langextract
Version:
A TypeScript library for extracting structured and grounded information from text using LLMs
106 lines • 4.26 kB
JavaScript
;
/**
* Copyright 2025 kmbro.
*
* This is a TypeScript translation of the original Python LangExtract library
* by Google LLC (https://github.com/google/langextract).
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
Object.defineProperty(exports, "__esModule", { value: true });
exports.GeminiSchemaImpl = exports.EXTRACTIONS_KEY = exports.ConstraintType = void 0;
var ConstraintType;
(function (ConstraintType) {
ConstraintType["NONE"] = "none";
})(ConstraintType || (exports.ConstraintType = ConstraintType = {}));
exports.EXTRACTIONS_KEY = "extractions";
class GeminiSchemaImpl {
constructor(schemaDict) {
this._schemaDict = schemaDict;
}
get schemaDict() {
return this._schemaDict;
}
set schemaDict(schemaDict) {
this._schemaDict = schemaDict;
}
static fromExamples(examplesData, attributeSuffix = "_attributes") {
// Track attribute types for each category
const extractionCategories = {};
for (const example of examplesData) {
for (const extraction of example.extractions) {
const category = extraction.extractionClass;
if (!extractionCategories[category]) {
extractionCategories[category] = {};
}
if (extraction.attributes) {
for (const [attrName, attrValue] of Object.entries(extraction.attributes)) {
if (!extractionCategories[category][attrName]) {
extractionCategories[category][attrName] = new Set();
}
extractionCategories[category][attrName].add(Array.isArray(attrValue) ? "array" : "string");
}
}
}
}
const extractionProperties = {};
for (const [category, attrs] of Object.entries(extractionCategories)) {
extractionProperties[category] = { type: "string" };
const attributesField = `${category}${attributeSuffix}`;
const attrProperties = {};
// If no attributes were found for this category, add a default property
if (Object.keys(attrs).length === 0) {
attrProperties["_unused"] = { type: "string" };
}
else {
for (const [attrName, attrTypes] of Object.entries(attrs)) {
// If we see array type, use array of strings
if (attrTypes.has("array")) {
attrProperties[attrName] = {
type: "array",
items: { type: "string" },
};
}
else {
attrProperties[attrName] = { type: "string" };
}
}
}
extractionProperties[attributesField] = {
type: "object",
properties: attrProperties,
nullable: true,
};
}
const extractionSchema = {
type: "object",
properties: extractionProperties,
};
const schemaDict = {
type: "object",
properties: {
[exports.EXTRACTIONS_KEY]: {
type: "array",
items: extractionSchema,
},
},
required: [exports.EXTRACTIONS_KEY],
};
return new GeminiSchemaImpl(schemaDict);
}
fromExamples(examplesData, attributeSuffix = "_attributes") {
return GeminiSchemaImpl.fromExamples(examplesData, attributeSuffix);
}
}
exports.GeminiSchemaImpl = GeminiSchemaImpl;
//# sourceMappingURL=schema.js.map