house-gov-page-scraper/dist/house/gov/getDistrictsInZip.js

a node module for scraping info from the US House of Representatives website: house.gov

'use strict';

Object.defineProperty(exports, "__esModule", {
    value: true
});
exports.endpoint = exports.INVALID_DISTRICT_CODE_STRING = exports.NO_RESULTS_ZIP = exports.INVALID_ZIP = undefined;
exports.parseDistrictCode = parseDistrictCode;
exports.scrapePage = scrapePage;
exports.default = getDistrictsInZip;

var _requestPromiseNative = require('request-promise-native');

var _requestPromiseNative2 = _interopRequireDefault(_requestPromiseNative);

function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }

var INVALID_ZIP = exports.INVALID_ZIP = function INVALID_ZIP(zip) {
    return 'Invalid zip code: "' + zip + '".';
};

var NO_RESULTS_ZIP = exports.NO_RESULTS_ZIP = function NO_RESULTS_ZIP(zip) {
    return 'No results for zip code: "' + zip + '".';
};

var INVALID_DISTRICT_CODE_STRING = exports.INVALID_DISTRICT_CODE_STRING = function INVALID_DISTRICT_CODE_STRING(val) {
    return 'District code substring: "' + val + '" is an invalid format.';
};

var endpoint = exports.endpoint = function endpoint(zip) {
    return 'http://ziplook.house.gov/htbin/findrep?ZIP=' + zip;
};

function parseDistrictCode(val) {

    if (!val) {
        // eslint-disable-next-line babel/new-cap
        throw new Error(INVALID_DISTRICT_CODE_STRING(val));
    }

    var num = val.match(/\d+/g);
    var st = val.substring(1, 3);

    if (!num || !st || st.length !== 2 || num.length !== 1 || isNaN(num[0]) || num[0].length > 2) {
        // eslint-disable-next-line babel/new-cap
        throw new Error(INVALID_DISTRICT_CODE_STRING(val));
    }

    return st + '-' + Number(num);
}

function scrapePage(html, zip) {

    var expected404Msg = 'The ZIP code ' + zip + '  was not found.';

    if (html.indexOf(expected404Msg) > -1) {
        return [];
    }

    if (html.indexOf('invalid Zip Code') > -1) {
        // eslint-disable-next-line babel/new-cap
        throw new Error(INVALID_ZIP(zip));
    }

    var startStr = 'districts=[';
    var start = html.indexOf(startStr) + startStr.length;
    var end = html.indexOf('];', start);
    var data = html.substring(start, end).split(',');
    return data.map(parseDistrictCode);
}

/**
 * Retrieves an array of Congressional districts based on zip
 * by scraping house.gov's district finder.
 * @see http://ziplook.house.gov/htbin/findrep
 * @param {string} zip -    A 5 digit zip code.
 *                          Prefix 4 digit zip codes w/ "0".
 * @returns {Promise}
 *      Rejects with a 404 if no district codes are found.
 *      Resolves with an array of hyphen delimited district codes.
 *      ex: ['AL-1','AL-2']
 */
function getDistrictsInZip(zip) {
    return _requestPromiseNative2.default.get({ uri: endpoint(zip) }).then(function (html) {
        return scrapePage(html, zip);
    }).then(function (result) {
        if (result.length) {
            return result;
        }
        return Promise.reject({
            statusCode: 404,
            // eslint-disable-next-line babel/new-cap
            message: NO_RESULTS_ZIP(zip)
        });
    });
}