dipenvekaria commited on
Commit
c9f26ef
·
verified ·
1 Parent(s): 5d27271

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +7 -8
app.py CHANGED
@@ -5,7 +5,7 @@ from datetime import datetime
5
  from typing import List
6
  import gradio as gr
7
  from dotenv import load_dotenv
8
- #from docling.document import DoclingDocument # Import Docling
9
  from langchain_openai import OpenAIEmbeddings, ChatOpenAI
10
  from langchain.prompts import PromptTemplate
11
  from qdrant_client import QdrantClient
@@ -13,8 +13,6 @@ from qdrant_client.http import models
13
  from sklearn.metrics.pairwise import cosine_similarity
14
  import numpy as np
15
  import re
16
- import docling
17
- docling.load_model("all")
18
 
19
  load_dotenv()
20
 
@@ -38,11 +36,12 @@ class PdfProcessor:
38
  )
39
 
40
  def _extract_text(self, file_path):
41
- """Extract text from a PDF using Docling."""
42
- doc = DoclingDocument.from_file(file_path)
43
- # Assuming Docling provides a text attribute or method; adjust if needed
44
- text = doc.text if hasattr(doc, 'text') else " ".join([str(item) for item in doc.content])
45
- return text if text else ""
 
46
 
47
  def _chunk_text(self, text, max_chunk_size=500, min_chunk_size=100, similarity_threshold=0.5):
48
  """Performs semantic chunking using embeddings and cosine similarity."""
 
5
  from typing import List
6
  import gradio as gr
7
  from dotenv import load_dotenv
8
+ import fitz # Import PyMuPDF as fitz
9
  from langchain_openai import OpenAIEmbeddings, ChatOpenAI
10
  from langchain.prompts import PromptTemplate
11
  from qdrant_client import QdrantClient
 
13
  from sklearn.metrics.pairwise import cosine_similarity
14
  import numpy as np
15
  import re
 
 
16
 
17
  load_dotenv()
18
 
 
36
  )
37
 
38
  def _extract_text(self, file_path):
39
+ """Extract text from a PDF using PyMuPDF."""
40
+ text = ""
41
+ with fitz.open(file_path) as pdf:
42
+ for page in pdf:
43
+ text += page.get_text("text") or "" # Extract text from each page
44
+ return text
45
 
46
  def _chunk_text(self, text, max_chunk_size=500, min_chunk_size=100, similarity_threshold=0.5):
47
  """Performs semantic chunking using embeddings and cosine similarity."""