Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,41 +1,18 @@
|
|
| 1 |
-
|
| 2 |
-
"""API prog
|
| 3 |
-
|
| 4 |
-
Automatically generated by Colab.
|
| 5 |
-
|
| 6 |
-
Original file is located at
|
| 7 |
-
https://colab.research.google.com/drive/1dwnXx2hUYegX54dlWrXEF8uF_Jj7QxX5
|
| 8 |
-
"""
|
| 9 |
-
|
| 10 |
-
from fastapi import FastAPI
|
| 11 |
from pydantic import BaseModel
|
| 12 |
from groq import Groq
|
|
|
|
| 13 |
|
| 14 |
app = FastAPI()
|
| 15 |
|
| 16 |
-
|
| 17 |
-
text: str
|
| 18 |
-
|
| 19 |
-
client = Groq(api_key="YOUR_KEY")
|
| 20 |
-
|
| 21 |
-
@app.post("/analyze")
|
| 22 |
-
def analyze_document(data: DocInput):
|
| 23 |
-
return {"result": "ok"}
|
| 24 |
-
from groq import Groq
|
| 25 |
-
from pypdf import PdfReader
|
| 26 |
-
import gradio as gr
|
| 27 |
-
|
| 28 |
-
# ---------------------------------------
|
| 29 |
-
# 1. Initialize Groq Client
|
| 30 |
-
# ---------------------------------------
|
| 31 |
client = Groq(api_key="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")
|
| 32 |
|
| 33 |
-
|
| 34 |
-
#
|
| 35 |
-
#
|
| 36 |
-
# ---------------------------------------
|
| 37 |
def extract_pdf_text(pdf_file):
|
| 38 |
-
reader = PdfReader(pdf_file
|
| 39 |
text = ""
|
| 40 |
for page in reader.pages:
|
| 41 |
extracted = page.extract_text()
|
|
@@ -43,16 +20,13 @@ def extract_pdf_text(pdf_file):
|
|
| 43 |
text += extracted + "\n"
|
| 44 |
return text
|
| 45 |
|
| 46 |
-
|
| 47 |
-
#
|
| 48 |
-
#
|
| 49 |
-
|
| 50 |
-
def analyze_document(text):
|
| 51 |
prompt = f"""
|
| 52 |
You are an expert document analysis AI.
|
| 53 |
-
|
| 54 |
Given the following document text, do 3 things:
|
| 55 |
-
|
| 56 |
1. Identify the document type (invoice, receipt, contract, report, certificate, etc.)
|
| 57 |
2. Extract key fields in JSON format.
|
| 58 |
3. Provide a short summary.
|
|
@@ -61,7 +35,6 @@ Document text:
|
|
| 61 |
{text}
|
| 62 |
|
| 63 |
Return your answer in this JSON structure:
|
| 64 |
-
|
| 65 |
{{
|
| 66 |
"document_type": "",
|
| 67 |
"fields": {{}},
|
|
@@ -75,4 +48,18 @@ Return your answer in this JSON structure:
|
|
| 75 |
temperature=0
|
| 76 |
)
|
| 77 |
|
| 78 |
-
return response.choices[0].message.content
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from fastapi import FastAPI, UploadFile, File
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2 |
from pydantic import BaseModel
|
| 3 |
from groq import Groq
|
| 4 |
+
from pypdf import PdfReader
|
| 5 |
|
| 6 |
app = FastAPI()
|
| 7 |
|
| 8 |
+
# Initialize Groq client
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
client = Groq(api_key="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")
|
| 10 |
|
| 11 |
+
# -----------------------------
|
| 12 |
+
# Extract text from PDF
|
| 13 |
+
# -----------------------------
|
|
|
|
| 14 |
def extract_pdf_text(pdf_file):
|
| 15 |
+
reader = PdfReader(pdf_file)
|
| 16 |
text = ""
|
| 17 |
for page in reader.pages:
|
| 18 |
extracted = page.extract_text()
|
|
|
|
| 20 |
text += extracted + "\n"
|
| 21 |
return text
|
| 22 |
|
| 23 |
+
# -----------------------------
|
| 24 |
+
# Analyze text with Groq
|
| 25 |
+
# -----------------------------
|
| 26 |
+
def analyze_text_with_groq(text):
|
|
|
|
| 27 |
prompt = f"""
|
| 28 |
You are an expert document analysis AI.
|
|
|
|
| 29 |
Given the following document text, do 3 things:
|
|
|
|
| 30 |
1. Identify the document type (invoice, receipt, contract, report, certificate, etc.)
|
| 31 |
2. Extract key fields in JSON format.
|
| 32 |
3. Provide a short summary.
|
|
|
|
| 35 |
{text}
|
| 36 |
|
| 37 |
Return your answer in this JSON structure:
|
|
|
|
| 38 |
{{
|
| 39 |
"document_type": "",
|
| 40 |
"fields": {{}},
|
|
|
|
| 48 |
temperature=0
|
| 49 |
)
|
| 50 |
|
| 51 |
+
return response.choices[0].message.content
|
| 52 |
+
|
| 53 |
+
# -----------------------------
|
| 54 |
+
# FastAPI Endpoint
|
| 55 |
+
# -----------------------------
|
| 56 |
+
@app.post("/analyze")
|
| 57 |
+
async def analyze_document(file: UploadFile = File(...)):
|
| 58 |
+
# Extract text from uploaded PDF
|
| 59 |
+
pdf_bytes = await file.read()
|
| 60 |
+
text = extract_pdf_text(pdf_bytes)
|
| 61 |
+
|
| 62 |
+
# Analyze with Groq
|
| 63 |
+
result = analyze_text_with_groq(text)
|
| 64 |
+
|
| 65 |
+
return {"result": result}
|