diff --git a/README.md b/README.md
index 7bfe6193..c072c3e0 100644
--- a/README.md
+++ b/README.md
@@ -143,6 +143,27 @@ JSON output:
}
```
+#### Hybrid Search <!-- omit in toc -->
+
+Hybrid search combines traditional keyword search with semantic search for more relevant results. You need to have an embedder configured in your index settings to use this feature.
+
+```python
+# Using hybrid search with the search method
+index.search(
+ 'action movie',
+ {
+ "hybrid": {"semanticRatio": 0.5, "embedder": "default"}
+ }
+)
+```
+
+The `semanticRatio` parameter (between 0 and 1) controls the balance between keyword search and semantic search:
+- 0: Only keyword search
+- 1: Only semantic search
+- Values in between: A mix of both approaches
+
+The `embedder` parameter specifies which configured embedder to use for the semantic search component.
+
#### Custom Search With Filters <!-- omit in toc -->
If you want to enable filtering, you must add your attributes to the `filterableAttributes` index setting.
diff --git a/meilisearch/index.py b/meilisearch/index.py
index 9f6e176d..812c3a57 100644
--- a/meilisearch/index.py
+++ b/meilisearch/index.py
@@ -24,19 +24,22 @@
from meilisearch.config import Config
from meilisearch.errors import version_error_hint_message
from meilisearch.models.document import Document, DocumentsResults
-from meilisearch.models.index import (
+from meilisearch.models.embedders import (
Embedders,
- Faceting,
+ EmbedderType,
HuggingFaceEmbedder,
- IndexStats,
- LocalizedAttributes,
OllamaEmbedder,
OpenAiEmbedder,
+ RestEmbedder,
+ UserProvidedEmbedder,
+)
+from meilisearch.models.index import (
+ Faceting,
+ IndexStats,
+ LocalizedAttributes,
Pagination,
ProximityPrecision,
- RestEmbedder,
TypoTolerance,
- UserProvidedEmbedder,
)
from meilisearch.models.task import Task, TaskInfo, TaskResults
from meilisearch.task import TaskHandler
@@ -277,14 +280,21 @@ def get_stats(self) -> IndexStats:
def search(self, query: str, opt_params: Optional[Mapping[str, Any]] = None) -> Dict[str, Any]:
"""Search in the index.
+ https://www.meilisearch.com/docs/reference/api/search
+
Parameters
----------
query:
String containing the searched word(s)
opt_params (optional):
Dictionary containing optional query parameters.
- Note: The vector parameter is only available in Meilisearch >= v1.13.0
- https://www.meilisearch.com/docs/reference/api/search#search-in-an-index
+ Common parameters include:
+ - hybrid: Dict with 'semanticRatio' and 'embedder' fields for hybrid search
+ - vector: Array of numbers for vector search
+ - retrieveVectors: Boolean to include vector data in search results
+ - filter: Filter queries by an attribute's value
+ - limit: Maximum number of documents returned
+ - offset: Number of documents to skip
Returns
-------
@@ -298,7 +308,9 @@ def search(self, query: str, opt_params: Optional[Mapping[str, Any]] = None) ->
"""
if opt_params is None:
opt_params = {}
+
body = {"q": query, **opt_params}
+
return self.http.post(
f"{self.config.paths.index}/{self.uid}/{self.config.paths.search}",
body=body,
@@ -955,14 +967,7 @@ def get_settings(self) -> Dict[str, Any]:
)
if settings.get("embedders"):
- embedders: dict[
- str,
- OpenAiEmbedder
- | HuggingFaceEmbedder
- | OllamaEmbedder
- | RestEmbedder
- | UserProvidedEmbedder,
- ] = {}
+ embedders: dict[str, EmbedderType] = {}
for k, v in settings["embedders"].items():
if v.get("source") == "openAi":
embedders[k] = OpenAiEmbedder(**v)
@@ -988,6 +993,26 @@ def update_settings(self, body: MutableMapping[str, Any]) -> TaskInfo:
----------
body:
Dictionary containing the settings of the index.
+ Supported settings include:
+ - 'rankingRules': List of ranking rules
+ - 'distinctAttribute': Attribute for deduplication
+ - 'searchableAttributes': Attributes that can be searched
+ - 'displayedAttributes': Attributes to display in search results
+ - 'stopWords': Words ignored in search queries
+ - 'synonyms': Dictionary of synonyms
+ - 'filterableAttributes': Attributes that can be used for filtering
+ - 'sortableAttributes': Attributes that can be used for sorting
+ - 'typoTolerance': Settings for typo tolerance
+ - 'pagination': Settings for pagination
+ - 'faceting': Settings for faceting
+ - 'dictionary': List of custom dictionary words
+ - 'separatorTokens': List of separator tokens
+ - 'nonSeparatorTokens': List of non-separator tokens
+ - 'embedders': Dictionary of embedder configurations for AI-powered search
+ - 'searchCutoffMs': Maximum search time in milliseconds
+ - 'proximityPrecision': Precision for proximity ranking
+ - 'localizedAttributes': Settings for localized attributes
+
More information:
https://www.meilisearch.com/docs/reference/api/settings#update-settings
@@ -1000,7 +1025,8 @@ def update_settings(self, body: MutableMapping[str, Any]) -> TaskInfo:
Raises
------
MeilisearchApiError
- An error containing details about why Meilisearch can't process your request. Meilisearch error codes are described here: https://www.meilisearch.com/docs/reference/errors/error_codes#meilisearch-errors
+ An error containing details about why Meilisearch can't process your request.
+ Meilisearch error codes are described here: https://www.meilisearch.com/docs/reference/errors/error_codes#meilisearch-errors
"""
if body.get("embedders"):
for _, v in body["embedders"].items():
@@ -1879,10 +1905,13 @@ def reset_non_separator_tokens(self) -> TaskInfo:
def get_embedders(self) -> Embedders | None:
"""Get embedders of the index.
+ Retrieves the current embedder configuration from Meilisearch.
+
Returns
-------
- settings:
- The embedders settings of the index.
+ Embedders:
+ The embedders settings of the index, or None if no embedders are configured.
+ Contains a dictionary of embedder configurations, where keys are embedder names.
Raises
------
@@ -1894,24 +1923,21 @@ def get_embedders(self) -> Embedders | None:
if not response:
return None
- embedders: dict[
- str,
- OpenAiEmbedder
- | HuggingFaceEmbedder
- | OllamaEmbedder
- | RestEmbedder
- | UserProvidedEmbedder,
- ] = {}
+ embedders: dict[str, EmbedderType] = {}
for k, v in response.items():
- if v.get("source") == "openAi":
+ source = v.get("source")
+ if source == "openAi":
embedders[k] = OpenAiEmbedder(**v)
- elif v.get("source") == "ollama":
- embedders[k] = OllamaEmbedder(**v)
- elif v.get("source") == "huggingFace":
+ elif source == "huggingFace":
embedders[k] = HuggingFaceEmbedder(**v)
- elif v.get("source") == "rest":
+ elif source == "ollama":
+ embedders[k] = OllamaEmbedder(**v)
+ elif source == "rest":
embedders[k] = RestEmbedder(**v)
+ elif source == "userProvided":
+ embedders[k] = UserProvidedEmbedder(**v)
else:
+ # Default to UserProvidedEmbedder for unknown sources
embedders[k] = UserProvidedEmbedder(**v)
return Embedders(embedders=embedders)
@@ -1919,10 +1945,13 @@ def get_embedders(self) -> Embedders | None:
def update_embedders(self, body: Union[MutableMapping[str, Any], None]) -> TaskInfo:
"""Update embedders of the index.
+ Updates the embedder configuration for the index. The embedder configuration
+ determines how Meilisearch generates vector embeddings for documents.
+
Parameters
----------
body: dict
- Dictionary containing the embedders.
+ Dictionary containing the embedders configuration.
Returns
-------
@@ -1933,13 +1962,28 @@ def update_embedders(self, body: Union[MutableMapping[str, Any], None]) -> TaskI
Raises
------
MeilisearchApiError
- An error containing details about why Meilisearch can't process your request. Meilisearch error codes are described here: https://www.meilisearch.com/docs/reference/errors/error_codes#meilisearch-errors
+ An error containing details about why Meilisearch can't process your request.
+ Meilisearch error codes are described here: https://www.meilisearch.com/docs/reference/errors/error_codes#meilisearch-errors
"""
+ if body is not None and body.get("embedders"):
+ embedders: dict[str, EmbedderType] = {}
+ for k, v in body["embedders"].items():
+ source = v.get("source")
+ if source == "openAi":
+ embedders[k] = OpenAiEmbedder(**v)
+ elif source == "huggingFace":
+ embedders[k] = HuggingFaceEmbedder(**v)
+ elif source == "ollama":
+ embedders[k] = OllamaEmbedder(**v)
+ elif source == "rest":
+ embedders[k] = RestEmbedder(**v)
+ elif source == "userProvided":
+ embedders[k] = UserProvidedEmbedder(**v)
+ else:
+ # Default to UserProvidedEmbedder for unknown sources
+ embedders[k] = UserProvidedEmbedder(**v)
- if body:
- for _, v in body.items():
- if "documentTemplateMaxBytes" in v and v["documentTemplateMaxBytes"] is None:
- del v["documentTemplateMaxBytes"]
+ body = {"embedders": {k: v.model_dump(by_alias=True) for k, v in embedders.items()}}
task = self.http.patch(self.__settings_url_for(self.config.paths.embedders), body)
@@ -1948,6 +1992,8 @@ def update_embedders(self, body: Union[MutableMapping[str, Any], None]) -> TaskI
def reset_embedders(self) -> TaskInfo:
"""Reset embedders of the index to default values.
+ Removes all embedder configurations from the index.
+
Returns
-------
task_info:
diff --git a/meilisearch/models/embedders.py b/meilisearch/models/embedders.py
new file mode 100644
index 00000000..01ba7b3c
--- /dev/null
+++ b/meilisearch/models/embedders.py
@@ -0,0 +1,213 @@
+from __future__ import annotations
+
+from typing import Any, Dict, Optional, Union
+
+from camel_converter.pydantic_base import CamelBase
+
+
+class Distribution(CamelBase):
+ """Distribution settings for embedders.
+
+ Parameters
+ ----------
+ mean: float
+ Mean value between 0 and 1
+ sigma: float
+ Sigma value between 0 and 1
+ """
+
+ mean: float
+ sigma: float
+
+
+class OpenAiEmbedder(CamelBase):
+ """OpenAI embedder configuration.
+
+ Parameters
+ ----------
+ source: str
+ The embedder source, must be "openAi"
+ url: Optional[str]
+ The URL Meilisearch contacts when querying the embedder
+ api_key: Optional[str]
+ Authentication token Meilisearch should send with each request to the embedder
+ model: Optional[str]
+ The model your embedder uses when generating vectors (defaults to text-embedding-3-small)
+ dimensions: Optional[int]
+ Number of dimensions in the chosen model
+ document_template: Optional[str]
+ Template defining the data Meilisearch sends to the embedder
+ document_template_max_bytes: Optional[int]
+ Maximum allowed size of rendered document template (defaults to 400)
+ distribution: Optional[Distribution]
+ Describes the natural distribution of search results
+ binary_quantized: Optional[bool]
+ Once set to true, irreversibly converts all vector dimensions to 1-bit values
+ """
+
+ source: str = "openAi"
+ url: Optional[str] = None
+ api_key: Optional[str] = None
+ model: Optional[str] = None # Defaults to text-embedding-3-small
+ dimensions: Optional[int] = None # Uses the model default
+ document_template: Optional[str] = None
+ document_template_max_bytes: Optional[int] = None # Default to 400
+ distribution: Optional[Distribution] = None
+ binary_quantized: Optional[bool] = None
+
+
+class HuggingFaceEmbedder(CamelBase):
+ """HuggingFace embedder configuration.
+
+ Parameters
+ ----------
+ source: str
+ The embedder source, must be "huggingFace"
+ url: Optional[str]
+ The URL Meilisearch contacts when querying the embedder
+ model: Optional[str]
+ The model your embedder uses when generating vectors (defaults to BAAI/bge-base-en-v1.5)
+ dimensions: Optional[int]
+ Number of dimensions in the chosen model
+ revision: Optional[str]
+ Model revision hash
+ document_template: Optional[str]
+ Template defining the data Meilisearch sends to the embedder
+ document_template_max_bytes: Optional[int]
+ Maximum allowed size of rendered document template (defaults to 400)
+ distribution: Optional[Distribution]
+ Describes the natural distribution of search results
+ binary_quantized: Optional[bool]
+ Once set to true, irreversibly converts all vector dimensions to 1-bit values
+ """
+
+ source: str = "huggingFace"
+ url: Optional[str] = None
+ model: Optional[str] = None # Defaults to BAAI/bge-base-en-v1.5
+ dimensions: Optional[int] = None
+ revision: Optional[str] = None
+ document_template: Optional[str] = None
+ document_template_max_bytes: Optional[int] = None # Default to 400
+ distribution: Optional[Distribution] = None
+ binary_quantized: Optional[bool] = None
+
+
+class OllamaEmbedder(CamelBase):
+ """Ollama embedder configuration.
+
+ Parameters
+ ----------
+ source: str
+ The embedder source, must be "ollama"
+ url: Optional[str]
+ The URL Meilisearch contacts when querying the embedder (defaults to http://localhost:11434/api/embeddings)
+ api_key: Optional[str]
+ Authentication token Meilisearch should send with each request to the embedder
+ model: Optional[str]
+ The model your embedder uses when generating vectors
+ dimensions: Optional[int]
+ Number of dimensions in the chosen model
+ document_template: Optional[str]
+ Template defining the data Meilisearch sends to the embedder
+ document_template_max_bytes: Optional[int]
+ Maximum allowed size of rendered document template (defaults to 400)
+ distribution: Optional[Distribution]
+ Describes the natural distribution of search results
+ binary_quantized: Optional[bool]
+ Once set to true, irreversibly converts all vector dimensions to 1-bit values
+ """
+
+ source: str = "ollama"
+ url: Optional[str] = None
+ api_key: Optional[str] = None
+ model: Optional[str] = None
+ dimensions: Optional[int] = None
+ document_template: Optional[str] = None
+ document_template_max_bytes: Optional[int] = None
+ distribution: Optional[Distribution] = None
+ binary_quantized: Optional[bool] = None
+
+
+class RestEmbedder(CamelBase):
+ """REST API embedder configuration.
+
+ Parameters
+ ----------
+ source: str
+ The embedder source, must be "rest"
+ url: Optional[str]
+ The URL Meilisearch contacts when querying the embedder
+ api_key: Optional[str]
+ Authentication token Meilisearch should send with each request to the embedder
+ dimensions: Optional[int]
+ Number of dimensions in the embeddings
+ document_template: Optional[str]
+ Template defining the data Meilisearch sends to the embedder
+ document_template_max_bytes: Optional[int]
+ Maximum allowed size of rendered document template (defaults to 400)
+ request: Dict[str, Any]
+ A JSON value representing the request Meilisearch makes to the remote embedder
+ response: Dict[str, Any]
+ A JSON value representing the request Meilisearch expects from the remote embedder
+ headers: Optional[Dict[str, str]]
+ Custom headers to send with the request
+ distribution: Optional[Distribution]
+ Describes the natural distribution of search results
+ binary_quantized: Optional[bool]
+ Once set to true, irreversibly converts all vector dimensions to 1-bit values
+ """
+
+ source: str = "rest"
+ url: Optional[str] = None
+ api_key: Optional[str] = None
+ dimensions: Optional[int] = None
+ document_template: Optional[str] = None
+ document_template_max_bytes: Optional[int] = None
+ request: Dict[str, Any]
+ response: Dict[str, Any]
+ headers: Optional[Dict[str, str]] = None
+ distribution: Optional[Distribution] = None
+ binary_quantized: Optional[bool] = None
+
+
+class UserProvidedEmbedder(CamelBase):
+ """User-provided embedder configuration.
+
+ Parameters
+ ----------
+ source: str
+ The embedder source, must be "userProvided"
+ dimensions: int
+ Number of dimensions in the embeddings
+ distribution: Optional[Distribution]
+ Describes the natural distribution of search results
+ binary_quantized: Optional[bool]
+ Once set to true, irreversibly converts all vector dimensions to 1-bit values
+ """
+
+ source: str = "userProvided"
+ dimensions: int
+ distribution: Optional[Distribution] = None
+ binary_quantized: Optional[bool] = None
+
+
+# Type alias for the embedder union type
+EmbedderType = Union[
+ OpenAiEmbedder,
+ HuggingFaceEmbedder,
+ OllamaEmbedder,
+ RestEmbedder,
+ UserProvidedEmbedder,
+]
+
+
+class Embedders(CamelBase):
+ """Container for embedder configurations.
+
+ Parameters
+ ----------
+ embedders: Dict[str, Union[OpenAiEmbedder, HuggingFaceEmbedder, OllamaEmbedder, RestEmbedder, UserProvidedEmbedder]]
+ Dictionary of embedder configurations, where keys are embedder names
+ """
+
+ embedders: Dict[str, EmbedderType]
diff --git a/meilisearch/models/index.py b/meilisearch/models/index.py
index 5457be56..f23cb458 100644
--- a/meilisearch/models/index.py
+++ b/meilisearch/models/index.py
@@ -1,7 +1,7 @@
from __future__ import annotations
from enum import Enum
-from typing import Any, Dict, Iterator, List, Optional, Union
+from typing import Any, Dict, Iterator, List, Optional
from camel_converter import to_snake
from camel_converter.pydantic_base import CamelBase
@@ -62,65 +62,3 @@ class EmbedderDistribution(CamelBase):
class LocalizedAttributes(CamelBase):
attribute_patterns: List[str]
locales: List[str]
-
-
-class OpenAiEmbedder(CamelBase):
- source: str = "openAi"
- url: Optional[str] = None
- model: Optional[str] = None # Defaults to text-embedding-3-small
- dimensions: Optional[int] = None # Uses the model default
- api_key: Optional[str] = None # Can be provided through a CLI option or environment variable
- document_template: Optional[str] = None
- document_template_max_bytes: Optional[int] = None # Default to 400
- distribution: Optional[EmbedderDistribution] = None
- binary_quantized: Optional[bool] = None
-
-
-class HuggingFaceEmbedder(CamelBase):
- source: str = "huggingFace"
- model: Optional[str] = None # Defaults to BAAI/bge-base-en-v1.5
- revision: Optional[str] = None
- document_template: Optional[str] = None
- document_template_max_bytes: Optional[int] = None # Default to 400
- distribution: Optional[EmbedderDistribution] = None
- binary_quantized: Optional[bool] = None
-
-
-class OllamaEmbedder(CamelBase):
- source: str = "ollama"
- url: Optional[str] = None
- api_key: Optional[str] = None
- model: str
- document_template: Optional[str] = None
- document_template_max_bytes: Optional[int] = None # Default to 400
- distribution: Optional[EmbedderDistribution] = None
- binary_quantized: Optional[bool] = None
-
-
-class RestEmbedder(CamelBase):
- source: str = "rest"
- url: str
- api_key: Optional[str] # required for protected APIs
- document_template: Optional[str] = None
- document_template_max_bytes: Optional[int] = None # Default to 400
- request: Dict[str, Any]
- response: Dict[str, Any]
- distribution: Optional[EmbedderDistribution] = None
- headers: Optional[Dict[str, Any]] = None
- binary_quantized: Optional[bool] = None
-
-
-class UserProvidedEmbedder(CamelBase):
- source: str = "userProvided"
- dimensions: int
- distribution: Optional[EmbedderDistribution] = None
- binary_quantized: Optional[bool] = None
-
-
-class Embedders(CamelBase):
- embedders: Dict[
- str,
- Union[
- OpenAiEmbedder, HuggingFaceEmbedder, OllamaEmbedder, RestEmbedder, UserProvidedEmbedder
- ],
- ]
diff --git a/tests/conftest.py b/tests/conftest.py
index 7a814ff7..54e53e4e 100644
--- a/tests/conftest.py
+++ b/tests/conftest.py
@@ -7,7 +7,7 @@
import meilisearch
from meilisearch.errors import MeilisearchApiError
-from meilisearch.models.index import OpenAiEmbedder, UserProvidedEmbedder
+from meilisearch.models.embedders import OpenAiEmbedder, UserProvidedEmbedder
from tests import common
diff --git a/tests/index/test_index_search_meilisearch.py b/tests/index/test_index_search_meilisearch.py
index 8939e9c5..8b087d1c 100644
--- a/tests/index/test_index_search_meilisearch.py
+++ b/tests/index/test_index_search_meilisearch.py
@@ -503,11 +503,28 @@ def test_show_ranking_score(index_with_documents):
def test_vector_search(index_with_documents_and_vectors):
+ """Tests vector search with hybrid parameters."""
response = index_with_documents_and_vectors().search(
"",
opt_params={"vector": [0.1, 0.2], "hybrid": {"semanticRatio": 1.0, "embedder": "default"}},
)
assert len(response["hits"]) > 0
+ # Check that semanticHitCount field is present in the response
+ assert "semanticHitCount" in response
+ # With semanticRatio = 1.0, all hits should be semantic
+ assert response["semanticHitCount"] == len(response["hits"])
+
+
+def test_hybrid_search(index_with_documents_and_vectors):
+ """Tests hybrid search with semantic ratio and embedder."""
+ response = index_with_documents_and_vectors().search(
+ "movie", opt_params={"hybrid": {"semanticRatio": 0.5, "embedder": "default"}}
+ )
+ assert len(response["hits"]) > 0
+ # Check that semanticHitCount field is present in the response
+ assert "semanticHitCount" in response
+ # semanticHitCount should be an integer
+ assert isinstance(response["semanticHitCount"], int)
def test_search_distinct(index_with_documents):
@@ -534,3 +551,70 @@ def test_search_ranking_threshold(query, ranking_score_threshold, expected, inde
query, {"rankingScoreThreshold": ranking_score_threshold}
)
assert len(response["hits"]) == expected
+
+
+def test_vector_search_with_retrieve_vectors(index_with_documents_and_vectors):
+ """Tests vector search with retrieveVectors parameter."""
+ response = index_with_documents_and_vectors().search(
+ "",
+ opt_params={
+ "vector": [0.1, 0.2],
+ "retrieveVectors": True,
+ "hybrid": {"semanticRatio": 1.0, "embedder": "default"},
+ },
+ )
+ assert len(response["hits"]) > 0
+ # Check that the first hit has a _vectors field
+ assert "_vectors" in response["hits"][0]
+ # Check that the _vectors field contains the default embedder
+ assert "default" in response["hits"][0]["_vectors"]
+
+
+def test_get_similar_documents_with_identical_vectors(empty_index):
+ """Tests get_similar_documents method with documents having identical vectors."""
+ # Create documents with identical vector embeddings
+ identical_vector = [0.5, 0.5]
+ documents = [
+ {"id": "doc1", "title": "Document 1", "_vectors": {"default": identical_vector}},
+ {"id": "doc2", "title": "Document 2", "_vectors": {"default": identical_vector}},
+ {"id": "doc3", "title": "Document 3", "_vectors": {"default": identical_vector}},
+ # Add a document with a different vector to verify it's not returned first
+ {"id": "doc4", "title": "Document 4", "_vectors": {"default": [0.1, 0.1]}},
+ ]
+
+ # Set up the index with the documents
+ index = empty_index()
+
+ # Configure the embedder
+ settings_update_task = index.update_embedders(
+ {
+ "default": {
+ "source": "userProvided",
+ "dimensions": 2,
+ }
+ }
+ )
+ index.wait_for_task(settings_update_task.task_uid)
+
+ # Add the documents
+ document_addition_task = index.add_documents(documents)
+ index.wait_for_task(document_addition_task.task_uid)
+
+ # Test get_similar_documents with doc1
+ response = index.get_similar_documents({"id": "doc1", "embedder": "default"})
+
+ # Verify response structure
+ assert isinstance(response, dict)
+ assert "hits" in response
+ assert len(response["hits"]) >= 2 # Should find at least doc2 and doc3
+ assert "id" in response
+ assert response["id"] == "doc1"
+
+ # Verify that doc2 and doc3 are in the results (they have identical vectors to doc1)
+ result_ids = [hit["id"] for hit in response["hits"]]
+ assert "doc2" in result_ids
+ assert "doc3" in result_ids
+
+ # Verify that doc4 is not the first result (it has a different vector)
+ if "doc4" in result_ids:
+ assert result_ids[0] != "doc4"
diff --git a/tests/settings/test_settings.py b/tests/settings/test_settings.py
index 8bbdafa2..147001de 100644
--- a/tests/settings/test_settings.py
+++ b/tests/settings/test_settings.py
@@ -1,7 +1,7 @@
# pylint: disable=redefined-outer-name
import pytest
-from meilisearch.models.index import OpenAiEmbedder, UserProvidedEmbedder
+from meilisearch.models.embedders import OpenAiEmbedder, UserProvidedEmbedder
@pytest.fixture
diff --git a/tests/settings/test_settings_embedders.py b/tests/settings/test_settings_embedders.py
index f932bbae..5baf2e09 100644
--- a/tests/settings/test_settings_embedders.py
+++ b/tests/settings/test_settings_embedders.py
@@ -1,6 +1,6 @@
# pylint: disable=redefined-outer-name
-from meilisearch.models.index import OpenAiEmbedder, UserProvidedEmbedder
+from meilisearch.models.embedders import OpenAiEmbedder, UserProvidedEmbedder
def test_get_default_embedders(empty_index):
@@ -42,3 +42,144 @@ def test_reset_embedders(new_embedders, empty_index):
assert isinstance(response_get.embedders["open_ai"], OpenAiEmbedder)
response_last = index.get_embedders()
assert response_last is None
+
+
+def test_openai_embedder_format(empty_index):
+ """Tests that OpenAi embedder has the required fields and proper format."""
+ index = empty_index()
+
+ openai_embedder = {
+ "openai": {
+ "source": "openAi",
+ "apiKey": "test-key",
+ "model": "text-embedding-3-small",
+ "dimensions": 1536,
+ "documentTemplateMaxBytes": 400,
+ "distribution": {"mean": 0.5, "sigma": 0.1},
+ "binaryQuantized": False,
+ }
+ }
+ response = index.update_embedders(openai_embedder)
+ index.wait_for_task(response.task_uid)
+ embedders = index.get_embedders()
+ assert embedders.embedders["openai"].source == "openAi"
+ assert embedders.embedders["openai"].model == "text-embedding-3-small"
+ assert embedders.embedders["openai"].dimensions == 1536
+ assert hasattr(embedders.embedders["openai"], "document_template")
+ assert embedders.embedders["openai"].document_template_max_bytes == 400
+ assert embedders.embedders["openai"].distribution.mean == 0.5
+ assert embedders.embedders["openai"].distribution.sigma == 0.1
+ assert embedders.embedders["openai"].binary_quantized is False
+
+
+def test_huggingface_embedder_format(empty_index):
+ """Tests that HuggingFace embedder has the required fields and proper format."""
+ index = empty_index()
+
+ huggingface_embedder = {
+ "huggingface": {
+ "source": "huggingFace",
+ "model": "BAAI/bge-base-en-v1.5",
+ "revision": "main",
+ "documentTemplateMaxBytes": 400,
+ "distribution": {"mean": 0.5, "sigma": 0.1},
+ "binaryQuantized": False,
+ }
+ }
+ response = index.update_embedders(huggingface_embedder)
+ index.wait_for_task(response.task_uid)
+ embedders = index.get_embedders()
+ assert embedders.embedders["huggingface"].source == "huggingFace"
+ assert embedders.embedders["huggingface"].model == "BAAI/bge-base-en-v1.5"
+ assert embedders.embedders["huggingface"].revision == "main"
+ assert hasattr(embedders.embedders["huggingface"], "document_template")
+ assert embedders.embedders["huggingface"].document_template_max_bytes == 400
+ assert embedders.embedders["huggingface"].distribution.mean == 0.5
+ assert embedders.embedders["huggingface"].distribution.sigma == 0.1
+ assert embedders.embedders["huggingface"].binary_quantized is False
+
+
+def test_ollama_embedder_format(empty_index):
+ """Tests that Ollama embedder has the required fields and proper format."""
+ index = empty_index()
+
+ ollama_embedder = {
+ "ollama": {
+ "source": "ollama",
+ "url": "http://localhost:11434/api/embeddings",
+ "apiKey": "test-key",
+ "model": "llama2",
+ "dimensions": 4096,
+ "documentTemplateMaxBytes": 400,
+ "distribution": {"mean": 0.5, "sigma": 0.1},
+ "binaryQuantized": False,
+ }
+ }
+ response = index.update_embedders(ollama_embedder)
+ index.wait_for_task(response.task_uid)
+ embedders = index.get_embedders()
+ assert embedders.embedders["ollama"].source == "ollama"
+ assert embedders.embedders["ollama"].url == "http://localhost:11434/api/embeddings"
+ assert embedders.embedders["ollama"].model == "llama2"
+ assert embedders.embedders["ollama"].dimensions == 4096
+ assert hasattr(embedders.embedders["ollama"], "document_template")
+ assert embedders.embedders["ollama"].document_template_max_bytes == 400
+ assert embedders.embedders["ollama"].distribution.mean == 0.5
+ assert embedders.embedders["ollama"].distribution.sigma == 0.1
+ assert embedders.embedders["ollama"].binary_quantized is False
+
+
+def test_rest_embedder_format(empty_index):
+ """Tests that Rest embedder has the required fields and proper format."""
+ index = empty_index()
+
+ rest_embedder = {
+ "rest": {
+ "source": "rest",
+ "url": "http://localhost:8000/embed",
+ "apiKey": "test-key",
+ "dimensions": 512,
+ "documentTemplateMaxBytes": 400,
+ "request": {"model": "MODEL_NAME", "input": "{{text}}"},
+ "response": {"result": {"data": ["{{embedding}}"]}},
+ "headers": {"Authorization": "Bearer test-key"},
+ "distribution": {"mean": 0.5, "sigma": 0.1},
+ "binaryQuantized": False,
+ }
+ }
+ response = index.update_embedders(rest_embedder)
+ index.wait_for_task(response.task_uid)
+ embedders = index.get_embedders()
+ assert embedders.embedders["rest"].source == "rest"
+ assert embedders.embedders["rest"].url == "http://localhost:8000/embed"
+ assert embedders.embedders["rest"].dimensions == 512
+ assert hasattr(embedders.embedders["rest"], "document_template")
+ assert embedders.embedders["rest"].document_template_max_bytes == 400
+ assert embedders.embedders["rest"].request == {"model": "MODEL_NAME", "input": "{{text}}"}
+ assert embedders.embedders["rest"].response == {"result": {"data": ["{{embedding}}"]}}
+ assert embedders.embedders["rest"].headers == {"Authorization": "Bearer test-key"}
+ assert embedders.embedders["rest"].distribution.mean == 0.5
+ assert embedders.embedders["rest"].distribution.sigma == 0.1
+ assert embedders.embedders["rest"].binary_quantized is False
+
+
+def test_user_provided_embedder_format(empty_index):
+ """Tests that UserProvided embedder has the required fields and proper format."""
+ index = empty_index()
+
+ user_provided_embedder = {
+ "user_provided": {
+ "source": "userProvided",
+ "dimensions": 512,
+ "distribution": {"mean": 0.5, "sigma": 0.1},
+ "binaryQuantized": False,
+ }
+ }
+ response = index.update_embedders(user_provided_embedder)
+ index.wait_for_task(response.task_uid)
+ embedders = index.get_embedders()
+ assert embedders.embedders["user_provided"].source == "userProvided"
+ assert embedders.embedders["user_provided"].dimensions == 512
+ assert embedders.embedders["user_provided"].distribution.mean == 0.5
+ assert embedders.embedders["user_provided"].distribution.sigma == 0.1
+ assert embedders.embedders["user_provided"].binary_quantized is False