documente_shared 0.1.188__tar.gz → 0.1.189__tar.gz

{documente_shared-0.1.188 → documente_shared-0.1.189}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: documente_shared
-Version: 0.1.188
+Version: 0.1.189
 Summary: Shared utilities for LlamitAI projects
 Author-email: Tech <tech@llamitai.com>
 License: MIT

documente_shared-0.1.189/documente_shared/domain/entities/layout.py

@@ -0,0 +1,77 @@
+from typing import TypedDict, Literal, cast, Any
+
+from pydantic import BaseModel, ConfigDict
+
+
+class Vertex(BaseModel):
+    x: float
+    y: float
+
+    model_config = ConfigDict(extra="ignore")
+
+
+class TextSegment(BaseModel):
+    start: int
+    end: int
+
+    model_config = ConfigDict(extra="ignore")
+
+
+class TableAttrs(TypedDict):
+    entity_type: Literal["STRUCTURED_TABLE", "SEMI_STRUCTURED_TABLE"] | None
+    row_count: int
+    column_count: int
+    title_id: str | None
+    footer_ids: list[str]
+    merged_cell_ids: list[str]
+
+
+class CellAttrs(TypedDict):
+    row_index: int
+    column_index: int
+    row_span: int
+    column_span: int
+    entity_types: list[str]
+
+
+class MergedCellAttrs(CellAttrs):
+    member_cell_ids: list[str]
+
+
+class LayoutBlock(BaseModel):
+    id: str
+    block_type: str
+    bbox: list[Vertex]
+    confidence: float | None
+    text_segments: list[TextSegment]
+    children: list[str] = []
+    attributes: dict[str, Any] | None = None
+
+    model_config = ConfigDict(extra="ignore")
+
+    def as_table_attrs(self) -> TableAttrs:
+        return cast(TableAttrs, self.attributes)
+
+    def as_cell_attrs(self) -> CellAttrs:
+        return cast(CellAttrs, self.attributes)
+
+    def as_merged_cell_attrs(self) -> MergedCellAttrs:
+        return cast(MergedCellAttrs, self.attributes)
+
+
+class LayoutPage(BaseModel):
+    page_number: int
+    text: str
+    formatted_text: str = ""
+    width: int | None = None
+    height: int | None = None
+    blocks: list[LayoutBlock] = []
+
+    model_config = ConfigDict(extra="ignore")
+
+
+class PageRange(BaseModel):
+    start: int
+    end: int
+
+    model_config = ConfigDict(extra="ignore")

documente_shared-0.1.189/documente_shared/domain/enums/vnext.py

@@ -0,0 +1,15 @@
+from documente_shared.domain.base_enum import BaseEnum
+
+
+class DocumentExtractorType(str, BaseEnum):
+    TEXTRACT = "textract"
+    TEXTRACT_LAYOUT = "textract_layout"
+    DOCUMENTAI = "documentai"
+    DOCUMENTAI_LAYOUT = "documentai_layout"
+
+
+class ExtractorFeatureType(str, BaseEnum):
+    LAYOUT = "LAYOUT"
+    TABLES = "TABLES"
+    FORMS = "FORMS"
+    SIGNATURES = "SIGNATURE"

{documente_shared-0.1.188 → documente_shared-0.1.189}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "documente_shared"
-version = "0.1.188"
+version = "0.1.189"
 description = "Shared utilities for LlamitAI projects"
 authors = [{ name = "Tech", email = "tech@llamitai.com" }]
 license = { text = "MIT" }

{documente_shared-0.1.188 → documente_shared-0.1.189}/uv.lock

@@ -138,7 +138,7 @@ wheels = [
 
 [[package]]
 name = "documente-shared"
-version = "0.1.188"
+version = "0.1.189"
 source = { editable = "." }
 dependencies = [
     { name = "boto3" },