documente_shared 0.1.26__py3-none-any.whl → 0.1.28__py3-none-any.whl

documente_shared/domain/{entities.py → entities/document_process.py}

@@ -4,6 +4,7 @@ from decimal import Decimal
 from typing import Optional, List
 
 from documente_shared.application.time_utils import get_datetime_from_data
+from documente_shared.domain.entities.document_process_metadata import DocumentProcessMetadata
 from documente_shared.domain.enums import (
     DocumentProcessStatus,
     DocumentProcessSubCategory,
@@ -33,6 +34,10 @@ class DocumentProcess(object):
     started_at: Optional[datetime] = None
     failed_at: Optional[datetime] = None
     completed_at: Optional[datetime] = None
+    metadata_items: Optional[List[DocumentProcessMetadata]] = None
+
+    def __post_init__(self):
+        self.metadata_items = self.metadata_items or []
 
     @property
     def is_pending(self) -> bool:
@@ -127,6 +132,7 @@ class DocumentProcess(object):
             'started_at': self.started_at.isoformat() if self.started_at else None,
             'failed_at': self.failed_at.isoformat() if self.failed_at else None,
             'completed_at': self.completed_at.isoformat() if self.completed_at else None,
+            'metadata_items': [metadata.to_dict for metadata in self.metadata_items],
         }
 
     def overload(
@@ -135,7 +141,7 @@ class DocumentProcess(object):
         properties: List[str] = None,
     ):
         instance_properties = properties or [
-            'status', 'file_path', 'file_bytes', 'category', 'sub_category',
+            'status', 'metadata', 'file_path', 'file_bytes', 'category', 'sub_category',
             'processed_csv_path', 'processed_csv_bytes', 'processed_xlsx_path',
             'processed_xlsx_bytes', 'processing_time', 'uploaded_at',
             'enqueued_at', 'started_at', 'failed_at', 'completed_at',
@@ -172,4 +178,8 @@ class DocumentProcess(object):
             started_at=get_datetime_from_data(input_datetime=data.get('started_at')),
             failed_at=get_datetime_from_data(input_datetime=data.get('failed_at')),
             completed_at=get_datetime_from_data(input_datetime=data.get('completed_at')),
+            metadata_items=[
+                DocumentProcessMetadata.from_dict(metadata)
+                for metadata in data.get('metadata_items', [])
+            ],
         )

documente_shared/domain/entities/document_process_metadata.py

@@ -0,0 +1,65 @@
+from dataclasses import dataclass
+from datetime import datetime
+from typing import Optional
+
+
+@dataclass
+class DocumentProcessMetadata(object):
+    publication_date: Optional[datetime] = None
+    num_circular: Optional[str] = None
+    asfi_identifier: Optional[str] = None
+    contains_tables: Optional[bool] = None
+    text_content: Optional[str] = None
+    case_name: Optional[str] = None
+    starting_office: Optional[str] = None
+    output_json: Optional[dict] = None
+    processing_time: Optional[float] = None
+    llm_model: Optional[str] = None
+    num_pages: Optional[float] = None
+    num_tokens: Optional[float] = None
+    citcular_type: Optional[str] = None
+
+
+    @property
+    def to_dict(self):
+        return {
+            'publication_date': (
+                self.publication_date.isoformat()
+                if self.publication_date
+                else None
+            ),
+            'num_circular': self.num_circular,
+            'asfi_identifier': self.asfi_identifier,
+            'contains_tables': self.contains_tables,
+            'text_content': self.text_content,
+            'case_name': self.case_name,
+            'starting_office': self.starting_office,
+            'output_json': self.output_json,
+            'processing_time': self.processing_time,
+            'llm_model': self.llm_model,
+            'num_pages': self.num_pages,
+            'num_tokens': self.num_tokens,
+            'citcular_type': self.citcular_type
+        }
+
+    @classmethod
+    def from_dict(cls, data: dict):
+        return cls(
+            publication_date=(
+                datetime.fromisoformat(data.get('publication_date'))
+                if data.get('publication_date')
+                else None
+            ),
+            num_circular=data.get('num_circular'),
+            asfi_identifier=data.get('asfi_identifier'),
+            contains_tables=data.get('contains_tables'),
+            text_content=data.get('text_content'),
+            case_name=data.get('case_name'),
+            starting_office=data.get('starting_office'),
+            output_json=data.get('output_json'),
+            processing_time=data.get('processing_time'),
+            llm_model=data.get('llm_model'),
+            num_pages=data.get('num_pages'),
+            num_tokens=data.get('num_tokens'),
+            citcular_type=data.get('citcular_type')
+        )

{documente_shared-0.1.26.dist-info → documente_shared-0.1.28.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: documente_shared
-Version: 0.1.26
+Version: 0.1.28
 Summary: Shared utilities for Documente AI projects
 License: MIT
 Author: Tech

{documente_shared-0.1.26.dist-info → documente_shared-0.1.28.dist-info}/RECORD

@@ -4,7 +4,9 @@ documente_shared/application/digest.py,sha256=Um6E8WfFri2_lly4RFWydJyvSfPZGFcOX-
 documente_shared/application/time_utils.py,sha256=XDH27cKgoTFO8ad1JgrxKaeT7sZ1fduuJqLkvHUjy-Q,309
 documente_shared/domain/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 documente_shared/domain/base_enum.py,sha256=DojAfn-zQdtjtImeHUpBzE6TBTm07XrbMOdW3h8RVd8,1449
-documente_shared/domain/entities.py,sha256=J4B5fMBcwyiVB5xQToPFsOvF5yWIchsmyicC6iq3eRs,6222
+documente_shared/domain/entities/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+documente_shared/domain/entities/document_process.py,sha256=ioWAyXvz-ORgkaZZzPsBibSOM6Kpo_ScIhQInNjHf7M,6735
+documente_shared/domain/entities/document_process_metadata.py,sha256=rBHkDkoKwrxReKtXIScU1vrCO_6bg2LwWhrtXMcQ8TA,2351
 documente_shared/domain/enums.py,sha256=s3bFDkpplWvJWNpUwPOkC9a3OYUYsx8uTusN_FrzNQk,463
 documente_shared/domain/repositories.py,sha256=EekvB2BE4AnT5myEhU3DfwHfF9MkNJZRbqRVGFxCEWM,509
 documente_shared/infrastructure/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -12,6 +14,6 @@ documente_shared/infrastructure/dynamo_repositories.py,sha256=WULK8_N7PjA397vXTZ
 documente_shared/infrastructure/dynamo_table.py,sha256=Lod8vwUCGX65SNH_8hvlpG6rRUYb4jfZqoFHD1YloAA,2096
 documente_shared/infrastructure/s3_bucket.py,sha256=RwcEyYPUUXQ0fFR-Zr1KfkFGGpO6qAIfZ70CWHu1V8M,1923
 documente_shared/infrastructure/sqs_queue.py,sha256=PSiTAnjXvQ-W-9mzLpH2UjbQJTvYkMiaxNaMecF-cR4,1505
-documente_shared-0.1.26.dist-info/METADATA,sha256=-gSLnP_ENue8Qg0k1lg9wp8lLuDmorBNLhRVEwUznxA,640
-documente_shared-0.1.26.dist-info/WHEEL,sha256=Nq82e9rUAnEjt98J6MlVmMCZb-t9cYE2Ir1kpBmnWfs,88
-documente_shared-0.1.26.dist-info/RECORD,,
+documente_shared-0.1.28.dist-info/METADATA,sha256=2Q2xK6OmOJzCJjFWg-sJ_Cyoz0a_ZcLoq0RKptXQ4q8,640
+documente_shared-0.1.28.dist-info/WHEEL,sha256=Nq82e9rUAnEjt98J6MlVmMCZb-t9cYE2Ir1kpBmnWfs,88
+documente_shared-0.1.28.dist-info/RECORD,,