nucliadb-models 6.3.0.post3415__py3-none-any.whl → 6.3.0.post3421__py3-none-any.whl

nucliadb_models/{filter.py → filters.py}

@@ -26,6 +26,7 @@ from pydantic import BaseModel, Discriminator, Tag, model_validator
 from typing_extensions import Annotated, Self
 
 from .common import FieldTypeName, Paragraph
+from .metadata import ResourceProcessingStatus
 from .utils import DateTime
 
 F = TypeVar("F")
@@ -217,6 +218,20 @@ class OriginPath(BaseModel, extra="forbid"):
     )
 
 
+class OriginSource(BaseModel, extra="forbid"):
+    """Matches the origin source id"""
+
+    prop: Literal["origin_source"]
+    id: str = pydantic.Field(description=("Source ID"))
+
+
+class OriginCollaborator(BaseModel, extra="forbid"):
+    """Matches the origin collaborators"""
+
+    prop: Literal["origin_collaborator"]
+    collaborator: str = pydantic.Field(description=("Collaborator"))
+
+
 class Generated(BaseModel, extra="forbid"):
     """Matches if the field was generated by the given source"""
 
@@ -236,6 +251,13 @@ class Kind(BaseModel, extra="forbid"):
     kind: Paragraph.TypeParagraph = pydantic.Field(description="The kind of paragraph to match")
 
 
+class Status(BaseModel, extra="forbid"):
+    """Matches resource in a certain processing status"""
+
+    prop: Literal["status"]
+    status: ResourceProcessingStatus = pydantic.Field(description="The status of the resource")
+
+
 # The discriminator function is optional, everything works without it.
 # We implement it because it makes pydantic produce more user-friendly errors
 def filter_discriminator(v: Any) -> Optional[str]:
@@ -269,14 +291,16 @@ FieldFilterExpression = Annotated[
         Annotated[Keyword, Tag("keyword")],
         Annotated[DateCreated, Tag("created")],
         Annotated[DateModified, Tag("modified")],
-        Annotated[OriginTag, Tag("origin_tag")],
         Annotated[Label, Tag("label")],
         Annotated[ResourceMimetype, Tag("resource_mimetype")],
         Annotated[FieldMimetype, Tag("field_mimetype")],
         Annotated[Entity, Tag("entity")],
         Annotated[Language, Tag("language")],
+        Annotated[OriginTag, Tag("origin_tag")],
         Annotated[OriginMetadata, Tag("origin_metadata")],
         Annotated[OriginPath, Tag("origin_path")],
+        Annotated[OriginSource, Tag("origin_source")],
+        Annotated[OriginCollaborator, Tag("origin_collaborator")],
         Annotated[Generated, Tag("generated")],
     ],
     Discriminator(filter_discriminator),
@@ -293,10 +317,31 @@ ParagraphFilterExpression = Annotated[
     Discriminator(filter_discriminator),
 ]
 
+ResourceFilterExpression = Annotated[
+    Union[
+        Annotated[And["ResourceFilterExpression"], Tag("and")],
+        Annotated[Or["ResourceFilterExpression"], Tag("or")],
+        Annotated[Not["ResourceFilterExpression"], Tag("not")],
+        Annotated[Resource, Tag("resource")],
+        Annotated[DateCreated, Tag("created")],
+        Annotated[DateModified, Tag("modified")],
+        Annotated[Label, Tag("label")],
+        Annotated[ResourceMimetype, Tag("resource_mimetype")],
+        Annotated[Language, Tag("language")],
+        Annotated[OriginTag, Tag("origin_tag")],
+        Annotated[OriginMetadata, Tag("origin_metadata")],
+        Annotated[OriginPath, Tag("origin_path")],
+        Annotated[OriginSource, Tag("origin_source")],
+        Annotated[OriginCollaborator, Tag("origin_collaborator")],
+        Annotated[Status, Tag("status")],
+    ],
+    Discriminator(filter_discriminator),
+]
+
 
 class FilterExpression(BaseModel, extra="forbid"):
     """Returns only documents that match this filter expression.
-    Filtering examples can be found here: https://docs.nuclia.dev/docs/rag/advanced/search/#filters
+    Filtering examples can be found here: https://docs.nuclia.dev/docs/rag/advanced/search_filters
 
     This allows building complex filtering expressions and replaces the following parameters:
     `fields`, `filters`, `range_*`, `resource_filters`, `keyword_filters`.
@@ -321,3 +366,14 @@ class FilterExpression(BaseModel, extra="forbid"):
             "OR returns text_blocks that match one of the two filters"
         ),
     )
+
+
+class CatalogFilterExpression(BaseModel, extra="forbid"):
+    """Returns only documents that match this filter expression.
+    Filtering examples can be found here: https://docs.nuclia.dev/docs/rag/advanced/search_filters
+
+    This allows building complex filtering expressions and replaces the following parameters:
+    `filters`, `range_*`, `with_status`.
+    """
+
+    resource: ResourceFilterExpression = pydantic.Field(description="Filter to apply to resources")

nucliadb_models/search.py

@@ -45,7 +45,7 @@ from nucliadb_models.internal.shards import (  # noqa isort: skip
     ShardReplica,
     KnowledgeboxShards,
 )
-from nucliadb_models.filter import FilterExpression
+from nucliadb_models.filters import CatalogFilterExpression, FilterExpression
 
 ANSWER_JSON_SCHEMA_EXAMPLE = {
     "name": "structred_response",
@@ -649,11 +649,21 @@ class SearchParamDefaults:
         title="Filter resource by an expression",
         description=(
             "Returns only documents that match this filter expression."
-            "Filtering examples can be found here: https://docs.nuclia.dev/docs/rag/advanced/search/#filters"
+            "Filtering examples can be found here: https://docs.nuclia.dev/docs/rag/advanced/search_filters"
             "This allows building complex filtering expressions and replaces the following parameters:"
             "`fields`, `filters`, `range_*`, `resource_filters`, `keyword_filters`."
         ),
     )
+    catalog_filter_expression = ParamDefault(
+        default=None,
+        title="Filter resource by an expression",
+        description=(
+            "Returns only documents that match this filter expression."
+            "Filtering examples can be found here: https://docs.nuclia.dev/docs/rag/advanced/search_filters"
+            "This allows building complex filtering expressions and replaces the following parameters:"
+            "`filters`, `range_*`, `with_status`."
+        ),
+    )
 
 
 class Filter(BaseModel):
@@ -671,6 +681,9 @@ class Filter(BaseModel):
 
 class CatalogRequest(BaseModel):
     query: str = SearchParamDefaults.query.to_pydantic_field()
+    filter_expression: SkipJsonSchema[Optional[CatalogFilterExpression]] = (
+        SearchParamDefaults.catalog_filter_expression.to_pydantic_field()
+    )
     filters: Union[list[str], list[Filter]] = Field(
         default=[],
         title="Filters",

{nucliadb_models-6.3.0.post3415.dist-info → nucliadb_models-6.3.0.post3421.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: nucliadb_models
-Version: 6.3.0.post3415
+Version: 6.3.0.post3421
 Author-email: Nuclia <nucliadb@nuclia.com>
 License: AGPL
 Project-URL: Homepage, https://nuclia.com

{nucliadb_models-6.3.0.post3415.dist-info → nucliadb_models-6.3.0.post3421.dist-info}/RECORD

@@ -8,7 +8,7 @@ nucliadb_models/export_import.py,sha256=A1KTjwQCRtyVAWcgabXsdltI78rauXBmZX1ie6Rx
 nucliadb_models/external_index_providers.py,sha256=aVyj-P4kVqfqPjF13E_lUM0FZsq8-DTbIsh-kHOgt2s,1787
 nucliadb_models/extracted.py,sha256=wnTjMsSPk1iZFtn4eFrBC9fSOZkNTeHQ_B9CRyaL0cA,6444
 nucliadb_models/file.py,sha256=4pDfQtXaBNB-ExeXC7NIdt33RbJp_u53_x8ACVkHXCM,2174
-nucliadb_models/filter.py,sha256=3CxqwhhVBy8a-aFhwxv0DLu1h3--mSz0wgiNCK9YvBU,10984
+nucliadb_models/filters.py,sha256=_umUEUs4QZEaqPaD3-cmxxKPEKVCONJLyFxTF1XDhag,13142
 nucliadb_models/labels.py,sha256=OUlX-apmFkibEN9bWThRJlbCD84hzJdddN1YYUV2Y3w,4201
 nucliadb_models/link.py,sha256=NRfsjLQpjZXndkb5o8qnSVPqb2knqk2kk5_iQB4AkaY,2785
 nucliadb_models/metadata.py,sha256=fiIJfht0Eg5a65ud2FdmHzElZ8VGdrDQ-F65-VJI4IE,8151
@@ -16,7 +16,7 @@ nucliadb_models/notifications.py,sha256=jr2J3zncs880jYf2oZHYt0VFcnlZevsbkyX69ovT
 nucliadb_models/processing.py,sha256=UeU-VxbBlOzkNxviOS3a0X_k7Ye-jYu3UOdGuu21M8M,971
 nucliadb_models/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nucliadb_models/resource.py,sha256=cjYloaRuCJFc3lGIxLZcX959oOq_N1f3V9bpPMYv4WA,9255
-nucliadb_models/search.py,sha256=ErfXrZsQe4b2iKrOE5dEzspj_1omHSeL2EpeiAEZEJc,80994
+nucliadb_models/search.py,sha256=nKd1v6enzkLBykUBxDGyChUiEIElmKOQoayFY2UYOxw,81669
 nucliadb_models/security.py,sha256=RewdzQ55nPZ9V7B0NX9KHeWg6B4Hg_RkeiFv2TQyrjs,1402
 nucliadb_models/synonyms.py,sha256=qXTPHfspMgw22hCjAOdFOIoUsRZ7Ju3JW-Lw9Nz4VaI,942
 nucliadb_models/text.py,sha256=RHN55PzQjyC0ghbf0r5GvVjTbFUTWzEDSCCkHkgnfig,3491
@@ -29,7 +29,7 @@ nucliadb_models/agents/ingestion.py,sha256=mV7gV6VpYg4VNpc59K3275TMUJZbUzeUnp3SZ
 nucliadb_models/internal/__init__.py,sha256=cp15ZcFnHvpcu_5-aK2A4uUyvuZVV_MJn4bIXMa20ks,835
 nucliadb_models/internal/predict.py,sha256=5rgUPrH_98gerySOZ-TR2PX_qzCGF1_8VxyOu3bGhis,2281
 nucliadb_models/internal/shards.py,sha256=uZLsMkYWrJDHq3xy_w7snSeV2X3aDBuht9GC_MG3sKc,1976
-nucliadb_models-6.3.0.post3415.dist-info/METADATA,sha256=nVLyPUsirO1nA5ZFanGErUS_gUsKrvq592pmLvgQQ_U,759
-nucliadb_models-6.3.0.post3415.dist-info/WHEEL,sha256=nn6H5-ilmfVryoAQl3ZQ2l8SH5imPWFpm1A5FgEuFV4,91
-nucliadb_models-6.3.0.post3415.dist-info/top_level.txt,sha256=UrY1I8oeovIRwkXLYplssTrxQdUjhSEFDFbnwaIV3tA,16
-nucliadb_models-6.3.0.post3415.dist-info/RECORD,,
+nucliadb_models-6.3.0.post3421.dist-info/METADATA,sha256=vh0jgWzUxIDDR3kphSO6m_9cfYITCVacm3e58DLnKGs,759
+nucliadb_models-6.3.0.post3421.dist-info/WHEEL,sha256=nn6H5-ilmfVryoAQl3ZQ2l8SH5imPWFpm1A5FgEuFV4,91
+nucliadb_models-6.3.0.post3421.dist-info/top_level.txt,sha256=UrY1I8oeovIRwkXLYplssTrxQdUjhSEFDFbnwaIV3tA,16
+nucliadb_models-6.3.0.post3421.dist-info/RECORD,,