structured data extraction

docs/modules/chains/examples/structured_data_extraction.ipynb

@@ -0,0 +1,101 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "id": "9bb1293b",
+   "metadata": {},
+   "source": [
+    "# Structured Data Extraction\n",
+    "\n",
+    "This notebook goes over how to use a chain to extract structured data from text.\n",
+    "\n",
+    "This heavily utilizes the fabulous [kor library](https://eyurtsev.github.io/kor/index.html). As a result, it only works for Python 3.10+"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "82ba478c",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!pip install kor"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "id": "d4b12abd",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from kor.extraction import Extractor\n",
+    "from kor.nodes import Object, Text, Number\n",
+    "from kor.llms import OpenAIChatCompletion, OpenAICompletion"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "id": "15726529",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "llm = OpenAIChatCompletion(model=\"gpt-3.5-turbo\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "id": "8c49e1ba",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "model = Extractor(llm)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "id": "752e408d",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "schema = Text(\n",
+    "    id=\"first_name\",\n",
+    "    description=\"The first name of a person\",\n",
+    "    examples=[(\"I am billy.\", \"billy\"), (\"John Smith is 33 years old\", \"John\")],\n",
+    ")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "50fe683e",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "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": "3.10.1"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}

langchain/chains/__init__.py

@@ -25,6 +25,7 @@ from langchain.chains.sql_database.base import (
 )
 from langchain.chains.transform import TransformChain
 from langchain.chains.vector_db_qa.base import VectorDBQA
+from langchain.chains.structured_data_extraction.base import StructuredDataExtractionChain
 
 __all__ = [
     "ConversationChain",
@@ -52,4 +53,5 @@ __all__ = [
     "ChatVectorDBChain",
     "GraphQAChain",
     "ConstitutionalChain",
+    "StructuredDataExtractionChain"
 ]

langchain/chains/structured_data_extraction/base.py

@@ -0,0 +1,45 @@
+from __future__ import annotations
+from typing import Dict, List, Any
+
+from langchain.chains.base import Chain
+from pydantic import root_validator
+
+
+
+class StructuredDataExtractionChain(Chain):
+
+
+    kor_extractor: Any
+    kor_schema: Any
+    input_key: str = "text"
+    output_key: str = "info"
+    @property
+    def _chain_type(self) -> str:
+        raise NotImplementedError
+
+    @property
+    def input_keys(self) -> List[str]:
+        return [self.input_key]
+
+    @property
+    def output_keys(self) -> List[str]:
+        return [self.output_key]
+
+    @root_validator()
+    def validate_environment(cls, values: Dict) -> Dict:
+        """Validate that api key and python package exists in environment."""
+        try:
+            import kor
+        except ImportError:
+            raise ValueError(
+                "Could not import kor python package. "
+                "Please it install it with `pip install kor`."
+            )
+        return values
+
+    def _call(self, inputs: Dict[str, str]) -> Dict[str, str]:
+        result = self.kor_extractor(inputs[self.input_key], self.kor_schema)
+        return {self.output_key: result}
+
+    async def _acall(self, inputs: Dict[str, str]) -> Dict[str, str]:
+        pass