Spaces:
Running
Running
File size: 4,114 Bytes
176bc9a b5553ae 219d24c b5553ae 176bc9a b5553ae 219d24c b5553ae 176bc9a 219d24c 176bc9a 219d24c b5553ae 219d24c b5553ae 219d24c b5553ae 176bc9a b5553ae |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 |
from flask import Flask, request, jsonify, send_from_directory
import requests
import os
from dotenv import load_dotenv
import traceback
from PIL import Image
from pdf2image import convert_from_bytes
import base64
from io import BytesIO
load_dotenv()
app = Flask(__name__)
API_URL = "/static-proxy?url=https%3A%2F%2Fapi-inference.huggingface.co%2Fmodels%2F%26quot%3B%3C%2Fspan%3E
headers = {"Authorization": f"Bearer {os.getenv('HUGGINGFACE_API_KEY')}"}
HF_REPO_ID = os.getenv('HF_REPO_ID') # Your Hugging Face repo ID
def query(payload, model):
response = requests.post(API_URL + model, headers=headers, json=payload)
return response.json()
# Process PDFs using Hugging Face's PDF processing model
def process_pdfs():
pdf_url = f"https://huggingface.co/spaces/{HF_REPO_ID}/resolve/main/data/your_pdf_file.pdf"
try:
# Download PDF
pdf_response = requests.get(pdf_url)
pdf_response.raise_for_status() # This will raise an exception for HTTP errors
print(f"PDF downloaded successfully. Content length: {len(pdf_response.content)} bytes")
# Convert PDF to images
images = convert_from_bytes(pdf_response.content)
print(f"Converted PDF to {len(images)} images")
# Process each image
vision_model = "google/vit-base-patch16-224"
summaries = []
for i, image in enumerate(images):
# Convert image to base64
buffered = BytesIO()
image.save(buffered, format="PNG")
img_str = base64.b64encode(buffered.getvalue()).decode()
# Process image with vision model
payload = {
"inputs": [
{
"image": img_str,
"text": "Describe the content of this image in detail."
}
]
}
response = query(payload, vision_model)
print(f"Page {i+1} processing response:", json.dumps(response, indent=2))
if isinstance(response, list) and len(response) > 0 and 'generated_text' in response[0]:
summaries.append(response[0]['generated_text'])
else:
summaries.append(f"Error processing page {i+1}")
return " ".join(summaries)
except Exception as e:
print(f"Error in process_pdfs: {str(e)}")
print(traceback.format_exc())
return f"Error processing PDF: {str(e)}"
# Get the summary of PDFs
pdf_summary = process_pdfs()
print("PDF Summary:", pdf_summary)
# Get embeddings for the summary
embedding_model = "sentence-transformers/all-MiniLM-L6-v2"
if not pdf_summary.startswith("Error"):
try:
summary_embedding = query({"inputs": pdf_summary}, embedding_model)[0]
print("Successfully created summary embedding")
except Exception as e:
print(f"Error getting embedding: {str(e)}")
print(traceback.format_exc())
summary_embedding = None
else:
print("Skipping embedding due to PDF processing error")
summary_embedding = None
if summary_embedding is None:
print("WARNING: summary_embedding is None. The chatbot will not be able to provide meaningful responses.")
@app.route('/')
def home():
return send_from_directory('.', 'index.html')
@app.route('/ask', methods=['POST'])
def ask():
prompt = request.json['question']
# Get embedding for the question
query_embedding = query({"inputs": prompt}, embedding_model)[0]
# Calculate similarity
similarity = sum(a*b for a, b in zip(query_embedding, summary_embedding))
# Generate character response
character_traits = "I am a knowledgeable and friendly AI assistant based on the content of the provided PDFs."
input_text = f"Character: {character_traits}\nContext: {pdf_summary}\nUser: {prompt}\nCharacter's response:"
generator_model = "google/flan-t5-base"
response = query({"inputs": input_text}, generator_model)[0]["generated_text"]
return jsonify({'response': response})
if __name__ == '__main__':
app.run(debug=True) |