@zaml/extract
v0.7.1
Published
Entity extracting helper for ZAML
Downloads
1
Keywords
Readme
zaml-extract
Entity extracting helper for ZAML.
Thanks to Github ❤️ ci is integrated with Github Actions
Install
npm install @zaml/extract
Usage
Plugin
A plugin is a function takes a text or Object expose methods for single text or text array
// extracting result structure
interface EntityInfo {
start: number;
end: number;
text: string;
type: string;
data: {};
}
// function plugin
type SingleExtractor = (text: string) => Promise<EntityInfo[]>;
// instance plugin
interface ExtractorInterface {
extract(text: string): Promise<EntityInfo[]>;
extractArray(textList: string[]): Promise<EntityInfo[][]>;
}
Build-in plugin
| Plugin | Type | Options | Description |
| ----------------------------------- | -------- | -------------------------------------- | --------------------------------------------- |
| link | function | none
| Extracting domain names, emails |
| mention | function | { users: string[], pattern: RegExp }
| Extract user mention |
| rest | Class | { url: string }
| Extract entities from third party RESTful API |
Extractor
type PluginOptions = string | {name: string, options: any} | SingleExtractor | ExtractorInterface;
interface {
constructor(options: {plugins: ExtractorOptions[]}});
extract(text: string | string[]): Promise<EntityInfo[] | EntityInfo[][]>
}
Extractor
instance can be created by extract.create()
,
It takes a option with plugins
as an array of plugin options, each of options can be:
- A single string indicates build-in plugin name.
- A object with
name
of build-in plugin name and aoptions
for plugin options. - A simple function of type
SingleExtractor
. - A object of type
ExtractorInterface
.
You can specify one plugin multi-times with different options, like rest
plugin for different API.
Each plugin will take turns to execute the extraction process, in case there is some entity overlap, the former ones will be kept.
Example
import zaml from '@zaml/parser';
import extract from '@zaml/extract';
const extractor = extract.create({
plugins: ['link'],
});
const node = zaml.tokenize('Hello @everyone! this is our web site: www.example.com');
node.extractEntities(extractor);
console.log(JSON.stringify(node, null, 2));
output:
API Documentation
Checkout API Documentation
Test
npm test
Test with more info of tokenizing
DEBUG=verbose npm test