Spaces:
Sleeping
Sleeping
Update agents/document_agent.py
Browse files- agents/document_agent.py +2 -1
agents/document_agent.py
CHANGED
|
@@ -346,6 +346,7 @@ class DocumentAgent(BaseAgent):
|
|
| 346 |
|
| 347 |
async def _extract_invoice_with_donut(self, file_name: str):
|
| 348 |
images = self._pdf_to_images(file_name)
|
|
|
|
| 349 |
|
| 350 |
# Assuming first page invoice
|
| 351 |
image = images[0]
|
|
@@ -373,7 +374,7 @@ class DocumentAgent(BaseAgent):
|
|
| 373 |
sequence = self.processor.batch_decode(outputs.sequences)[0]
|
| 374 |
sequence = sequence.replace(self.processor.tokenizer.eos_token, "").strip()
|
| 375 |
sequence = sequence.replace(task_prompt, "").strip()
|
| 376 |
-
|
| 377 |
return safe_json_parse(sequence)
|
| 378 |
|
| 379 |
async def _enhance_invoice_data(self, invoice_data: InvoiceData, raw_text: str) -> InvoiceData:
|
|
|
|
| 346 |
|
| 347 |
async def _extract_invoice_with_donut(self, file_name: str):
|
| 348 |
images = self._pdf_to_images(file_name)
|
| 349 |
+
print('images from donut', images)
|
| 350 |
|
| 351 |
# Assuming first page invoice
|
| 352 |
image = images[0]
|
|
|
|
| 374 |
sequence = self.processor.batch_decode(outputs.sequences)[0]
|
| 375 |
sequence = sequence.replace(self.processor.tokenizer.eos_token, "").strip()
|
| 376 |
sequence = sequence.replace(task_prompt, "").strip()
|
| 377 |
+
print('seq from donut:', sequence)
|
| 378 |
return safe_json_parse(sequence)
|
| 379 |
|
| 380 |
async def _enhance_invoice_data(self, invoice_data: InvoiceData, raw_text: str) -> InvoiceData:
|