documente_shared 0.1.105__tar.gz → 0.1.107__tar.gz

{documente_shared-0.1.105 → documente_shared-0.1.107}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: documente_shared
-Version: 0.1.105
+Version: 0.1.107
 Summary: Shared utilities for Documente AI projects
 License: MIT
 Author: Tech

{documente_shared-0.1.105 → documente_shared-0.1.107}/documente_shared/application/files.py

@@ -20,3 +20,8 @@ def remove_slash_from_path(file_path: str) -> str:
     if file_path and file_path.startswith('/'):
         return file_path[1:]
     return file_path
+
+def remove_extension(filename: str) -> str:
+    if filename and '.' in filename:
+        return filename.rsplit('.', 1)[0]
+    return filename

{documente_shared-0.1.105 → documente_shared-0.1.107}/documente_shared/domain/entities/document.py

@@ -4,7 +4,7 @@ from datetime import datetime, tzinfo
 from decimal import Decimal
 from typing import Optional, List
 
-from documente_shared.application.files import remove_slash_from_path, get_filename_from_path
+from documente_shared.application.files import remove_slash_from_path, get_filename_from_path, remove_extension
 from documente_shared.application.numbers import normalize_number
 from documente_shared.application.time_utils import get_datetime_from_data
 from documente_shared.domain.constants import la_paz_tz
@@ -140,6 +140,10 @@ class DocumentProcessing(object):
     def extended_filename(self) -> str:
         return self.file_path.split('/')[-1]
 
+    @property
+    def raw_file_name(self) -> str:
+        return remove_extension(self.extended_filename)
+
     @property
     def filename(self) -> str:
         filename_with_extension = self.extended_filename

{documente_shared-0.1.105 → documente_shared-0.1.107}/documente_shared/domain/entities/in_memory_document.py

@@ -2,7 +2,7 @@ import base64
 from typing import Optional
 from dataclasses import dataclass
 
-from documente_shared.application.files import get_filename_from_path
+from documente_shared.application.files import get_filename_from_path, remove_extension
 from documente_shared.domain.exceptions import InMemoryDocumentContentError
 
 
@@ -33,6 +33,10 @@ class InMemoryDocument(object):
     def file_name(self) -> Optional[str]:
         return get_filename_from_path(self.file_path) if self.file_path else None
 
+    @property
+    def raw_file_name(self) -> str:
+        return remove_extension(self.file_name)
+
     @property
     def file_key(self) -> Optional[str]:
         return self.file_name

{documente_shared-0.1.105 → documente_shared-0.1.107}/documente_shared/domain/entities/processing_case_item.py

@@ -36,34 +36,25 @@ class ProcessingCaseItem(object):
         self.feedback = self.feedback or []
         self.metadata = self.metadata or {}
 
-    @property
-    def has_processed_csv(self) -> bool:
-        return self.processed_csv and self.processed_csv.is_valid
 
-    @property
-    def has_processed_xlsx(self) -> bool:
-        return self.processed_xlsx and self.processed_xlsx.is_valid
+    def __eq__(self, other: 'ProcessingCaseItem') -> bool:
+        if not other:
+            return False
 
-    @property
-    def has_processed_json(self) -> bool:
-        return self.processed_json and self.processed_json.is_valid
-    
-    @property
-    def is_procesable(self) -> bool:
         return (
-            (self.status.is_pending or self.status.is_enqueued)
-            and self.digest
-            and self.document
-            and self.document.is_procesable
+            self.uuid == other.uuid
+            and self.digest == other.digest
+            and self.status == other.status
+            and self.document_type == other.document_type
+            and self.document == other.document
+            and self.processing_time == other.processing_time
+            and self.processing_confidence == other.processing_confidence
+            and self.uploaded_at == other.uploaded_at
+            and self.started_at == other.started_at
+            and self.failed_at == other.failed_at
+            and self.completed_at == other.completed_at
         )
 
-    @property
-    def is_finished(self) -> bool:
-        return self.status in [
-            ProcessingStatus.COMPLETED,
-            ProcessingStatus.FAILED,
-        ]
-
     def pending(self, timezone: tzinfo = la_paz_tz):
         self.status = ProcessingStatus.PENDING
         self.started_at = None
@@ -73,9 +64,9 @@ class ProcessingCaseItem(object):
         self.started_at = datetime.now(tz=timezone)
 
     def failed(
-        self,
-        error_message: Optional[str] = None,
-        timezone: tzinfo = la_paz_tz,
+            self,
+            error_message: Optional[str] = None,
+            timezone: tzinfo = la_paz_tz,
     ):
         self.status = ProcessingStatus.FAILED
         self.failed_at = datetime.now(tz=timezone)
@@ -90,28 +81,71 @@ class ProcessingCaseItem(object):
 
     def deleted(self):
         self.status = ProcessingStatus.DELETED
-    
+
     def in_review(self):
         self.status = ProcessingStatus.IN_REVIEW
 
-    def __eq__(self, other: 'ProcessingCaseItem') -> bool:
-        if not other:
-            return False
+    def overload(
+        self,
+        new_instance: 'ProcessingCaseItem',
+        properties: List[str] = None,
+    ):
+        instance_properties = properties or [
+            'status',
+            'document',
+            'document_type',
+            'uploaded_from',
+            'processed_csv',
+            'processed_xlsx',
+            'processed_json',
+            'processing_time',
+            'processing_confidence',
+            'uploaded_at',
+            'started_at',
+            'failed_at',
+            'completed_at',
+            'feedback',
+            'metadata',
+        ]
+        for _property in instance_properties:
+            property_value = getattr(new_instance, _property)
+            if not hasattr(self, _property):
+                continue
+            setattr(self, _property, property_value)
+        return self
+
+    @property
+    def combined_id(self) -> str:
+        return f"{self.case_id}_{self.uuid}"
+
+    @property
+    def has_processed_csv(self) -> bool:
+        return self.processed_csv and self.processed_csv.is_valid
+
+    @property
+    def has_processed_xlsx(self) -> bool:
+        return self.processed_xlsx and self.processed_xlsx.is_valid
 
+    @property
+    def has_processed_json(self) -> bool:
+        return self.processed_json and self.processed_json.is_valid
+    
+    @property
+    def is_procesable(self) -> bool:
         return (
-            self.uuid == other.uuid
-            and self.digest == other.digest
-            and self.status == other.status
-            and self.document_type == other.document_type
-            and self.document == other.document
-            and self.processing_time == other.processing_time
-            and self.processing_confidence == other.processing_confidence
-            and self.uploaded_at == other.uploaded_at
-            and self.started_at == other.started_at
-            and self.failed_at == other.failed_at
-            and self.completed_at == other.completed_at
+            (self.status.is_pending or self.status.is_enqueued)
+            and self.digest
+            and self.document
+            and self.document.is_procesable
         )
 
+    @property
+    def is_finished(self) -> bool:
+        return self.status in [
+            ProcessingStatus.COMPLETED,
+            ProcessingStatus.FAILED,
+        ]
+
     @property
     def to_dict(self) -> dict:
         return {
@@ -185,35 +219,6 @@ class ProcessingCaseItem(object):
         simple_dict["processed_json_path"] = self.processed_json.file_path if self.processed_json else None
         return simple_dict
 
-    def overload(
-        self,
-        new_instance: 'ProcessingCaseItem',
-        properties: List[str] = None,
-    ):
-        instance_properties = properties or [
-            'status',
-            'document',
-            'document_type',
-            'uploaded_from',
-            'processed_csv',
-            'processed_xlsx',
-            'processed_json',
-            'processing_time',
-            'processing_confidence',
-            'uploaded_at',
-            'started_at',
-            'failed_at',
-            'completed_at',
-            'feedback',
-            'metadata',
-        ]
-        for _property in instance_properties:
-            property_value = getattr(new_instance, _property)
-            if not hasattr(self, _property):
-                continue
-            setattr(self, _property, property_value)
-        return self
-
     @classmethod
     def from_dict(cls, data: dict) -> 'ProcessingCaseItem':
         return cls(

{documente_shared-0.1.105 → documente_shared-0.1.107}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "documente_shared"
-version = "0.1.105"
+version = "0.1.107"
 description = "Shared utilities for Documente AI projects"
 authors = ["Tech <tech@llamitai.com>"]
 license = "MIT"