simon987/hexlib
1
0
Fork 0
mirror of https://github.com/simon987/hexlib.git synced 2025-04-04 02:12:59 +00:00
Code Issues Projects Releases Wiki Activity

Add strip_dashes option in preprocess()

Browse Source
This commit is contained in:
simon987 2022-02-26 19:31:22 -05:00
parent c9fac7151a
commit 4befc3973d
3 changed files with 23 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
hexlib/text.py
View File

@ -1,3 +1,4 @@
import re
from functools import partial
from itertools import chain, repeat
from multiprocessing.pool import Pool
@ -53,6 +54,8 @@ SINGLE_QUOTE_TRANS = str.maketrans("".join(SINGLE_QUOTES), "".join(repeat("'", l
DASHES = ("–", "⸺", "–", "—")
DASHES_TRANS = str.maketrans("".join(DASHES), "".join(repeat("-", len(DASHES))))
DASHES_RE = re.compile(r"-+")
SPECIAL_PUNCTUATION = ";:\"/()|*=>"
SPECIAL_PUNCTUATION_TRANS = str.maketrans(SPECIAL_PUNCTUATION, " " * len(SPECIAL_PUNCTUATION))
@ -62,6 +65,7 @@ PUNCTUATION_TRANS = str.maketrans(PUNCTUATION, " " * len(PUNCTUATION))
def preprocess(text, lowercase=False, clean_html=False, remove_punctuation=False, remove_special_punctuation=False,
remove_stopwords_en=False, lemmatize=False, fix_single_quotes=False, strip_quotes=False,
strip_dashes=False,
remove_urls=False, bigrams: set = None, trigrams: set = None, remove_numbers=False):
if lowercase:
text = text.lower()
@ -71,6 +75,9 @@ def preprocess(text, lowercase=False, clean_html=False, remove_punctuation=False
text = text.translate(DASHES_TRANS)
if strip_dashes:
text = DASHES_RE.sub("-", text)
if remove_urls:
text = LINK_RE.sub(" ", text)
@ -96,6 +103,9 @@ def preprocess(text, lowercase=False, clean_html=False, remove_punctuation=False
if strip_quotes:
words = map(lambda w: w.strip("\"'“”"), words)
if strip_dashes:
words = map(lambda w: w.strip("-"), words)
if bigrams:
words = _transform_bigram(nltk.bigrams(chain(words, ("*",))), bigrams)

2
setup.py
View File

@ -2,7 +2,7 @@ from setuptools import setup
setup(
name="hexlib",
version="1.76",
version="1.77",
description="Misc utility methods",
author="simon987",
author_email="me@simon987.net",

14
test/test_text.py
View File

@ -152,7 +152,7 @@ class TestText(TestCase):
remove_stopwords_en=True,
remove_urls=True
)
expected = "hello world"
expected = "hello world |"
self.assertEqual(" ".join(cleaned), expected)
@ -170,7 +170,7 @@ class TestText(TestCase):
remove_urls=False
)
expected = "217709510 is there a servant that is against civilization and humanity literally instant summon"
expected = ">>217709510 is there a servant that is against civilization and humanity literally instant summon"
self.assertEqual(" ".join(cleaned), expected)
def test_html_entity(self):
@ -257,3 +257,13 @@ class TestText(TestCase):
expected = "hi test hello"
self.assertEqual(" ".join(cleaned), expected)
def test_strip_dashes(self):
text = "yes -But something-something -- hello aa--bb"
cleaned = preprocess(
text,
strip_dashes=True
)
expected = "yes But something-something hello aa-bb"
self.assertEqual(" ".join(cleaned), expected)