my-project-public

repository

loading code, commits, and activity

repositories

loading repo index

#1	---
#2	title: '🔍 search'
#3	---
#4
#5	`.search()` enables you to uncover the most pertinent context by performing a semantic search across your data sources based on a given query. Refer to the function signature below:
#6
#7	### Parameters
#8
#9	<ParamField path="query" type="str">
#10	Question
#11	</ParamField>
#12	<ParamField path="num_documents" type="int" optional>
#13	Number of relevant documents to fetch. Defaults to `3`
#14	</ParamField>
#15	<ParamField path="where" type="dict" optional>
#16	Key value pair for metadata filtering.
#17	</ParamField>
#18	<ParamField path="raw_filter" type="dict" optional>
#19	Pass raw filter query based on your vector database.
#20	Currently, `raw_filter` param is only supported for Pinecone vector database.
#21	</ParamField>
#22
#23	### Returns
#24
#25	<ResponseField name="answer" type="dict">
#26	Return list of dictionaries that contain the relevant chunk and their source information.
#27	</ResponseField>
#28
#29	## Usage
#30
#31	### Basic
#32
#33	Refer to the following example on how to use the search api:
#34
#35	```python Code example
#36	from embedchain import App
#37
#38	app = App()
#39	app.add("https://www.forbes.com/profile/elon-musk")
#40
#41	context = app.search("What is the net worth of Elon?", num_documents=2)
#42	print(context)
#43	```
#44
#45	### Advanced
#46
#47	#### Metadata filtering using `where` params
#48
#49	Here is an advanced example of `search()` API with metadata filtering on pinecone database:
#50
#51	```python
#52	import os
#53
#54	from embedchain import App
#55
#56	os.environ["PINECONE_API_KEY"] = "xxx"
#57
#58	config = {
#59	"vectordb": {
#60	"provider": "pinecone",
#61	"config": {
#62	"metric": "dotproduct",
#63	"vector_dimension": 1536,
#64	"index_name": "ec-test",
#65	"serverless_config": {"cloud": "aws", "region": "us-west-2"},
#66	},
#67	}
#68	}
#69
#70	app = App.from_config(config=config)
#71
#72	app.add("https://www.forbes.com/profile/bill-gates", metadata={"type": "forbes", "person": "gates"})
#73	app.add("https://en.wikipedia.org/wiki/Bill_Gates", metadata={"type": "wiki", "person": "gates"})
#74
#75	results = app.search("What is the net worth of Bill Gates?", where={"person": "gates"})
#76	print("Num of search results: ", len(results))
#77	```
#78
#79	#### Metadata filtering using `raw_filter` params
#80
#81	Following is an example of metadata filtering by passing the raw filter query that pinecone vector database follows:
#82
#83	```python
#84	import os
#85
#86	from embedchain import App
#87
#88	os.environ["PINECONE_API_KEY"] = "xxx"
#89
#90	config = {
#91	"vectordb": {
#92	"provider": "pinecone",
#93	"config": {
#94	"metric": "dotproduct",
#95	"vector_dimension": 1536,
#96	"index_name": "ec-test",
#97	"serverless_config": {"cloud": "aws", "region": "us-west-2"},
#98	},
#99	}
#100	}
#101
#102	app = App.from_config(config=config)
#103
#104	app.add("https://www.forbes.com/profile/bill-gates", metadata={"year": 2022, "person": "gates"})
#105	app.add("https://en.wikipedia.org/wiki/Bill_Gates", metadata={"year": 2024, "person": "gates"})
#106
#107	print("Filter with person: gates and year > 2023")
#108	raw_filter = {"$and": [{"person": "gates"}, {"year": {"$gt": 2023}}]}
#109	results = app.search("What is the net worth of Bill Gates?", raw_filter=raw_filter)
#110	print("Num of search results: ", len(results))
#111	```
#112

z6Mkq5mY3JWtxoxUobWcfNHm7AkRubgSWEZTkBVqZXJviFZ5/my-project-public