42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397 | class MongoDBAtlasVectorSearch(BasePydanticVectorStore):
"""MongoDB Atlas Vector Store.
To use, you should have both:
- the ``pymongo`` python package installed
- a connection string associated with a MongoDB Atlas Cluster
that has an Atlas Vector Search index
To get started head over to the [Atlas quick start](https://www.mongodb.com/docs/atlas/getting-started/).
Once your store is created, be sure to enable indexing in the Atlas GUI.
Please refer to the [documentation](https://www.mongodb.com/docs/atlas/atlas-vector-search/create-index/)
to get more details on how to define an Atlas Vector Search index. You can name the index {ATLAS_VECTOR_SEARCH_INDEX_NAME}
and create the index on the namespace {DB_NAME}.{COLLECTION_NAME}.
Finally, write the following definition in the JSON editor on MongoDB Atlas:
```
{
"name": "vector_index",
"type": "vectorSearch",
"fields":[
{
"type": "vector",
"path": "embedding",
"numDimensions": 1536,
"similarity": "cosine"
}
]
}
```
Examples:
`pip install llama-index-vector-stores-mongodb`
```python
import pymongo
from llama_index.vector_stores.mongodb import MongoDBAtlasVectorSearch
# Ensure you have the MongoDB URI with appropriate credentials
mongo_uri = "mongodb+srv://<username>:<password>@<host>?retryWrites=true&w=majority"
mongodb_client = pymongo.MongoClient(mongo_uri)
# Create an instance of MongoDBAtlasVectorSearch
vector_store = MongoDBAtlasVectorSearch(mongodb_client)
```
"""
stores_text: bool = True
flat_metadata: bool = False
_mongodb_client: Any = PrivateAttr()
_collection: Any = PrivateAttr()
_vector_index_name: str = PrivateAttr()
_embedding_key: str = PrivateAttr()
_id_key: str = PrivateAttr()
_text_key: str = PrivateAttr()
_metadata_key: str = PrivateAttr()
_fulltext_index_name: str = PrivateAttr()
_insert_kwargs: Dict = PrivateAttr()
_index_name: str = PrivateAttr() # DEPRECATED
_oversampling_factor: int = PrivateAttr()
def __init__(
self,
mongodb_client: Optional[Any] = None,
db_name: str = "default_db",
collection_name: str = "default_collection",
vector_index_name: str = "vector_index",
id_key: str = "_id",
embedding_key: str = "embedding",
text_key: str = "text",
metadata_key: str = "metadata",
fulltext_index_name: str = "fulltext_index",
index_name: str = None,
insert_kwargs: Optional[Dict] = None,
oversampling_factor: int = 10,
**kwargs: Any,
) -> None:
"""Initialize the vector store.
Args:
mongodb_client: A MongoDB client.
db_name: A MongoDB database name.
collection_name: A MongoDB collection name.
vector_index_name: A MongoDB Atlas *Vector* Search index name. ($vectorSearch)
id_key: The data field to use as the id.
embedding_key: A MongoDB field that will contain
the embedding for each document.
text_key: A MongoDB field that will contain the text for each document.
metadata_key: A MongoDB field that will contain
the metadata for each document.
insert_kwargs: The kwargs used during `insert`.
fulltext_index_name: A MongoDB Atlas *full-text* Search index name. ($search)
oversampling_factor: This times n_results is 'ef' in the HNSW algorithm.
'ef' determines the number of nearest neighbor candidates to consider during the search phase.
A higher value leads to more accuracy, but is slower. Default = 10
index_name: DEPRECATED: Please use vector_index_name.
"""
super().__init__()
if mongodb_client is not None:
self._mongodb_client = cast(MongoClient, mongodb_client)
else:
if "MONGODB_URI" not in os.environ:
raise ValueError(
"Must specify MONGODB_URI via env variable "
"if not directly passing in client."
)
self._mongodb_client = MongoClient(
os.environ["MONGODB_URI"],
driver=DriverInfo(name="llama-index", version=version("llama-index")),
)
if index_name is not None:
logger.warning("index_name is deprecated. Please use vector_index_name")
if vector_index_name is None:
vector_index_name = index_name
else:
logger.warning(
"vector_index_name and index_name both specified. Will use vector_index_name"
)
self._collection: Collection = self._mongodb_client[db_name][collection_name]
self._vector_index_name = vector_index_name
self._embedding_key = embedding_key
self._id_key = id_key
self._text_key = text_key
self._metadata_key = metadata_key
self._fulltext_index_name = fulltext_index_name
self._insert_kwargs = insert_kwargs or {}
self._oversampling_factor = oversampling_factor
def add(
self,
nodes: List[BaseNode],
**add_kwargs: Any,
) -> List[str]:
"""Add nodes to index.
Args:
nodes: List[BaseNode]: list of nodes with embeddings
Returns:
A List of ids for successfully added nodes.
"""
ids = []
data_to_insert = []
for node in nodes:
metadata = node_to_metadata_dict(
node, remove_text=True, flat_metadata=self.flat_metadata
)
entry = {
self._id_key: node.node_id,
self._embedding_key: node.get_embedding(),
self._text_key: node.get_content(metadata_mode=MetadataMode.NONE) or "",
self._metadata_key: metadata,
}
data_to_insert.append(entry)
ids.append(node.node_id)
logger.debug("Inserting data into MongoDB: %s", data_to_insert)
insert_result = self._collection.insert_many(
data_to_insert, **self._insert_kwargs
)
logger.debug("Result of insert: %s", insert_result)
return ids
def delete(self, ref_doc_id: str, **delete_kwargs: Any) -> None:
"""
Delete nodes using with ref_doc_id.
Args:
ref_doc_id (str): The doc_id of the document to delete.
"""
# delete by filtering on the doc_id metadata
self._collection.delete_many(
filter={self._metadata_key + ".ref_doc_id": ref_doc_id}, **delete_kwargs
)
@property
def client(self) -> Any:
"""Return MongoDB client."""
return self._mongodb_client
def _query(self, query: VectorStoreQuery) -> VectorStoreQueryResult:
hybrid_top_k = query.hybrid_top_k or query.similarity_top_k
sparse_top_k = query.sparse_top_k or query.similarity_top_k
dense_top_k = query.similarity_top_k
if query.mode == VectorStoreQueryMode.DEFAULT:
if not query.query_embedding:
raise ValueError("query_embedding in VectorStoreQueryMode.DEFAULT")
# Atlas Vector Search, potentially with filter
logger.debug(f"Running {query.mode} mode query pipeline")
filter = filters_to_mql(query.filters, metadata_key=self._metadata_key)
pipeline = [
vector_search_stage(
query_vector=query.query_embedding,
search_field=self._embedding_key,
index_name=self._vector_index_name,
limit=dense_top_k,
filter=filter,
oversampling_factor=self._oversampling_factor,
),
{"$set": {"score": {"$meta": "vectorSearchScore"}}},
]
elif query.mode == VectorStoreQueryMode.TEXT_SEARCH:
# Atlas Full-Text Search, potentially with filter
if not query.query_str:
raise ValueError("query_str in VectorStoreQueryMode.TEXT_SEARCH ")
logger.debug(f"Running {query.mode} mode query pipeline")
filter = filters_to_mql(query.filters, metadata_key=self._metadata_key)
pipeline = fulltext_search_stage(
query=query.query_str,
search_field=self._text_key,
index_name=self._fulltext_index_name,
operator="text",
filter=filter,
limit=sparse_top_k,
)
pipeline.append({"$set": {"score": {"$meta": "searchScore"}}})
elif query.mode == VectorStoreQueryMode.HYBRID:
# Combines Vector and Full-Text searches with Reciprocal Rank Fusion weighting
logger.debug(f"Running {query.mode} mode query pipeline")
scores_fields = ["vector_score", "fulltext_score"]
filter = filters_to_mql(query.filters, metadata_key=self._metadata_key)
pipeline = []
# Vector Search pipeline
if query.query_embedding:
vector_pipeline = [
vector_search_stage(
query_vector=query.query_embedding,
search_field=self._embedding_key,
index_name=self._vector_index_name,
limit=dense_top_k,
filter=filter,
oversampling_factor=self._oversampling_factor,
)
]
vector_pipeline.extend(reciprocal_rank_stage("vector_score"))
combine_pipelines(pipeline, vector_pipeline, self._collection.name)
# Full-Text Search pipeline
if query.query_str:
text_pipeline = fulltext_search_stage(
query=query.query_str,
search_field=self._text_key,
index_name=self._fulltext_index_name,
operator="text",
filter=filter,
limit=sparse_top_k,
)
text_pipeline.extend(reciprocal_rank_stage("fulltext_score"))
combine_pipelines(pipeline, text_pipeline, self._collection.name)
# Compute weighted sum and sort pipeline
alpha = (
query.alpha or 0.5
) # If no alpha is given, equal weighting is applied
pipeline += final_hybrid_stage(
scores_fields=scores_fields, limit=hybrid_top_k, alpha=alpha
)
# Remove embeddings unless requested.
if (
query.output_fields is None
or self._embedding_key not in query.output_fields
):
pipeline.append({"$project": {self._embedding_key: 0}})
else:
raise NotImplementedError(
f"{VectorStoreQueryMode.DEFAULT} (vector), "
f"{VectorStoreQueryMode.HYBRID} and {VectorStoreQueryMode.TEXT_SEARCH} "
f"are available. {query.mode} is not."
)
# Execution
logger.debug("Running query pipeline: %s", pipeline)
cursor = self._collection.aggregate(pipeline) # type: ignore
# Post-processing
top_k_nodes = []
top_k_ids = []
top_k_scores = []
for res in cursor:
text = res.pop(self._text_key)
score = res.pop("score")
id = res.pop(self._id_key)
metadata_dict = res.pop(self._metadata_key)
try:
node = metadata_dict_to_node(metadata_dict)
node.set_content(text)
except Exception:
# NOTE: deprecated legacy logic for backward compatibility
metadata, node_info, relationships = legacy_metadata_dict_to_node(
metadata_dict
)
node = TextNode(
text=text,
id_=id,
metadata=metadata,
start_char_idx=node_info.get("start", None),
end_char_idx=node_info.get("end", None),
relationships=relationships,
)
top_k_ids.append(id)
top_k_nodes.append(node)
top_k_scores.append(score)
result = VectorStoreQueryResult(
nodes=top_k_nodes, similarities=top_k_scores, ids=top_k_ids
)
logger.debug("Result of query: %s", result)
return result
def query(self, query: VectorStoreQuery, **kwargs: Any) -> VectorStoreQueryResult:
r"""Query index for top k most similar nodes.
The type of search to be performed is based on the VectorStoreQuery.mode.
Choose from DEFAULT (vector), HYBRID (hybrid), or TEXT_SEARCH (full-text).
When the mode is one of HYBRID or TEXT_SEARCH,
VectorStoreQuery.query_str is used for the full-text search.
See MongoDB Atlas documentation for full details on these.
For details on VectorStoreQueryMode.DEFAULT == 'default',
which does vector search, see:
https://www.mongodb.com/docs/atlas/atlas-vector-search/vector-search-stage/
For details on VectorStoreQueryMode.TEXT_SEARCH == "text_search",
which performs full-text search, see:
https://www.mongodb.com/docs/atlas/atlas-search/aggregation-stages/search/#mongodb-pipeline-pipe.-search
For details on VectorStoreQueryMode.HYBRID == "hybrid",
which combines the two with Reciprocal Rank Fusion, see the following.
https://www.mongodb.com/docs/atlas/atlas-vector-search/tutorials/reciprocal-rank-fusion/
In the scoring algorithm used, Reciprocal Rank Fusion,
scores := \frac{1}{rank + penalty} with rank in [1,2,..,n]
Args:
query: a VectorStoreQuery object.
Returns:
A VectorStoreQueryResult containing the results of the query.
"""
return self._query(query)
|
