nucliadb-models 6.9.3.post5346__py3-none-any.whl → 6.10.0.post5788__py3-none-any.whl

nucliadb_models/metadata.py

@@ -15,7 +15,7 @@
 import warnings
 from datetime import datetime
 from enum import Enum
-from typing import Any, Dict, List, Optional
+from typing import Any
 
 from pydantic import BaseModel, Field, field_validator, model_validator
 from typing_extensions import Self
@@ -30,7 +30,7 @@ class EntityRelation(BaseModel):
     entity_type: str
 
 
-class RelationType(Enum):
+class RelationType(str, Enum):
     ABOUT = "ABOUT"
     CHILD = "CHILD"
     COLAB = "COLAB"
@@ -49,7 +49,7 @@ class RelationNodeType(str, Enum):
 class RelationEntity(BaseModel):
     value: str
     type: RelationNodeType
-    group: Optional[str] = None
+    group: str | None = None
 
     @model_validator(mode="after")
     def check_relation_is_valid(self) -> Self:
@@ -60,20 +60,20 @@ class RelationEntity(BaseModel):
 
 
 class RelationMetadata(BaseModel):
-    paragraph_id: Optional[str] = None
-    source_start: Optional[int] = None
-    source_end: Optional[int] = None
-    to_start: Optional[int] = None
-    to_end: Optional[int] = None
-    data_augmentation_task_id: Optional[str] = None
+    paragraph_id: str | None = None
+    source_start: int | None = None
+    source_end: int | None = None
+    to_start: int | None = None
+    to_end: int | None = None
+    data_augmentation_task_id: str | None = None
 
 
 class Relation(BaseModel):
     relation: RelationType
-    label: Optional[str] = None
-    metadata: Optional[RelationMetadata] = None
+    label: str | None = None
+    metadata: RelationMetadata | None = None
 
-    from_: Optional[RelationEntity] = Field(default=None, alias="from")
+    from_: RelationEntity | None = Field(default=None, alias="from")
     to: RelationEntity
 
     @model_validator(mode="after")
@@ -100,9 +100,9 @@ class Relation(BaseModel):
 
 
 class InputMetadata(BaseModel):
-    metadata: Dict[str, str] = {}
-    language: Optional[str] = None
-    languages: Optional[List[str]] = Field(default=None, max_length=1024)
+    metadata: dict[str, str] = {}
+    language: str | None = None
+    languages: list[str] | None = Field(default=None, max_length=1024)
 
 
 class ResourceProcessingStatus(Enum):
@@ -120,7 +120,7 @@ class Metadata(InputMetadata):
 
 class FieldClassification(BaseModel):
     field: FieldID
-    classifications: List[Classification] = []
+    classifications: list[Classification] = []
 
 
 class ComputedMetadata(BaseModel):
@@ -129,12 +129,12 @@ class ComputedMetadata(BaseModel):
     without having to load the whole computed metadata field.
     """
 
-    field_classifications: List[FieldClassification] = []
+    field_classifications: list[FieldClassification] = []
 
 
 class UserMetadata(BaseModel):
-    classifications: List[UserClassification] = []
-    relations: List[Relation] = []
+    classifications: list[UserClassification] = []
+    relations: list[Relation] = []
 
 
 class TokenSplit(BaseModel):
@@ -154,7 +154,7 @@ class TokenSplit(BaseModel):
 
 
 class ParagraphAnnotation(BaseModel):
-    classifications: List[UserClassification] = []
+    classifications: list[UserClassification] = []
     key: str
 
 
@@ -169,12 +169,12 @@ class VisualSelection(BaseModel):
     left: float
     right: float
     bottom: float
-    token_ids: List[int]
+    token_ids: list[int]
 
 
 class PageSelections(BaseModel):
     page: int
-    visual: List[VisualSelection]
+    visual: list[VisualSelection]
 
     def __init__(self, **data):
         warnings.warn(
@@ -190,57 +190,69 @@ class UserFieldMetadata(BaseModel):
     Field-level metadata set by the user via the rest api
     """
 
-    paragraphs: List[ParagraphAnnotation] = []
-    question_answers: List[QuestionAnswerAnnotation] = []
+    paragraphs: list[ParagraphAnnotation] = []
+    question_answers: list[QuestionAnswerAnnotation] = []
     field: FieldID
 
 
 class Basic(BaseModel):
-    icon: Optional[str] = None
-    title: Optional[str] = None
-    summary: Optional[str] = None
-    thumbnail: Optional[str] = None
-    created: Optional[datetime] = None
-    modified: Optional[datetime] = None
-    metadata: Optional[Metadata] = None
-    usermetadata: Optional[UserMetadata] = None
-    fieldmetadata: Optional[List[UserFieldMetadata]] = None
-    computedmetadata: Optional[ComputedMetadata] = None
-    uuid: Optional[str] = None
-    last_seqid: Optional[int] = None
-    last_account_seq: Optional[int] = None
+    icon: str | None = None
+    title: str | None = None
+    summary: str | None = None
+    thumbnail: str | None = None
+    created: datetime | None = None
+    modified: datetime | None = None
+    metadata: Metadata | None = None
+    usermetadata: UserMetadata | None = None
+    fieldmetadata: list[UserFieldMetadata] | None = None
+    computedmetadata: ComputedMetadata | None = None
+    uuid: str | None = None
+    last_seqid: int | None = None
+    last_account_seq: int | None = None
+
+
+class SyncMetadata(BaseModel):
+    file_id: str = Field(description="Identifier of the file in the origin cloud storage system")
+    auth_provider: str = Field(
+        description="Authentication provider used to access the origin cloud storage system"
+    )
 
 
 class InputOrigin(BaseModel):
-    source_id: Optional[str] = None
-    url: Optional[str] = None
-    created: Optional[DateTime] = Field(
+    source_id: str | None = None
+    url: str | None = None
+    created: DateTime | None = Field(
         default=None,
         description="Creation date of the resource at the origin system. This can be later used for date range filtering on search endpoints. Have a look at the advanced search documentation page: https://docs.nuclia.dev/docs/rag/advanced/search/#date-filtering",
     )
-    modified: Optional[DateTime] = Field(
+    modified: DateTime | None = Field(
         default=None,
         description="Modification date of the resource at the origin system. This can be later used for date range filtering on search endpoints.  Have a look at the advanced search documentation page: https://docs.nuclia.dev/docs/rag/advanced/search/#date-filtering",
     )
-    metadata: Dict[str, str] = Field(
+    metadata: dict[str, str] = Field(
         default={},
         title="Metadata",
         description="Generic metadata from the resource at the origin system. It can later be used for filtering on search endpoints with '/origin.metadata/{key}/{value}'",
     )
-    tags: List[str] = Field(
+    tags: list[str] = Field(
         default=[],
         title="Tags",
         description="Resource tags about the origin system. It can later be used for filtering on search endpoints with '/origin.tags/{tag}'",
         max_length=300,
     )
-    collaborators: List[str] = Field(default=[], max_length=100)
-    filename: Optional[str] = None
-    related: List[str] = Field(default=[], max_length=100)
-    path: Optional[str] = Field(
+    collaborators: list[str] = Field(default=[], max_length=100)
+    filename: str | None = None
+    related: list[str] = Field(default=[], max_length=100)
+    path: str | None = Field(
         default=None,
         description="Path of the original resource. Typically used to store folder structure information of the resource at the origin system. It can be later used for filtering on search endpoints with '/origin.path/{path}'",
         max_length=2048,
     )
+    sync_metadata: SyncMetadata | None = Field(
+        default=None,
+        title="Sync Metadata",
+        description="Metadata related to the resource from the origin system fetched by the Progress Agentic RAG's Cloud Storage Sync service.",
+    )
 
     @field_validator("tags")
     def validate_tag_length(cls, tags):
@@ -253,10 +265,10 @@ class InputOrigin(BaseModel):
 class Origin(InputOrigin):
     # Created and modified are redefined to
     # use native datetime objects and skip validation
-    created: Optional[datetime] = None
-    modified: Optional[datetime] = None
+    created: datetime | None = None
+    modified: datetime | None = None
 
-    tags: List[str] = Field(
+    tags: list[str] = Field(
         default=[],
         title="Tags",
         description="Resource tags about the origin system. It can later be used for filtering on search endpoints with '/origin.tags/{tag}'",
@@ -268,16 +280,16 @@ class Origin(InputOrigin):
         API = "API"
         PYSDK = "PYSDK"
 
-    source: Optional[Source] = Source.API
+    source: Source | None = Source.API
 
 
 class Extra(BaseModel):
-    metadata: Dict[Any, Any] = Field(
+    metadata: dict[Any, Any] = Field(
         ...,
         title="Metadata",
-        description="Arbitrary JSON metadata provided by the user that is not meant to be searchable, but can be serialized on results.",  # noqa
+        description="Arbitrary JSON metadata provided by the user that is not meant to be searchable, but can be serialized on results.",
     )
 
 
 class Relations(BaseModel):
-    relations: Optional[List[Relation]] = None
+    relations: list[Relation] | None = None

nucliadb_models/notifications.py

@@ -49,7 +49,7 @@ class ResourceIndexed(BaseModel):
     seqid: int = Field(
         ...,
         title="Sequence ID",
-        description="Sequence ID of the resource operation. This can be used to track completion of specific operations.",  # noqa: E501
+        description="Sequence ID of the resource operation. This can be used to track completion of specific operations.",
     )
 
 
@@ -59,7 +59,7 @@ class ResourceWritten(BaseModel):
     seqid: int = Field(
         ...,
         title="Sequence ID",
-        description="Sequence ID of the resource operation. This can be used to track completion of specific operations.",  # noqa: E501
+        description="Sequence ID of the resource operation. This can be used to track completion of specific operations.",
     )
     operation: ResourceOperationType = Field(
         ..., title="Operation", description="Type of resource write operation."
@@ -77,7 +77,7 @@ class ResourceProcessed(BaseModel):
     seqid: int = Field(
         ...,
         title="Sequence ID",
-        description="Sequence ID of the resource operation. This can be used to track completion of specific operations.",  # noqa: E501
+        description="Sequence ID of the resource operation. This can be used to track completion of specific operations.",
     )
     ingestion_succeeded: bool = Field(
         default=True,

nucliadb_models/processing.py

@@ -12,10 +12,9 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from typing import Optional
 
 from pydantic import BaseModel
 
 
 class PushProcessingOptions(BaseModel):
-    ml_text: Optional[bool] = True
+    ml_text: bool | None = True

nucliadb_models/resource.py

@@ -16,7 +16,7 @@
 import string
 from datetime import datetime
 from enum import Enum
-from typing import Any, Dict, List, Optional, Union
+from typing import Any
 
 from pydantic import BaseModel, Field, field_validator, model_validator
 
@@ -72,44 +72,35 @@ class ExtractedDataTypeName(str, Enum):
     QA = "question_answers"
 
 
-class ReleaseChannel(str, Enum):
-    """
-    Deprecated. No longer used.
-    """
-
-    STABLE = "STABLE"
-    EXPERIMENTAL = "EXPERIMENTAL"
-
-
 class KnowledgeBoxConfig(BaseModel):
-    slug: Optional[SlugString] = Field(
+    slug: SlugString | None = Field(
         default=None, title="Slug", description="Slug for the Knowledge Box."
     )
-    title: Optional[str] = Field(default=None, title="Title", description="Title for the Knowledge Box.")
-    description: Optional[str] = Field(
+    title: str | None = Field(default=None, title="Title", description="Title for the Knowledge Box.")
+    description: str | None = Field(
         default=None,
         title="Description",
         description="Description for the Knowledge Box.",
     )
-    learning_configuration: Optional[Dict[str, Any]] = Field(
+    learning_configuration: dict[str, Any] | None = Field(
         default=None,
         title="Learning Configuration",
-        description="Learning configuration for the Knowledge Box. If provided, NucliaDB will set the learning configuration for the Knowledge Box.",  # noqa: E501
+        description="Learning configuration for the Knowledge Box. If provided, NucliaDB will set the learning configuration for the Knowledge Box.",
     )
 
-    external_index_provider: Optional[ExternalIndexProvider] = Field(
+    external_index_provider: ExternalIndexProvider | None = Field(
         default=None,
         title="External Index Provider",
         description="External index provider for the Knowledge Box.",
     )
 
-    configured_external_index_provider: Optional[dict[str, Any]] = Field(
+    configured_external_index_provider: dict[str, Any] | None = Field(
         default=None,
         title="Configured External Index Provider",
         description="Metadata for the configured external index provider (if any)",
     )
 
-    similarity: Optional[VectorSimilarity] = Field(
+    similarity: VectorSimilarity | None = Field(
         default=None,
         description="This field is deprecated. Use 'learning_configuration' instead.",
     )
@@ -126,7 +117,7 @@ class KnowledgeBoxConfig(BaseModel):
 
     @field_validator("slug")
     @classmethod
-    def id_check(cls, v: Optional[str]) -> Optional[str]:
+    def id_check(cls, v: str | None) -> str | None:
         if v is None:
             return v
 
@@ -145,7 +136,7 @@ class KnowledgeBoxConfig(BaseModel):
 
 
 class KnowledgeBoxObjSummary(BaseModel):
-    slug: Optional[SlugString] = None
+    slug: SlugString | None = None
     uuid: str
 
 
@@ -158,25 +149,25 @@ class KnowledgeBoxObj(BaseModel):
     The API representation of a Knowledge Box object.
     """
 
-    slug: Optional[SlugString] = None
+    slug: SlugString | None = None
     uuid: str
-    config: Optional[KnowledgeBoxConfig] = None
-    model: Optional[SemanticModelMetadata] = None
+    config: KnowledgeBoxConfig | None = None
+    model: SemanticModelMetadata | None = None
 
 
 class KnowledgeBoxList(BaseModel):
-    kbs: List[KnowledgeBoxObjSummary] = []
+    kbs: list[KnowledgeBoxObjSummary] = []
 
 
 # Resources
 
 
 class ExtractedData(BaseModel):
-    text: Optional[ExtractedText] = None
-    metadata: Optional[FieldComputedMetadata] = None
-    large_metadata: Optional[LargeComputedMetadata] = None
-    vectors: Optional[VectorObject] = None
-    question_answers: Optional[FieldQuestionAnswers] = None
+    text: ExtractedText | None = None
+    metadata: FieldComputedMetadata | None = None
+    large_metadata: LargeComputedMetadata | None = None
+    vectors: VectorObject | None = None
+    question_answers: FieldQuestionAnswers | None = None
 
 
 class TextFieldExtractedData(ExtractedData):
@@ -184,32 +175,31 @@ class TextFieldExtractedData(ExtractedData):
 
 
 class FileFieldExtractedData(ExtractedData):
-    file: Optional[FileExtractedData] = None
+    file: FileExtractedData | None = None
 
 
 class LinkFieldExtractedData(ExtractedData):
-    link: Optional[LinkExtractedData] = None
+    link: LinkExtractedData | None = None
 
 
 class ConversationFieldExtractedData(ExtractedData):
     pass
 
 
-ExtractedDataType = Optional[
-    Union[
-        TextFieldExtractedData,
-        FileFieldExtractedData,
-        LinkFieldExtractedData,
-        ConversationFieldExtractedData,
-    ]
-]
+ExtractedDataType = (
+    TextFieldExtractedData
+    | FileFieldExtractedData
+    | LinkFieldExtractedData
+    | ConversationFieldExtractedData
+    | None
+)
 
 
 class Error(BaseModel):
     body: str
     code: int
     code_str: str
-    created: Optional[datetime]
+    created: datetime | None
     severity: str
 
 
@@ -217,51 +207,51 @@ class FieldData(BaseModel): ...
 
 
 class TextFieldData(BaseModel):
-    value: Optional[FieldText] = None
-    extracted: Optional[TextFieldExtractedData] = None
-    error: Optional[Error] = None
-    status: Optional[str] = None
-    errors: Optional[list[Error]] = None
+    value: FieldText | None = None
+    extracted: TextFieldExtractedData | None = None
+    error: Error | None = None
+    status: str | None = None
+    errors: list[Error] | None = None
 
 
 class FileFieldData(BaseModel):
-    value: Optional[FieldFile] = None
-    extracted: Optional[FileFieldExtractedData] = None
-    error: Optional[Error] = None
-    status: Optional[str] = None
-    errors: Optional[list[Error]] = None
+    value: FieldFile | None = None
+    extracted: FileFieldExtractedData | None = None
+    error: Error | None = None
+    status: str | None = None
+    errors: list[Error] | None = None
 
 
 class LinkFieldData(BaseModel):
-    value: Optional[FieldLink] = None
-    extracted: Optional[LinkFieldExtractedData] = None
-    error: Optional[Error] = None
-    status: Optional[str] = None
-    errors: Optional[list[Error]] = None
+    value: FieldLink | None = None
+    extracted: LinkFieldExtractedData | None = None
+    error: Error | None = None
+    status: str | None = None
+    errors: list[Error] | None = None
 
 
 class ConversationFieldData(BaseModel):
-    value: Optional[FieldConversation] = None
-    extracted: Optional[ConversationFieldExtractedData] = None
-    error: Optional[Error] = None
-    status: Optional[str] = None
-    errors: Optional[list[Error]] = None
+    value: FieldConversation | None = None
+    extracted: ConversationFieldExtractedData | None = None
+    error: Error | None = None
+    status: str | None = None
+    errors: list[Error] | None = None
 
 
 class GenericFieldData(BaseModel):
-    value: Optional[str] = None
-    extracted: Optional[TextFieldExtractedData] = None
-    error: Optional[Error] = None
-    status: Optional[str] = None
-    errors: Optional[list[Error]] = None
+    value: str | None = None
+    extracted: TextFieldExtractedData | None = None
+    error: Error | None = None
+    status: str | None = None
+    errors: list[Error] | None = None
 
 
 class ResourceData(BaseModel):
-    texts: Optional[Dict[str, TextFieldData]] = None
-    files: Optional[Dict[str, FileFieldData]] = None
-    links: Optional[Dict[str, LinkFieldData]] = None
-    conversations: Optional[Dict[str, ConversationFieldData]] = None
-    generics: Optional[Dict[str, GenericFieldData]] = None
+    texts: dict[str, TextFieldData] | None = None
+    files: dict[str, FileFieldData] | None = None
+    links: dict[str, LinkFieldData] | None = None
+    conversations: dict[str, ConversationFieldData] | None = None
+    generics: dict[str, GenericFieldData] | None = None
 
 
 class QueueType(str, Enum):
@@ -273,29 +263,29 @@ class Resource(BaseModel):
     id: str
 
     # This first block of attributes correspond to Basic fields
-    slug: Optional[str] = None
-    title: Optional[str] = None
-    summary: Optional[str] = None
-    icon: Optional[str] = None
-    thumbnail: Optional[str] = None
-    metadata: Optional[Metadata] = None
-    usermetadata: Optional[UserMetadata] = None
-    fieldmetadata: Optional[List[UserFieldMetadata]] = None
-    computedmetadata: Optional[ComputedMetadata] = None
-    created: Optional[datetime] = None
-    modified: Optional[datetime] = None
-    last_seqid: Optional[int] = None
-    last_account_seq: Optional[int] = None
-    queue: Optional[QueueType] = None
-    hidden: Optional[bool] = None
-
-    origin: Optional[Origin] = None
-    extra: Optional[Extra] = None
-    relations: Optional[List[Relation]] = None
-
-    data: Optional[ResourceData] = None
-
-    security: Optional[ResourceSecurity] = Field(
+    slug: str | None = None
+    title: str | None = None
+    summary: str | None = None
+    icon: str | None = None
+    thumbnail: str | None = None
+    metadata: Metadata | None = None
+    usermetadata: UserMetadata | None = None
+    fieldmetadata: list[UserFieldMetadata] | None = None
+    computedmetadata: ComputedMetadata | None = None
+    created: datetime | None = None
+    modified: datetime | None = None
+    last_seqid: int | None = None
+    last_account_seq: int | None = None
+    queue: QueueType | None = None
+    hidden: bool | None = None
+
+    origin: Origin | None = None
+    extra: Extra | None = None
+    relations: list[Relation] | None = None
+
+    data: ResourceData | None = None
+
+    security: ResourceSecurity | None = Field(
         default=None,
         title="Security",
         description="Resource security metadata",
@@ -309,19 +299,12 @@ class ResourcePagination(BaseModel):
 
 
 class ResourceList(BaseModel):
-    resources: List[Resource]
+    resources: list[Resource]
     pagination: ResourcePagination
 
 
 class ResourceField(BaseModel):
     field_type: FieldTypeName
     field_id: str
-    value: Optional[
-        Union[
-            FieldText,
-            FieldFile,
-            FieldLink,
-            Conversation,
-        ]
-    ] = None
+    value: FieldText | FieldFile | FieldLink | Conversation | None = None
     extracted: ExtractedDataType = None