instamancer/src/api/instagram.d.ts

Scrape the Instagram API with Puppeteer

import { Type } from "io-ts";
import { Page } from "puppeteer";
import * as winston from "winston";
import { IOptions } from "./api";
/**
 * Instagram API wrapper
 */
export declare class Instagram<PostType> {
    /**
     * Apply defaults to undefined options
     */
    private static defaultOptions;
    id: string;
    url: string;
    started: boolean;
    paused: boolean;
    finished: boolean;
    finishedReason: FinishedReasons;
    catchURL: string;
    postURL: string;
    defaultPostURL: string;
    jumpMod: number;
    jumpSize: number;
    page: Page;
    logger: winston.Logger;
    defaultPageFunctions: (() => void)[];
    private readonly strict;
    private readonly validator;
    private browser;
    private browserDisconnected;
    private readonly browserInstance?;
    private readonly headless;
    private postBuffer;
    private postBufferLock;
    private requestBuffer;
    private requestBufferLock;
    private responseBuffer;
    private responseBufferLock;
    private readonly fullAPI;
    private pagePromises;
    private readonly enableGrafting;
    private readonly sameBrowser;
    private graft;
    private graftURL;
    private graftHeaders;
    private foundGraft;
    private hibernate;
    private readonly hibernationTime;
    private failedJumps;
    private responseFromAPI;
    private readonly pageQuery;
    private readonly edgeQuery;
    private postIds;
    private readonly total;
    private index;
    private jumps;
    private readonly maxPageUrlAttempts;
    private pageUrlAttempts;
    private postPageRetries;
    private readonly silent;
    private writeLock;
    private sleepRemaining;
    private readonly sleepTime;
    private readonly proxyURL;
    private readonly executablePath;
    private pluginFunctions;
    /**
     * Create API wrapper instance
     * @param endpoint the url for the type of resource to scrape
     * @param id the identifier for the resource
     * @param pageQuery the query to identify future pages in the nested API structure
     * @param edgeQuery the query to identify posts in the nested API structure
     * @param options configuration details
     * @param validator response type validator
     */
    constructor(endpoint: string, id: string, pageQuery: string, edgeQuery: string, options: IOptions, validator: Type<unknown>);
    /**
     * Toggle pausing data collection
     */
    pause(): void;
    /**
     * Toggle prolonged pausing
     */
    toggleHibernation(): void;
    /**
     * Force the API to stop
     */
    forceStop(force?: boolean): Promise<void>;
    /**
     * Generator of posts on page
     */
    generator(): AsyncIterableIterator<PostType>;
    /**
     * Construct page and add listeners
     */
    start(): Promise<void>;
    /**
     * Match the url to the url used in API requests
     */
    matchURL(url: string): boolean;
    /**
     * Close the page and browser
     */
    protected stop(): Promise<void>;
    /**
     * Finish retrieving data for the generator
     */
    protected finish(reason: FinishedReasons): void;
    /**
     * Process the requests in the request buffer
     */
    protected processRequests(): Promise<void>;
    /**
     * Process the responses in the response buffer
     */
    protected processResponses(): Promise<void>;
    protected processResponseData(data: unknown): Promise<void>;
    /**
     * Open a post in a new page, then extract its metadata
     */
    protected postPage(post: string, retries: number): Promise<void>;
    private handlePostPageError;
    protected validatePost(post: PostType): Promise<void>;
    /**
     * Stimulate the page until responses gathered
     */
    protected getNext(): Promise<void>;
    /**
     * Halt execution
     * @param time Seconds
     */
    protected sleep(time: number): Promise<void>;
    /**
     * Create the browser and page, then visit the url
     */
    private constructPage;
    /***
     * Handle errors that occur during page construction
     */
    private handleConstructionError;
    /**
     * Pause and wait until resumed
     */
    private waitResume;
    /**
     * Pop a post off the postBuffer (using locks). Returns null if no posts in buffer
     */
    private postPop;
    /**
     * Print progress to stderr
     */
    private progress;
    /**
     * Add request to the request buffer
     */
    private interceptRequest;
    /**
     * Add the response to the response buffer
     */
    private interceptResponse;
    /**
     * Log failed requests
     */
    private interceptFailure;
    /**
     * Add post to buffer
     */
    private addToPostBuffer;
    /**
     * Manipulate the page to stimulate a request
     */
    private jump;
    /**
     * Clear request and response buffers
     */
    private initiateGraft;
    /**
     * Read the posts that are pre-loaded on the page
     */
    private scrapeDefaultPosts;
    private addPlugins;
    private executePlugins;
}
/**
 * Reasons why the collection finished
 */
declare enum FinishedReasons {
    FORCED_STOP = 0,
    API_FINISHED = 1,
    TOTAL_REACHED_API = 2,
    TOTAL_REACHED_PAGE = 3,
    NO_RESPONSE = 4,
    NO_INCREMENT = 5
}
export {};