = dl_manager.download_and_extract(_URLs[self.config.name])
|
186 |
|
187 |
-
if self.config.name
|
188 |
return [
|
189 |
datasets.SplitGenerator(
|
190 |
name=datasets.Split.TRAIN,
|
@@ -192,9 +188,7 @@ class Alt(datasets.GeneratorBasedBuilder):
|
|
192 |
)
|
193 |
]
|
194 |
else:
|
195 |
-
data_split =
|
196 |
-
for k in _SPLIT:
|
197 |
-
data_split[k] = dl_manager.download_and_extract(_SPLIT[k])
|
198 |
|
199 |
return [
|
200 |
datasets.SplitGenerator(
|
@@ -220,95 +214,86 @@ class Alt(datasets.GeneratorBasedBuilder):
|
|
220 |
urlid = sp[0].replace("URL.", "")
|
221 |
allow_urls[urlid] = {"SNT.URLID": urlid, "url": sp[1]}
|
222 |
|
223 |
-
data = {}
|
224 |
if self.config.name.startswith("alt-parallel"):
|
225 |
-
files = self.config.languages
|
226 |
-
|
227 |
data = {}
|
228 |
-
for lang in
|
229 |
file_path = os.path.join(basepath, "ALT-Parallel-Corpus-20191206", f"data_{lang}.txt")
|
230 |
-
|
231 |
-
|
232 |
-
|
233 |
-
|
234 |
-
|
235 |
-
|
236 |
-
|
237 |
-
|
238 |
-
|
239 |
-
|
240 |
-
|
241 |
-
|
242 |
-
|
243 |
-
|
244 |
-
|
245 |
-
|
246 |
-
|
247 |
-
|
248 |
-
|
249 |
-
|
250 |
-
|
251 |
-
|
|
|
|
|
|
|
252 |
|
253 |
elif self.config.name == "alt-en":
|
254 |
data = {}
|
255 |
for fname in ["English-ALT-Draft.txt", "English-ALT-Reviewed.txt"]:
|
256 |
file_path = os.path.join(basepath, "English-ALT-20210218", fname)
|
257 |
-
|
258 |
-
|
259 |
-
|
260 |
-
|
261 |
|
262 |
-
|
263 |
-
|
264 |
-
|
265 |
|
266 |
-
|
267 |
-
|
268 |
-
|
269 |
-
|
270 |
-
|
271 |
-
|
272 |
-
|
|
|
|
|
273 |
|
274 |
-
|
275 |
-
|
276 |
-
d["status"] = "draft"
|
277 |
-
else:
|
278 |
-
d["status"] = "reviewed"
|
279 |
|
280 |
-
data[sntid] = d
|
281 |
-
fin.close()
|
282 |
elif self.config.name == "alt-jp":
|
283 |
data = {}
|
284 |
for fname in ["Japanese-ALT-Draft.txt", "Japanese-ALT-Reviewed.txt"]:
|
285 |
file_path = os.path.join(basepath, "Japanese-ALT-20210218", fname)
|
286 |
-
|
287 |
-
|
288 |
-
|
289 |
-
|
290 |
-
|
291 |
-
|
292 |
-
|
293 |
-
|
294 |
-
|
295 |
-
|
296 |
-
|
297 |
-
|
298 |
-
|
299 |
-
|
300 |
-
|
301 |
-
|
302 |
-
|
303 |
-
|
304 |
|
305 |
-
|
306 |
-
if fname == "Japanese-ALT-Draft.txt":
|
307 |
-
d["status"] = "draft"
|
308 |
-
else:
|
309 |
-
d["status"] = "reviewed"
|
310 |
-
data[sntid] = d
|
311 |
-
fin.close()
|
312 |
|
313 |
keys = {
|
314 |
"word_alignment": "word-alignment/data_ja.en-ja",
|
@@ -317,108 +302,105 @@ class Alt(datasets.GeneratorBasedBuilder):
|
|
317 |
}
|
318 |
for k in keys:
|
319 |
file_path = os.path.join(basepath, "Japanese-ALT-20210218", keys[k])
|
320 |
-
|
321 |
-
|
322 |
-
|
323 |
-
|
324 |
|
325 |
-
|
326 |
-
|
327 |
-
|
328 |
|
329 |
-
|
330 |
-
|
331 |
-
|
332 |
|
333 |
-
|
|
|
334 |
|
335 |
-
|
336 |
-
|
337 |
|
338 |
elif self.config.name == "alt-my":
|
339 |
-
|
340 |
for fname in ["data"]:
|
341 |
file_path = os.path.join(basepath, "my-alt-190530", fname)
|
342 |
-
|
343 |
-
|
344 |
-
|
345 |
-
|
346 |
-
|
347 |
-
|
348 |
-
|
349 |
-
|
350 |
-
|
351 |
-
|
352 |
-
|
353 |
-
|
354 |
-
|
355 |
-
|
356 |
-
|
357 |
|
358 |
elif self.config.name == "alt-km":
|
359 |
data = {}
|
360 |
for fname in ["data_km.km-tag.nova", "data_km.km-tok.nova"]:
|
361 |
file_path = os.path.join(basepath, "km-nova-181101", fname)
|
362 |
-
|
363 |
-
|
364 |
-
|
365 |
-
|
366 |
-
|
367 |
-
|
368 |
-
|
369 |
-
|
370 |
-
|
371 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
372 |
data[sntid][k] = sp[1]
|
373 |
-
|
374 |
-
|
375 |
-
|
376 |
-
"SNT.URLID.SNTID": sntid,
|
377 |
-
"url": allow_urls[urlid]["url"],
|
378 |
-
"km_pos_tag": None,
|
379 |
-
"km_tokenized": None,
|
380 |
-
}
|
381 |
-
data[sntid][k] = sp[1]
|
382 |
-
fin.close()
|
383 |
|
384 |
elif self.config.name == "alt-my-transliteration":
|
385 |
file_path = os.path.join(basepath, "my-en-transliteration", "data.txt")
|
386 |
# Need to set errors='ignore' because of the unknown error
|
387 |
# UnicodeDecodeError: 'utf-8' codec can't decode byte 0xff in position 0: invalid start byte
|
388 |
# It might due to some issues related to Myanmar alphabets
|
389 |
-
|
390 |
-
|
391 |
-
|
392 |
-
line = line.strip()
|
393 |
|
394 |
-
|
395 |
-
|
396 |
-
|
397 |
|
398 |
-
|
399 |
-
|
400 |
-
|
|
|
|
|
|
|
|
|
|
|
401 |
|
402 |
-
data[_id] = {"en": sp[0].strip(), "my": [sp[1].strip()]}
|
403 |
-
_id += 1
|
404 |
-
fin.close()
|
405 |
elif self.config.name == "alt-my-west-transliteration":
|
406 |
file_path = os.path.join(basepath, "western-myanmar-transliteration", "321.txt")
|
407 |
# Need to set errors='ignore' because of the unknown error
|
408 |
# UnicodeDecodeError: 'utf-8' codec can't decode byte 0xff in position 0: invalid start byte
|
409 |
# It might due to some issues related to Myanmar alphabets
|
410 |
-
|
411 |
-
|
412 |
-
|
413 |
-
|
414 |
-
|
415 |
-
|
416 |
-
|
417 |
-
|
418 |
-
|
419 |
-
|
420 |
-
|
421 |
-
_id = 1
|
422 |
-
for k in data:
|
423 |
-
yield _id, data[k]
|
424 |
-
_id += 1
|
|
|
45 |
|
46 |
|
47 |
class AltParallelConfig(datasets.BuilderConfig):
|
48 |
+
"""BuilderConfig for ALT Parallel."""
|
49 |
|
50 |
def __init__(self, languages, **kwargs):
|
51 |
+
"""BuilderConfig for ALT Parallel.
|
52 |
|
53 |
Args:
|
54 |
+
languages: languages that will be used for translation.
|
|
|
|
|
55 |
**kwargs: keyword arguments forwarded to super.
|
56 |
"""
|
57 |
|
58 |
name = "alt-parallel"
|
59 |
|
60 |
description = "ALT Parallel Corpus"
|
61 |
+
super().__init__(
|
62 |
name=name,
|
63 |
description=description,
|
64 |
version=datasets.Version("1.0.0", ""),
|
65 |
**kwargs,
|
66 |
)
|
67 |
|
68 |
+
available_langs = {"bg", "en", "en_tok", "fil", "hi", "id", "ja", "khm", "lo", "ms", "my", "th", "vi", "zh"}
|
|
|
|
|
69 |
for language in languages:
|
70 |
assert language in available_langs
|
71 |
|
|
|
180 |
else:
|
181 |
data_path = dl_manager.download_and_extract(_URLs[self.config.name])
|
182 |
|
183 |
+
if self.config.name in {"alt-my-transliteration", "alt-my-west-transliteration"}:
|
184 |
return [
|
185 |
datasets.SplitGenerator(
|
186 |
name=datasets.Split.TRAIN,
|
|
|
188 |
)
|
189 |
]
|
190 |
else:
|
191 |
+
data_split = dl_manager.download(_SPLIT)
|
|
|
|
|
192 |
|
193 |
return [
|
194 |
datasets.SplitGenerator(
|
|
|
214 |
urlid = sp[0].replace("URL.", "")
|
215 |
allow_urls[urlid] = {"SNT.URLID": urlid, "url": sp[1]}
|
216 |
|
|
|
217 |
if self.config.name.startswith("alt-parallel"):
|
|
|
|
|
218 |
data = {}
|
219 |
+
for lang in self.config.languages:
|
220 |
file_path = os.path.join(basepath, "ALT-Parallel-Corpus-20191206", f"data_{lang}.txt")
|
221 |
+
with open(file_path, encoding="utf-8") as fin:
|
222 |
+
for line in fin:
|
223 |
+
line = line.strip()
|
224 |
+
sp = line.split("\t")
|
225 |
+
|
226 |
+
_, urlid, sntid = sp[0].split(".")
|
227 |
+
# Some lines have a trailing blank space: "SNT.102053.5598 " in data_fil.txt
|
228 |
+
sntid = sntid.strip()
|
229 |
+
if urlid not in allow_urls:
|
230 |
+
continue
|
231 |
+
|
232 |
+
if sntid not in data:
|
233 |
+
data[sntid] = {
|
234 |
+
"SNT.URLID": urlid,
|
235 |
+
"SNT.URLID.SNTID": sntid,
|
236 |
+
"url": allow_urls[urlid]["url"],
|
237 |
+
"translation": {},
|
238 |
+
}
|
239 |
+
|
240 |
+
# Note that Japanese and Myanmar texts have empty sentence fields in this release.
|
241 |
+
if len(sp) >= 2:
|
242 |
+
data[sntid]["translation"][lang] = sp[1]
|
243 |
+
|
244 |
+
for _id, item in enumerate(data.values()):
|
245 |
+
yield _id, item
|
246 |
|
247 |
elif self.config.name == "alt-en":
|
248 |
data = {}
|
249 |
for fname in ["English-ALT-Draft.txt", "English-ALT-Reviewed.txt"]:
|
250 |
file_path = os.path.join(basepath, "English-ALT-20210218", fname)
|
251 |
+
with open(file_path, encoding="utf-8") as fin:
|
252 |
+
for line in fin:
|
253 |
+
line = line.strip()
|
254 |
+
sp = line.split("\t")
|
255 |
|
256 |
+
_, urlid, sntid = sp[0].split(".")
|
257 |
+
if urlid not in allow_urls:
|
258 |
+
continue
|
259 |
|
260 |
+
d = {
|
261 |
+
"SNT.URLID": urlid,
|
262 |
+
"SNT.URLID.SNTID": sntid,
|
263 |
+
"url": allow_urls[urlid]["url"],
|
264 |
+
"status": "draft" if fname == "English-ALT-Draft.txt" else "reviewed",
|
265 |
+
"value": sp[1],
|
266 |
+
}
|
267 |
+
|
268 |
+
data[sntid] = d
|
269 |
|
270 |
+
for _id, item in enumerate(data.values()):
|
271 |
+
yield _id, item
|
|
|
|
|
|
|
272 |
|
|
|
|
|
273 |
elif self.config.name == "alt-jp":
|
274 |
data = {}
|
275 |
for fname in ["Japanese-ALT-Draft.txt", "Japanese-ALT-Reviewed.txt"]:
|
276 |
file_path = os.path.join(basepath, "Japanese-ALT-20210218", fname)
|
277 |
+
with open(file_path, encoding="utf-8") as fin:
|
278 |
+
for line in fin:
|
279 |
+
line = line.strip()
|
280 |
+
sp = line.split("\t")
|
281 |
+
_, urlid, sntid = sp[0].split(".")
|
282 |
+
if urlid not in allow_urls:
|
283 |
+
continue
|
284 |
+
|
285 |
+
d = {
|
286 |
+
"SNT.URLID": urlid,
|
287 |
+
"SNT.URLID.SNTID": sntid,
|
288 |
+
"url": allow_urls[urlid]["url"],
|
289 |
+
"value": sp[1],
|
290 |
+
"status": "draft" if fname == "Japanese-ALT-Draft.txt" else "reviewed",
|
291 |
+
"word_alignment": None,
|
292 |
+
"en_tokenized": None,
|
293 |
+
"jp_tokenized": None,
|
294 |
+
}
|
295 |
|
296 |
+
data[sntid] = d
|
|
|
|
|
|
|
|
|
|
|
|
|
297 |
|
298 |
keys = {
|
299 |
"word_alignment": "word-alignment/data_ja.en-ja",
|
|
|
302 |
}
|
303 |
for k in keys:
|
304 |
file_path = os.path.join(basepath, "Japanese-ALT-20210218", keys[k])
|
305 |
+
with open(file_path, encoding="utf-8") as fin:
|
306 |
+
for line in fin:
|
307 |
+
line = line.strip()
|
308 |
+
sp = line.split("\t")
|
309 |
|
310 |
+
# Note that Japanese and Myanmar texts have empty sentence fields in this release.
|
311 |
+
if len(sp) < 2:
|
312 |
+
continue
|
313 |
|
314 |
+
_, urlid, sntid = sp[0].split(".")
|
315 |
+
if urlid not in allow_urls:
|
316 |
+
continue
|
317 |
|
318 |
+
if sntid in data:
|
319 |
+
data[sntid][k] = sp[1]
|
320 |
|
321 |
+
for _id, item in enumerate(data.values()):
|
322 |
+
yield _id, item
|
323 |
|
324 |
elif self.config.name == "alt-my":
|
325 |
+
_id = 0
|
326 |
for fname in ["data"]:
|
327 |
file_path = os.path.join(basepath, "my-alt-190530", fname)
|
328 |
+
with open(file_path, encoding="utf-8") as fin:
|
329 |
+
for line in fin:
|
330 |
+
line = line.strip()
|
331 |
+
sp = line.split("\t")
|
332 |
+
_, urlid, sntid = sp[0].split(".")
|
333 |
+
if urlid not in allow_urls:
|
334 |
+
continue
|
335 |
+
|
336 |
+
yield _id, {
|
337 |
+
"SNT.URLID": urlid,
|
338 |
+
"SNT.URLID.SNTID": sntid,
|
339 |
+
"url": allow_urls[urlid]["url"],
|
340 |
+
"value": sp[1],
|
341 |
+
}
|
342 |
+
_id += 1
|
343 |
|
344 |
elif self.config.name == "alt-km":
|
345 |
data = {}
|
346 |
for fname in ["data_km.km-tag.nova", "data_km.km-tok.nova"]:
|
347 |
file_path = os.path.join(basepath, "km-nova-181101", fname)
|
348 |
+
with open(file_path, encoding="utf-8") as fin:
|
349 |
+
for line in fin:
|
350 |
+
line = line.strip()
|
351 |
+
sp = line.split("\t")
|
352 |
+
_, urlid, sntid = sp[0].split(".")
|
353 |
+
if urlid not in allow_urls:
|
354 |
+
continue
|
355 |
+
|
356 |
+
k = "km_pos_tag" if fname == "data_km.km-tag.nova" else "km_tokenized"
|
357 |
+
if sntid not in data:
|
358 |
+
data[sntid] = {
|
359 |
+
"SNT.URLID": urlid,
|
360 |
+
"SNT.URLID.SNTID": sntid,
|
361 |
+
"url": allow_urls[urlid]["url"],
|
362 |
+
"km_pos_tag": None,
|
363 |
+
"km_tokenized": None,
|
364 |
+
}
|
365 |
data[sntid][k] = sp[1]
|
366 |
+
|
367 |
+
for _id, item in enumerate(data.values()):
|
368 |
+
yield _id, item
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
369 |
|
370 |
elif self.config.name == "alt-my-transliteration":
|
371 |
file_path = os.path.join(basepath, "my-en-transliteration", "data.txt")
|
372 |
# Need to set errors='ignore' because of the unknown error
|
373 |
# UnicodeDecodeError: 'utf-8' codec can't decode byte 0xff in position 0: invalid start byte
|
374 |
# It might due to some issues related to Myanmar alphabets
|
375 |
+
with open(file_path, encoding="utf-8", errors="ignore") as fin:
|
376 |
+
for _id, line in enumerate(fin):
|
377 |
+
line = line.strip()
|
|
|
378 |
|
379 |
+
# I don't know why there are \x00 between |||. They don't show in the editor.
|
380 |
+
line = line.replace("\x00", "")
|
381 |
+
sp = line.split("|||")
|
382 |
|
383 |
+
# When I read data, it seems to have empty sentence betweem the actual sentence. Don't know why?
|
384 |
+
if len(sp) < 2:
|
385 |
+
continue
|
386 |
+
|
387 |
+
yield _id, {
|
388 |
+
"en": sp[0].strip(),
|
389 |
+
"my": [sp[1].strip()],
|
390 |
+
}
|
391 |
|
|
|
|
|
|
|
392 |
elif self.config.name == "alt-my-west-transliteration":
|
393 |
file_path = os.path.join(basepath, "western-myanmar-transliteration", "321.txt")
|
394 |
# Need to set errors='ignore' because of the unknown error
|
395 |
# UnicodeDecodeError: 'utf-8' codec can't decode byte 0xff in position 0: invalid start byte
|
396 |
# It might due to some issues related to Myanmar alphabets
|
397 |
+
with open(file_path, encoding="utf-8", errors="ignore") as fin:
|
398 |
+
for _id, line in enumerate(fin):
|
399 |
+
line = line.strip()
|
400 |
+
line = line.replace("\x00", "")
|
401 |
+
sp = line.split("|||")
|
402 |
+
|
403 |
+
yield _id, {
|
404 |
+
"en": sp[0].strip(),
|
405 |
+
"my": [k.strip() for k in sp[1].split("|")],
|
406 |
+
}
|
|
|
|
|
|
|
|
|
|