docling-core 0.2.0__py3-none-any.whl → 1.0.0__py3-none-any.whl

docling_core/search/package.py

@@ -5,6 +5,7 @@
 
 """Models and methods to define a package model."""
 
+import importlib.metadata
 import re
 from typing import Final
 
@@ -27,7 +28,7 @@ class Package(BaseModel, extra="forbid"):
 
     name: StrictStr
     version: Annotated[str, StringConstraints(strict=True, pattern=VERSION_PATTERN)] = (
-        "0.1.0"
+        importlib.metadata.version("docling-core")
     )
 
     def __hash__(self):

docling_core/types/base.py

@@ -39,6 +39,10 @@ PredicateKeyNameT = TypeVar("PredicateKeyNameT", bound=str)
 PredicateKeyTypeT = TypeVar("PredicateKeyTypeT", bound=str)
 ProvenanceTypeT = TypeVar("ProvenanceTypeT", bound=str)
 CollectionNameTypeT = TypeVar("CollectionNameTypeT", bound=str)
+Coordinates = Annotated[
+    list[float],
+    Field(min_length=2, max_length=2, json_schema_extra=es_field(type="geo_point")),
+]
 T = TypeVar("T", bound=Hashable)
 
 UniqueList = Annotated[
@@ -61,7 +65,7 @@ ACQUISITION_TYPE = Literal[
 
 
 class Identifier(AliasModel, Generic[IdentifierTypeT], extra="forbid"):
-    """Unique identifier of a Deep Search data object."""
+    """Unique identifier of a Docling data object."""
 
     type_: IdentifierTypeT = Field(
         alias="type",
@@ -81,7 +85,7 @@ class Identifier(AliasModel, Generic[IdentifierTypeT], extra="forbid"):
         alias="_name",
         title="_Name",
         description=(
-            "A unique identifier of the data object across Deep Search, consisting of "
+            "A unique identifier of the data object across Docling, consisting of "
             "the concatenation of type and value in lower case, separated by hash "
             "(#)."
         ),
@@ -118,7 +122,7 @@ class Log(AliasModel, extra="forbid"):
         json_schema_extra=es_field(type="keyword", ignore_above=8191),
     )
     agent: StrictStr = Field(
-        description="The Deep Search agent that performed the task, e.g., CCS or CXS.",
+        description="The Docling agent that performed the task, e.g., CCS or CXS.",
         json_schema_extra=es_field(type="keyword", ignore_above=8191),
     )
     type_: StrictStr = Field(
@@ -138,7 +142,7 @@ class Log(AliasModel, extra="forbid"):
 
 
 class FileInfoObject(AliasModel):
-    """Filing information for any data object to be stored in a Deep Search database."""
+    """Filing information for any data object to be stored in a Docling database."""
 
     filename: StrictStr = Field(
         description="The name of a persistent object that created this data object",
@@ -156,7 +160,7 @@ class FileInfoObject(AliasModel):
     document_hash: StrictStr = Field(
         description=(
             "A unique identifier of this data object within a collection of a "
-            "Deep Search database"
+            "Docling database"
         ),
         alias="document-hash",
         json_schema_extra=es_field(type="keyword", ignore_above=8191),
@@ -164,7 +168,7 @@ class FileInfoObject(AliasModel):
 
 
 class CollectionTypeEnum(str, Enum):
-    """Enumeration of valid Deep Search collection types."""
+    """Enumeration of valid Docling collection types."""
 
     generic = "Generic"
     document = "Document"

docling_core/types/doc/document.py

@@ -3,7 +3,7 @@
 # SPDX-License-Identifier: MIT
 #
 
-"""Models for the Deep Search Document data type."""
+"""Models for the Docling Document data type."""
 
 from datetime import datetime
 from typing import Generic, Optional, Union
@@ -352,7 +352,7 @@ class ExportedCCSDocument(
         CollectionNameTypeT,
     ],
 ):
-    """Document model for Deep Search."""
+    """Document model for Docling."""
 
     obj_type: StrictStr = Field(
         "pdf-document",

docling_core/types/rec/attribute.py

@@ -6,7 +6,7 @@
 """Define the model Attribute."""
 from typing import Generic, Optional
 
-from pydantic import BaseModel, Field
+from pydantic import Field
 from typing_extensions import Annotated
 
 from docling_core.search.mapping import es_field
@@ -16,23 +16,20 @@ from docling_core.types.base import (
     PredicateKeyTypeT,
     PredicateValueTypeT,
     ProvenanceTypeT,
-    SubjectNameTypeT,
-    SubjectTypeT,
 )
 from docling_core.types.rec.base import ProvenanceItem
 from docling_core.types.rec.predicate import Predicate
+from docling_core.utils.alias import AliasModel
 
 
 class Attribute(
-    BaseModel,
+    AliasModel,
     Generic[
         IdentifierTypeT,
         PredicateValueTypeT,
         PredicateKeyNameT,
         PredicateKeyTypeT,
         ProvenanceTypeT,
-        SubjectTypeT,
-        SubjectNameTypeT,
     ],
     extra="forbid",
 ):

docling_core/types/rec/predicate.py

@@ -5,7 +5,7 @@
 
 """Define the model Predicate."""
 from datetime import datetime
-from typing import Annotated, Generic, Optional, TypeVar
+from typing import Annotated, Generic, Optional
 
 from pydantic import (
     BaseModel,
@@ -17,16 +17,14 @@ from pydantic import (
 )
 
 from docling_core.search.mapping import es_field
+from docling_core.types.base import (
+    Coordinates,
+    PredicateKeyNameT,
+    PredicateKeyTypeT,
+    PredicateValueTypeT,
+)
 from docling_core.utils.alias import AliasModel
 
-PredicateValueTypeT = TypeVar("PredicateValueTypeT", bound=str)
-PredicateKeyNameT = TypeVar("PredicateKeyNameT", bound=str)
-PredicateKeyTypeT = TypeVar("PredicateKeyTypeT", bound=str)
-Coordinates = Annotated[
-    list[float],
-    Field(min_length=2, max_length=2, json_schema_extra=es_field(type="geo_point")),
-]
-
 
 class NumericalValue(BaseModel, extra="forbid"):
     """Model for numerical values."""
@@ -117,7 +115,7 @@ class PredicateValue(AliasModel, Generic[PredicateValueTypeT], extra="forbid"):
 
 
 class Predicate(
-    BaseModel,
+    AliasModel,
     Generic[PredicateValueTypeT, PredicateKeyNameT, PredicateKeyTypeT],
     extra="forbid",
 ):

docling_core/types/rec/record.py

@@ -80,8 +80,6 @@ class Record(
                 PredicateKeyNameT,
                 PredicateKeyTypeT,
                 ProvenanceTypeT,
-                SubjectTypeT,
-                SubjectNameTypeT,
             ]
         ]
     ] = None

docling_core/types/rec/subject.py

@@ -19,6 +19,10 @@ from docling_core.types.doc.base import S3Reference
 from docling_core.utils.alias import AliasModel
 
 
+class SubjectNameIdentifier(Identifier[SubjectNameTypeT], Generic[SubjectNameTypeT]):
+    """Identifier of subject names.""" ""
+
+
 class Subject(
     AliasModel,
     Generic[IdentifierTypeT, SubjectTypeT, SubjectNameTypeT],
@@ -53,7 +57,7 @@ class Subject(
         ),
         json_schema_extra=es_field(type="keyword", ignore_above=8191),
     )
-    names: list[Identifier[SubjectNameTypeT]] = Field(
+    names: list[SubjectNameIdentifier[SubjectNameTypeT]] = Field(
         description=(
             "List of given names for this subject. They may not be unique across "
             "different subjects."

docling_core/utils/ds_generate_docs.py

@@ -44,7 +44,7 @@ def _prepare_directory(folder: str, clean: bool = False) -> None:
 
 
 def generate_collection_jsonschema(folder: str):
-    """Generate the JSON schema of Deep Search collections and export them to a folder.
+    """Generate the JSON schema of Docling collections and export them to a folder.
 
     Args:
         folder: The name of the directory.
@@ -58,7 +58,7 @@ def generate_collection_jsonschema(folder: str):
 
 
 def generate_collection_html(folder: str):
-    """Generate HTML pages documenting the data model of Deep Search collections.
+    """Generate HTML pages documenting the data model of Docling collections.
 
     The JSON schemas files need to be in a folder and the generated HTML pages will be
     written in the same folder.
@@ -79,7 +79,7 @@ def generate_collection_html(folder: str):
 
 
 def generate_collection_markdown(folder: str):
-    """Generate Markdown pages documenting the data model of Deep Search collections.
+    """Generate Markdown pages documenting the data model of Docling collections.
 
     The JSON schemas files need to be in a folder and the generated markdown pages will
     be written in the same folder.
@@ -101,7 +101,7 @@ def generate_collection_markdown(folder: str):
 
 
 def main() -> None:
-    """Generate the JSON Schema of Deep Search collections and export documentation."""
+    """Generate the JSON Schema of Docling collections and export documentation."""
     argparser = argparse.ArgumentParser()
     argparser.add_argument(
         "directory",

{docling_core-0.2.0.dist-info → docling_core-1.0.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: docling-core
-Version: 0.2.0
+Version: 1.0.0
 Summary: A python library to define and validate data types in Docling.
 Home-page: https://ds4sd.github.io/
 License: MIT

{docling_core-0.2.0.dist-info → docling_core-1.0.0.dist-info}/RECORD

@@ -12,35 +12,35 @@ docling_core/search/__init__.py,sha256=RucCUQjDlTZ7VfgbfnKDRBL-A-_Lcc2JWWeiVWHto
 docling_core/search/json_schema_to_search_mapper.py,sha256=9crSFuSbcXrJej7j1rYWK6b0x37cHDmPF6va5j3gknA,12933
 docling_core/search/mapping.py,sha256=6rqG7LgYSeWmooKNEcRa5gFDLp1ZdzPqDGlwTA5gpOk,724
 docling_core/search/meta.py,sha256=wSurrsqdP1N3gQKx027fVdzVmc33a7Y6rPl-FClQvtA,3318
-docling_core/search/package.py,sha256=S2inXEf9MUIqeFzQ3Vwi04rGD-7ouf_c7wyfTJD15FM,1763
+docling_core/search/package.py,sha256=Q0_FAWFt71_g0ifcFkCuXEpVAgpVFiT9mOdzq1fqeDM,1824
 docling_core/types/__init__.py,sha256=6mrAEKRW85uHJwNQBufwjPcMWCjm3oocA6MaO4_NLgg,805
-docling_core/types/base.py,sha256=u5F94ePswh-iIGcHX753WAHNRXcTCv_VfHnG2P6U4L8,8065
+docling_core/types/base.py,sha256=fNtfQ20NKa_RBNBWbq0DfO8o0zC1Cec8UAMu0Znsltk,8170
 docling_core/types/doc/__init__.py,sha256=Pzj_8rft6SJTVTCHgXRwHtuZjL6LK_6dcBWjikL9biY,125
 docling_core/types/doc/base.py,sha256=-j4vVs3JZuaUjm0fHIkLU9TD_4IZXQuGouLrddEAwPw,5508
 docling_core/types/doc/doc_ann.py,sha256=8pV2efUglw19jxl4_oqB__mSxjWvtGIcllyCdqA-b2s,1196
 docling_core/types/doc/doc_ocr.py,sha256=6PC0C-OczF-MyfgRxEI1xs3PWgNOzi7i2yEQbTqZz0I,1387
 docling_core/types/doc/doc_raw.py,sha256=Y69G6IiauNDaoT-5el4xo1ypWpnBJQ75akGGkCMTZSc,3888
-docling_core/types/doc/document.py,sha256=tHRlSCKy--mMIzp1Bu9kw7ZULd1MAE8zViarNk4WEMg,12557
+docling_core/types/doc/document.py,sha256=cMduCiFkPVCmXQehvNkXqXtDiXJJtB72o7_LZXz_S6I,12549
 docling_core/types/gen/__init__.py,sha256=C6TuCfvpSnSL5XDOFMcYHUY2-i08vvfOGRcdu6Af0pI,124
 docling_core/types/gen/generic.py,sha256=l4CZ4_Lb8ONG36WNJWbKX5hGKvTh_yU-hXp5hsm7uVU,844
 docling_core/types/nlp/__init__.py,sha256=hGcztAeVK7xkRBqRRvc4zbY4PGeJ0r0QrEsetnSx9nI,119
 docling_core/types/nlp/qa.py,sha256=TyZjubqkEoREv0YzmuLKlq4WW_TnJNj7BoBY1_r2a1E,2731
 docling_core/types/nlp/qa_labels.py,sha256=YLW2SYM9M1riktCUYctsg83Msb988NV2I754w4ibWzA,5880
 docling_core/types/rec/__init__.py,sha256=JVcjGAc7FsIryBmlE1syiOJYWhh3hpJIpp2o7VrX_vE,123
-docling_core/types/rec/attribute.py,sha256=RGFTBKGhsY8E_Yp0oEu96XGnEExheIDp0q_6BtBRBMc,1601
+docling_core/types/rec/attribute.py,sha256=PzPdaPhP5NWbFo8rYOoBl3Vfyx4zJUxN6ZpXl8UY7FM,1551
 docling_core/types/rec/base.py,sha256=jhTfInNGyB9NUw7o33PElrFGL80TqhU8MLcLZNZYj3E,3222
-docling_core/types/rec/predicate.py,sha256=1MY6L58ftv33RZzO8EmJqqoXSPspTBj1tTm2bamLYoc,4153
-docling_core/types/rec/record.py,sha256=PRAwgqAUVsuECPedbA1daeNt1q85iYYG003-rluWgOY,2814
+docling_core/types/rec/predicate.py,sha256=4iDwXl9c4jzHTDIlRNE88yvDzKA9_od0xjPUUUP5IjI,3959
+docling_core/types/rec/record.py,sha256=r1QgPepwH3YjmMHlwwmeK00ZHEJnAsvyOMeXFY_D9_Q,2750
 docling_core/types/rec/statement.py,sha256=BXkuKBz0BL7eiowL_aaYxsz_WBLfR4hfgiqTby4TRnk,920
-docling_core/types/rec/subject.py,sha256=QG_0_aNFr3bCbpYLGFGipwODfZt4VXzVdc6WSZotFNk,2424
+docling_core/types/rec/subject.py,sha256=wX9qsihwDbR7ZNSzY3vQymxi0eN1nxxsonrhSZzsMhA,2565
 docling_core/utils/__init__.py,sha256=VauNNpWRHG0_ISKrsy5-gTxicrdQZSau6qMfuMl3iqk,120
 docling_core/utils/alias.py,sha256=B6Lqvss8CbaNARHLR4qSmNh9OkB6LvqTpxfsFmkLAFo,874
-docling_core/utils/ds_generate_docs.py,sha256=tPzu_qgGXsqxfAWteun0_gwiBoGKfz9CEyBAwydHdM4,4264
+docling_core/utils/ds_generate_docs.py,sha256=0xGBagdC_PGjyeHXYZo90VnVrSTMZgHb0SYhFa6X7bQ,4248
 docling_core/utils/ds_generate_jsonschema.py,sha256=EhNQutqWJFWuN-yl9UUPFZ7DJTvGqg54qBIvUMHTHdA,1647
 docling_core/utils/validate.py,sha256=3FmnxnKTDZC5J9OGxCL3U3DGRl0t0bBV1NcySXswdas,2031
 docling_core/utils/validators.py,sha256=fBdyWX4PvFh7o_d25ZTs4iwmeo75QTbrxsvXv2kXkTg,2777
-docling_core-0.2.0.dist-info/LICENSE,sha256=2M9-6EoQ1sxFztTOkXGAtwUDJvnWaAHdB9BYWVwGkIw,1087
-docling_core-0.2.0.dist-info/METADATA,sha256=KrDlR-PeLf4yMhZr90iTjDtWBKF7QD0VgXa59DgY2g0,5174
-docling_core-0.2.0.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
-docling_core-0.2.0.dist-info/entry_points.txt,sha256=XHhtJEkdUuLxXSNxLdFIzx_siQ3z2UFQEKp-P8VYAE4,189
-docling_core-0.2.0.dist-info/RECORD,,
+docling_core-1.0.0.dist-info/LICENSE,sha256=2M9-6EoQ1sxFztTOkXGAtwUDJvnWaAHdB9BYWVwGkIw,1087
+docling_core-1.0.0.dist-info/METADATA,sha256=RPsZbjVEs0aIfMYDmK25CxR1b77iRCqbu8WbodN4l9g,5174
+docling_core-1.0.0.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
+docling_core-1.0.0.dist-info/entry_points.txt,sha256=XHhtJEkdUuLxXSNxLdFIzx_siQ3z2UFQEKp-P8VYAE4,189
+docling_core-1.0.0.dist-info/RECORD,,