diff --git a/src/grammar_checker_google_jeremy.py b/src/grammar_checker_google_jeremy.py
index 153e23a5e38e7e67d263b0c2d3e234cbfbab44d9..530a3c44f08c181873e6b978379bbf5b9a91c15a 100644
--- a/src/grammar_checker_google_jeremy.py
+++ b/src/grammar_checker_google_jeremy.py
@@ -6,6 +6,7 @@ import requests
from pyinflect import getAllInflections, getInflection
from nltk.stem import WordNetLemmatizer
from nltk.corpus import wordnet as wn
+import numpy as np
from time import sleep
@@ -142,9 +143,10 @@ class GrammarCheckerGoogle:
probs = [self.float_min]
chained_probs = math.prod(probs)
diff_list = [(j-i) for i, j in zip(probs[:-1], probs[1:])]
+ logged_chained_prob = -np.log10((chained_probs)**(1/len(n_grams))) if x != 0 else (chained_probs)**(1/len(n_grams))
- if (chained_probs)**(1/len(n_grams)) <= self.threshold:
+ if logged_chained_prob <= self.threshold:
return probs.index(min(probs))
else:
return None
\ No newline at end of file
diff --git a/src/main.py b/src/main.py
index 59e4e6d0be55a6664a63c3dcc1af777dfd645079..5dcc6ca4f25256f804752147ff8b3a782b1d4b85 100644
--- a/src/main.py
+++ b/src/main.py
@@ -1,7 +1,9 @@
+from re import L
import wx
import nltk
from wx.richtext import RichTextCtrl
from grammar_checker import GrammarChecker
+from grammar_checker_google_jeremy import *
class StatisticalGrammarChecker(wx.App):
@@ -51,7 +53,9 @@ class Panel(wx.Panel):
self.create_ui()
# grammar checker
- self.grammar_checker = GrammarChecker(n=3)
+ # self.grammar_checker = GrammarChecker(n=3)
+ self.grammar_checker2 = GrammarCheckerGoogle(2,1e-200, 1e-6)
+ self.grammar_checker3 = GrammarCheckerGoogle(3,1e-200, 1e-6)
# FRONTEND
def create_ui(self):
@@ -76,18 +80,75 @@ class Panel(wx.Panel):
# THREADS
def grammar_check_thread(self, event):
- output_text = ""
+ # output_text = ""
# get input
input_text = self.text_input.GetValue()
# for each sentence
- sentences = nltk.sent_tokenize(input_text)
- for i, sentence in enumerate(sentences):
- # grammar check
- output_text += self.grammar_checker.check(sentence)
- if i != len(sentences):
- output_text += " "
+
+ # sentences = nltk.sent_tokenize(input_text)
+ # for i, sentence in enumerate(sentences):
+ # # grammar check
+ # output_text += self.grammar_checker.check(sentence)
+ # if i != len(sentences):
+ # output_text += " "
+
+ error_grammar_2 = self.grammar_checker2.check(input_text)
+ error_grammar_3 = self.grammar_checker3.check(input_text)
+ if error_grammar_2 is not None:
+ error_grammar_2_list = list(list(nltk.ngrams(self.grammar_checker2.tokenizer.tokenize(input_text),self.grammar_checker2.n))[error_grammar_2])
+ elif error_grammar_2 is None:
+ error_grammar_2_list = None
+ if error_grammar_3 is not None:
+ error_grammar_3_list = list(list(nltk.ngrams(self.grammar_checker3.tokenizer.tokenize(input_text),self.grammar_checker3.n))[error_grammar_3])
+ elif error_grammar_3 is None:
+ error_grammar_3_list = None
+ if (error_grammar_3_list is not None and error_grammar_2_list is not None) or (error_grammar_3_list is not None and error_grammar_2_list is None):
+ list_of_error = error_grammar_3_list
+ elif error_grammar_3_list is None and error_grammar_2_list is not None:
+ list_of_error = error_grammar_2_list
+ elif error_grammar_3_list is None and error_grammar_2_list is None:
+ list_of_error = None
+ output_text = "Nothing to declare"
+
+ if list_of_error is not None:
+ if len(list_of_error) == 2:
+ list_error_pos_1 = [list_of_error[0]] if suggest_inflection(list_of_error[0].lower()) is None else suggest_inflection(list_of_error[0].lower())
+ list_error_pos_2 = [list_of_error[1]] if suggest_inflection(list_of_error[1].lower()) is None else suggest_inflection(list_of_error[1].lower())
+ list_of_suggestions =[]
+ for i in list_error_pos_1:
+ for j in list_error_pos_2:
+ list_of_suggestions.append((i,j))
+ list_of_prob = []
+ for i in list_of_suggestions:
+ prob = self.grammar_checker2.get_google_ngram_prob(i) if self.grammar_checker2.get_google_ngram_prob(i) is not None else 0.0
+ list_of_prob.append(prob)
+ output_bigram = list(list_of_suggestions.index(max(list_of_prob)))
+ output_text = " ".join(output_bigram)
+
+ elif len(list_of_error) == 3:
+ list_error_pos_1 = [list_of_error[0]] if suggest_inflection(list_of_error[0].lower()) is None else suggest_inflection(list_of_error[0].lower())
+ list_error_pos_2 = [list_of_error[1]] if suggest_inflection(list_of_error[1].lower()) is None else suggest_inflection(list_of_error[1].lower())
+ list_error_pos_3 = [list_of_error[2]] if suggest_inflection(list_of_error[2].lower()) is None else suggest_inflection(list_of_error[2].lower())
+ list_of_suggestions =[]
+ for i in list_error_pos_1:
+ for j in list_error_pos_2:
+ for x in list_error_pos_3:
+ list_of_suggestions.append((i,j,x))
+ list_of_prob = []
+ for i in list_of_suggestions:
+ prob = self.grammar_checker3.get_google_ngram_prob(i) if self.grammar_checker3.get_google_ngram_prob(i) is not None else 0.0
+ list_of_prob.append(prob)
+ output_trigram = list(list_of_suggestions.index(max(list_of_prob)))
+ output_text = " ".join(output_trigram)
+ else:
+ output_text = "Something's wrong :("
+ elif list_of_error is None:
+ output_text = "Nothing to declare"
+
+
+
# set output text
- self.text_output.SetValue(output_text)
+ self.text_output.SetValue(str(output_text))
# highlight errors
#for error_location in text_error_locations:
# self.text_output.SetStyle(error_location[0], error_location[1], wx.TextAttr(colText=wx.WHITE, colBack=wx.RED))