Merge in RR/cv-analysis from pdf2image to master
Squashed commit of the following:
commit 1353f54d2dceb0a79b1f81bfa2c035f5a454275a
Author: Julius Unverfehrt <julius.unverfehrt@iqser.com>
Date: Wed Aug 10 09:07:31 2022 +0200
add deRotation and transformation vie rectanglePlus
commit 51459dbf57a86e3eac66ec0da02de40dc1b68796
Author: Julius Unverfehrt <julius.unverfehrt@iqser.com>
Date: Tue Aug 9 08:53:50 2022 +0200
add derotation and to pdf coords transformation to cv-analysis output
commit 733991e2f5a4664205b2f7cc756cebcbc9ee3930
Author: Julius Unverfehrt <julius.unverfehrt@iqser.com>
Date: Mon Aug 8 15:15:13 2022 +0200
update pipline with detrotation logic WIP
43 lines
1.5 KiB
Python
43 lines
1.5 KiB
Python
import gzip
|
|
import json
|
|
import logging
|
|
from operator import itemgetter
|
|
|
|
from cv_analysis.server.pipeline import get_analysis_pipeline
|
|
from cv_analysis.utils.banner import make_art
|
|
from pyinfra import config as pyinfra_config
|
|
from pyinfra.queue.queue_manager import QueueManager
|
|
from pyinfra.storage.storage import get_storage
|
|
|
|
PYINFRA_CONFIG = pyinfra_config.get_config()
|
|
|
|
logging.basicConfig(level=PYINFRA_CONFIG.logging_level_root)
|
|
|
|
|
|
def analysis_callback(queue_message: dict):
|
|
|
|
dossier_id, file_id, target_file_ext, response_file_ext, operation = itemgetter(
|
|
"dossierId", "fileId", "targetFileExtension", "responseFileExtension", "operation"
|
|
)(queue_message)
|
|
logging.info(f"Processing {dossier_id=}/{file_id=}, {operation=}.")
|
|
storage = get_storage(PYINFRA_CONFIG)
|
|
object_name = f"{dossier_id}/{file_id}.{target_file_ext}"
|
|
object_bytes = gzip.decompress(storage.get_object(PYINFRA_CONFIG.storage_bucket, object_name))
|
|
analysis_fn = get_analysis_pipeline(operation)
|
|
|
|
results = analysis_fn(object_bytes)
|
|
response = {**queue_message, "data": list(results)}
|
|
response = gzip.compress(json.dumps(response).encode())
|
|
response_name = f"{dossier_id}/{file_id}.{response_file_ext}"
|
|
|
|
storage.put_object(PYINFRA_CONFIG.storage_bucket, response_name, response)
|
|
return {"dossierId": dossier_id, "fileId": file_id}
|
|
|
|
|
|
if __name__ == "__main__":
|
|
|
|
logging.info(make_art())
|
|
|
|
queue_manager = QueueManager(PYINFRA_CONFIG)
|
|
queue_manager.start_consuming(analysis_callback)
|