def doc_classify(self):
tmp = self.table_widget.control1_wid.combo_box.currentText()
self.statusBar().showMessage('Classifying files')
if tmp == "docx":
SimilarityRatio(self.files_path.docx,"docx")
elif tmp == "pptx":
SimilarityRatio(self.files_path.pptx,"pptx")
elif tmp == "xlsx":
SimilarityRatio(self.files_path.xlsx,"xlsx")
def doc_classify2(self):
tmp = self.table_widget.control1_wid.combo_box.currentText()
self.statusBar().showMessage('Classifying new file')
f1 = self.open_file(type=tmp)
if not f1:
self.statusBar().showMessage('Action Cancelled')
else:
if tmp == "docx":
SimilarityRatio([f1],"docx", method="inference")
elif tmp == "pptx":
SimilarityRatio([f1],"pptx", method="inference")
elif tmp == "xlsx":
SimilarityRatio([f1],"xlsx", method="inference")
def doc_similarity(self):
tmp = self.table_widget.control1_wid.combo_box.currentText()
self.statusBar().showMessage('Calculating Similarity of 2 files')
f1 = self.open_file(type=tmp)
f2 = self.open_file(type=tmp)
if not f1 or not f2:
self.statusBar().showMessage('Action Cancelled')
else:
SimilarityRatio([f1,f2],tmp,method="fuzzywuzzy")
elif op == "docx_compare":
startTime = time.time()
CompareFiles(files_path.docx, "docx")
elif op == "xlsx_compare":
startTime = time.time()
CompareFiles(files_path.xlsx, "xlsx")
elif op == "pptx_compare":
startTime = time.time()
CompareFiles(files_path.pptx, "pptx")
elif op == "docx_similarity":
startTime = time.time()
SimilarityRatio(files_path.docx, "docx", method="None")
elif op == "xlsx_similarity":
startTime = time.time()
SimilarityRatio(files_path.xlsx, "xlsx", method="fuzzywuzzy")
elif op == "pptx_similarity":
startTime = time.time()
SimilarityRatio(files_path.pptx, "pptx", method="fuzzywuzzy")
elif op == "txt_class_txt":
startTime = time.time()
for term in search_terms:
c(files_path.docx, search=term)
c(files_path.pptx, search=term)
c(files_path.xlsx, search=term)