https://v2-app.gladia.io/auth/login
npm i @gladiaio/sdkimport gladia from '@gladiaio/sdk';
const gladiaClient = gladia({ apiKey: 'XXXXXXXX' });
const result = await gladiaClient.autocorrect({ sentence: 'thnik' });
// result = { prediction: 'think' }import { gladia } from '@gladiaio/sdk';
const gladiaClient = gladia({ apiKey: 'XXXXXXXX' });
const result = await gladiaClient.autocorrect({ sentence: 'thnik' });
// result = { prediction: 'think' }const { gladia } = require('@gladiaio/sdk');
const gladiaClient = gladia({ apiKey: 'XXXXXXXX' });
const result = await gladiaClient.autocorrect({ sentence: 'thnik' });
// result = { prediction: 'think' }Notes: ES5 version is built for IE targets when you have no bundler, prefer other methods
const { gladia } = require('@gladiaio/sdk/lib/es5');
const gladiaClient = gladia({ apiKey: 'XXXXXXXX' });
const result = await gladiaClient.autocorrect({ sentence: 'thnik' });
// result = { prediction: 'think' }<!DOCTYPE html>
<html>
<body>
<script src="./gladiaio-sdk.js"></script>
<script>
const gladiaClient = gladiaio_sdk({ apiKey: 'XXXXXXXX' });
(async () => {
const result = await gladiaClient.autocorrect({ sentence: 'thnik' });
// result = { prediction: 'think' }
})();
</script>
</body>
</html>There is multiple bundle choices:
gladiaio-sdk.js: contain bundled ES5 JS + source mapgladiaio-sdk.min.js: contain bundled ES5 JS but not source mapgladiaio-sdk.min.js.map: contain only source map forgladiaio-sdk.min.js
For production, we recommand gladiaio-sdk.min.js;
There is 2 ways to use gladia tasks: full path and shortcuts.
The full path will help you to find what you can do with some kind of input, targetting some kind of output.
Example:
const ocrResult = await gladiaClient.fromImage().toText().ocr({ image: readImageAsArrayBuffer() });Here we can see explicitly we want to start from an image, end to text and only tasks matching this couple input/output will be possible through autocompletion.
The shortcut is more convinient when you know what you want to do.
Example:
const ocrResult = await gladiaClient.ocr({ image: readImageAsArrayBuffer() });Here you can see we can directly call the task we want on the GladiaClient without indicating input/output. If you know what you want: it's for you!
Note: full path and shortcut are strictly equivalent.
For each task, we chose for you a model we recommand you to use. The recommanded model is automatically used if you specify no model.
To choose yourself the model you want, you can specify it like that:
const ocrResult = await gladiaClient.ocr({ image: readImageAsArrayBuffer(), model: 'tesseract-default' });const inputText = 'I love you';
const result = await gladiaClient.sentimentAnalysis({ text: inputText });
const sentimentAnalysis = JSON.parse(result).label;
// => POSITIVEconst imageInput = fileInputRef.files[0];
const imageOutput = await gladiaClient.fromImage().toImage().backgroundRemoval({ image: imageInput });
const imageUrl = URL.createObjectURL(new Blob([ imageOutput ]));
// => imageUrl can be used to display the image by setting it as srcYou have 2 choices: adding a global header or adding a task specific header.
import gladia from '@gladiaio/sdk';
const gladiaClient = gladia({ apiKey: 'XXXXXXXX', header: { 'X-Custom-Header': 'value' } });import gladia from '@gladiaio/sdk';
const gladiaClient = gladia({ apiKey: 'XXXXXXXX' });
const result = await gladiaClient.autocorrect({ sentence: 'thnik', header: { 'X-Custom-Header': 'value' } });Under the hood, GladIA SDK use Axios to make http calls. By default Axios use XMLHttpRequest on browser and use request on node.js. If you prefer use fetch on browser, you can use the useFetch param when create the client:
import gladia from '@gladiaio/sdk';
const gladiaClient = gladia({ apiKey: 'XXXXXXXX', useFetch: true });On some case you may want to use a fully custom http client. This is possible like that:
import { gladia, HttpClientFactory, HttpClientFactoryParams, PostParams } from '@gladiaio/sdk';
const myHttpClientFactory: HttpClientFactory = (params: HttpClientFactoryParams) => {
return {
post(postParams: PostParams) {
// ...
}
};
}
const gladiaClient = gladia({ apiKey: 'XXXXXXXX', customHttpClient: myHttpClientFactory });You can use default http client as inspiration: https://github.com/gladiaio/gladia-sdk-js/blob/main/src/internal/http-client.ts#L40
By default, timeout is set to 300000 (5 minutes). You can override this by passing the timeout property to Gladia SDK option object and set any number you want starting from 0 (no timeout).
Here, we are setting a 2 seconds timeout:
import gladia from '@gladiaio/sdk';
const gladiaClient = gladia({ apiKey: 'XXXXXXXX', useFetch: true, httpClientTimeout: 2000 });npm i
npm run generate-metadata
npm run generate-sdk
npm run buildNotes:
generate-metadatascript will download the latestopenapi.jsonfile from production api and generategenerate-sdkwill generate most parts of the SDK using theopenapi.jsondefinition- If you need a custom sdk, you can provide a
openapi.jsonfile like thisnpm run generate-metadata -- --from-file=./path/to/openapi.json, then call the scriptsgenerate-sdkandbuildlike before - You can download only the current
openapi.jsonby callingnpm run download-openapi
Change version in package.json then:
npm install
npm run generate-metadata
npm run generate-sdk
npm run test
npm run buildThen create a commit with every modified files with name [release] x.y.z (where x.y.z is replace with the version specified in package.json).
Create an npm link
npm link
On the other project (ex: Office Addin)
npm link @gladiaio/sdk
After your test, dont forget to do
npm cache clean -f