Upload app.py
Browse files
app.py
ADDED
@@ -0,0 +1,67 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import docx # Importing the required module for DOCX extraction
|
2 |
+
from datasketch import MinHash, MinHashLSH # Importing MinHash and LSH from datasketch
|
3 |
+
import gradio as gr # Importing Gradio for creating the web interface
|
4 |
+
|
5 |
+
# Function to extract text from DOCX files
|
6 |
+
def extract_text_from_docx(docx_path):
|
7 |
+
try:
|
8 |
+
doc = docx.Document(docx_path) # Open the DOCX file
|
9 |
+
text = "\n".join([para.text for para in doc.paragraphs]) # Extract the text from paragraphs
|
10 |
+
return text
|
11 |
+
except Exception as e:
|
12 |
+
print(f"Error extracting text from DOCX: {str(e)}")
|
13 |
+
return ""
|
14 |
+
|
15 |
+
# Function to calculate MinHash-based similarity between two texts
|
16 |
+
def calculate_similarity(doc1, doc2):
|
17 |
+
def text_to_shingles(text, k=5):
|
18 |
+
# Split the text into k-grams (shingles) of length k
|
19 |
+
shingles = set()
|
20 |
+
for i in range(len(text) - k + 1):
|
21 |
+
shingles.add(text[i:i + k])
|
22 |
+
return shingles
|
23 |
+
|
24 |
+
# Generate shingles for both documents
|
25 |
+
shingles1 = text_to_shingles(doc1)
|
26 |
+
shingles2 = text_to_shingles(doc2)
|
27 |
+
|
28 |
+
# Compute MinHash signatures
|
29 |
+
minhash1 = MinHash(num_perm=128)
|
30 |
+
minhash2 = MinHash(num_perm=128)
|
31 |
+
|
32 |
+
for shingle in shingles1:
|
33 |
+
minhash1.update(shingle.encode('utf8'))
|
34 |
+
|
35 |
+
for shingle in shingles2:
|
36 |
+
minhash2.update(shingle.encode('utf8'))
|
37 |
+
|
38 |
+
# Compute Jaccard similarity using MinHash
|
39 |
+
similarity_score = minhash1.jaccard(minhash2)
|
40 |
+
return similarity_score
|
41 |
+
|
42 |
+
# Function to handle the file upload and similarity calculation
|
43 |
+
def similarity(file1, file2):
|
44 |
+
if file1.name.endswith('.docx'):
|
45 |
+
text1 = extract_text_from_docx(file1.name)
|
46 |
+
else:
|
47 |
+
return "File type not supported. Please upload a DOCX file."
|
48 |
+
|
49 |
+
if file2.name.endswith('.docx'):
|
50 |
+
text2 = extract_text_from_docx(file2.name)
|
51 |
+
else:
|
52 |
+
return "File type not supported. Please upload a DOCX file."
|
53 |
+
|
54 |
+
return calculate_similarity(text1, text2)
|
55 |
+
|
56 |
+
# Create a Gradio interface
|
57 |
+
with gr.Blocks() as demo:
|
58 |
+
gr.Markdown("## Document Similarity Checker for DOCX files")
|
59 |
+
file1 = gr.File(label="Upload Document 1 (DOCX)")
|
60 |
+
file2 = gr.File(label="Upload Document 2 (DOCX)")
|
61 |
+
output = gr.Textbox(label="Similarity Score")
|
62 |
+
submit = gr.Button("Submit")
|
63 |
+
|
64 |
+
submit.click(fn=similarity, inputs=[file1, file2], outputs=output)
|
65 |
+
|
66 |
+
# Launching the Streamlit app
|
67 |
+
demo.launch()
|