documente_shared 0.1.164b2__tar.gz → 0.1.165__tar.gz

{documente_shared-0.1.164b2 → documente_shared-0.1.165}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: documente_shared
-Version: 0.1.164b2
+Version: 0.1.165
 Summary: Shared utilities for LlamitAI projects
 Author-email: Tech <tech@llamitai.com>
 License: MIT

{documente_shared-0.1.164b2 → documente_shared-0.1.165}/documente_shared/domain/entities/workspace_document.py

@@ -9,7 +9,7 @@ from documente_shared.application.time_utils import get_datetime_from_data
 from documente_shared.domain.entities.in_memory_document import InMemoryDocument
 from documente_shared.domain.enums.common import ProcessingStatus
 from documente_shared.domain.enums.workspace import WorkspaceSource
-from documente_shared.domain.helpers.values import optional_str
+from documente_shared.domain.helpers.values import optional_str, optional_datetime_str
 
 
 class WorkspaceDocument(BaseModel):
@@ -34,7 +34,6 @@ class WorkspaceDocument(BaseModel):
     completed_at: datetime | None = Field(default=None)
     failed_at: datetime | None = Field(default=None)
 
-
     @property
     def to_persist_dict(self) -> dict:
         return {
@@ -45,10 +44,12 @@ class WorkspaceDocument(BaseModel):
             "digest": self.digest,
             "status": str(self.status),
             "source": optional_str(self.source),
-            "uploaded_at": (self.uploaded_at.isoformat() if self.uploaded_at else None),
-            "started_at": (self.started_at.isoformat() if self.started_at else None),
-            "completed_at": (self.completed_at.isoformat() if self.completed_at else None),
-            "failed_at": (self.failed_at.isoformat() if self.failed_at else None),
+            "uploaded_at": optional_datetime_str(self.uploaded_at),
+            "created_at": optional_datetime_str(self.created_at),
+            "updated_at": optional_datetime_str(self.updated_at),
+            "started_at": optional_datetime_str(self.started_at),
+            "completed_at": optional_datetime_str(self.completed_at),
+            "failed_at": optional_datetime_str(self.failed_at),
             "processing_time": self.processing_time,
             "metadata": self.metadata or {},
             "extraction": self.extraction or {},
@@ -71,14 +72,10 @@ class WorkspaceDocument(BaseModel):
             "processing_time": optional_str(self.processing_time),
             "metadata": self.metadata,
             "extraction": self.extraction,
-            "uploaded_at": (self.uploaded_at.isoformat() if self.uploaded_at else None),
-            "created_at": (self.created_at.isoformat() if self.created_at else None),
-            "updated_at": (self.updated_at.isoformat() if self.updated_at else None),
-            "started_at": (self.started_at.isoformat() if self.started_at else None),
-            "completed_at": (
-                self.completed_at.isoformat() if self.completed_at else None
-            ),
-            "failed_at": (self.failed_at.isoformat() if self.failed_at else None),
+            "uploaded_at": optional_datetime_str(self.uploaded_at),
+            "started_at": optional_datetime_str(self.started_at),
+            "completed_at": optional_datetime_str(self.completed_at),
+            "failed_at": optional_datetime_str(self.failed_at),
         }
 
     @classmethod

{documente_shared-0.1.164b2 → documente_shared-0.1.165}/documente_shared/domain/entities/workspace_document_page.py

@@ -17,6 +17,8 @@ class WorkspaceDocumentPage(BaseModel):
     tenant_slug: str | None = Field(default=None)
     page_file: InMemoryDocument | None = Field(default=None)
     page_file_url: str | None = Field(default=None)
+    cleaned_page_file: InMemoryDocument | None = Field(default=None)
+    cleaned_page_file_url: str | None = Field(default=None)
     created_at: datetime | None = Field(default=None)
     synced_at: datetime | None = Field(default=None)
 
@@ -44,6 +46,11 @@ class WorkspaceDocumentPage(BaseModel):
                 if self.page_file else None
             ),
             "page_file_url": self.page_file_url,
+            "cleaned_page_file": (
+                self.cleaned_page_file.to_dict
+                if self.cleaned_page_file else None
+            ),
+            "cleaned_page_file_url": self.cleaned_page_file_url,
             "created_at": (
                 self.created_at.isoformat()
                 if self.created_at else None
@@ -68,6 +75,11 @@ class WorkspaceDocumentPage(BaseModel):
                 if data.get('page_file') else None
             ),
             page_file_url=data.get('page_file_url'),
+            cleaned_page_file=(
+                InMemoryDocument.from_dict(data.get('cleaned_page_file'))
+                if data.get('cleaned_page_file') else None
+            ),
+            cleaned_page_file_url=data.get('cleaned_page_file_url'),
             created_at=get_datetime_from_data(input_datetime=data.get('created_at')),
             synced_at=get_datetime_from_data(input_datetime=data.get('synced_at')),
         )

{documente_shared-0.1.164b2 → documente_shared-0.1.165}/documente_shared/domain/enums/processing_case.py

@@ -6,7 +6,7 @@ class ProcessingCaseType(BaseEnum):
     UNIVIDA_SOAT = 'UNIVIDA_SOAT'
     AGNOSTIC = 'AGNOSTIC'
     CREDISEGURO_PARSE_CI = 'CREDISEGURO_PARSE_CI'
-    WORKSPACE = 'WORKSPACE'
+    WORKSPACE_DOCUMENT = 'WORKSPACE_DOCUMENT'
 
     @property
     def is_bcp_microcredito(self):
@@ -25,8 +25,8 @@ class ProcessingCaseType(BaseEnum):
         return self == ProcessingCaseType.CREDISEGURO_PARSE_CI
 
     @property
-    def is_workspace(self):
-        return self == ProcessingCaseType.WORKSPACE
+    def is_workspace_document(self):
+        return self == ProcessingCaseType.WORKSPACE_DOCUMENT
 
 
 class ProcessingDocumentType(BaseEnum):

{documente_shared-0.1.164b2 → documente_shared-0.1.165}/documente_shared/domain/filters/workspace_document.py

@@ -9,7 +9,6 @@ from documente_shared.domain.pagination.entities import ListFilters
 
 
 class WorkspaceDocumentFilters(ListFilters):
-    workspace_id: UUID | None = Field(default=None)
     tenant_ids: list[UUID] | None = Field(default_factory=list)
     workspace_ids: list[UUID] | None = Field(default_factory=list)
     statuses: list[ProcessingStatus] | None = Field(default_factory=list)
@@ -33,7 +32,6 @@ class WorkspaceDocumentFilters(ListFilters):
         return cls(
             cursor=params.get("cursor", None),
             limit=params.get_int(key="limit", default=None),
-            workspace_id=params.get_uuid(key="workspace_id", default=None),
             tenant_ids=params.get_uuid_list(key="tenant_ids", default=None),
             workspace_ids=params.get_uuid_list(key="workspace_ids", default=None),
             search=search_term.strip() if search_term else None,

{documente_shared-0.1.164b2 → documente_shared-0.1.165}/documente_shared/domain/pagination/entities.py

@@ -9,6 +9,7 @@ from documente_shared.domain.helpers.encoding import decode_base64, encode_base6
 
 
 class ListFilters(BaseModel):
+    search: str | None = Field(default=None)
     cursor: str | None = Field(default=None)
     limit: int = Field(default=DEFAULT_PAGINATION_PAGE_SIZE)
 
@@ -24,6 +25,7 @@ class ListFilters(BaseModel):
     @classmethod
     def from_query_params(cls, params: dict) -> "ListFilters":
         return cls(
+            search=params.get("search"),
             cursor=params.get("cursor"),
             limit=params.get("limit"),
         )
@@ -31,6 +33,7 @@ class ListFilters(BaseModel):
     @property
     def to_dict(self) -> dict:
         return {
+            "search": self.search,
             "cursor": self.cursor,
             "limit": self.limit,
         }

{documente_shared-0.1.164b2 → documente_shared-0.1.165}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "documente_shared"
-version = "0.1.164b2"
+version = "0.1.165"
 description = "Shared utilities for LlamitAI projects"
 authors = [{ name = "Tech", email = "tech@llamitai.com" }]
 license = { text = "MIT" }

{documente_shared-0.1.164b2 → documente_shared-0.1.165}/uv.lock

@@ -138,7 +138,7 @@ wheels = [
 
 [[package]]
 name = "documente-shared"
-version = "0.1.164b2"
+version = "0.1.165"
 source = { editable = "." }
 dependencies = [
     { name = "boto3" },

documente_shared-0.1.164b2/documente_shared/domain/entities/workspace_filters.py

@@ -1,36 +0,0 @@
-from dataclasses import dataclass
-from typing import Optional
-
-from documente_shared.application.query_params import QueryParams
-from documente_shared.domain.helpers.dicts import filter_none_values
-
-
-@dataclass
-class WorkspaceFilters(object):
-    sort_order: Optional[str] = None
-    search: Optional[str] = None
-    include_archived: bool = False
-
-    def __post_init__(self):
-        self.sort_order = self.sort_order or "desc"
-
-    @property
-    def to_params(self) -> dict:
-        params = {
-            "sort": self.sort_order,
-            "search": self.search,
-            "include_archived": str(self.include_archived).lower(),
-        }
-        return filter_none_values(params)
-
-    @classmethod
-    def from_params(cls, params: QueryParams) -> "WorkspaceFilters":
-        search_term = params.get_str(key="search", default=None)
-        return cls(
-            sort_order=params.get(key="sort", default="desc"),
-            search=search_term.strip() if search_term else None,
-            include_archived=params.get_bool(
-                key="include_archived",
-                default=False,
-            ),
-        )