Spaces:
Running
Running
Merge pull request #104 from borisdayma/feat-hf_hub
Browse files- dalle_mini/text.py +32 -43
dalle_mini/text.py
CHANGED
@@ -2,36 +2,28 @@
|
|
2 |
Utilities for processing text.
|
3 |
"""
|
4 |
|
5 |
-
import requests
|
6 |
from pathlib import Path
|
7 |
from unidecode import unidecode
|
8 |
|
9 |
import re, math, random, html
|
10 |
import ftfy
|
11 |
|
12 |
-
|
13 |
-
WIKI_STATS_LOCAL = Path(WIKI_STATS_URL).parts[-1]
|
14 |
|
15 |
# based on wiki word occurence
|
16 |
person_token = [("a person", 282265), ("someone", 121194), ("somebody", 12219)]
|
17 |
temp_token = "xtokx" # avoid repeating chars
|
18 |
|
19 |
|
20 |
-
def get_wiki_file():
|
21 |
-
if not Path(WIKI_STATS_LOCAL).exists():
|
22 |
-
r = requests.get(WIKI_STATS_URL, stream=True)
|
23 |
-
with open(WIKI_STATS_LOCAL, "wb") as fd:
|
24 |
-
for chunk in r.iter_content(chunk_size=128):
|
25 |
-
fd.write(chunk)
|
26 |
-
return WIKI_STATS_LOCAL
|
27 |
-
|
28 |
-
|
29 |
class HashtagProcessor:
|
30 |
# Adapted from wordninja library
|
31 |
# We use our wikipedia word count + a good heuristic to make it work
|
32 |
def __init__(self):
|
|
|
|
|
|
|
33 |
self._word_cost = (
|
34 |
-
l.split()[0] for l in Path(
|
35 |
)
|
36 |
self._word_cost = {
|
37 |
str(k): math.log(float(i + 1)) for i, k in enumerate(self._word_cost)
|
@@ -158,7 +150,7 @@ def handle_special_chars(t):
|
|
158 |
|
159 |
def expand_hashtags(t, hashtag_processor):
|
160 |
"Remove # and try to split words"
|
161 |
-
return re.sub("#(\w+)", lambda m: hashtag_processor(m.group(1)), t)
|
162 |
|
163 |
|
164 |
_re_ignore_chars = """[_#\/\\%]"""
|
@@ -205,15 +197,13 @@ class TextNormalizer:
|
|
205 |
def __init__(self):
|
206 |
self._hashtag_processor = HashtagProcessor()
|
207 |
|
208 |
-
def __call__(self, t
|
209 |
-
|
210 |
# fix some characters
|
211 |
t = ftfy.fix_text(t)
|
212 |
# fix html
|
213 |
t = fix_html(t)
|
214 |
-
|
215 |
-
|
216 |
-
t = unidecode(t)
|
217 |
# lower case
|
218 |
t = t.lower()
|
219 |
# replace <PERSON> (for CC12M)
|
@@ -226,32 +216,31 @@ class TextNormalizer:
|
|
226 |
t = remove_urls(t)
|
227 |
# remove commas in numbers
|
228 |
t = remove_comma_numbers(t)
|
229 |
-
|
230 |
-
|
231 |
-
|
232 |
-
|
233 |
-
|
234 |
-
|
235 |
-
|
236 |
-
|
237 |
-
|
238 |
-
|
239 |
-
|
240 |
-
|
241 |
-
|
242 |
-
|
243 |
-
|
244 |
-
|
245 |
-
|
246 |
-
|
247 |
-
|
248 |
-
|
249 |
-
|
250 |
-
|
251 |
-
t = merge_quotes(t)
|
252 |
# remove multiple spaces
|
253 |
t = remove_extra_spaces(t)
|
254 |
# remove first and last comma
|
255 |
t = remove_first_last_commas(t)
|
256 |
# always start with a space
|
257 |
-
return f" {t}"
|
|
|
2 |
Utilities for processing text.
|
3 |
"""
|
4 |
|
|
|
5 |
from pathlib import Path
|
6 |
from unidecode import unidecode
|
7 |
|
8 |
import re, math, random, html
|
9 |
import ftfy
|
10 |
|
11 |
+
from huggingface_hub import hf_hub_download
|
|
|
12 |
|
13 |
# based on wiki word occurence
|
14 |
person_token = [("a person", 282265), ("someone", 121194), ("somebody", 12219)]
|
15 |
temp_token = "xtokx" # avoid repeating chars
|
16 |
|
17 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
18 |
class HashtagProcessor:
|
19 |
# Adapted from wordninja library
|
20 |
# We use our wikipedia word count + a good heuristic to make it work
|
21 |
def __init__(self):
|
22 |
+
wiki_word_frequency = hf_hub_download(
|
23 |
+
"dalle-mini/dalle-mini", filename="enwiki-words-frequency.txt"
|
24 |
+
)
|
25 |
self._word_cost = (
|
26 |
+
l.split()[0] for l in Path(wiki_word_frequency).read_text().splitlines()
|
27 |
)
|
28 |
self._word_cost = {
|
29 |
str(k): math.log(float(i + 1)) for i, k in enumerate(self._word_cost)
|
|
|
150 |
|
151 |
def expand_hashtags(t, hashtag_processor):
|
152 |
"Remove # and try to split words"
|
153 |
+
return re.sub("#(\w+)", lambda m: " , " + hashtag_processor(m.group(1)), t)
|
154 |
|
155 |
|
156 |
_re_ignore_chars = """[_#\/\\%]"""
|
|
|
197 |
def __init__(self):
|
198 |
self._hashtag_processor = HashtagProcessor()
|
199 |
|
200 |
+
def __call__(self, t):
|
|
|
201 |
# fix some characters
|
202 |
t = ftfy.fix_text(t)
|
203 |
# fix html
|
204 |
t = fix_html(t)
|
205 |
+
# decode and simplify text: see unidecode library
|
206 |
+
t = unidecode(t)
|
|
|
207 |
# lower case
|
208 |
t = t.lower()
|
209 |
# replace <PERSON> (for CC12M)
|
|
|
216 |
t = remove_urls(t)
|
217 |
# remove commas in numbers
|
218 |
t = remove_comma_numbers(t)
|
219 |
+
# handle dots in numbers and quotes - Part 1
|
220 |
+
t = pre_process_dot_numbers(t)
|
221 |
+
t = pre_process_quotes(t)
|
222 |
+
# handle special characters
|
223 |
+
t = handle_special_chars(t)
|
224 |
+
# handle hashtags
|
225 |
+
t = expand_hashtags(t, self._hashtag_processor)
|
226 |
+
# ignore useless characters
|
227 |
+
t = ignore_chars(t)
|
228 |
+
# simplify quotes
|
229 |
+
t = simplify_quotes(t)
|
230 |
+
# all punctuation becomes commas
|
231 |
+
t = replace_punctuation_with_commas(t)
|
232 |
+
# handle dots in numbers and quotes - Part 2
|
233 |
+
t = post_process_dot_numbers(t)
|
234 |
+
t = post_process_quotes(t)
|
235 |
+
# handle repeating characters
|
236 |
+
t = remove_repeating_chars(t)
|
237 |
+
# merge quotes
|
238 |
+
t = merge_quotes(t)
|
239 |
+
# merge commas
|
240 |
+
t = merge_commas(t)
|
|
|
241 |
# remove multiple spaces
|
242 |
t = remove_extra_spaces(t)
|
243 |
# remove first and last comma
|
244 |
t = remove_first_last_commas(t)
|
245 |
# always start with a space
|
246 |
+
return f" {t}"
|