Datasets:
Upload 2 files
Browse files- wikitext.py +47 -0
- wikitext_en.py +36 -0
wikitext.py
ADDED
@@ -0,0 +1,47 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from bs4 import BeautifulSoup
|
2 |
+
from datasets import load_dataset
|
3 |
+
|
4 |
+
def get_titles(file_path):
|
5 |
+
# get the titles from html file (html file is downloaded from https://de.wikipedia.org/wiki/Wikipedia:Exzellente_Artikel)
|
6 |
+
with open(file_path, 'r') as f:
|
7 |
+
html_content = f.read()
|
8 |
+
soup = BeautifulSoup(html_content, 'html.parser')
|
9 |
+
# Find the <tbody> element (you can modify this based on your HTML structure)
|
10 |
+
tbody = soup.find('tbody')
|
11 |
+
# Extract all <tr> elements within the <tbody> element
|
12 |
+
trs = tbody.find_all('tr')
|
13 |
+
# Remove the first two <tr> elements
|
14 |
+
trs = trs[2:]
|
15 |
+
# Extract title attributes from the remaining <tr> elements
|
16 |
+
titles = []
|
17 |
+
for tr in trs:
|
18 |
+
if tr is None or tr.find('a') is None:
|
19 |
+
continue
|
20 |
+
a_tags = tr.find_all('a')
|
21 |
+
for a_tag in a_tags:
|
22 |
+
if a_tag and 'title' in a_tag.attrs:
|
23 |
+
titles.append(a_tag['title'])
|
24 |
+
return titles
|
25 |
+
|
26 |
+
if __name__ == '__main__':
|
27 |
+
titles_exzellent = get_titles('exzellent.txt')
|
28 |
+
#titles_lesenswert = get_titles('lesenswert.txt')
|
29 |
+
titles = titles_exzellent #+ titles_lesenswert
|
30 |
+
titles = list(set(titles))
|
31 |
+
with open('titles.txt', 'w') as f:
|
32 |
+
for title in titles:
|
33 |
+
f.write(title + '\n')
|
34 |
+
|
35 |
+
# Get wikipedia dataset
|
36 |
+
dataset = load_dataset("graelo/wikipedia", "20230901.de", split="train")
|
37 |
+
# Filter dataset
|
38 |
+
dataset = dataset.filter(lambda example: example['title'] in titles, num_proc=64)
|
39 |
+
dataset.map(lambda x: {'text': f"# {x['title']}\n\n{x['text']}"}, remove_columns=['title'], num_proc=64)
|
40 |
+
# Save dataset
|
41 |
+
used_title = [example['title'] for example in dataset]
|
42 |
+
non_used_title = [title for title in titles if title not in used_title]
|
43 |
+
print(f'Number of used titles: {len(used_title)}')
|
44 |
+
print(f'Number of non used titles: {len(non_used_title)}')
|
45 |
+
print(non_used_title[:20])
|
46 |
+
|
47 |
+
dataset.push_to_hub("LeoLM/wiki_de_exzellent", private=True)
|
wikitext_en.py
ADDED
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from bs4 import BeautifulSoup
|
2 |
+
from datasets import load_dataset
|
3 |
+
|
4 |
+
def get_titles(file_path):
|
5 |
+
# Get titles from html file (html file is downloaded from https://en.wikipedia.org/wiki/Wikipedia:Featured_articles)
|
6 |
+
with open(file_path, 'r') as f:
|
7 |
+
html_content = f.read()
|
8 |
+
soup = BeautifulSoup(html_content, 'html.parser')
|
9 |
+
div = soup.find_all('div', attrs={'class': 'wp-fa-contents'})
|
10 |
+
titles = []
|
11 |
+
for d in div:
|
12 |
+
if d is None or d.find('a') is None:
|
13 |
+
continue
|
14 |
+
a_tags = d.find_all('a')
|
15 |
+
for a_tag in a_tags:
|
16 |
+
if a_tag and 'title' in a_tag.attrs:
|
17 |
+
titles.append(a_tag['title'])
|
18 |
+
return titles
|
19 |
+
|
20 |
+
|
21 |
+
if __name__ == '__main__':
|
22 |
+
titles = get_titles('featured.html')
|
23 |
+
titles = list(set(titles))
|
24 |
+
|
25 |
+
# Get wikipedia dataset
|
26 |
+
dataset = load_dataset("graelo/wikipedia", "20230901.en", split="train")
|
27 |
+
# Filter dataset
|
28 |
+
dataset = dataset.filter(lambda example: example['title'] in titles, num_proc=64)
|
29 |
+
dataset.map(lambda x: {'text': f"# {x['title']}\n\n{x['text']}"}, remove_columns=['title'], num_proc=64)
|
30 |
+
# Save dataset
|
31 |
+
used_title = [example['title'] for example in dataset]
|
32 |
+
non_used_title = [title for title in titles if title not in used_title]
|
33 |
+
print(f'Number of used titles: {len(used_title)}')
|
34 |
+
print(f'Number of non used titles: {len(non_used_title)}')
|
35 |
+
print(non_used_title[:20])
|
36 |
+
dataset.push_to_hub("LeoLM/wiki_en_featured", private=True)
|