chore: Types fixes (#267)

* docs: Recommend installing all groups for mypy to work properly

Signed-off-by: Eugene <[email protected]>

* chore: Fix some incomplete type defs

Signed-off-by: Eugene <[email protected]>

---------

Signed-off-by: Eugene <[email protected]>

Author: Fogapod

CONTRIBUTING.md

@@ -119,7 +119,7 @@ poetry shell
 To spawn a shell with the Virtual Environment activated. If the Virtual Environment doesn't exist, Poetry will create one for you. Then, to install dependencies, run:
 
 ```bash
-poetry install
+poetry install --all-extras
 ```
 
 **(Advanced) Use a Specific Python Version**

docling_core/experimental/serializer/base.py

@@ -6,7 +6,7 @@
 """Define base classes for serialization."""
 from abc import ABC, abstractmethod
 from pathlib import Path
-from typing import Optional, Union
+from typing import Any, Optional, Union
 
 from pydantic import AnyUrl, BaseModel
 
@@ -51,7 +51,7 @@ def serialize(
         item: TextItem,
         doc_serializer: "BaseDocSerializer",
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         ...
@@ -67,7 +67,7 @@ def serialize(
         item: TableItem,
         doc_serializer: "BaseDocSerializer",
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         ...
@@ -83,7 +83,7 @@ def serialize(
         item: PictureItem,
         doc_serializer: "BaseDocSerializer",
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         ...
@@ -99,7 +99,7 @@ def serialize(
         item: KeyValueItem,
         doc_serializer: "BaseDocSerializer",
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         ...
@@ -115,7 +115,7 @@ def serialize(
         item: FormItem,
         doc_serializer: "BaseDocSerializer",
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         ...
@@ -131,7 +131,7 @@ def serialize(
         item: Union[UnorderedList, OrderedList],
         doc_serializer: "BaseDocSerializer",
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         ...
@@ -147,7 +147,7 @@ def serialize(
         item: InlineGroup,
         doc_serializer: "BaseDocSerializer",
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         ...
@@ -163,7 +163,7 @@ def serialize(
         item: NodeItem,
         doc_serializer: "BaseDocSerializer",
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         ...
@@ -174,34 +174,40 @@ class BaseDocSerializer(ABC):
 
     @abstractmethod
     def serialize(
-        self, *, item: Optional[NodeItem] = None, **kwargs
+        self,
+        *,
+        item: Optional[NodeItem] = None,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Run the serialization."""
         ...
 
     @abstractmethod
-    def serialize_bold(self, text: str, **kwargs) -> str:
+    def serialize_bold(self, text: str, **kwargs: Any) -> str:
         """Hook for bold formatting serialization."""
         ...
 
     @abstractmethod
-    def serialize_italic(self, text: str, **kwargs) -> str:
+    def serialize_italic(self, text: str, **kwargs: Any) -> str:
         """Hook for italic formatting serialization."""
         ...
 
     @abstractmethod
-    def serialize_underline(self, text: str, **kwargs) -> str:
+    def serialize_underline(self, text: str, **kwargs: Any) -> str:
         """Hook for underline formatting serialization."""
         ...
 
     @abstractmethod
-    def serialize_strikethrough(self, text: str, **kwargs) -> str:
+    def serialize_strikethrough(self, text: str, **kwargs: Any) -> str:
         """Hook for strikethrough formatting serialization."""
         ...
 
     @abstractmethod
     def serialize_hyperlink(
-        self, text: str, hyperlink: Union[AnyUrl, Path], **kwargs
+        self,
+        text: str,
+        hyperlink: Union[AnyUrl, Path],
+        **kwargs: Any,
     ) -> str:
         """Hook for hyperlink serialization."""
         ...
@@ -210,7 +216,7 @@ def serialize_hyperlink(
     def get_parts(
         self,
         item: Optional[NodeItem] = None,
-        **kwargs,
+        **kwargs: Any,
     ) -> list[SerializationResult]:
         """Get the components to be combined for serializing this node."""
         ...
@@ -219,7 +225,7 @@ def get_parts(
     def post_process(
         self,
         text: str,
-        **kwargs,
+        **kwargs: Any,
     ) -> str:
         """Apply some text post-processing steps."""
         ...
@@ -228,13 +234,13 @@ def post_process(
     def serialize_captions(
         self,
         item: FloatingItem,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serialize the item's captions."""
         ...
 
     @abstractmethod
-    def get_excluded_refs(self, **kwargs) -> set[str]:
+    def get_excluded_refs(self, **kwargs: Any) -> set[str]:
         """Get references to excluded items."""
         ...
 

docling_core/experimental/serializer/common.py

@@ -214,7 +214,7 @@ def _captions_of_some_item(self) -> set[str]:
         return refs
 
     @override
-    def get_excluded_refs(self, **kwargs) -> set[str]:
+    def get_excluded_refs(self, **kwargs: Any) -> set[str]:
         """References to excluded items."""
         params = self.params.merge_with_patch(patch=kwargs)
         params_json = params.model_dump_json()
@@ -252,7 +252,10 @@ def get_excluded_refs(self, **kwargs) -> set[str]:
 
     @abstractmethod
     def serialize_doc(
-        self, *, parts: list[SerializationResult], **kwargs
+        self,
+        *,
+        parts: list[SerializationResult],
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serialize a document out of its pages."""
         ...
@@ -271,7 +274,7 @@ def serialize(
         list_level: int = 0,
         is_inline_scope: bool = False,
         visited: Optional[set[str]] = None,  # refs of visited items
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serialize a given node."""
         my_visited: set[str] = visited if visited is not None else set()
@@ -380,7 +383,7 @@ def get_parts(
         list_level: int = 0,
         is_inline_scope: bool = False,
         visited: Optional[set[str]] = None,  # refs of visited items
-        **kwargs,
+        **kwargs: Any,
     ) -> list[SerializationResult]:
         """Get the components to be combined for serializing this node."""
         parts: list[SerializationResult] = []
@@ -415,7 +418,7 @@ def post_process(
         *,
         formatting: Optional[Formatting] = None,
         hyperlink: Optional[Union[AnyUrl, Path]] = None,
-        **kwargs,
+        **kwargs: Any,
     ) -> str:
         """Apply some text post-processing steps."""
         params = self.params.merge_with_patch(patch=kwargs)
@@ -434,28 +437,31 @@ def post_process(
         return res
 
     @override
-    def serialize_bold(self, text: str, **kwargs) -> str:
+    def serialize_bold(self, text: str, **kwargs: Any) -> str:
         """Hook for bold formatting serialization."""
         return text
 
     @override
-    def serialize_italic(self, text: str, **kwargs) -> str:
+    def serialize_italic(self, text: str, **kwargs: Any) -> str:
         """Hook for italic formatting serialization."""
         return text
 
     @override
-    def serialize_underline(self, text: str, **kwargs) -> str:
+    def serialize_underline(self, text: str, **kwargs: Any) -> str:
         """Hook for underline formatting serialization."""
         return text
 
     @override
-    def serialize_strikethrough(self, text: str, **kwargs) -> str:
+    def serialize_strikethrough(self, text: str, **kwargs: Any) -> str:
         """Hook for strikethrough formatting serialization."""
         return text
 
     @override
     def serialize_hyperlink(
-        self, text: str, hyperlink: Union[AnyUrl, Path], **kwargs
+        self,
+        text: str,
+        hyperlink: Union[AnyUrl, Path],
+        **kwargs: Any,
     ) -> str:
         """Hook for hyperlink serialization."""
         return text
@@ -464,7 +470,7 @@ def serialize_hyperlink(
     def serialize_captions(
         self,
         item: FloatingItem,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serialize the item's captions."""
         params = self.params.merge_with_patch(patch=kwargs)

docling_core/experimental/serializer/doctags.py

@@ -1,7 +1,7 @@
 """Define classes for Doctags serialization."""
 
 from enum import Enum
-from typing import Dict, List, Optional, Union
+from typing import Any, Dict, List, Optional, Union
 
 from pydantic import BaseModel
 from typing_extensions import override
@@ -91,7 +91,7 @@ def serialize(
         item: TextItem,
         doc_serializer: BaseDocSerializer,
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         from docling_core.types.doc.document import SectionHeaderItem
@@ -154,7 +154,7 @@ def serialize(
         item: TableItem,
         doc_serializer: BaseDocSerializer,
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         params = DocTagsParams(**kwargs)
@@ -201,7 +201,7 @@ def serialize(
         item: PictureItem,
         doc_serializer: BaseDocSerializer,
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         params = DocTagsParams(**kwargs)
@@ -284,7 +284,7 @@ def serialize(
         item: KeyValueItem,
         doc_serializer: "BaseDocSerializer",
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         params = DocTagsParams(**kwargs)
@@ -356,7 +356,7 @@ def serialize(
         item: FormItem,
         doc_serializer: "BaseDocSerializer",
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         # TODO add actual implementation
@@ -378,7 +378,7 @@ def serialize(
         list_level: int = 0,
         is_inline_scope: bool = False,
         visited: Optional[set[str]] = None,  # refs of visited items
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         my_visited = visited if visited is not None else set()
@@ -423,7 +423,7 @@ def serialize(
         doc: DoclingDocument,
         list_level: int = 0,
         visited: Optional[set[str]] = None,  # refs of visited items
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         my_visited = visited if visited is not None else set()
@@ -454,7 +454,7 @@ def serialize(
         item: NodeItem,
         doc_serializer: "BaseDocSerializer",
         doc: DoclingDocument,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serializes the passed item."""
         return create_ser_result()
@@ -477,7 +477,10 @@ class DocTagsDocSerializer(DocSerializer):
 
     @override
     def serialize_doc(
-        self, *, parts: list[SerializationResult], **kwargs
+        self,
+        *,
+        parts: list[SerializationResult],
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serialize a document out of its pages."""
         delim = _get_delim(params=self.params)
@@ -496,7 +499,7 @@ def serialize_doc(
     def serialize_captions(
         self,
         item: FloatingItem,
-        **kwargs,
+        **kwargs: Any,
     ) -> SerializationResult:
         """Serialize the item's captions."""
         params = DocTagsParams(**kwargs)