Archives
/
langchain
mirror of https://github.com/hwchase17/langchain
2
0
Fork 1
Code Issues Packages Projects Releases Wiki Activity

Elasticsearch self-query retriever (#9248)

Browse Source 
Now with ElasticsearchStore VectorStore merged, i've added support for
the self-query retriever.

I've added a notebook also to demonstrate capability. I've also added
unit tests.

**Credit**
@elastic and @phoey1 on twitter.
pull/9254/head
Joseph McElroy 1 year ago committed by GitHub
parent 0a04e63811
commit 5e9687a196
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
4 changed files with 678 additions and 0 deletions
Show Stats Download Patch File Download Diff File

362
docs/extras/modules/data_connection/retrievers/self_query/elasticsearch_self_query.ipynb
Unescape Escape View File

@ -0,0 +1,362 @@
{
"cells": [
{
"cell_type": "markdown",
"id": "13afcae7",
"metadata": {},
"source": [
"# Elasticsearch self-querying "
]
},
{
"cell_type": "markdown",
"id": "68e75fb9",
"metadata": {},
"source": [
"## Creating a Elasticsearch vectorstore\n",
"First we'll want to create a Elasticsearch VectorStore and seed it with some data. We've created a small demo set of documents that contain summaries of movies.\n",
"\n",
"NOTE: The self-query retriever requires you to have `lark` installed (`pip install lark`). We also need the `elasticsearch` package."
]
},
{
"cell_type": "code",
"execution_count": 1,
"id": "63a8af5b",
"metadata": {
"tags": []
},
"outputs": [],
"source": [
"#!pip install lark elasticsearch"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "cb4a5787",
"metadata": {
"tags": []
},
"outputs": [],
"source": [
"from langchain.schema import Document\n",
"from langchain.embeddings.openai import OpenAIEmbeddings\n",
"from langchain.vectorstores import ElasticsearchStore\n",
"import os\n",
"import getpass\n",
"\n",
"os.environ[\"OPENAI_API_KEY\"] = getpass.getpass(\"OpenAI API Key:\")\n",
"\n",
"embeddings = OpenAIEmbeddings()"
]
},
{
"cell_type": "code",
"execution_count": 9,
"id": "bcbe04d9",
"metadata": {
"tags": []
},
"outputs": [],
"source": [
"docs = [\n",
" Document(\n",
" page_content=\"A bunch of scientists bring back dinosaurs and mayhem breaks loose\",\n",
" metadata={\"year\": 1993, \"rating\": 7.7, \"genre\": \"science fiction\"},\n",
" ),\n",
" Document(\n",
" page_content=\"Leo DiCaprio gets lost in a dream within a dream within a dream within a ...\",\n",
" metadata={\"year\": 2010, \"director\": \"Christopher Nolan\", \"rating\": 8.2},\n",
" ),\n",
" Document(\n",
" page_content=\"A psychologist / detective gets lost in a series of dreams within dreams within dreams and Inception reused the idea\",\n",
" metadata={\"year\": 2006, \"director\": \"Satoshi Kon\", \"rating\": 8.6},\n",
" ),\n",
" Document(\n",
" page_content=\"A bunch of normal-sized women are supremely wholesome and some men pine after them\",\n",
" metadata={\"year\": 2019, \"director\": \"Greta Gerwig\", \"rating\": 8.3},\n",
" ),\n",
" Document(\n",
" page_content=\"Toys come alive and have a blast doing so\",\n",
" metadata={\"year\": 1995, \"genre\": \"animated\"},\n",
" ),\n",
" Document(\n",
" page_content=\"Three men walk into the Zone, three men walk out of the Zone\",\n",
" metadata={\n",
" \"year\": 1979,\n",
" \"rating\": 9.9,\n",
" \"director\": \"Andrei Tarkovsky\",\n",
" \"genre\": \"science fiction\",\n",
" \"rating\": 9.9,\n",
" },\n",
" ),\n",
"]\n",
"vectorstore = ElasticsearchStore.from_documents(\n",
" docs, embeddings, index_name=\"elasticsearch-self-query-demo\", es_url=\"http://localhost:9200\"\n",
")"
]
},
{
"cell_type": "markdown",
"id": "5ecaab6d",
"metadata": {},
"source": [
"## Creating our self-querying retriever\n",
"Now we can instantiate our retriever. To do this we'll need to provide some information upfront about the metadata fields that our documents support and a short description of the document contents."
]
},
{
"cell_type": "code",
"execution_count": 6,
"id": "86e34dbf",
"metadata": {
"tags": []
},
"outputs": [],
"source": [
"from langchain.llms import OpenAI\n",
"from langchain.retrievers.self_query.base import SelfQueryRetriever\n",
"from langchain.chains.query_constructor.base import AttributeInfo\n",
"\n",
"metadata_field_info = [\n",
" AttributeInfo(\n",
" name=\"genre\",\n",
" description=\"The genre of the movie\",\n",
" type=\"string or list[string]\",\n",
" ),\n",
" AttributeInfo(\n",
" name=\"year\",\n",
" description=\"The year the movie was released\",\n",
" type=\"integer\",\n",
" ),\n",
" AttributeInfo(\n",
" name=\"director\",\n",
" description=\"The name of the movie director\",\n",
" type=\"string\",\n",
" ),\n",
" AttributeInfo(\n",
" name=\"rating\", description=\"A 1-10 rating for the movie\", type=\"float\"\n",
" ),\n",
"]\n",
"document_content_description = \"Brief summary of a movie\"\n",
"llm = OpenAI(temperature=0)\n",
"retriever = SelfQueryRetriever.from_llm(\n",
" llm, vectorstore, document_content_description, metadata_field_info, verbose=True\n",
")"
]
},
{
"cell_type": "markdown",
"id": "ea9df8d4",
"metadata": {},
"source": [
"## Testing it out\n",
"And now we can try actually using our retriever!"
]
},
{
"cell_type": "code",
"execution_count": 10,
"id": "38a126e9",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"query='dinosaur' filter=None limit=None\n"
]
},
{
"data": {
"text/plain": [
"[Document(page_content='A bunch of scientists bring back dinosaurs and mayhem breaks loose', metadata={'year': 1993, 'rating': 7.7, 'genre': 'science fiction'}),\n",
" Document(page_content='Toys come alive and have a blast doing so', metadata={'year': 1995, 'genre': 'animated'}),\n",
" Document(page_content='Three men walk into the Zone, three men walk out of the Zone', metadata={'year': 1979, 'rating': 9.9, 'director': 'Andrei Tarkovsky', 'genre': 'science fiction'}),\n",
" Document(page_content='A psychologist / detective gets lost in a series of dreams within dreams within dreams and Inception reused the idea', metadata={'year': 2006, 'director': 'Satoshi Kon', 'rating': 8.6})]"
]
},
"execution_count": 10,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# This example only specifies a relevant query\n",
"retriever.get_relevant_documents(\"What are some movies about dinosaurs\")"
]
},
{
"cell_type": "code",
"execution_count": 11,
"id": "b19d4da0",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"query='women' filter=Comparison(comparator=<Comparator.EQ: 'eq'>, attribute='director', value='Greta Gerwig') limit=None\n"
]
},
{
"data": {
"text/plain": [
"[Document(page_content='A bunch of normal-sized women are supremely wholesome and some men pine after them', metadata={'year': 2019, 'director': 'Greta Gerwig', 'rating': 8.3})]"
]
},
"execution_count": 11,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# This example specifies a query and a filter\n",
"retriever.get_relevant_documents(\"Has Greta Gerwig directed any movies about women\")"
]
},
{
"cell_type": "markdown",
"id": "39bd1de1-b9fe-4a98-89da-58d8a7a6ae51",
"metadata": {},
"source": [
"## Filter k\n",
"\n",
"We can also use the self query retriever to specify `k`: the number of documents to fetch.\n",
"\n",
"We can do this by passing `enable_limit=True` to the constructor."
]
},
{
"cell_type": "code",
"execution_count": 12,
"id": "bff36b88-b506-4877-9c63-e5a1a8d78e64",
"metadata": {
"tags": []
},
"outputs": [],
"source": [
"retriever = SelfQueryRetriever.from_llm(\n",
" llm,\n",
" vectorstore,\n",
" document_content_description,\n",
" metadata_field_info,\n",
" enable_limit=True,\n",
" verbose=True,\n",
")"
]
},
{
"cell_type": "code",
"execution_count": 13,
"id": "2758d229-4f97-499c-819f-888acaf8ee10",
"metadata": {
"tags": []
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"query='dinosaur' filter=None limit=2\n"
]
},
{
"data": {
"text/plain": [
"[Document(page_content='A bunch of scientists bring back dinosaurs and mayhem breaks loose', metadata={'year': 1993, 'rating': 7.7, 'genre': 'science fiction'}),\n",
" Document(page_content='Toys come alive and have a blast doing so', metadata={'year': 1995, 'genre': 'animated'})]"
]
},
"execution_count": 13,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# This example only specifies a relevant query\n",
"retriever.get_relevant_documents(\"what are two movies about dinosaurs\")"
]
},
{
"cell_type": "markdown",
"id": "61a10294",
"metadata": {},
"source": [
"## Complex queries in Action!\n",
"We've tried out some simple queries, but what about more complex ones? Let's try out a few more complex queries that utilize the full power of Elasticsearch."
]
},
{
"cell_type": "code",
"execution_count": 18,
"id": "e460da93",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"query='animated toys' filter=Operation(operator=<Operator.AND: 'and'>, arguments=[Operation(operator=<Operator.OR: 'or'>, arguments=[Comparison(comparator=<Comparator.EQ: 'eq'>, attribute='genre', value='animated'), Comparison(comparator=<Comparator.EQ: 'eq'>, attribute='genre', value='comedy')]), Comparison(comparator=<Comparator.GTE: 'gte'>, attribute='year', value=1990)]) limit=None\n"
]
},
{
"data": {
"text/plain": [
"[Document(page_content='Toys come alive and have a blast doing so', metadata={'year': 1995, 'genre': 'animated'})]"
]
},
"execution_count": 18,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"retriever.get_relevant_documents(\"what animated or comedy movies have been released in the last 30 years about animated toys?\")"
]
},
{
"cell_type": "code",
"execution_count": 8,
"id": "0851fc42",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"ObjectApiResponse({'acknowledged': True})"
]
},
"execution_count": 8,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"vectorstore.client.indices.delete(index=\"elasticsearch-self-query-demo\")"
]
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3 (ipykernel)",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.10.3"
}
},
"nbformat": 4,
"nbformat_minor": 5
}

3
libs/langchain/langchain/retrievers/self_query/base.py
Unescape Escape View File

@ -11,6 +11,7 @@ from langchain.chains.query_constructor.ir import StructuredQuery, Visitor
from langchain.chains.query_constructor.schema import AttributeInfo
from langchain.retrievers.self_query.chroma import ChromaTranslator
from langchain.retrievers.self_query.deeplake import DeepLakeTranslator
from langchain.retrievers.self_query.elasticsearch import ElasticsearchTranslator
from langchain.retrievers.self_query.myscale import MyScaleTranslator
from langchain.retrievers.self_query.pinecone import PineconeTranslator
from langchain.retrievers.self_query.qdrant import QdrantTranslator
@ -20,6 +21,7 @@ from langchain.schema.language_model import BaseLanguageModel
from langchain.vectorstores import (
Chroma,
DeepLake,
ElasticsearchStore,
MyScale,
Pinecone,
Qdrant,
@ -38,6 +40,7 @@ def _get_builtin_translator(vectorstore: VectorStore) -> Visitor:
Qdrant: QdrantTranslator,
MyScale: MyScaleTranslator,
DeepLake: DeepLakeTranslator,
ElasticsearchStore: ElasticsearchTranslator,
}
if vectorstore_cls not in BUILTIN_TRANSLATORS:
raise ValueError(

93
libs/langchain/langchain/retrievers/self_query/elasticsearch.py
Unescape Escape View File

@ -0,0 +1,93 @@
from typing import Dict, Tuple, Union
from langchain.chains.query_constructor.ir import (
Comparator,
Comparison,
Operation,
Operator,
StructuredQuery,
Visitor,
)
class ElasticsearchTranslator(Visitor):
"""Translate the internal query language elements to valid filters."""
allowed_comparators = [
Comparator.EQ,
Comparator.GT,
Comparator.GTE,
Comparator.LT,
Comparator.LTE,
Comparator.CONTAIN,
Comparator.LIKE,
]
"""Subset of allowed logical comparators."""
allowed_operators = [Operator.AND, Operator.OR, Operator.NOT]
"""Subset of allowed logical operators."""
def _format_func(self, func: Union[Operator, Comparator]) -> str:
self._validate_func(func)
map_dict = {
Operator.OR: "should",
Operator.NOT: "must_not",
Operator.AND: "must",
Comparator.EQ: "term",
Comparator.GT: "gt",
Comparator.GTE: "gte",
Comparator.LT: "lt",
Comparator.LTE: "lte",
Comparator.CONTAIN: "match",
Comparator.LIKE: "fuzzy",
}
return map_dict[func]
def visit_operation(self, operation: Operation) -> Dict:
args = [arg.accept(self) for arg in operation.arguments]
return {"bool": {self._format_func(operation.operator): args}}
def visit_comparison(self, comparison: Comparison) -> Dict:
# ElasticsearchStore filters require to target
# the metadata object field
field = f"metadata.{comparison.attribute}"
is_range_comparator = comparison.comparator in [
Comparator.GT,
Comparator.GTE,
Comparator.LT,
Comparator.LTE,
]
if is_range_comparator:
return {
"range": {
field: {self._format_func(comparison.comparator): comparison.value}
}
}
if comparison.comparator == Comparator.LIKE:
return {
self._format_func(comparison.comparator): {
field: {"value": comparison.value, "fuzziness": "AUTO"}
}
}
if comparison.comparator == Comparator.CONTAIN:
return {self._format_func(comparison.comparator): {field: comparison.value}}
# we assume that if the value is a string,
# we want to use the keyword field
field = f"{field}.keyword" if isinstance(comparison.value, str) else field
return {self._format_func(comparison.comparator): {field: comparison.value}}
def visit_structured_query(
self, structured_query: StructuredQuery
) -> Tuple[str, dict]:
if structured_query.filter is None:
kwargs = {}
else:
kwargs = {"filter": [structured_query.filter.accept(self)]}
return structured_query.query, kwargs

220
libs/langchain/tests/unit_tests/retrievers/self_query/test_elasticsearch.py
Unescape Escape View File

@ -0,0 +1,220 @@
from typing import Dict, Tuple
from langchain.chains.query_constructor.ir import (
Comparator,
Comparison,
Operation,
Operator,
StructuredQuery,
)
from langchain.retrievers.self_query.elasticsearch import ElasticsearchTranslator
DEFAULT_TRANSLATOR = ElasticsearchTranslator()
def test_visit_comparison() -> None:
comp = Comparison(comparator=Comparator.EQ, attribute="foo", value="1")
expected = {"term": {"metadata.foo.keyword": "1"}}
actual = DEFAULT_TRANSLATOR.visit_comparison(comp)
assert expected == actual
def test_visit_comparison_range_gt() -> None:
comp = Comparison(comparator=Comparator.GT, attribute="foo", value=1)
expected = {"range": {"metadata.foo": {"gt": 1}}}
actual = DEFAULT_TRANSLATOR.visit_comparison(comp)
assert expected == actual
def test_visit_comparison_range_gte() -> None:
comp = Comparison(comparator=Comparator.GTE, attribute="foo", value=1)
expected = {"range": {"metadata.foo": {"gte": 1}}}
actual = DEFAULT_TRANSLATOR.visit_comparison(comp)
assert expected == actual
def test_visit_comparison_range_lt() -> None:
comp = Comparison(comparator=Comparator.LT, attribute="foo", value=1)
expected = {"range": {"metadata.foo": {"lt": 1}}}
actual = DEFAULT_TRANSLATOR.visit_comparison(comp)
assert expected == actual
def test_visit_comparison_range_lte() -> None:
comp = Comparison(comparator=Comparator.LTE, attribute="foo", value=1)
expected = {"range": {"metadata.foo": {"lte": 1}}}
actual = DEFAULT_TRANSLATOR.visit_comparison(comp)
assert expected == actual
def test_visit_comparison_range_match() -> None:
comp = Comparison(comparator=Comparator.CONTAIN, attribute="foo", value="1")
expected = {"match": {"metadata.foo": "1"}}
actual = DEFAULT_TRANSLATOR.visit_comparison(comp)
assert expected == actual
def test_visit_comparison_range_like() -> None:
comp = Comparison(comparator=Comparator.LIKE, attribute="foo", value="bar")
expected = {"fuzzy": {"metadata.foo": {"value": "bar", "fuzziness": "AUTO"}}}
actual = DEFAULT_TRANSLATOR.visit_comparison(comp)
assert expected == actual
def test_visit_operation() -> None:
op = Operation(
operator=Operator.AND,
arguments=[
Comparison(comparator=Comparator.EQ, attribute="foo", value=2),
Comparison(comparator=Comparator.EQ, attribute="bar", value="baz"),
],
)
expected = {
"bool": {
"must": [
{"term": {"metadata.foo": 2}},
{"term": {"metadata.bar.keyword": "baz"}},
]
}
}
actual = DEFAULT_TRANSLATOR.visit_operation(op)
assert expected == actual
def test_visit_operation_or() -> None:
op = Operation(
operator=Operator.OR,
arguments=[
Comparison(comparator=Comparator.EQ, attribute="foo", value=2),
Comparison(comparator=Comparator.EQ, attribute="bar", value="baz"),
],
)
expected = {
"bool": {
"should": [
{"term": {"metadata.foo": 2}},
{"term": {"metadata.bar.keyword": "baz"}},
]
}
}
actual = DEFAULT_TRANSLATOR.visit_operation(op)
assert expected == actual
def test_visit_operation_not() -> None:
op = Operation(
operator=Operator.NOT,
arguments=[
Comparison(comparator=Comparator.EQ, attribute="foo", value=2),
Comparison(comparator=Comparator.EQ, attribute="bar", value="baz"),
],
)
expected = {
"bool": {
"must_not": [
{"term": {"metadata.foo": 2}},
{"term": {"metadata.bar.keyword": "baz"}},
]
}
}
actual = DEFAULT_TRANSLATOR.visit_operation(op)
assert expected == actual
def test_visit_structured_query() -> None:
query = "What is the capital of France?"
structured_query = StructuredQuery(query=query, filter=None, limit=None)
expected: Tuple[str, Dict] = (query, {})
actual = DEFAULT_TRANSLATOR.visit_structured_query(structured_query)
assert expected == actual
def test_visit_structured_query_filter() -> None:
query = "What is the capital of France?"
comp = Comparison(comparator=Comparator.EQ, attribute="foo", value="1")
structured_query = StructuredQuery(query=query, filter=comp, limit=None)
expected = (
query,
{"filter": [{"term": {"metadata.foo.keyword": "1"}}]},
)
actual = DEFAULT_TRANSLATOR.visit_structured_query(structured_query)
assert expected == actual
def test_visit_structured_query_filter_and() -> None:
query = "What is the capital of France?"
op = Operation(
operator=Operator.AND,
arguments=[
Comparison(comparator=Comparator.EQ, attribute="foo", value=2),
Comparison(comparator=Comparator.EQ, attribute="bar", value="baz"),
],
)
structured_query = StructuredQuery(query=query, filter=op, limit=None)
expected = (
query,
{
"filter": [
{
"bool": {
"must": [
{"term": {"metadata.foo": 2}},
{"term": {"metadata.bar.keyword": "baz"}},
]
}
}
]
},
)
actual = DEFAULT_TRANSLATOR.visit_structured_query(structured_query)
assert expected == actual
def test_visit_structured_query_complex() -> None:
query = "What is the capital of France?"
op = Operation(
operator=Operator.AND,
arguments=[
Comparison(comparator=Comparator.EQ, attribute="foo", value=2),
Operation(
operator=Operator.OR,
arguments=[
Comparison(comparator=Comparator.LT, attribute="bar", value=1),
Comparison(comparator=Comparator.LIKE, attribute="bar", value="10"),
],
),
],
)
structured_query = StructuredQuery(query=query, filter=op, limit=None)
expected = (
query,
{
"filter": [
{
"bool": {
"must": [
{"term": {"metadata.foo": 2}},
{
"bool": {
"should": [
{"range": {"metadata.bar": {"lt": 1}}},
{
"fuzzy": {
"metadata.bar": {
"value": "10",
"fuzziness": "AUTO",
}
}
},
]
}
},
]
}
}
]
},
)
actual = DEFAULT_TRANSLATOR.visit_structured_query(structured_query)
assert expected == actual
Write Preview
Loading…
Cancel
Save