Spaces:
Sleeping
Sleeping
Modified based on JSON Schema
Browse files
app.py
CHANGED
|
@@ -19,6 +19,7 @@ from dotenv import load_dotenv
|
|
| 19 |
import base64
|
| 20 |
from bson.objectid import ObjectId
|
| 21 |
|
|
|
|
| 22 |
load_dotenv()
|
| 23 |
|
| 24 |
# Set up logging
|
|
@@ -30,28 +31,28 @@ MONGODB_URI = os.getenv("MONGODB_URI")
|
|
| 30 |
DATABASE_NAME = os.getenv("DATABASE_NAME")
|
| 31 |
COLLECTION_NAME = os.getenv("COLLECTION_NAME", "invoice_collection")
|
| 32 |
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
output_dir = 'output'
|
| 36 |
|
| 37 |
# Check if environment variables are set
|
| 38 |
if not MONGODB_URI:
|
| 39 |
raise ValueError("MONGODB_URL is not set. Please add it to Hugging Face secrets.")
|
| 40 |
|
| 41 |
# Initialize MongoDB Connection
|
| 42 |
-
|
| 43 |
-
db =
|
| 44 |
invoice_collection = db[COLLECTION_NAME]
|
|
|
|
|
|
|
|
|
|
| 45 |
|
| 46 |
@app.on_event("startup")
|
| 47 |
def startup_db():
|
| 48 |
try:
|
| 49 |
-
|
| 50 |
-
|
| 51 |
-
server_info = client.server_info() # This is a blocking call
|
| 52 |
-
print("MongoDB connection successful:", server_info)
|
| 53 |
except Exception as e:
|
| 54 |
-
|
| 55 |
|
| 56 |
# AWS S3 Configuration
|
| 57 |
API_KEY = os.getenv("API_KEY")
|
|
@@ -70,7 +71,7 @@ s3_client = boto3.client(
|
|
| 70 |
)
|
| 71 |
|
| 72 |
# Function to fetch file from S3
|
| 73 |
-
def
|
| 74 |
try:
|
| 75 |
response = s3_client.get_object(Bucket=S3_BUCKET_NAME, Key=file_key)
|
| 76 |
content_type = response['ContentType'] # Retrieve MIME type
|
|
@@ -80,146 +81,153 @@ def fetch_file_from_s3_file(file_key):
|
|
| 80 |
raise Exception(f"Failed to fetch file from S3: {str(e)}")
|
| 81 |
|
| 82 |
# Function to summarize text using OpenAI GPT
|
| 83 |
-
def
|
| 84 |
-
system_prompt = "
|
| 85 |
-
|
| 86 |
-
|
| 87 |
-
|
| 88 |
-
|
| 89 |
-
|
| 90 |
-
|
| 91 |
-
|
| 92 |
-
|
| 93 |
-
|
| 94 |
-
|
| 95 |
-
|
| 96 |
-
|
| 97 |
-
- Total Invoice Amount
|
| 98 |
-
- Type of Invoice (e.g., "Tax Invoice", "Proforma Invoice", etc.)
|
| 99 |
-
|
| 100 |
-
- **Customer Information**:
|
| 101 |
-
- Customer Name
|
| 102 |
-
- Customer Address
|
| 103 |
-
- Customer GST No.
|
| 104 |
-
|
| 105 |
-
- **Shipping and References**:
|
| 106 |
-
- MBL No./HBL No./Container No./Shipping Bill No./Shipper Invoice No./Manifest No./MAWB/HAWB/OBL No./Bill of Lading Number/REF/Ocean Bill of Lading/House Bill of Lading/BL No./Job No. → Considered as RefNo.
|
| 107 |
-
- Shipping Order
|
| 108 |
-
|
| 109 |
-
You should extract this data and structure it into the following JSON format:
|
| 110 |
-
|
| 111 |
-
{
|
| 112 |
-
"response_format": {
|
| 113 |
-
"type": "json_schema",
|
| 114 |
-
"json_schema": {
|
| 115 |
-
"name": "invoice",
|
| 116 |
-
"strict": true,
|
| 117 |
-
"schema": {
|
| 118 |
-
"type": "object",
|
| 119 |
-
"title": "Invoice Information Extractor",
|
| 120 |
-
"$schema": "http://json-schema.org/draft-07/schema#",
|
| 121 |
-
"properties": {
|
| 122 |
-
"LineItems": {
|
| 123 |
-
"type": "array",
|
| 124 |
-
"items": {
|
| 125 |
-
"type": "object",
|
| 126 |
-
"required": ["ProductCode", "Description", "Amount"],
|
| 127 |
-
"properties": {
|
| 128 |
-
"Amount": {
|
| 129 |
-
"type": "number",
|
| 130 |
-
"description": "Amount for the line item"
|
| 131 |
-
},
|
| 132 |
-
"Description": {
|
| 133 |
-
"type": "string",
|
| 134 |
-
"description": "Description of the line item"
|
| 135 |
-
},
|
| 136 |
-
"ProductCode": {
|
| 137 |
-
"type": "string",
|
| 138 |
-
"description": "Product or service code for the line item"
|
| 139 |
-
}
|
| 140 |
-
},
|
| 141 |
-
"additionalProperties": false
|
| 142 |
-
},
|
| 143 |
-
"title": "Line Items"
|
| 144 |
-
},
|
| 145 |
-
"TaxAmount": {
|
| 146 |
-
"type": "number",
|
| 147 |
-
"description": "Total tax amount for the invoice"
|
| 148 |
-
},
|
| 149 |
-
"VendorGST": {
|
| 150 |
-
"type": "string",
|
| 151 |
-
"description": "Vendor's GST number"
|
| 152 |
-
},
|
| 153 |
-
"VendorName": {
|
| 154 |
-
"type": "string",
|
| 155 |
-
"description": "Name of the vendor"
|
| 156 |
-
},
|
| 157 |
-
"InvoiceDate": {
|
| 158 |
-
"type": "string",
|
| 159 |
-
"format": "date",
|
| 160 |
-
"description": "Invoice date in dd-MMM-yyyy format"
|
| 161 |
-
},
|
| 162 |
-
"TotalAmount": {
|
| 163 |
-
"type": "number",
|
| 164 |
-
"description": "Total amount for the invoice"
|
| 165 |
-
},
|
| 166 |
-
"InvoiceNumber": {
|
| 167 |
-
"type": "string",
|
| 168 |
-
"description": "Invoice number"
|
| 169 |
-
},
|
| 170 |
-
"VendorAddress": {
|
| 171 |
-
"type": "string",
|
| 172 |
-
"description": "Vendor's address"
|
| 173 |
-
},
|
| 174 |
-
"InvoiceCurrency": {
|
| 175 |
-
"type": "string",
|
| 176 |
-
"description": "Currency used in the invoice"
|
| 177 |
-
}
|
| 178 |
-
},
|
| 179 |
-
"required": [
|
| 180 |
-
"LineItems", "TaxAmount", "VendorGST", "VendorName",
|
| 181 |
-
"InvoiceDate", "TotalAmount", "InvoiceNumber", "VendorAddress", "InvoiceCurrency"
|
| 182 |
-
],
|
| 183 |
-
"additionalProperties": false,
|
| 184 |
-
"description": "Schema for extracting specific information from invoices"
|
| 185 |
-
}
|
| 186 |
-
}
|
| 187 |
-
}
|
| 188 |
-
}
|
| 189 |
-
|
| 190 |
-
### Guidelines for Processing:
|
| 191 |
-
- Extract all relevant data from the invoice using naming conventions such as "Bill to" or "Taxpayer Name" for Vendor and Customer info.
|
| 192 |
-
- Convert the **Invoice Date** to dd-MMM-yyyy format.
|
| 193 |
-
- Ensure correct handling of amounts (e.g., **Amount**, **Tax Amount**, **Total Invoice Amount**).
|
| 194 |
-
- For line items, include descriptions, tax percentages, tax amounts, and amounts.
|
| 195 |
-
- If certain values are missing, leave them empty or set them as null.
|
| 196 |
-
- This JSON format will be used to store and manage invoices in a structured and uniform way. Please ensure only return JSON format. No extra content should not provide."""
|
| 197 |
try:
|
| 198 |
response = openai.ChatCompletion.create(
|
| 199 |
model="gpt-4o-mini",
|
| 200 |
messages=[
|
| 201 |
{"role": "system", "content": system_prompt},
|
| 202 |
-
{
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 203 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 204 |
temperature=0.5,
|
| 205 |
max_tokens=16384
|
| 206 |
)
|
|
|
|
|
|
|
| 207 |
content = response.choices[0].message.content.strip()
|
| 208 |
-
|
| 209 |
-
cleaned_content = re.sub(r'^.*```json\n', '', content) # Remove '```json\n' at the beginning
|
| 210 |
-
cleaned_content = re.sub(r'\n```$', '', cleaned_content) # Remove '\n```' at the end
|
| 211 |
|
| 212 |
-
# Step 2: Parse the cleaned content as JSON
|
| 213 |
try:
|
| 214 |
parsed_content = json.loads(cleaned_content)
|
| 215 |
return parsed_content
|
| 216 |
except json.JSONDecodeError as e:
|
| 217 |
-
|
| 218 |
-
# Optionally, print the cleaned content to debug
|
| 219 |
-
print("Cleaned content:", cleaned_content)
|
| 220 |
return None
|
|
|
|
| 221 |
except Exception as e:
|
| 222 |
-
|
|
|
|
| 223 |
|
| 224 |
# Dependency to check API Key
|
| 225 |
def verify_api_key(api_key: str = Header(...)):
|
|
@@ -239,50 +247,27 @@ def extract_text_from_file(
|
|
| 239 |
):
|
| 240 |
"""Extract text from a PDF or Image stored in S3 and process it based on document size."""
|
| 241 |
try:
|
| 242 |
-
existing_document = invoice_collection.find_one({"entityrefkey":entity_ref_key})
|
| 243 |
|
| 244 |
-
if
|
| 245 |
existing_document["_id"] = str(existing_document["_id"])
|
| 246 |
return {
|
| 247 |
"message": "Document Retrieved from MongoDB.",
|
| 248 |
"document": existing_document
|
| 249 |
}
|
| 250 |
-
# Fetch file from S3
|
| 251 |
-
file_data, content_type = fetch_file_from_s3_file(file_key)
|
| 252 |
-
|
| 253 |
-
extracted_text = []
|
| 254 |
-
base64DataResp = None
|
| 255 |
-
summary = None
|
| 256 |
-
|
| 257 |
-
if content_type.startswith("image/"): # Image file
|
| 258 |
-
image = Image.open(io.BytesIO(file_data)).convert("RGB")
|
| 259 |
-
extracted_text.append(pytesseract.image_to_string(image))
|
| 260 |
-
|
| 261 |
-
# If single image, store Base64
|
| 262 |
-
base64Data = base64.b64encode(file_data).decode('utf-8')
|
| 263 |
-
base64DataResp = f"data:image/{content_type.lower()};base64,{base64Data}"
|
| 264 |
|
| 265 |
-
|
| 266 |
-
|
| 267 |
-
num_pages = len(pdf_document)
|
| 268 |
|
| 269 |
-
|
| 270 |
-
|
| 271 |
-
extracted_text.append(page.get_text("text"))
|
| 272 |
|
| 273 |
-
|
| 274 |
-
|
| 275 |
-
|
| 276 |
-
|
| 277 |
-
|
| 278 |
-
|
| 279 |
-
|
| 280 |
-
# If 2 pages or less, generate summary
|
| 281 |
-
if num_pages <= 2:
|
| 282 |
-
full_text = " ".join(extracted_text)
|
| 283 |
-
summary = summarize_text(full_text)
|
| 284 |
-
else:
|
| 285 |
-
return {"error": f"Unsupported file type: {content_type}"}
|
| 286 |
|
| 287 |
# Store extracted data in MongoDB
|
| 288 |
document = {
|
|
@@ -290,20 +275,20 @@ def extract_text_from_file(
|
|
| 290 |
"file_type": content_type,
|
| 291 |
"document_type": document_type,
|
| 292 |
"entityrefkey": entity_ref_key,
|
| 293 |
-
"num_pages":
|
| 294 |
"base64DataResp": base64DataResp, # Only for small files
|
| 295 |
-
"extracted_text":
|
| 296 |
"summary": summary, # Only for small files
|
| 297 |
}
|
| 298 |
|
| 299 |
inserted_doc = invoice_collection.insert_one(document)
|
| 300 |
-
document_id = str(inserted_doc.inserted_id)
|
| 301 |
|
| 302 |
return {
|
| 303 |
"message": "Document successfully stored in MongoDB",
|
| 304 |
"document_id": document_id,
|
| 305 |
"file_key": file_key,
|
| 306 |
-
"num_pages":
|
| 307 |
"summary": summary if summary else "Skipped for large documents"
|
| 308 |
}
|
| 309 |
|
|
@@ -314,7 +299,6 @@ def extract_text_from_file(
|
|
| 314 |
"traceback": traceback.format_exc()
|
| 315 |
}
|
| 316 |
return {"error": error_details}
|
| 317 |
-
|
| 318 |
|
| 319 |
# Serve the output folder as static files
|
| 320 |
app.mount("/output", StaticFiles(directory="output", follow_symlink=True, html=True), name="output")
|
|
|
|
| 19 |
import base64
|
| 20 |
from bson.objectid import ObjectId
|
| 21 |
|
| 22 |
+
db_client = None
|
| 23 |
load_dotenv()
|
| 24 |
|
| 25 |
# Set up logging
|
|
|
|
| 31 |
DATABASE_NAME = os.getenv("DATABASE_NAME")
|
| 32 |
COLLECTION_NAME = os.getenv("COLLECTION_NAME", "invoice_collection")
|
| 33 |
|
| 34 |
+
# use_gpu = False
|
| 35 |
+
# output_dir = 'output'
|
|
|
|
| 36 |
|
| 37 |
# Check if environment variables are set
|
| 38 |
if not MONGODB_URI:
|
| 39 |
raise ValueError("MONGODB_URL is not set. Please add it to Hugging Face secrets.")
|
| 40 |
|
| 41 |
# Initialize MongoDB Connection
|
| 42 |
+
db_client = MongoClient(MONGODB_URI)
|
| 43 |
+
db = db_client[DATABASE_NAME]
|
| 44 |
invoice_collection = db[COLLECTION_NAME]
|
| 45 |
+
openai.api_key = OPENAI_API_KEY
|
| 46 |
+
|
| 47 |
+
app = FastAPI(docs_url='/')
|
| 48 |
|
| 49 |
@app.on_event("startup")
|
| 50 |
def startup_db():
|
| 51 |
try:
|
| 52 |
+
db_client.server_info()
|
| 53 |
+
logger.info("MongoDB connection successful")
|
|
|
|
|
|
|
| 54 |
except Exception as e:
|
| 55 |
+
logger.error(f"MongoDB connection failed: {str(e)}")
|
| 56 |
|
| 57 |
# AWS S3 Configuration
|
| 58 |
API_KEY = os.getenv("API_KEY")
|
|
|
|
| 71 |
)
|
| 72 |
|
| 73 |
# Function to fetch file from S3
|
| 74 |
+
def fetch_file_from_s3(file_key):
|
| 75 |
try:
|
| 76 |
response = s3_client.get_object(Bucket=S3_BUCKET_NAME, Key=file_key)
|
| 77 |
content_type = response['ContentType'] # Retrieve MIME type
|
|
|
|
| 81 |
raise Exception(f"Failed to fetch file from S3: {str(e)}")
|
| 82 |
|
| 83 |
# Function to summarize text using OpenAI GPT
|
| 84 |
+
def extract_invoice_data(file_data, content_type):
|
| 85 |
+
system_prompt = "You are an expert in document data extraction."
|
| 86 |
+
|
| 87 |
+
# Convert file to Base64
|
| 88 |
+
base64_encoded = base64.b64encode(file_data).decode('utf-8')
|
| 89 |
+
|
| 90 |
+
# Determine the correct MIME type for OpenAI
|
| 91 |
+
if content_type.startswith("image/"):
|
| 92 |
+
mime_type = content_type # e.g., image/png, image/jpeg
|
| 93 |
+
elif content_type == "application/pdf":
|
| 94 |
+
mime_type = "application/pdf"
|
| 95 |
+
else:
|
| 96 |
+
raise ValueError(f"Unsupported content type: {content_type}")
|
| 97 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 98 |
try:
|
| 99 |
response = openai.ChatCompletion.create(
|
| 100 |
model="gpt-4o-mini",
|
| 101 |
messages=[
|
| 102 |
{"role": "system", "content": system_prompt},
|
| 103 |
+
{
|
| 104 |
+
"role": "user",
|
| 105 |
+
"content": [
|
| 106 |
+
{
|
| 107 |
+
"type": "image_url",
|
| 108 |
+
"image_url": {
|
| 109 |
+
"url": f"data:{mime_type};base64,{base64_encoded}"
|
| 110 |
+
}
|
| 111 |
+
}
|
| 112 |
+
]
|
| 113 |
+
}
|
| 114 |
],
|
| 115 |
+
response_format={
|
| 116 |
+
"type": "json_schema",
|
| 117 |
+
"json_schema": {
|
| 118 |
+
"name": "invoice",
|
| 119 |
+
"strict": True,
|
| 120 |
+
"schema": {
|
| 121 |
+
"type": "object",
|
| 122 |
+
"title": "Invoice Information Extractor",
|
| 123 |
+
"$schema": "http://json-schema.org/draft-07/schema#",
|
| 124 |
+
"properties": {
|
| 125 |
+
"LineItems": {
|
| 126 |
+
"type": "array",
|
| 127 |
+
"items": {
|
| 128 |
+
"type": "object",
|
| 129 |
+
"required": [
|
| 130 |
+
"ProductCode",
|
| 131 |
+
"Description",
|
| 132 |
+
"Amount"
|
| 133 |
+
],
|
| 134 |
+
"properties": {
|
| 135 |
+
"ProductCode": {
|
| 136 |
+
"type": "string",
|
| 137 |
+
"title": "Product Code",
|
| 138 |
+
"description": "The code of the product"
|
| 139 |
+
},
|
| 140 |
+
"Description": {
|
| 141 |
+
"type": "string",
|
| 142 |
+
"title": "Description",
|
| 143 |
+
"description": "Description of the product"
|
| 144 |
+
},
|
| 145 |
+
"Amount": {
|
| 146 |
+
"type": "number",
|
| 147 |
+
"title": "Amount",
|
| 148 |
+
"description": "The amount of the product"
|
| 149 |
+
}
|
| 150 |
+
},
|
| 151 |
+
"additionalProperties": False
|
| 152 |
+
},
|
| 153 |
+
"title": "Line Items",
|
| 154 |
+
"description": "List of line items on the invoice"
|
| 155 |
+
},
|
| 156 |
+
"TaxAmount": {
|
| 157 |
+
"type": "number",
|
| 158 |
+
"title": "Tax Amount",
|
| 159 |
+
"description": "The tax amount on the invoice"
|
| 160 |
+
},
|
| 161 |
+
"VendorGST": {
|
| 162 |
+
"type": "string",
|
| 163 |
+
"title": "Vendor GST",
|
| 164 |
+
"description": "The GST number of the vendor"
|
| 165 |
+
},
|
| 166 |
+
"VendorName": {
|
| 167 |
+
"type": "string",
|
| 168 |
+
"title": "Vendor Name",
|
| 169 |
+
"description": "The name of the vendor"
|
| 170 |
+
},
|
| 171 |
+
"InvoiceDate": {
|
| 172 |
+
"type": "string",
|
| 173 |
+
"title": "Invoice Date",
|
| 174 |
+
"description": "The date of the invoice (format: dd-MMM-yyyy)"
|
| 175 |
+
},
|
| 176 |
+
"TotalAmount": {
|
| 177 |
+
"type": "number",
|
| 178 |
+
"title": "Total Amount",
|
| 179 |
+
"description": "The total amount on the invoice"
|
| 180 |
+
},
|
| 181 |
+
"InvoiceNumber": {
|
| 182 |
+
"type": "string",
|
| 183 |
+
"title": "Invoice Number",
|
| 184 |
+
"description": "The number of the invoice"
|
| 185 |
+
},
|
| 186 |
+
"VendorAddress": {
|
| 187 |
+
"type": "string",
|
| 188 |
+
"title": "Vendor Address",
|
| 189 |
+
"description": "The address of the vendor"
|
| 190 |
+
},
|
| 191 |
+
"InvoiceCurrency": {
|
| 192 |
+
"type": "string",
|
| 193 |
+
"title": "Invoice Currency",
|
| 194 |
+
"description": "The currency used in the invoice (e.g., USD, INR, AUD)"
|
| 195 |
+
}
|
| 196 |
+
},
|
| 197 |
+
"required": [
|
| 198 |
+
"LineItems",
|
| 199 |
+
"TaxAmount",
|
| 200 |
+
"VendorGST",
|
| 201 |
+
"VendorName",
|
| 202 |
+
"InvoiceDate",
|
| 203 |
+
"TotalAmount",
|
| 204 |
+
"InvoiceNumber",
|
| 205 |
+
"VendorAddress",
|
| 206 |
+
"InvoiceCurrency"
|
| 207 |
+
],
|
| 208 |
+
"additionalProperties": False,
|
| 209 |
+
"description": "Schema for extracting structured invoice data"
|
| 210 |
+
}
|
| 211 |
+
}
|
| 212 |
+
},
|
| 213 |
temperature=0.5,
|
| 214 |
max_tokens=16384
|
| 215 |
)
|
| 216 |
+
|
| 217 |
+
# Clean and parse JSON output
|
| 218 |
content = response.choices[0].message.content.strip()
|
| 219 |
+
cleaned_content = content.strip().strip('```json').strip('```')
|
|
|
|
|
|
|
| 220 |
|
|
|
|
| 221 |
try:
|
| 222 |
parsed_content = json.loads(cleaned_content)
|
| 223 |
return parsed_content
|
| 224 |
except json.JSONDecodeError as e:
|
| 225 |
+
logger.error(f"JSON Parse Error: {e}")
|
|
|
|
|
|
|
| 226 |
return None
|
| 227 |
+
|
| 228 |
except Exception as e:
|
| 229 |
+
logger.error(f"Error in data extraction: {e}")
|
| 230 |
+
return {"error": str(e)}
|
| 231 |
|
| 232 |
# Dependency to check API Key
|
| 233 |
def verify_api_key(api_key: str = Header(...)):
|
|
|
|
| 247 |
):
|
| 248 |
"""Extract text from a PDF or Image stored in S3 and process it based on document size."""
|
| 249 |
try:
|
| 250 |
+
existing_document = invoice_collection.find_one({"entityrefkey": entity_ref_key})
|
| 251 |
|
| 252 |
+
if existing_document:
|
| 253 |
existing_document["_id"] = str(existing_document["_id"])
|
| 254 |
return {
|
| 255 |
"message": "Document Retrieved from MongoDB.",
|
| 256 |
"document": existing_document
|
| 257 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 258 |
|
| 259 |
+
# Retrieve file from S3 and determine content type (Ensure this step is implemented)
|
| 260 |
+
content_type = get_content_type_from_s3(file_key) # Implement this function
|
|
|
|
| 261 |
|
| 262 |
+
# Extract text (Ensure Extraction function is implemented)
|
| 263 |
+
extracted_text, num_pages = extract_text_from_s3(file_key, content_type)
|
|
|
|
| 264 |
|
| 265 |
+
# Define values for small/large files
|
| 266 |
+
base64DataResp = None
|
| 267 |
+
summary = None
|
| 268 |
+
if num_pages <= 2:
|
| 269 |
+
base64DataResp = convert_to_base64(file_key) # Implement this function
|
| 270 |
+
summary = generate_summary(extracted_text) # Implement this function
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 271 |
|
| 272 |
# Store extracted data in MongoDB
|
| 273 |
document = {
|
|
|
|
| 275 |
"file_type": content_type,
|
| 276 |
"document_type": document_type,
|
| 277 |
"entityrefkey": entity_ref_key,
|
| 278 |
+
"num_pages": num_pages,
|
| 279 |
"base64DataResp": base64DataResp, # Only for small files
|
| 280 |
+
"extracted_text": extracted_text,
|
| 281 |
"summary": summary, # Only for small files
|
| 282 |
}
|
| 283 |
|
| 284 |
inserted_doc = invoice_collection.insert_one(document)
|
| 285 |
+
document_id = str(inserted_doc.inserted_id)
|
| 286 |
|
| 287 |
return {
|
| 288 |
"message": "Document successfully stored in MongoDB",
|
| 289 |
"document_id": document_id,
|
| 290 |
"file_key": file_key,
|
| 291 |
+
"num_pages": num_pages,
|
| 292 |
"summary": summary if summary else "Skipped for large documents"
|
| 293 |
}
|
| 294 |
|
|
|
|
| 299 |
"traceback": traceback.format_exc()
|
| 300 |
}
|
| 301 |
return {"error": error_details}
|
|
|
|
| 302 |
|
| 303 |
# Serve the output folder as static files
|
| 304 |
app.mount("/output", StaticFiles(directory="output", follow_symlink=True, html=True), name="output")
|