|
| 1 | +import os |
| 2 | +from datetime import datetime, timezone |
| 3 | + |
| 4 | +from botocore.exceptions import ClientError |
| 5 | +from enums.document_review_status import DocumentReviewStatus |
| 6 | +from models.document_reference import DocumentReferenceMetadataFields |
| 7 | +from models.document_review import ( |
| 8 | + DocumentReviewFileDetails, |
| 9 | + DocumentUploadReviewReference, |
| 10 | +) |
| 11 | +from models.sqs.review_message_body import ReviewMessageBody |
| 12 | +from services.base.dynamo_service import DynamoDBService |
| 13 | +from services.base.s3_service import S3Service |
| 14 | +from utils.audit_logging_setup import LoggingService |
| 15 | +from utils.request_context import request_context |
| 16 | + |
| 17 | +logger = LoggingService(__name__) |
| 18 | + |
| 19 | + |
| 20 | +class ReviewProcessorService: |
| 21 | + """ |
| 22 | + Service for processing single SQS messages from the document review queue. |
| 23 | + """ |
| 24 | + |
| 25 | + def __init__(self): |
| 26 | + """Initialize the review processor service with required AWS services.""" |
| 27 | + self.dynamo_service = DynamoDBService() |
| 28 | + self.s3_service = S3Service() |
| 29 | + |
| 30 | + self.review_table_name = os.environ["DOCUMENT_REVIEW_DYNAMODB_NAME"] |
| 31 | + self.staging_bucket_name = os.environ["STAGING_STORE_BUCKET_NAME"] |
| 32 | + self.review_bucket_name = os.environ["PENDING_REVIEW_BUCKET_NAME"] |
| 33 | + |
| 34 | + def process_review_message(self, review_message: ReviewMessageBody) -> None: |
| 35 | + """ |
| 36 | + Process a single SQS message from the review queue. |
| 37 | +
|
| 38 | + Args: |
| 39 | + sqs_message: SQS message record containing file and failure information |
| 40 | +
|
| 41 | + Raises: |
| 42 | + InvalidMessageException: If message format is invalid or required fields missing |
| 43 | + S3FileNotFoundException: If file doesn't exist in staging bucket |
| 44 | + ClientError: For AWS service errors (DynamoDB, S3) |
| 45 | + """ |
| 46 | + |
| 47 | + logger.info("Processing review queue message") |
| 48 | + |
| 49 | + request_context.patient_nhs_no = review_message.nhs_number |
| 50 | + |
| 51 | + review_id = review_message.upload_id |
| 52 | + review_files = self._move_files_to_review_bucket(review_message, review_id) |
| 53 | + document_upload_review = self._build_review_record( |
| 54 | + review_message, review_id, review_files |
| 55 | + ) |
| 56 | + try: |
| 57 | + self.dynamo_service.create_item( |
| 58 | + table_name=self.review_table_name, |
| 59 | + item=document_upload_review.model_dump( |
| 60 | + by_alias=True, exclude_none=True |
| 61 | + ), |
| 62 | + key_name=DocumentReferenceMetadataFields.ID.value, |
| 63 | + ) |
| 64 | + |
| 65 | + logger.info(f"Created review record {document_upload_review.id}") |
| 66 | + except ClientError as e: |
| 67 | + if e.response["Error"]["Code"] == "ConditionalCheckFailedException": |
| 68 | + logger.info("Entry already exists on Document Review table") |
| 69 | + else: |
| 70 | + raise e |
| 71 | + |
| 72 | + self._delete_files_from_staging(review_message) |
| 73 | + |
| 74 | + def _build_review_record( |
| 75 | + self, |
| 76 | + message_data: ReviewMessageBody, |
| 77 | + review_id: str, |
| 78 | + review_files: list[DocumentReviewFileDetails], |
| 79 | + ) -> DocumentUploadReviewReference: |
| 80 | + return DocumentUploadReviewReference( |
| 81 | + id=review_id, |
| 82 | + nhs_number=message_data.nhs_number, |
| 83 | + review_status=DocumentReviewStatus.PENDING_REVIEW, |
| 84 | + review_reason=message_data.failure_reason, |
| 85 | + author=message_data.uploader_ods, |
| 86 | + custodian=message_data.current_gp, |
| 87 | + files=review_files, |
| 88 | + upload_date=int(datetime.now(tz=timezone.utc).timestamp()), |
| 89 | + ) |
| 90 | + |
| 91 | + def _move_files_to_review_bucket( |
| 92 | + self, message_data: ReviewMessageBody, review_record_id: str |
| 93 | + ) -> list[DocumentReviewFileDetails]: |
| 94 | + """ |
| 95 | + Move file from staging to review bucket. |
| 96 | +
|
| 97 | + Args: |
| 98 | + message_data: Review queue message data |
| 99 | + review_record_id: ID of the review record being created |
| 100 | +
|
| 101 | + Returns: |
| 102 | + List of DocumentReviewFileDetails with new file locations in review bucket |
| 103 | + """ |
| 104 | + new_file_keys: list[DocumentReviewFileDetails] = [] |
| 105 | + |
| 106 | + for file in message_data.files: |
| 107 | + new_file_key = ( |
| 108 | + f"{message_data.nhs_number}/{review_record_id}/{file.file_name}" |
| 109 | + ) |
| 110 | + |
| 111 | + logger.info( |
| 112 | + f"Copying file from ({file.file_path}) in staging to review bucket: {new_file_key}" |
| 113 | + ) |
| 114 | + try: |
| 115 | + |
| 116 | + self.s3_service.copy_across_bucket( |
| 117 | + source_bucket=self.staging_bucket_name, |
| 118 | + source_file_key=file.file_path, |
| 119 | + dest_bucket=self.review_bucket_name, |
| 120 | + dest_file_key=new_file_key, |
| 121 | + if_none_match="*", |
| 122 | + ) |
| 123 | + logger.info("File successfully copied to review bucket") |
| 124 | + logger.info(f"Successfully moved file to: {new_file_key}") |
| 125 | + |
| 126 | + except ClientError as e: |
| 127 | + if e.response["Error"]["Code"] == "PreconditionFailed": |
| 128 | + logger.info("File already exists in the Review Bucket") |
| 129 | + else: |
| 130 | + raise e |
| 131 | + |
| 132 | + new_file_keys.append( |
| 133 | + DocumentReviewFileDetails( |
| 134 | + file_name=file.file_name, |
| 135 | + file_location=new_file_key, |
| 136 | + ) |
| 137 | + ) |
| 138 | + |
| 139 | + return new_file_keys |
| 140 | + |
| 141 | + def _delete_files_from_staging(self, message_data: ReviewMessageBody) -> None: |
| 142 | + for file in message_data.files: |
| 143 | + try: |
| 144 | + logger.info(f"Deleting file from staging bucket: {file.file_path}") |
| 145 | + self.s3_service.delete_object( |
| 146 | + s3_bucket_name=self.staging_bucket_name, file_key=file.file_path |
| 147 | + ) |
| 148 | + except Exception as e: |
| 149 | + logger.error(f"Error deleting files from staging: {str(e)}") |
| 150 | + # Continue processing as files |
0 commit comments