Spaces:
Running
Running
File size: 8,293 Bytes
79c8ea7 0540355 0c9548a 0540355 b1622cb d74850e 0540355 239c804 0540355 8e24199 0540355 1be9899 0540355 4c11732 753db53 0540355 2be14bd 0540355 4c11732 0540355 2be14bd 0540355 753db53 0540355 753db53 0540355 d74850e 0540355 d74850e 0540355 d74850e 0540355 0b363e7 4c11732 0540355 0b363e7 4c11732 0b363e7 4c11732 0b363e7 4c11732 2be14bd d2931fe 0540355 2be14bd d2931fe 0540355 7e5ddc3 753db53 2852c90 2be14bd 0540355 01cb6f1 0540355 d74850e 79c8ea7 d74850e 0540355 d74850e |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 |
"""import gradio as gr
import uvicorn
import numpy as np
import fitz # PyMuPDF
import tika
import torch
from fastapi import FastAPI
from transformers import pipeline
from PIL import Image
from io import BytesIO
from starlette.responses import RedirectResponse
from tika import parser
from openpyxl import load_workbook
# Initialize Tika for DOCX & PPTX parsing
tika.initVM()
# Initialize FastAPI
app = FastAPI()
# Load models
device = "cuda" if torch.cuda.is_available() else "cpu"
qa_pipeline = pipeline("text-generation", model="TinyLlama/TinyLlama-1.1B-Chat-v1.0", device=device)
image_captioning_pipeline = pipeline("image-to-text", model="Salesforce/blip-image-captioning-base")
ALLOWED_EXTENSIONS = {"pdf", "docx", "pptx", "xlsx"}
# ✅ Function to Validate File Type
def validate_file_type(file):
if isinstance(file, str): # Text-based input (NamedString)
return None
if hasattr(file, "name"):
ext = file.name.split(".")[-1].lower()
if ext not in ALLOWED_EXTENSIONS:
return f"❌ Unsupported file format: {ext}"
return None
return "❌ Invalid file format!"
# ✅ Extract Text from PDF
def extract_text_from_pdf(pdf_bytes):
doc = fitz.open(stream=pdf_bytes, filetype="pdf")
return "\n".join([page.get_text() for page in doc])
# ✅ Extract Text from DOCX & PPTX using Tika
def extract_text_with_tika(file_bytes):
return parser.from_buffer(file_bytes)["content"]
# ✅ Extract Text from Excel
def extract_text_from_excel(file_bytes):
wb = load_workbook(BytesIO(file_bytes), data_only=True)
text = []
for sheet in wb.worksheets:
for row in sheet.iter_rows(values_only=True):
text.append(" ".join(str(cell) for cell in row if cell))
return "\n".join(text)
# ✅ Truncate Long Text for Model
def truncate_text(text, max_length=2048):
return text[:max_length] if len(text) > max_length else text
# ✅ Answer Questions from Image or Document
def answer_question(file, question: str):
# Image Processing (Gradio sends images as NumPy arrays)
if isinstance(file, np.ndarray):
image = Image.fromarray(file)
caption = image_captioning_pipeline(image)[0]['generated_text']
response = qa_pipeline(f"Question: {question}\nContext: {caption}")
return response[0]["generated_text"]
# Validate File
validation_error = validate_file_type(file)
if validation_error:
return validation_error
file_ext = file.name.split(".")[-1].lower() if hasattr(file, "name") else None
file_bytes = file.read() if hasattr(file, "read") else None
if not file_bytes:
return "❌ Could not read file content!"
# Extract Text from Supported Documents
if file_ext == "pdf":
text = extract_text_from_pdf(file_bytes)
elif file_ext in ["docx", "pptx"]:
text = extract_text_with_tika(file_bytes)
elif file_ext == "xlsx":
text = extract_text_from_excel(file_bytes)
else:
return "❌ Unsupported file format!"
if not text:
return "⚠️ No text extracted from the document."
truncated_text = truncate_text(text)
response = qa_pipeline(f"Question: {question}\nContext: {truncated_text}")
return response[0]["generated_text"]
# ✅ Gradio Interface (Unified for Images & Documents)
with gr.Blocks() as demo:
gr.Markdown("## 📄 AI-Powered Document & Image QA")
with gr.Row():
file_input = gr.File(label="Upload Document / Image")
question_input = gr.Textbox(label="Ask a Question", placeholder="What is this document about?")
answer_output = gr.Textbox(label="Answer")
submit_btn = gr.Button("Get Answer")
submit_btn.click(answer_question, inputs=[file_input, question_input], outputs=answer_output)
# ✅ Mount Gradio with FastAPI
app = gr.mount_gradio_app(app, demo, path="/")
@app.get("/")
def home():
return RedirectResponse(url="/")
# ✅ Run FastAPI + Gradio
if __name__ == "__main__":
uvicorn.run(app, host="0.0.0.0", port=7860)
""" import gradio as gr
import uvicorn
import numpy as np
import fitz # PyMuPDF
import tika
import torch
from fastapi import FastAPI
from transformers import pipeline
from PIL import Image
from io import BytesIO
from starlette.responses import RedirectResponse
from tika import parser
from openpyxl import load_workbook
# Initialize Tika for DOCX & PPTX parsing
tika.initVM()
# Initialize FastAPI
app = FastAPI()
# Load models
device = "cuda" if torch.cuda.is_available() else "cpu"
qa_pipeline = pipeline("text-generation", model="TinyLlama/TinyLlama-1.1B-Chat-v1.0", device=device)
image_captioning_pipeline = pipeline("image-to-text", model="Salesforce/blip-image-captioning-base")
ALLOWED_EXTENSIONS = {"pdf", "docx", "pptx", "xlsx"}
# ✅ Function to Validate File Type
def validate_file_type(file):
if file is None:
return "❌ No file uploaded!"
if isinstance(file, str): # Text-based input (NamedString)
return None
if hasattr(file, "name"):
ext = file.name.split(".")[-1].lower()
if ext not in ALLOWED_EXTENSIONS:
return f"❌ Unsupported file format: {ext}"
return None
return "❌ Invalid file format!"
# ✅ Extract Text from PDF
def extract_text_from_pdf(file):
try:
doc = fitz.open(stream=file, filetype="pdf")
return "\n".join([page.get_text() for page in doc])
except Exception:
return None
# ✅ Extract Text from DOCX & PPTX using Tika
def extract_text_with_tika(file):
try:
return parser.from_buffer(file)["content"]
except Exception:
return None
# ✅ Extract Text from Excel
def extract_text_from_excel(file):
try:
wb = load_workbook(BytesIO(file), data_only=True)
text = []
for sheet in wb.worksheets:
for row in sheet.iter_rows(values_only=True):
text.append(" ".join(str(cell) for cell in row if cell))
return "\n".join(text)
except Exception:
return None
# ✅ Truncate Long Text for Model
def truncate_text(text, max_length=2048):
return text[:max_length] if len(text) > max_length else text
# ✅ Answer Questions from Image or Document
def answer_question(file, question: str):
# Image Processing (Gradio sends images as NumPy arrays)
if isinstance(file, np.ndarray):
image = Image.fromarray(file)
caption = image_captioning_pipeline(image)[0]['generated_text']
response = qa_pipeline(f"Question: {question}\nContext: {caption}")
return response[0]["generated_text"]
# Validate File
validation_error = validate_file_type(file)
if validation_error:
return validation_error
# ✅ Read File Bytes Properly
file_ext = file.name.split(".")[-1].lower() if hasattr(file, "name") else None
file_bytes = file.read() if hasattr(file, "read") else None
if not file_bytes:
return "❌ Could not read file content!"
# Extract Text from Supported Documents
text = None
if file_ext == "pdf":
text = extract_text_from_pdf(file_bytes)
elif file_ext in ["docx", "pptx"]:
text = extract_text_with_tika(file_bytes)
elif file_ext == "xlsx":
text = extract_text_from_excel(file_bytes)
if not text:
return "⚠️ No text extracted from the document."
truncated_text = truncate_text(text)
response = qa_pipeline(f"Question: {question}\nContext: {truncated_text}")
return response[0]["generated_text"]
# ✅ Gradio Interface (Unified for Images & Documents)
with gr.Blocks() as demo:
gr.Markdown("## 📄 AI-Powered Document & Image QA")
with gr.Row():
file_input = gr.File(label="Upload Document / Image")
question_input = gr.Textbox(label="Ask a Question", placeholder="What is this document about?")
answer_output = gr.Textbox(label="Answer")
submit_btn = gr.Button("Get Answer")
submit_btn.click(answer_question, inputs=[file_input, question_input], outputs=answer_output)
# ✅ Mount Gradio with FastAPI
app = gr.mount_gradio_app(app, demo, path="/")
@app.get("/")
def home():
return RedirectResponse(url="/")
# ✅ Run FastAPI + Gradio
if __name__ == "__main__":
uvicorn.run(app, host="0.0.0.0", port=7860)
|