Commit
·
be37136
1
Parent(s):
9282835
Switched to pyspellchecker for simpler spell checking
Browse files- HTR/spell_and_gramer_check.py +10 -34
- main.py +1 -4
- requirements.txt +1 -1
HTR/spell_and_gramer_check.py
CHANGED
@@ -1,46 +1,22 @@
|
|
1 |
-
import os
|
2 |
-
import language_tool_python
|
3 |
-
from pathlib import Path
|
4 |
from spellchecker import SpellChecker
|
5 |
-
|
6 |
-
# Use the cache directory from environment variable
|
7 |
-
cache_dir = os.getenv('LANGUAGE_TOOL_CACHE_DIR', '/tmp/language_tool')
|
8 |
-
os.makedirs(cache_dir, exist_ok=True)
|
9 |
|
10 |
-
|
11 |
-
tool = language_tool_python.LanguageTool('en-US', cache_dir=cache_dir)
|
12 |
|
13 |
-
def
|
14 |
-
|
15 |
-
my_matches = tool.check(answer)
|
16 |
-
corrected_text = tool.correct(answer)
|
17 |
-
return corrected_text
|
18 |
-
|
19 |
-
def correct_spelling(text):
|
20 |
-
spell = SpellChecker()
|
21 |
words = text.split()
|
22 |
-
|
23 |
# Find misspelled words
|
24 |
misspelled = spell.unknown(words)
|
25 |
-
|
26 |
# Correct misspelled words
|
27 |
-
|
28 |
for word in words:
|
29 |
if word in misspelled:
|
30 |
-
|
31 |
-
# If no correction found or correction is None, keep the original word
|
32 |
-
if correction is None:
|
33 |
-
corrected_text.append(word)
|
34 |
-
else:
|
35 |
-
corrected_text.append(correction)
|
36 |
else:
|
37 |
-
|
38 |
-
|
39 |
-
return " ".join(map(str, corrected_text))
|
40 |
-
|
41 |
-
def spell_grammer(text):
|
42 |
-
spell_check_text = correct_spelling(text)
|
43 |
-
|
44 |
-
corrected_text = check_grammar(spell_check_text)
|
45 |
|
|
|
|
|
46 |
return corrected_text
|
|
|
|
|
|
|
|
|
1 |
from spellchecker import SpellChecker
|
|
|
|
|
|
|
|
|
2 |
|
3 |
+
spell = SpellChecker()
|
|
|
4 |
|
5 |
+
def spell_grammer(text):
|
6 |
+
# Split text into words
|
|
|
|
|
|
|
|
|
|
|
|
|
7 |
words = text.split()
|
8 |
+
|
9 |
# Find misspelled words
|
10 |
misspelled = spell.unknown(words)
|
11 |
+
|
12 |
# Correct misspelled words
|
13 |
+
corrected_words = []
|
14 |
for word in words:
|
15 |
if word in misspelled:
|
16 |
+
corrected_words.append(spell.correction(word))
|
|
|
|
|
|
|
|
|
|
|
17 |
else:
|
18 |
+
corrected_words.append(word)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
19 |
|
20 |
+
# Join words back into text
|
21 |
+
corrected_text = ' '.join(corrected_words)
|
22 |
return corrected_text
|
main.py
CHANGED
@@ -1,12 +1,9 @@
|
|
1 |
import os
|
2 |
import tempfile
|
3 |
|
4 |
-
#
|
5 |
cache_dir = tempfile.mkdtemp()
|
6 |
-
os.environ['TRANSFORMERS_CACHE'] = cache_dir
|
7 |
os.environ['HF_HOME'] = cache_dir
|
8 |
-
# Set language tool cache directory
|
9 |
-
os.environ['LANGUAGE_TOOL_CACHE_DIR'] = cache_dir
|
10 |
|
11 |
from flask import Flask, request, jsonify, render_template
|
12 |
import json
|
|
|
1 |
import os
|
2 |
import tempfile
|
3 |
|
4 |
+
# Set up cache directory for transformers
|
5 |
cache_dir = tempfile.mkdtemp()
|
|
|
6 |
os.environ['HF_HOME'] = cache_dir
|
|
|
|
|
7 |
|
8 |
from flask import Flask, request, jsonify, render_template
|
9 |
import json
|
requirements.txt
CHANGED
@@ -9,6 +9,6 @@ chromadb==0.4.6
|
|
9 |
PyMuPDF==1.24.12
|
10 |
nltk==3.8.1
|
11 |
gensim==4.3.2
|
12 |
-
language-tool-python==2.8
|
13 |
pyspellchecker==0.8.1
|
14 |
werkzeug==3.0.0
|
|
|
9 |
PyMuPDF==1.24.12
|
10 |
nltk==3.8.1
|
11 |
gensim==4.3.2
|
12 |
+
# language-tool-python==2.8
|
13 |
pyspellchecker==0.8.1
|
14 |
werkzeug==3.0.0
|