File size: 15,170 Bytes
4a51346
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
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
from typing import TYPE_CHECKING, Optional, Tuple, cast, List
from pydantic import BaseModel, PrivateAttr
from uuid import UUID
import chromadb.utils.embedding_functions as ef

from chromadb.api.types import (
    CollectionMetadata,
    Embedding,
    Include,
    Metadata,
    Document,
    Where,
    IDs,
    EmbeddingFunction,
    GetResult,
    QueryResult,
    ID,
    OneOrMany,
    WhereDocument,
    maybe_cast_one_to_many,
    validate_ids,
    validate_include,
    validate_metadatas,
    validate_where,
    validate_where_document,
    validate_n_results,
    validate_embeddings,
)
import logging

logger = logging.getLogger(__name__)

if TYPE_CHECKING:
    from chromadb.api import API


class Collection(BaseModel):
    name: str
    id: UUID
    metadata: Optional[CollectionMetadata] = None
    _client: "API" = PrivateAttr()
    _embedding_function: Optional[EmbeddingFunction] = PrivateAttr()

    def __init__(
        self,
        client: "API",
        name: str,
        id: UUID,
        embedding_function: Optional[EmbeddingFunction] = ef.DefaultEmbeddingFunction(),
        metadata: Optional[CollectionMetadata] = None,
    ):
        self._client = client
        self._embedding_function = embedding_function
        super().__init__(name=name, metadata=metadata, id=id)

    def __repr__(self) -> str:
        return f"Collection(name={self.name})"

    def count(self) -> int:
        """The total number of embeddings added to the database

        Returns:
            int: The total number of embeddings added to the database

        """
        return self._client._count(collection_id=self.id)

    def add(
        self,
        ids: OneOrMany[ID],
        embeddings: Optional[OneOrMany[Embedding]] = None,
        metadatas: Optional[OneOrMany[Metadata]] = None,
        documents: Optional[OneOrMany[Document]] = None,
        increment_index: bool = True,
    ) -> None:
        """Add embeddings to the data store.
        Args:
            ids: The ids of the embeddings you wish to add
            embedding: The embeddings to add. If None, embeddings will be computed based on the documents using the embedding_function set for the Collection. Optional.
            metadata: The metadata to associate with the embeddings. When querying, you can filter on this metadata. Optional.
            documents: The documents to associate with the embeddings. Optional.
            ids: The ids to associate with the embeddings. Optional.

        Returns:
            None

        Raises:
            ValueError: If you don't provide either embeddings or documents
            ValueError: If the length of ids, embeddings, metadatas, or documents don't match
            ValueError: If you don't provide an embedding function and don't provide embeddings
            ValueError: If you provide both embeddings and documents
            ValueError: If you provide an id that already exists

        """

        ids, embeddings, metadatas, documents = self._validate_embedding_set(
            ids, embeddings, metadatas, documents
        )

        self._client._add(
            ids, self.id, embeddings, metadatas, documents, increment_index
        )

    def get(
        self,
        ids: Optional[OneOrMany[ID]] = None,
        where: Optional[Where] = None,
        limit: Optional[int] = None,
        offset: Optional[int] = None,
        where_document: Optional[WhereDocument] = None,
        include: Include = ["metadatas", "documents"],
    ) -> GetResult:
        """Get embeddings and their associate data from the data store. If no ids or where filter is provided returns
        all embeddings up to limit starting at offset.

        Args:
            ids: The ids of the embeddings to get. Optional.
            where: A Where type dict used to filter results by. E.g. `{"color" : "red", "price": 4.20}`. Optional.
            limit: The number of documents to return. Optional.
            offset: The offset to start returning results from. Useful for paging results with limit. Optional.
            where_document: A WhereDocument type dict used to filter by the documents. E.g. `{$contains: {"text": "hello"}}`. Optional.
            include: A list of what to include in the results. Can contain `"embeddings"`, `"metadatas"`, `"documents"`. Ids are always included. Defaults to `["metadatas", "documents"]`. Optional.

        Returns:
            GetResult: A GetResult object containing the results.

        """
        where = validate_where(where) if where else None
        where_document = (
            validate_where_document(where_document) if where_document else None
        )
        ids = validate_ids(maybe_cast_one_to_many(ids)) if ids else None
        include = validate_include(include, allow_distances=False)
        return self._client._get(
            self.id,
            ids,
            where,
            None,
            limit,
            offset,
            where_document=where_document,
            include=include,
        )

    def peek(self, limit: int = 10) -> GetResult:
        """Get the first few results in the database up to limit

        Args:
            limit: The number of results to return.

        Returns:
            GetResult: A GetResult object containing the results.
        """
        return self._client._peek(self.id, limit)

    def query(
        self,
        query_embeddings: Optional[OneOrMany[Embedding]] = None,
        query_texts: Optional[OneOrMany[Document]] = None,
        n_results: int = 10,
        where: Optional[Where] = None,
        where_document: Optional[WhereDocument] = None,
        include: Include = ["metadatas", "documents", "distances"],
    ) -> QueryResult:
        """Get the n_results nearest neighbor embeddings for provided query_embeddings or query_texts.

        Args:
            query_embeddings: The embeddings to get the closes neighbors of. Optional.
            query_texts: The document texts to get the closes neighbors of. Optional.
            n_results: The number of neighbors to return for each query_embedding or query_texts. Optional.
            where: A Where type dict used to filter results by. E.g. `{"color" : "red", "price": 4.20}`. Optional.
            where_document: A WhereDocument type dict used to filter by the documents. E.g. `{$contains: {"text": "hello"}}`. Optional.
            include: A list of what to include in the results. Can contain `"embeddings"`, `"metadatas"`, `"documents"`, `"distances"`. Ids are always included. Defaults to `["metadatas", "documents", "distances"]`. Optional.

        Returns:
            QueryResult: A QueryResult object containing the results.

        Raises:
            ValueError: If you don't provide either query_embeddings or query_texts
            ValueError: If you provide both query_embeddings and query_texts

        """
        where = validate_where(where) if where else None
        where_document = (
            validate_where_document(where_document) if where_document else None
        )
        query_embeddings = (
            validate_embeddings(maybe_cast_one_to_many(query_embeddings))
            if query_embeddings is not None
            else None
        )
        query_texts = (
            maybe_cast_one_to_many(query_texts) if query_texts is not None else None
        )
        include = validate_include(include, allow_distances=True)
        n_results = validate_n_results(n_results)

        # If neither query_embeddings nor query_texts are provided, or both are provided, raise an error
        if (query_embeddings is None and query_texts is None) or (
            query_embeddings is not None and query_texts is not None
        ):
            raise ValueError(
                "You must provide either query embeddings or query texts, but not both"
            )

        # If query_embeddings are not provided, we need to compute them from the query_texts
        if query_embeddings is None:
            if self._embedding_function is None:
                raise ValueError(
                    "You must provide embeddings or a function to compute them"
                )
            # We know query texts is not None at this point, cast for the typechecker
            query_embeddings = self._embedding_function(
                cast(List[Document], query_texts)
            )

        if where is None:
            where = {}

        if where_document is None:
            where_document = {}

        return self._client._query(
            collection_id=self.id,
            query_embeddings=query_embeddings,
            n_results=n_results,
            where=where,
            where_document=where_document,
            include=include,
        )

    def modify(
        self, name: Optional[str] = None, metadata: Optional[CollectionMetadata] = None
    ) -> None:
        """Modify the collection name or metadata

        Args:
            name: The updated name for the collection. Optional.
            metadata: The updated metadata for the collection. Optional.

        Returns:
            None
        """
        self._client._modify(id=self.id, new_name=name, new_metadata=metadata)
        if name:
            self.name = name
        if metadata:
            self.metadata = metadata

    def update(
        self,
        ids: OneOrMany[ID],
        embeddings: Optional[OneOrMany[Embedding]] = None,
        metadatas: Optional[OneOrMany[Metadata]] = None,
        documents: Optional[OneOrMany[Document]] = None,
    ) -> None:
        """Update the embeddings, metadatas or documents for provided ids.

        Args:
            ids: The ids of the embeddings to update
            embeddings: The embeddings to add. If None, embeddings will be computed based on the documents using the embedding_function set for the Collection. Optional.
            metadatas:  The metadata to associate with the embeddings. When querying, you can filter on this metadata. Optional.
            documents: The documents to associate with the embeddings. Optional.

        Returns:
            None
        """

        ids, embeddings, metadatas, documents = self._validate_embedding_set(
            ids, embeddings, metadatas, documents, require_embeddings_or_documents=False
        )

        self._client._update(self.id, ids, embeddings, metadatas, documents)

    def upsert(
        self,
        ids: OneOrMany[ID],
        embeddings: Optional[OneOrMany[Embedding]] = None,
        metadatas: Optional[OneOrMany[Metadata]] = None,
        documents: Optional[OneOrMany[Document]] = None,
        increment_index: bool = True,
    ) -> None:
        """Update the embeddings, metadatas or documents for provided ids, or create them if they don't exist.

        Args:
            ids: The ids of the embeddings to update
            embeddings: The embeddings to add. If None, embeddings will be computed based on the documents using the embedding_function set for the Collection. Optional.
            metadatas:  The metadata to associate with the embeddings. When querying, you can filter on this metadata. Optional.
            documents: The documents to associate with the embeddings. Optional.

        Returns:
            None
        """

        ids, embeddings, metadatas, documents = self._validate_embedding_set(
            ids, embeddings, metadatas, documents
        )

        self._client._upsert(
            collection_id=self.id,
            ids=ids,
            embeddings=embeddings,
            metadatas=metadatas,
            documents=documents,
            increment_index=increment_index,
        )

    def delete(
        self,
        ids: Optional[IDs] = None,
        where: Optional[Where] = None,
        where_document: Optional[WhereDocument] = None,
    ) -> None:
        """Delete the embeddings based on ids and/or a where filter

        Args:
            ids: The ids of the embeddings to delete
            where: A Where type dict used to filter the delection by. E.g. `{"color" : "red", "price": 4.20}`. Optional.
            where_document: A WhereDocument type dict used to filter the deletion by the document content. E.g. `{$contains: {"text": "hello"}}`. Optional.

        Returns:
            None
        """
        ids = validate_ids(maybe_cast_one_to_many(ids)) if ids else None
        where = validate_where(where) if where else None
        where_document = (
            validate_where_document(where_document) if where_document else None
        )
        self._client._delete(self.id, ids, where, where_document)

    def create_index(self) -> None:
        self._client.create_index(self.name)

    def _validate_embedding_set(
        self,
        ids: OneOrMany[ID],
        embeddings: Optional[OneOrMany[Embedding]],
        metadatas: Optional[OneOrMany[Metadata]],
        documents: Optional[OneOrMany[Document]],
        require_embeddings_or_documents: bool = True,
    ) -> Tuple[
        IDs,
        List[Embedding],
        Optional[List[Metadata]],
        Optional[List[Document]],
    ]:
        ids = validate_ids(maybe_cast_one_to_many(ids))
        embeddings = (
            validate_embeddings(maybe_cast_one_to_many(embeddings))
            if embeddings is not None
            else None
        )
        metadatas = (
            validate_metadatas(maybe_cast_one_to_many(metadatas))
            if metadatas is not None
            else None
        )
        documents = maybe_cast_one_to_many(documents) if documents is not None else None

        # Check that one of embeddings or documents is provided
        if require_embeddings_or_documents:
            if embeddings is None and documents is None:
                raise ValueError(
                    "You must provide either embeddings or documents, or both"
                )

        # Check that, if they're provided, the lengths of the arrays match the length of ids
        if embeddings is not None and len(embeddings) != len(ids):
            raise ValueError(
                f"Number of embeddings {len(embeddings)} must match number of ids {len(ids)}"
            )
        if metadatas is not None and len(metadatas) != len(ids):
            raise ValueError(
                f"Number of metadatas {len(metadatas)} must match number of ids {len(ids)}"
            )
        if documents is not None and len(documents) != len(ids):
            raise ValueError(
                f"Number of documents {len(documents)} must match number of ids {len(ids)}"
            )

        # If document embeddings are not provided, we need to compute them
        if embeddings is None and documents is not None:
            if self._embedding_function is None:
                raise ValueError(
                    "You must provide embeddings or a function to compute them"
                )
            embeddings = self._embedding_function(documents)

        # if embeddings is None:
        #     raise ValueError(
        #         "Something went wrong. Embeddings should be computed at this point"
        #     )

        return ids, embeddings, metadatas, documents  # type: ignore