chatgpt-api/legacy/packages/jigsawstack/src/jigsawstack-client.ts

279 wiersze
6.4 KiB
TypeScript

import { aiFunction, AIFunctionsProvider, assert, getEnv } from '@agentic/core'
import ky, { type KyInstance } from 'ky'
import { z } from 'zod'
export namespace jigsawstack {
export interface BaseResponse {
success: boolean
}
export const API_BASE_URL = 'https://api.jigsawstack.com/v1/'
export interface SearchParams {
query: string
ai_overview?: boolean
safe_search?: 'moderate' | 'strict' | 'off'
spell_check?: boolean
}
export interface SearchResponse extends BaseResponse {
query: string
spell_fixed: string
is_safe: boolean
ai_overview: string
results: {
title: string
url: string
description: string
content: string
is_safe: boolean
site_name: string
site_long_name: string
age: string
language: string
favicon: string
snippets: string[]
related_index: []
}[]
}
export interface CookieParameter {
name: string
value: string
url?: string
domain?: string
path?: string
secure?: boolean
httpOnly?: boolean
sameSite?: 'Strict' | 'Lax' | 'None'
expires?: boolean
priority?: string
sameParty?: boolean
}
export interface ScrapeParams {
url: string
element_prompts: string[]
http_headers?: object
reject_request_pattern?: string[]
goto_options?: {
timeout: number
wait_until: string
}
wait_for?: {
mode: string
value: string | number
}
advance_config?: {
console: boolean
network: boolean
cookies: boolean
}
size_preset?: string
is_mobile?: boolean
scale?: number
width?: number
height?: number
cookies?: Array<CookieParameter>
}
export interface ScrapeResponse extends BaseResponse {
data: any
}
export interface VOCRParams {
prompt: string | string[]
url?: string
file_store_key?: string
}
export interface VOCRResponse extends BaseResponse {
context: string
width: number
height: number
tags: string[]
has_text: boolean
sections: Array<any>
}
export interface TextToSqlParams {
prompt: string
sql_schema?: string
file_store_key?: string
}
export interface TextToSqlResponse extends BaseResponse {
sql: string
}
export interface SpeechToTextParams {
url?: string
file_store_key?: string
language?: string
translate?: boolean
by_speaker?: boolean
webhook_url?: string
}
export interface SpeechToTextResponse extends BaseResponse {
text: string
chunks: Array<{
timestamp: number[]
text: string
}>
status?: 'processing' | 'error'
id?: string
}
}
/**
* Basic JigsawStack API wrapper.
*/
export class JigsawStackClient extends AIFunctionsProvider {
protected readonly apiKey: string
protected readonly ky: KyInstance
constructor({
apiKey = getEnv('JIGSAWSTACK_API_KEY'),
timeoutMs = 60_000
}: {
apiKey?: string
throttle?: boolean
timeoutMs?: number
} = {}) {
assert(
apiKey,
'Please set the JIGSAWSTACK_API_KEY environment variable or pass it to the constructor as the apiKey field.'
)
super()
this.apiKey = apiKey
this.ky = ky.extend({
prefixUrl: jigsawstack.API_BASE_URL,
timeout: timeoutMs,
headers: {
'x-api-key': this.apiKey
}
})
}
/**
* Perform web searches and retrieve high-quality results of the given query
*/
@aiFunction({
name: 'jigsawstack_ai_search',
description:
'Perform web searches and retrieve high-quality results of the given query',
inputSchema: z.object({
query: z.string().describe('The search query')
})
})
async aiSearch(params: jigsawstack.SearchParams) {
return this.ky
.get('web/search', {
searchParams: {
...params
}
})
.json<jigsawstack.SearchResponse>()
}
/**
* Scrape any website
*/
@aiFunction({
name: 'jigsawstack_ai_scrape',
description: 'Scrape any website',
inputSchema: z.object({
url: z.string().describe('The url to scrape its content'),
element_prompts: z
.array(z.string())
.describe(
'The items to scrape (retrieve) from the given url. eg. Page title, price, etc'
)
})
})
async aiScrape(params: jigsawstack.ScrapeParams) {
return this.ky
.post('ai/scrape', {
json: {
...params
}
})
.json<jigsawstack.ScrapeResponse>()
}
/**
* Recognise, describe and retrieve data within an image with great accuracy.
*/
@aiFunction({
name: 'jigsawstack_vocr',
description:
'Recognise, describe and retrieve data within an image with great accuracy.',
inputSchema: z.object({
url: z.string().describe('The image url to OCR'),
prompt: z
.string()
.default('Describe the image in detail.')
.describe('What you want to know or retrieve from the image')
})
})
async vocr(params: jigsawstack.VOCRParams) {
return this.ky
.post('vocr', {
json: {
...params
}
})
.json<jigsawstack.VOCRResponse>()
}
/**
* Generate semantically correct SQL queries from text.
*/
@aiFunction({
name: 'jigsawstack_text_to_sql',
description: 'Generate semantically correct SQL queries from text.',
inputSchema: z.object({
prompt: z
.string()
.describe('The text that will be translated to an SQL query.'),
sql_schema: z
.string()
.describe('The valid sql schema where the query will be run')
})
})
async textToSql(
params: jigsawstack.TextToSqlParams
): Promise<jigsawstack.TextToSqlResponse> {
return this.ky
.post('ai/sql', {
json: {
...params
}
})
.json<jigsawstack.TextToSqlResponse>()
}
/**
* Convert audio/video files into accurate text transcriptions instantly.
*/
@aiFunction({
name: 'jigsawstack_speech_to_text',
description:
'Convert audio/video files into accurate text transcriptions instantly.',
inputSchema: z.object({
url: z.string().describe('The audio or video url')
})
})
async speechToText(
params: jigsawstack.SpeechToTextParams
): Promise<jigsawstack.SpeechToTextResponse> {
return this.ky
.post('ai/transcribe', {
json: {
...params
}
})
.json<jigsawstack.SpeechToTextResponse>()
}
}