Merge branch 'feat/updating-record' into develop

Author: frascuchon

argilla/docs/how_to_guides/record.md

@@ -436,7 +436,7 @@ updated_data = [
 dataset.records.log(records=updated_data)
 ```
 
-!!! note "Update the metadata"
+=== "Update the metadata"
     The `metadata` of `Record` object is a python dictionary. So to update the metadata of a record, you can iterate over the records and update the metadata by key or using `metadata.update`. After that, you should update the records in the dataset.
 
     ```python
@@ -452,6 +452,22 @@ dataset.records.log(records=updated_data)
     dataset.records.log(records=updated_records)
     ```
 
+=== "Update vectors"
+    When a new vector field is added to the dataset settings, or some value for the existing record vectors must updated, you can iterate over the records and update the vectors in the same way as the metadata. 
+
+    ```python
+    updated_records = []
+
+    for record in dataset.records():
+
+        record.vectors["new_vector"] = [...]
+        record.vector["v"] = [...]
+
+        updated_records.append(record)
+
+    dataset.records.log(records=updated_records)
+    ```
+
 ## Delete records
 
 You can delete records in a dataset calling the `delete` method on the `Dataset` object. To delete records, you need to retrieve them from the server and get a list with those that you want to delete.

argilla/src/argilla/_models/__init__.py

@@ -18,10 +18,10 @@
 from argilla._models._workspace import WorkspaceModel
 from argilla._models._user import UserModel, Role
 from argilla._models._dataset import DatasetModel
-from argilla._models._record._record import RecordModel
+from argilla._models._record._record import RecordModel, FieldValue
 from argilla._models._record._suggestion import SuggestionModel
 from argilla._models._record._response import UserResponseModel, ResponseStatus
-from argilla._models._record._vector import VectorModel
+from argilla._models._record._vector import VectorModel, VectorValue
 from argilla._models._record._metadata import MetadataModel, MetadataValue
 from argilla._models._search import (
     SearchQueryModel,

argilla/src/argilla/_models/_record/_record.py

@@ -16,17 +16,20 @@
 
 from pydantic import Field, field_serializer, field_validator
 
-from argilla._models._resource import ResourceModel
 from argilla._models._record._metadata import MetadataModel, MetadataValue
 from argilla._models._record._response import UserResponseModel
 from argilla._models._record._suggestion import SuggestionModel
 from argilla._models._record._vector import VectorModel
+from argilla._models._resource import ResourceModel
+
+__all__ = ["RecordModel", "FieldValue"]
 
+FieldValue = Union[str, None]
 
 class RecordModel(ResourceModel):
     """Schema for the records of a `Dataset`"""
 
-    fields: Optional[Dict[str, Union[str, None]]] = None
+    fields: Optional[Dict[str, FieldValue]] = None
     metadata: Optional[Union[List[MetadataModel], Dict[str, MetadataValue]]] = Field(default_factory=dict)
     vectors: Optional[List[VectorModel]] = Field(default_factory=list)
     responses: Optional[List[UserResponseModel]] = Field(default_factory=list)
@@ -49,7 +52,7 @@ def serialize_metadata(self, value: List[MetadataModel]) -> Dict[str, Any]:
         return {metadata.name: metadata.value for metadata in value}
 
     @field_serializer("fields", when_used="always")
-    def serialize_empty_fields(self, value: Dict[str, Union[str, None]]) -> Dict[str, Union[str, None]]:
+    def serialize_empty_fields(self, value: Dict[str, Union[str, None]]) -> Optional[Dict[str, Union[str, None]]]:
         """Serialize empty fields to None."""
         if isinstance(value, dict) and len(value) == 0:
             return None

argilla/src/argilla/_models/_record/_vector.py

@@ -11,19 +11,21 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+import re
 from typing import List
 
+from pydantic import field_validator
+
 from argilla._models import ResourceModel
 
-import re
-from pydantic import field_validator
+__all__ = ["VectorModel", "VectorValue"]
 
-__all__ = ["VectorModel"]
+VectorValue = List[float]
 
 
 class VectorModel(ResourceModel):
     name: str
-    vector_values: List[float]
+    vector_values: VectorValue
 
     @field_validator("name")
     @classmethod

argilla/src/argilla/records/_dataset_records.py

@@ -21,7 +21,7 @@
 
 from argilla._api import RecordsAPI
 from argilla._helpers import LoggingMixin
-from argilla._models import RecordModel, MetadataValue
+from argilla._models import RecordModel, MetadataValue, VectorValue, FieldValue
 from argilla.client import Argilla
 from argilla.records._io import GenericIO, HFDataset, HFDatasetsIO, JsonIO
 from argilla.records._resource import Record
@@ -414,13 +414,15 @@ def _infer_record_from_mapping(
         Returns:
             A Record object.
         """
-        fields: Dict[str, str] = {}
-        responses: List[Response] = []
         record_id: Optional[str] = None
-        suggestion_values = defaultdict(dict)
-        vectors: List[Vector] = []
+
+        fields: Dict[str, FieldValue] = {}
+        vectors: Dict[str, VectorValue] = {}
         metadata: Dict[str, MetadataValue] = {}
 
+        responses: List[Response] = []
+        suggestion_values: Dict[str, dict] = defaultdict(dict)
+
         schema = self.__dataset.schema
 
         for attribute, value in data.items():
@@ -475,7 +477,7 @@ def _infer_record_from_mapping(
                     {"value": value, "question_name": attribute, "question_id": schema_item.id}
                 )
             elif isinstance(schema_item, VectorField):
-                vectors.append(Vector(name=attribute, values=value))
+                vectors[attribute] = value
             elif isinstance(schema_item, MetadataPropertyBase):
                 metadata[attribute] = value
             else:
@@ -487,9 +489,9 @@ def _infer_record_from_mapping(
         return Record(
             id=record_id,
             fields=fields,
-            suggestions=suggestions,
-            responses=responses,
             vectors=vectors,
             metadata=metadata,
+            suggestions=suggestions,
+            responses=responses,
             _dataset=self.__dataset,
         )

argilla/src/argilla/records/_resource.py

@@ -23,6 +23,8 @@
     SuggestionModel,
     VectorModel,
     MetadataValue,
+    FieldValue,
+    VectorValue,
 )
 from argilla._resource import Resource
 from argilla.responses import Response, UserResponse
@@ -54,9 +56,9 @@ class Record(Resource):
     def __init__(
         self,
         id: Optional[Union[UUID, str]] = None,
-        fields: Optional[Dict[str, Union[str, None]]] = None,
+        fields: Optional[Dict[str, FieldValue]] = None,
         metadata: Optional[Dict[str, MetadataValue]] = None,
-        vectors: Optional[List[Vector]] = None,
+        vectors: Optional[Dict[str, VectorValue]] = None,
         responses: Optional[List[Response]] = None,
         suggestions: Optional[List[Suggestion]] = None,
         _server_id: Optional[UUID] = None,
@@ -93,7 +95,7 @@ def __init__(
         # Initialize the fields
         self.__fields = RecordFields(fields=self._model.fields)
         # Initialize the vectors
-        self.__vectors = RecordVectors(vectors=vectors, record=self)
+        self.__vectors = RecordVectors(vectors=vectors)
         # Initialize the metadata
         self.__metadata = RecordMetadata(metadata=metadata)
         self.__responses = RecordResponses(responses=responses, record=self)
@@ -158,8 +160,8 @@ def api_model(self) -> RecordModel:
             id=self._model.id,
             external_id=self._model.external_id,
             fields=self.fields.to_dict(),
-            metadata=self.metadata.models,
-            vectors=self.vectors.models,
+            metadata=self.metadata.api_models(),
+            vectors=self.vectors.api_models(),
             responses=self.responses.api_models(),
             suggestions=self.suggestions.api_models(),
         )
@@ -181,19 +183,22 @@ def to_dict(self) -> Dict[str, Dict]:
             represented as a key-value pair in the dictionary of the respective key. i.e.
             `{"fields": {"prompt": "...", "response": "..."}, "responses": {"rating": "..."},
         """
+        id = str(self.id) if self.id else None
+        server_id = str(self._model.id) if self._model.id else None
         fields = self.fields.to_dict()
-        metadata = dict(self.metadata)
+        metadata = self.metadata.to_dict()
         suggestions = self.suggestions.to_dict()
         responses = self.responses.to_dict()
         vectors = self.vectors.to_dict()
+
         return {
-            "id": self.id,
+            "id": id,
             "fields": fields,
             "metadata": metadata,
             "suggestions": suggestions,
             "responses": responses,
             "vectors": vectors,
-            "_server_id": str(self._model.id) if self._model.id else None,
+            "_server_id": server_id,
         }
 
     @classmethod
@@ -219,7 +224,6 @@ def from_dict(cls, data: Dict[str, Dict], dataset: Optional["Dataset"] = None) -
             for question_name, _responses in responses.items()
             for value in _responses
         ]
-        vectors = [Vector(name=vector_name, values=values) for vector_name, values in vectors.items()]
 
         return cls(
             id=record_id,
@@ -245,7 +249,7 @@ def from_model(cls, model: RecordModel, dataset: "Dataset") -> "Record":
             id=model.external_id,
             fields=model.fields,
             metadata={meta.name: meta.value for meta in model.metadata},
-            vectors=[Vector.from_model(model=vector) for vector in model.vectors],
+            vectors={vector.name: vector.vector_values for vector in model.vectors},
             # Responses and their models are not aligned 1-1.
             responses=[
                 response
@@ -258,27 +262,44 @@ def from_model(cls, model: RecordModel, dataset: "Dataset") -> "Record":
         )
 
 
-class RecordFields:
+class RecordFields(dict):
     """This is a container class for the fields of a Record.
-    It allows for accessing fields by attribute and iterating over them.
+    It allows for accessing fields by attribute and key name.
     """
 
-    def __init__(self, fields: Dict[str, Union[str, None]]) -> None:
-        self.__fields = fields or {}
-        for key, value in self.__fields.items():
-            setattr(self, key, value)
+    def __init__(self, fields: Optional[Dict[str, FieldValue]] = None) -> None:
+        super().__init__(fields or {})
 
-    def __getitem__(self, key: str) -> Optional[str]:
-        return self.__fields.get(key)
+    def to_dict(self) -> dict:
+        return dict(self.items())
 
-    def __iter__(self):
-        return iter(self.__fields)
 
-    def to_dict(self) -> Dict[str, Union[str, None]]:
-        return self.__fields
+class RecordMetadata(dict):
+    """This is a container class for the metadata of a Record."""
 
-    def __repr__(self) -> str:
-        return self.to_dict().__repr__()
+    def __init__(self, metadata: Optional[Dict[str, MetadataValue]] = None) -> None:
+        super().__init__(metadata or {})
+
+    def to_dict(self) -> dict:
+        return dict(self.items())
+
+    def api_models(self) -> List[MetadataModel]:
+        return [MetadataModel(name=key, value=value) for key, value in self.items()]
+
+
+class RecordVectors(dict):
+    """This is a container class for the vectors of a Record.
+    It allows for accessing suggestions by attribute and key name.
+    """
+
+    def __init__(self, vectors: Dict[str, VectorValue]) -> None:
+        super().__init__(vectors or {})
+
+    def to_dict(self) -> Dict[str, List[float]]:
+        return dict(self.items())
+
+    def api_models(self) -> List[VectorModel]:
+        return [Vector(name=name, values=value).api_model() for name, value in self.items()]
 
 
 class RecordResponses(Iterable[Response]):
@@ -309,6 +330,16 @@ def __getattr__(self, name) -> List[Response]:
     def __repr__(self) -> str:
         return {k: [{"value": v["value"]} for v in values] for k, values in self.to_dict().items()}.__repr__()
 
+    def to_dict(self) -> Dict[str, List[Dict]]:
+        """Converts the responses to a dictionary.
+        Returns:
+            A dictionary of responses.
+        """
+        response_dict = defaultdict(list)
+        for response in self.__responses:
+            response_dict[response.question_name].append({"value": response.value, "user_id": str(response.user_id)})
+        return response_dict
+
     def api_models(self) -> List[UserResponseModel]:
         """Returns a list of ResponseModel objects."""
 
@@ -321,16 +352,6 @@ def api_models(self) -> List[UserResponseModel]:
             for responses in responses_by_user_id.values()
         ]
 
-    def to_dict(self) -> Dict[str, List[Dict]]:
-        """Converts the responses to a dictionary.
-        Returns:
-            A dictionary of responses.
-        """
-        response_dict = defaultdict(list)
-        for response in self.__responses:
-            response_dict[response.question_name].append({"value": response.value, "user_id": response.user_id})
-        return response_dict
-
 
 class RecordSuggestions(Iterable[Suggestion]):
     """This is a container class for the suggestions of a Record.
@@ -345,15 +366,15 @@ def __init__(self, suggestions: List[Suggestion], record: Record) -> None:
             suggestion.record = self.record
             setattr(self, suggestion.question_name, suggestion)
 
-    def api_models(self) -> List[SuggestionModel]:
-        return [suggestion.api_model() for suggestion in self.__suggestions]
-
     def __iter__(self):
         return iter(self.__suggestions)
 
     def __getitem__(self, index: int):
         return self.__suggestions[index]
 
+    def __repr__(self) -> str:
+        return self.to_dict().__repr__()
+
     def to_dict(self) -> Dict[str, List[str]]:
         """Converts the suggestions to a dictionary.
         Returns:
@@ -368,48 +389,5 @@ def to_dict(self) -> Dict[str, List[str]]:
             }
         return suggestion_dict
 
-    def __repr__(self) -> str:
-        return self.to_dict().__repr__()
-
-
-class RecordVectors:
-    """This is a container class for the vectors of a Record.
-    It allows for accessing suggestions by attribute and iterating over them.
-    """
-
-    def __init__(self, vectors: List[Vector], record: Record) -> None:
-        self.__vectors = vectors or []
-        self.record = record
-        for vector in self.__vectors:
-            setattr(self, vector.name, vector.values)
-
-    def __repr__(self) -> str:
-        return {vector.name: f"{len(vector.values)}" for vector in self.__vectors}.__repr__()
-
-    @property
-    def models(self) -> List[VectorModel]:
-        return [vector.api_model() for vector in self.__vectors]
-
-    def to_dict(self) -> Dict[str, List[float]]:
-        """Converts the vectors to a dictionary.
-        Returns:
-            A dictionary of vectors.
-        """
-        return {vector.name: list(map(float, vector.values)) for vector in self.__vectors}
-
-
-class RecordMetadata(dict):
-    """This is a container class for the metadata of a Record."""
-
-    def __init__(self, metadata: Optional[Dict[str, MetadataValue]] = None) -> None:
-        super().__init__(metadata or {})
-
-    def __getattr__(self, item: str):
-        return self[item]
-
-    def __setattr__(self, key: str, value: MetadataValue):
-        self[key] = value
-
-    @property
-    def models(self) -> List[MetadataModel]:
-        return [MetadataModel(name=key, value=value) for key, value in self.items()]
+    def api_models(self) -> List[SuggestionModel]:
+        return [suggestion.api_model() for suggestion in self.__suggestions]