Spaces:
Runtime error
Runtime error
File size: 11,032 Bytes
105b369 |
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 |
from typing import Optional, Dict, Union, List
try:
from pinecone import Pinecone
from pinecone.config import Config
except ImportError:
raise ImportError(
"The `pinecone-client` package is not installed, please install using `pip install pinecone-client`."
)
from phi.document import Document
from phi.embedder import Embedder
from phi.vectordb.base import VectorDb
from phi.utils.log import logger
from pinecone.core.client.api.manage_indexes_api import ManageIndexesApi
from pinecone.models import ServerlessSpec, PodSpec
from pinecone.core.client.models import Vector
class PineconeDB(VectorDb):
"""A class representing a Pinecone database.
Args:
name (str): The name of the index.
dimension (int): The dimension of the embeddings.
spec (Union[Dict, ServerlessSpec, PodSpec]): The index spec.
metric (Optional[str], optional): The metric used for similarity search. Defaults to "cosine".
additional_headers (Optional[Dict[str, str]], optional): Additional headers to pass to the Pinecone client. Defaults to {}.
pool_threads (Optional[int], optional): The number of threads to use for the Pinecone client. Defaults to 1.
timeout (Optional[int], optional): The timeout for Pinecone operations. Defaults to None.
index_api (Optional[ManageIndexesApi], optional): The Index API object. Defaults to None.
api_key (Optional[str], optional): The Pinecone API key. Defaults to None.
host (Optional[str], optional): The Pinecone host. Defaults to None.
config (Optional[Config], optional): The Pinecone config. Defaults to None.
**kwargs: Additional keyword arguments.
Attributes:
client (Pinecone): The Pinecone client.
index: The Pinecone index.
api_key (Optional[str]): The Pinecone API key.
host (Optional[str]): The Pinecone host.
config (Optional[Config]): The Pinecone config.
additional_headers (Optional[Dict[str, str]]): Additional headers to pass to the Pinecone client.
pool_threads (Optional[int]): The number of threads to use for the Pinecone client.
index_api (Optional[ManageIndexesApi]): The Index API object.
name (str): The name of the index.
dimension (int): The dimension of the embeddings.
spec (Union[Dict, ServerlessSpec, PodSpec]): The index spec.
metric (Optional[str]): The metric used for similarity search.
timeout (Optional[int]): The timeout for Pinecone operations.
kwargs (Optional[Dict[str, str]]): Additional keyword arguments.
"""
def __init__(
self,
name: str,
dimension: int,
spec: Union[Dict, ServerlessSpec, PodSpec],
embedder: Optional[Embedder] = None,
metric: Optional[str] = "cosine",
additional_headers: Optional[Dict[str, str]] = None,
pool_threads: Optional[int] = 1,
namespace: Optional[str] = None,
timeout: Optional[int] = None,
index_api: Optional[ManageIndexesApi] = None,
api_key: Optional[str] = None,
host: Optional[str] = None,
config: Optional[Config] = None,
**kwargs,
):
self._client = None
self._index = None
self.api_key: Optional[str] = api_key
self.host: Optional[str] = host
self.config: Optional[Config] = config
self.additional_headers: Dict[str, str] = additional_headers or {}
self.pool_threads: Optional[int] = pool_threads
self.namespace: Optional[str] = namespace
self.index_api: Optional[ManageIndexesApi] = index_api
self.name: str = name
self.dimension: int = dimension
self.spec: Union[Dict, ServerlessSpec, PodSpec] = spec
self.metric: Optional[str] = metric
self.timeout: Optional[int] = timeout
self.kwargs: Optional[Dict[str, str]] = kwargs
# Embedder for embedding the document contents
_embedder = embedder
if _embedder is None:
from phi.embedder.openai import OpenAIEmbedder
_embedder = OpenAIEmbedder()
self.embedder: Embedder = _embedder
@property
def client(self) -> Pinecone:
"""The Pinecone client.
Returns:
Pinecone: The Pinecone client.
"""
if self._client is None:
logger.debug("Creating Pinecone Client")
self._client = Pinecone(
api_key=self.api_key,
host=self.host,
config=self.config,
additional_headers=self.additional_headers,
pool_threads=self.pool_threads,
index_api=self.index_api,
**self.kwargs,
)
return self._client
@property
def index(self):
"""The Pinecone index.
Returns:
Pinecone.Index: The Pinecone index.
"""
if self._index is None:
logger.debug(f"Connecting to Pinecone Index: {self.name}")
self._index = self.client.Index(self.name)
return self._index
def exists(self) -> bool:
"""Check if the index exists.
Returns:
bool: True if the index exists, False otherwise.
"""
list_indexes = self.client.list_indexes()
return self.name in list_indexes.names()
def create(self) -> None:
"""Create the index if it does not exist."""
if not self.exists():
logger.debug(f"Creating index: {self.name}")
self.client.create_index(
name=self.name,
dimension=self.dimension,
spec=self.spec,
metric=self.metric if self.metric is not None else "cosine",
timeout=self.timeout,
)
def delete(self) -> None:
"""Delete the index if it exists."""
if self.exists():
logger.debug(f"Deleting index: {self.name}")
self.client.delete_index(name=self.name, timeout=self.timeout)
def doc_exists(self, document: Document) -> bool:
"""Check if a document exists in the index.
Args:
document (Document): The document to check.
Returns:
bool: True if the document exists, False otherwise.
"""
response = self.index.fetch(ids=[document.id])
return len(response.vectors) > 0
def name_exists(self, name: str) -> bool:
"""Check if an index with the given name exists.
Args:
name (str): The name of the index.
Returns:
bool: True if the index exists, False otherwise.
"""
try:
self.client.describe_index(name)
return True
except Exception:
return False
def upsert(
self,
documents: List[Document],
namespace: Optional[str] = None,
batch_size: Optional[int] = None,
show_progress: bool = False,
) -> None:
"""insert documents into the index.
Args:
documents (List[Document]): The documents to upsert.
namespace (Optional[str], optional): The namespace for the documents. Defaults to None.
batch_size (Optional[int], optional): The batch size for upsert. Defaults to None.
show_progress (bool, optional): Whether to show progress during upsert. Defaults to False.
"""
vectors = []
for document in documents:
document.embed(embedder=self.embedder)
document.meta_data["text"] = document.content
vectors.append(
Vector(
id=document.id,
values=document.embedding,
metadata=document.meta_data,
)
)
self.index.upsert(
vectors=vectors,
namespace=namespace,
batch_size=batch_size,
show_progress=show_progress,
)
def upsert_available(self) -> bool:
"""Check if upsert operation is available.
Returns:
bool: True if upsert is available, False otherwise.
"""
return True
def insert(self, documents: List[Document]) -> None:
"""Insert documents into the index.
This method is not supported by Pinecone. Use `upsert` instead.
Args:
documents (List[Document]): The documents to insert.
Raises:
NotImplementedError: This method is not supported by Pinecone.
"""
raise NotImplementedError("Pinecone does not support insert operations. Use upsert instead.")
def search(
self,
query: str,
limit: int = 5,
namespace: Optional[str] = None,
filter: Optional[Dict[str, Union[str, float, int, bool, List, dict]]] = None,
include_values: Optional[bool] = None,
) -> List[Document]:
"""Search for similar documents in the index.
Args:
query (str): The query to search for.
limit (int, optional): The maximum number of results to return. Defaults to 5.
namespace (Optional[str], optional): The namespace to search in. Defaults to None.
filter (Optional[Dict[str, Union[str, float, int, bool, List, dict]]], optional): The filter for the search. Defaults to None.
include_values (Optional[bool], optional): Whether to include values in the search results. Defaults to None.
include_metadata (Optional[bool], optional): Whether to include metadata in the search results. Defaults to None.
Returns:
List[Document]: The list of matching documents.
"""
query_embedding = self.embedder.get_embedding(query)
if query_embedding is None:
logger.error(f"Error getting embedding for Query: {query}")
return []
response = self.index.query(
vector=query_embedding,
top_k=limit,
namespace=namespace,
filter=filter,
include_values=include_values,
include_metadata=True,
)
return [
Document(
content=(result.metadata.get("text", "") if result.metadata is not None else ""),
id=result.id,
embedding=result.values,
meta_data=result.metadata,
)
for result in response.matches
]
def optimize(self) -> None:
"""Optimize the index.
This method can be left empty as Pinecone automatically optimizes indexes.
"""
pass
def clear(self, namespace: Optional[str] = None) -> bool:
"""Clear the index.
Args:
namespace (Optional[str], optional): The namespace to clear. Defaults to None.
"""
try:
self.index.delete(delete_all=True, namespace=namespace)
return True
except Exception:
return False
|