data-extracter-extesion/scripts/background/extractor.js

var __EXTRACTOR_STATE__ = "";

class Extractor {
    constructor(options) {
        this._tasks = [];
        this._running = false;
        this._options = options;
    }
    /**
     * Save current state, in case we restore it later.
     */
    save() {
        saveFile(JSON.stringify(this), 'application/json', 'state.json');
    }
    /**
     * Restore previous state by loading from saved state.
     */
    load() {
        if (!__EXTRACTOR_STATE__) {
            console.log('No state found. \nPlease upload a saved state from the popup window first.');
            return;
        }
        let state = JSON.parse(__EXTRACTOR_STATE__);
        __EXTRACTOR_STATE__ = "";
        this._options = state._options;
        this._tasks = state._tasks.map(t => new Task(this._options, 'whaterver', ['whaterver']).load(t));
        return this;
    }
    /**
     * Add a task to Extractor. \n
     * One Extractor could has multiple tasks, which orgnized in a task chian.
     * If url arguments not given within later tasks, they will use previous task result as input (target url list).
     * @param {...any} args itemsSelector, fieldSelectors, and more args to specify target urls.
     */
    task(...args) {
        this._tasks.push(new Task(this._options, ...args));
        return this;
    }
    /**
     * Clear tasks and task caches.
     */
    clear() {
        this._tasks = [];
        return this;
    }
    /**
     * Start the task chain.
     */
    async start() {
        return this._startTasks(0);
    }
    /**
     * restart from specified task, but don't restart the previous tasks.
     * @param {number} from where to restart the tasks, begins with 0
     */
    async restart(from = 0) {
        let id = this._checkTaskId(from, 0);
        if (id < 0) return;
        for (let i = id; i < this._tasks.length; i++) {
            this._tasks[i].clean();
        }
        return this._startTasks(0);
    }
    async _startTasks(from) {
        if (this._running) {
            console.log('The Extractor is running. Please wait..');
            return;
        }
        if (!this._tasks.length) {
            console.log('No task to run.');
            return;
        }

        let tab;
        let task = this._tasks[0];
        if (task.urls.length) {
            // task specifies target urls, create new tab with first url for it
            tab = await createTab(task.urls[0], false);
        } else {
            tab = await getActiveTab(true) || await getActiveTab(false);
            let succ = await ping(tab);
            if (!succ) {
                console.log('Cannot contact with active tab.');
                return;
            }
        }
        this._running = true;
        return this._tasks.reduce((pms, task, i) => {
            return pms.then(
                () => {
                    if (i < from) return;
                    if (i > 0) {
                        let prevTask = this._tasks[i - 1];
                        return task.execute(tab, new ExtractResult(prevTask.results));
                    }
                    return task.execute(tab, undefined);
                });
        }, Promise.resolve(undefined)).then(
            () => {
                this._running = false;
                this.export();
            }
        ).catch(err => {
            this._running = false;
            console.log(err);
        });
    }
    /**
     * export result of a task to CSV
     * @param {number} taskid which task id to save, begins with 0
     */
    export(taskid) {
        let id = this._checkTaskId(taskid, this._tasks.length - 1);
        if (id < 0) return;
        let results = this._tasks[id].results
        results.unshift(this._tasks[id].fieldSelectors);

        let exResults = new ExtractResult(results);

        if (!results.length) {
            console.log(`No result for task #${id}. Forget to call ".start()"?`);
            return;
        }
        let msg = `
Please confirm to download (${results.length - 1} items)：

${exResults.toString(50) || "- Empty -"}
`.trim();
        if (confirm(msg)) {
            saveFile(exResults, "text/csv");
        }
    }
    _checkTaskId(id, defaultId) {
        if (!this._tasks.length) {
            console.log("No task found.");
            return -1;
        }
        if (!isNaN(defaultId) && id === undefined) id = defaultId;
        if (isNaN(id) || id < 0 || id >= this._tasks.length) {
            console.log(`Invalid task id. Rang(0-${this._tasks.length - 1})`);
            return -1;
        }
        return id
    }
}