documente_shared 0.1.47__tar.gz → 0.1.51__tar.gz

{documente_shared-0.1.47 → documente_shared-0.1.51}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: documente_shared
-Version: 0.1.47
+Version: 0.1.51
 Summary: Shared utilities for Documente AI projects
 License: MIT
 Author: Tech

documente_shared-0.1.51/documente_shared/domain/constants.py

@@ -0,0 +1,3 @@
+import pytz
+
+la_paz_tz = pytz.timezone("America/La_Paz")

documente_shared-0.1.47/documente_shared/domain/entities/document_process.py → documente_shared-0.1.51/documente_shared/domain/entities/document.py

@@ -1,14 +1,15 @@
 from dataclasses import dataclass
-from datetime import datetime
+from datetime import datetime, tzinfo
 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.constants import la_paz_tz
+from documente_shared.domain.entities.document_metadata import DocumentProcessingMetadata
 from documente_shared.domain.enums import (
-    DocumentProcessStatus,
-    DocumentProcessSubCategory,
-    DocumentProcessCategory,
+    DocumentProcessingStatus,
+    DocumentProcessingSubCategory,
+    DocumentProcessingCategory,
 )
 
 def remove_slash_from_path(path: str) -> str:
@@ -17,13 +18,13 @@ def remove_slash_from_path(path: str) -> str:
     return path
 
 @dataclass
-class DocumentProcess(object):
+class DocumentProcessing(object):
     digest: str
-    status: DocumentProcessStatus
+    status: DocumentProcessingStatus
     file_path: Optional[str] = None
     file_bytes: Optional[bytes] = None
-    category: Optional[DocumentProcessCategory] = None
-    sub_category: Optional[DocumentProcessSubCategory] = None
+    category: Optional[DocumentProcessingCategory] = None
+    sub_category: Optional[DocumentProcessingSubCategory] = None
     processed_csv_path: Optional[str] = None
     processed_csv_bytes: Optional[bytes] = None
     processed_xlsx_path: Optional[str] = None
@@ -39,30 +40,30 @@ class DocumentProcess(object):
     failed_at: Optional[datetime] = None
     failed_reason: Optional[str] = None
     completed_at: Optional[datetime] = None
-    metadata_items: Optional[List[DocumentProcessMetadata]] = None
+    metadata_items: Optional[List[DocumentProcessingMetadata]] = None
 
     def __post_init__(self):
         self.metadata_items = self.metadata_items or []
 
     @property
     def is_pending(self) -> bool:
-        return self.status == DocumentProcessStatus.PENDING
+        return self.status == DocumentProcessingStatus.PENDING
 
     @property
     def is_enqueued(self) -> bool:
-        return self.status == DocumentProcessStatus.ENQUEUED
+        return self.status == DocumentProcessingStatus.ENQUEUED
 
     @property
     def is_processing(self) -> bool:
-        return self.status == DocumentProcessStatus.PROCESSING
+        return self.status == DocumentProcessingStatus.PROCESSING
 
     @property
     def is_completed(self) -> bool:
-        return self.status == DocumentProcessStatus.COMPLETED
+        return self.status == DocumentProcessingStatus.COMPLETED
 
     @property
     def is_failed(self) -> bool:
-        return self.status == DocumentProcessStatus.FAILED
+        return self.status == DocumentProcessingStatus.FAILED
 
     @property
     def is_valid(self) -> bool:
@@ -75,29 +76,33 @@ class DocumentProcess(object):
     @property
     def is_finished(self) -> bool:
         return self.status in [
-            DocumentProcessStatus.COMPLETED,
-            DocumentProcessStatus.FAILED,
+            DocumentProcessingStatus.COMPLETED,
+            DocumentProcessingStatus.FAILED,
         ]
 
-    def enqueue(self):
-        self.status = DocumentProcessStatus.ENQUEUED
-        self.enqueued_at = datetime.now()
+    def enqueue(self, timezone: tzinfo = la_paz_tz):
+        self.status = DocumentProcessingStatus.ENQUEUED
+        self.enqueued_at = datetime.now(tz=timezone)
 
-    def processing(self):
-        self.status = DocumentProcessStatus.PROCESSING
-        self.started_at = datetime.now()
+    def processing(self, timezone: tzinfo = la_paz_tz):
+        self.status = DocumentProcessingStatus.PROCESSING
+        self.started_at = datetime.now(tz=timezone)
 
-    def failed(self, error_message: Optional[str] = None):
+    def failed(
+        self,
+        error_message: Optional[str] = None,
+        timezone: tzinfo = la_paz_tz,
+    ):
         self.failed_reason = error_message
-        self.status = DocumentProcessStatus.FAILED
-        self.failed_at = datetime.now()
+        self.status = DocumentProcessingStatus.FAILED
+        self.failed_at = datetime.now(tz=timezone)
 
-    def completed(self):
-        self.status = DocumentProcessStatus.COMPLETED
-        self.completed_at = datetime.now()
+    def completed(self, timezone: tzinfo = la_paz_tz):
+        self.status = DocumentProcessingStatus.COMPLETED
+        self.completed_at = datetime.now(tz=timezone)
 
     def deleted(self):
-        self.status = DocumentProcessStatus.DELETED
+        self.status = DocumentProcessingStatus.DELETED
 
     @property
     def file_key(self) -> str:
@@ -128,7 +133,7 @@ class DocumentProcess(object):
         filename_with_extension = self.extended_filename
         return filename_with_extension.split('.')[0]
 
-    def __eq__(self, other: 'DocumentProcess') -> bool:
+    def __eq__(self, other: 'DocumentProcessing') -> bool:
         if not other:
             return False
 
@@ -185,7 +190,7 @@ class DocumentProcess(object):
 
     def overload(
         self,
-        new_instance: 'DocumentProcess',
+        new_instance: 'DocumentProcessing',
         properties: List[str] = None,
     ):
         instance_properties = properties or [
@@ -202,6 +207,7 @@ class DocumentProcess(object):
             'processed_json_path',
             'processed_json_bytes',
             'processed_metadata_path',
+            'processed_metadata_bytes',
             'processing_time',
             'issued_at',
             'uploaded_at',
@@ -219,17 +225,17 @@ class DocumentProcess(object):
         return self
 
     @classmethod
-    def from_dict(cls, data: dict) -> 'DocumentProcess':
+    def from_dict(cls, data: dict) -> 'DocumentProcessing':
         return cls(
             digest=data.get('digest'),
-            status=DocumentProcessStatus.from_value(data.get('status')),
+            status=DocumentProcessingStatus.from_value(data.get('status')),
             file_path=data.get('file_path'),
             category=(
-                DocumentProcessCategory.from_value(data.get('category'))
+                DocumentProcessingCategory.from_value(data.get('category'))
                 if data.get('category') else None
             ),
             sub_category=(
-                DocumentProcessSubCategory.from_value(data.get('sub_category'))
+                DocumentProcessingSubCategory.from_value(data.get('sub_category'))
                 if data.get('sub_category') else None
             ),
             processed_csv_path=data.get('processed_csv_path'),
@@ -248,7 +254,7 @@ class DocumentProcess(object):
             failed_reason=data.get('failed_reason'),
             completed_at=get_datetime_from_data(input_datetime=data.get('completed_at')),
             metadata_items=[
-                DocumentProcessMetadata.from_dict(metadata)
+                DocumentProcessingMetadata.from_dict(metadata)
                 for metadata in data.get('metadata_items', [])
             ],
         )

documente_shared-0.1.47/documente_shared/domain/entities/document_process_metadata.py → documente_shared-0.1.51/documente_shared/domain/entities/document_metadata.py

@@ -4,7 +4,7 @@ from typing import Optional
 
 
 @dataclass
-class DocumentProcessMetadata(object):
+class DocumentProcessingMetadata(object):
     publication_date: Optional[datetime] = None
     num_circular: Optional[str] = None
     asfi_identifier: Optional[str] = None

{documente_shared-0.1.47 → documente_shared-0.1.51}/documente_shared/domain/enums.py

@@ -1,7 +1,7 @@
 from documente_shared.domain.base_enum import BaseEnum
 
 
-class DocumentProcessStatus(BaseEnum):
+class DocumentProcessingStatus(BaseEnum):
     PENDING = 'PENDING'
     ENQUEUED = 'ENQUEUED'
     PROCESSING = 'PROCESSING'
@@ -11,20 +11,20 @@ class DocumentProcessStatus(BaseEnum):
     CANCELLED = 'CANCELLED'
 
 
-class DocumentProcessCategory(BaseEnum):
+class DocumentProcessingCategory(BaseEnum):
     CIRCULAR = 'CIRCULAR'
     DESGRAVAMEN = 'DESGRAVAMEN'
 
     @property
     def is_circular(self):
-        return self == DocumentProcessCategory.CIRCULAR
+        return self == DocumentProcessingCategory.CIRCULAR
 
     @property
     def is_desgravamen(self):
-        return self == DocumentProcessCategory.DESGRAVAMEN
+        return self == DocumentProcessingCategory.DESGRAVAMEN
 
 
-class DocumentProcessSubCategory(BaseEnum):
+class DocumentProcessingSubCategory(BaseEnum):
     # Circulares
     CC_COMBINADA = 'CC_COMBINADA'
     CC_NORMATIVA = 'CC_NORMATIVA'
@@ -32,5 +32,5 @@ class DocumentProcessSubCategory(BaseEnum):
     CC_RETENCION_SUSPENSION_REMISION = 'CC_RETENCION_SUSPENSION_REMISION'
 
     # Desgravamenes
-    DS_CREDISEGUROS = 'DS_CREDISEGUROS'
+    DS_CREDISEGURO = 'DS_CREDISEGURO'
 

documente_shared-0.1.51/documente_shared/domain/repositories.py

@@ -0,0 +1,25 @@
+from abc import ABC, abstractmethod
+from typing import Optional, List
+
+from documente_shared.domain.entities.document import DocumentProcessing
+from documente_shared.domain.enums import DocumentProcessingStatus
+
+
+class DocumentProcessingRepository(ABC):
+
+    @abstractmethod
+    def find(self, digest: str) ->Optional[DocumentProcessing]:
+        raise NotImplementedError
+
+    @abstractmethod
+    def persist(self, instance: DocumentProcessing) -> DocumentProcessing:
+        raise NotImplementedError
+
+    @abstractmethod
+    def remove(self, instance: DocumentProcessing):
+        raise NotImplementedError
+
+
+    @abstractmethod
+    def filter(self, statuses: List[DocumentProcessingStatus]) -> List[DocumentProcessing]:
+        raise NotImplementedError

{documente_shared-0.1.47 → documente_shared-0.1.51}/documente_shared/infrastructure/dynamo_repositories.py

@@ -2,31 +2,31 @@ from typing import Optional, List
 
 from boto3.dynamodb.conditions import Key
 
-from documente_shared.domain.entities.document_process import DocumentProcess
-from documente_shared.domain.enums import DocumentProcessStatus
-from documente_shared.domain.repositories import DocumentProcessRepository
+from documente_shared.domain.entities.document import DocumentProcessing
+from documente_shared.domain.enums import DocumentProcessingStatus, DocumentProcessingCategory
+from documente_shared.domain.repositories import DocumentProcessingRepository
 from documente_shared.infrastructure.dynamo_table import DynamoDBTable
 
 
 
-class DynamoDocumentProcessRepository(
+class DynamoDocumentProcessingRepository(
     DynamoDBTable,
-    DocumentProcessRepository,
+    DocumentProcessingRepository,
 ):
-    def find(self, digest: str) -> Optional[DocumentProcess]:
+    def find(self, digest: str) -> Optional[DocumentProcessing]:
         item = self.get(key={'digest': digest})
         if item:
-            return DocumentProcess.from_dict(item)
+            return DocumentProcessing.from_dict(item)
         return None
 
-    def persist(self, instance: DocumentProcess) -> DocumentProcess:
+    def persist(self, instance: DocumentProcessing) -> DocumentProcessing:
         self.put(instance.to_simple_dict)
         return instance
 
-    def remove(self, instance: DocumentProcess):
+    def remove(self, instance: DocumentProcessing):
         self.delete(key={'digest': instance.digest})
 
-    def filter(self, statuses: List[DocumentProcessStatus]) -> List[DocumentProcess]:
+    def filter(self, statuses: List[DocumentProcessingStatus]) -> List[DocumentProcessing]:
         items = []
 
         for status in statuses:
@@ -38,6 +38,6 @@ class DynamoDocumentProcessRepository(
             items.extend(status_items)
 
         return [
-            DocumentProcess.from_dict(item)
+            DocumentProcessing.from_dict(item)
             for item in items
         ]

{documente_shared-0.1.47 → documente_shared-0.1.51}/documente_shared/infrastructure/s3_bucket.py

@@ -3,7 +3,7 @@ import boto3
 from dataclasses import dataclass
 from typing import Optional
 
-from documente_shared.domain.entities.document_process import remove_slash_from_path
+from documente_shared.domain.entities.document import remove_slash_from_path
 
 
 def remove_none_values(data: dict) -> dict:  # noqa: WPS110

{documente_shared-0.1.47 → documente_shared-0.1.51}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "documente_shared"
-version = "0.1.47"
+version = "0.1.51"
 description = "Shared utilities for Documente AI projects"
 authors = ["Tech <tech@llamitai.com>"]
 license = "MIT"

documente_shared-0.1.47/documente_shared/domain/repositories.py

@@ -1,25 +0,0 @@
-from abc import ABC, abstractmethod
-from typing import Optional, List
-
-from documente_shared.domain.entities.document_process import DocumentProcess
-from documente_shared.domain.enums import DocumentProcessStatus
-
-
-class DocumentProcessRepository(ABC):
-
-    @abstractmethod
-    def find(self, digest: str) ->Optional[DocumentProcess]:
-        raise NotImplementedError
-
-    @abstractmethod
-    def persist(self, instance: DocumentProcess) -> DocumentProcess:
-        raise NotImplementedError
-
-    @abstractmethod
-    def remove(self, instance: DocumentProcess):
-        raise NotImplementedError
-
-
-    @abstractmethod
-    def filter(self, statuses: List[DocumentProcessStatus]) -> List[DocumentProcess]:
-        raise NotImplementedError