conceptofmind
commited on
Delete megawika.py
Browse files- megawika.py +0 -197
megawika.py
DELETED
@@ -1,197 +0,0 @@
|
|
1 |
-
"""MegaWika dataset loading script for HuggingFace Datasets."""
|
2 |
-
|
3 |
-
import csv
|
4 |
-
import json
|
5 |
-
import os
|
6 |
-
import re
|
7 |
-
import pathlib
|
8 |
-
from pathlib import Path
|
9 |
-
import yaml
|
10 |
-
from ast import literal_eval
|
11 |
-
import urllib.request
|
12 |
-
import datasets
|
13 |
-
|
14 |
-
_CITATION = """\
|
15 |
-
@article{barham2023megawika,
|
16 |
-
title={MegaWika: Millions of reports and their sources across 50 diverse languages},
|
17 |
-
author={Barham, Samuel and Weller, Orion and
|
18 |
-
Yuan, Michelle and Murray, Kenton and
|
19 |
-
Yarmohammadi, Mahsa and Jiang, Zhengping and
|
20 |
-
Vashishtha, Siddharth and Martin, Alexander and
|
21 |
-
Liu, Anqi and White, Aaron Steven and
|
22 |
-
Boyd-Graber, Jordan and Van Durme, Benjamin
|
23 |
-
},
|
24 |
-
journal={INSERT ARXIV PREPRINT ID HERE},
|
25 |
-
year={2023}
|
26 |
-
}
|
27 |
-
"""
|
28 |
-
|
29 |
-
_DESCRIPTION = """\
|
30 |
-
MegaWika is a multi- and crosslingual text dataset containing 30 million
|
31 |
-
Wikipedia passages with their scraped and cleaned web citations across 50 languages.
|
32 |
-
"""
|
33 |
-
|
34 |
-
_HOMEPAGE = "https://huggingface.co/datasets/DataProvenanceInitiative/Megawika_subset"
|
35 |
-
_LICENSE = "cc-by-sa-4.0"
|
36 |
-
_URL = "https://huggingface.co/datasets/DataProvenanceInitiative/Megawika_subset"
|
37 |
-
|
38 |
-
def load_file_paths():
|
39 |
-
"""Load and parse the files.yml containing dataset file paths.
|
40 |
-
|
41 |
-
Expected YAML structure:
|
42 |
-
en:
|
43 |
-
- https://huggingface.co/datasets/DataProvenanceInitiative/Megawika_subset/resolve/main/en/en-00000-of-06154.jsonl
|
44 |
-
fr:
|
45 |
-
- https://huggingface.co/datasets/DataProvenanceInitiative/Megawika_subset/resolve/main/fr/fr-00000-of-00123.jsonl
|
46 |
-
...
|
47 |
-
|
48 |
-
Returns:
|
49 |
-
dict: Dictionary mapping language codes to lists of file URLs
|
50 |
-
"""
|
51 |
-
file_list_url = "https://huggingface.co/datasets/DataProvenanceInitiative/Megawika_subset/raw/main/files.yml"
|
52 |
-
try:
|
53 |
-
with urllib.request.urlopen(file_list_url) as f:
|
54 |
-
# Direct YAML parsing - the structure is already in the correct format
|
55 |
-
return yaml.safe_load(f)
|
56 |
-
except (yaml.YAMLError, urllib.error.URLError) as exc:
|
57 |
-
print(f"Error loading dataset file paths: {exc}")
|
58 |
-
return {}
|
59 |
-
|
60 |
-
class MegaWikaConfig(datasets.BuilderConfig):
|
61 |
-
"""BuilderConfig for MegaWika."""
|
62 |
-
|
63 |
-
def __init__(self, language=None, **kwargs):
|
64 |
-
"""BuilderConfig for MegaWika.
|
65 |
-
|
66 |
-
Args:
|
67 |
-
language: Language identifier for the dataset split
|
68 |
-
**kwargs: Keyword arguments forwarded to super.
|
69 |
-
"""
|
70 |
-
super().__init__(**kwargs)
|
71 |
-
self.language = language
|
72 |
-
|
73 |
-
class MegaWika(datasets.GeneratorBasedBuilder):
|
74 |
-
"""MegaWika dataset."""
|
75 |
-
|
76 |
-
VERSION = datasets.Version("1.0.0")
|
77 |
-
|
78 |
-
# Load available languages directly from YAML structure
|
79 |
-
LANGUAGES = list(load_file_paths().keys())
|
80 |
-
|
81 |
-
BUILDER_CONFIGS = ([
|
82 |
-
MegaWikaConfig(
|
83 |
-
name="all",
|
84 |
-
language=None,
|
85 |
-
version=VERSION,
|
86 |
-
description="Complete MegaWika dataset across all languages",
|
87 |
-
)
|
88 |
-
] + [
|
89 |
-
MegaWikaConfig(
|
90 |
-
name=lang,
|
91 |
-
language=lang,
|
92 |
-
version=VERSION,
|
93 |
-
description=f"MegaWika dataset for {lang} language",
|
94 |
-
)
|
95 |
-
for lang in LANGUAGES
|
96 |
-
])
|
97 |
-
|
98 |
-
DEFAULT_CONFIG_NAME = "all"
|
99 |
-
|
100 |
-
def _split_generators(self, dl_manager):
|
101 |
-
"""Returns SplitGenerators."""
|
102 |
-
# Load the file paths afresh to ensure we have the latest data
|
103 |
-
data_sources = load_file_paths()
|
104 |
-
|
105 |
-
if self.config.name == "all":
|
106 |
-
# Process all languages
|
107 |
-
selected_sources = data_sources
|
108 |
-
else:
|
109 |
-
# Process single language
|
110 |
-
if self.config.name not in data_sources:
|
111 |
-
raise ValueError(
|
112 |
-
f"Language '{self.config.name}' not found in available languages: {list(data_sources.keys())}"
|
113 |
-
)
|
114 |
-
selected_sources = {self.config.name: data_sources[self.config.name]}
|
115 |
-
|
116 |
-
return [
|
117 |
-
datasets.SplitGenerator(
|
118 |
-
name=datasets.Split.TRAIN,
|
119 |
-
gen_kwargs={
|
120 |
-
"filepaths": dl_manager.download(urls),
|
121 |
-
"language": lang
|
122 |
-
}
|
123 |
-
)
|
124 |
-
for lang, urls in selected_sources.items()
|
125 |
-
]
|
126 |
-
|
127 |
-
def _get_qa_pair_list_features(self, qa_pair, feature_name):
|
128 |
-
"""Helper function to extract QA pair features."""
|
129 |
-
if feature_name in qa_pair and qa_pair[feature_name]:
|
130 |
-
return qa_pair[feature_name]
|
131 |
-
elif feature_name.startswith('en'):
|
132 |
-
base_feature = '_'.join(feature_name.split('_')[1:])
|
133 |
-
if base_feature in qa_pair and qa_pair[base_feature]:
|
134 |
-
return qa_pair[base_feature]
|
135 |
-
return []
|
136 |
-
|
137 |
-
def _generate_examples(self, filepaths, language):
|
138 |
-
"""Yields examples."""
|
139 |
-
_id = 0
|
140 |
-
for filepath in filepaths:
|
141 |
-
try:
|
142 |
-
with open(filepath, "r", encoding="utf-8") as f:
|
143 |
-
for line in f:
|
144 |
-
if line.strip():
|
145 |
-
example = json.loads(line)
|
146 |
-
if isinstance(example, dict):
|
147 |
-
yield _id, {
|
148 |
-
"article_title": example.get("article_title", ""),
|
149 |
-
"article_text": example.get("article_text", ""),
|
150 |
-
"entries": [
|
151 |
-
{
|
152 |
-
"id": entry.get("id", "").lower(),
|
153 |
-
"passage": {
|
154 |
-
"text": entry['passage'].get("text", []),
|
155 |
-
"parse": json.dumps(entry['passage'].get("parse", [{}])),
|
156 |
-
"en_tokens": list(entry['passage'].get("en_tokens", {}).values()),
|
157 |
-
"lang_tokens": list(entry['passage'].get("lang_tokens", {}).values()),
|
158 |
-
"en_lang_token_map": [
|
159 |
-
[int(k), int(v)] for k, v in
|
160 |
-
entry['passage'].get("en_lang_token_map", {}).items()
|
161 |
-
]
|
162 |
-
},
|
163 |
-
"mt": {
|
164 |
-
"original": entry.get("original", ""),
|
165 |
-
"original_sents": entry.get("original_sents", []),
|
166 |
-
"translation": entry.get("translation", ""),
|
167 |
-
"translation_sents": entry.get("translation_sents", []),
|
168 |
-
"translation_probs": entry.get("translation_probs", [[]]),
|
169 |
-
"repetitious_translation": entry.get("repetitious_translation", False)
|
170 |
-
},
|
171 |
-
"source_lang": entry.get("source_lang", ""),
|
172 |
-
"source_url": entry.get("source_url", ""),
|
173 |
-
"source_text": entry.get("source_text", ""),
|
174 |
-
"qa_pairs": [
|
175 |
-
{
|
176 |
-
"question": qa_pair.get('question', ""),
|
177 |
-
"en_answer": qa_pair.get('en_answer', qa_pair.get('answer', "")),
|
178 |
-
'lang_answer': qa_pair.get('lang_answer', ''),
|
179 |
-
'frames': qa_pair.get('frames', []),
|
180 |
-
"en_matches_in_source": self._get_qa_pair_list_features(qa_pair, "en_matches_in_source"),
|
181 |
-
"en_match_in_passage": self._get_qa_pair_list_features(qa_pair, "en_match_in_passage"),
|
182 |
-
"lang_matches_in_source": self._get_qa_pair_list_features(qa_pair, "lang_matches_in_source"),
|
183 |
-
"lang_match_in_passage": self._get_qa_pair_list_features(qa_pair, "lang_match_in_passage"),
|
184 |
-
"passage": qa_pair.get('passage', []),
|
185 |
-
"en_answer_tokens": qa_pair.get('en_answer_tokens', qa_pair.get('answer_tokens', [])),
|
186 |
-
"match_disambiguated_question": qa_pair.get('match_disambiguated_question', ""),
|
187 |
-
}
|
188 |
-
for qa_pair in entry.get('qa_pairs', [])
|
189 |
-
]
|
190 |
-
}
|
191 |
-
for entry in example.get("entries", [])
|
192 |
-
]
|
193 |
-
}
|
194 |
-
_id += 1
|
195 |
-
except Exception as e:
|
196 |
-
print(f"Error reading file {filepath}: {str(e)}")
|
197 |
-
continue
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|