factbook
Version:
Serves as an independent data scraping module, complete with ontology and full scraping ability for the CIA World Factbook site
56 lines (55 loc) • 2.56 kB
JavaScript
;
Object.defineProperty(exports, "__esModule", { value: true });
var funktologies_1 = require("funktologies");
var getUuid = require("uuid-by-string");
var constants_1 = require("../constants/constants");
var globalStore_1 = require("../constants/globalStore");
function getFlag(cheerioElem, country, countryId) {
var objectProperties = globalStore_1.store.countries[countryId].objectProperties;
var flag = funktologies_1.getRelation(objectProperties, constants_1.consts.ONTOLOGY.HAS_FLAG);
var fId = constants_1.consts.ONTOLOGY.INST_FLAG + getUuid(country);
var objectProp = {};
var bailOut = true;
cheerioElem('div.flagBox').each(function () {
if (!flag) {
if (globalStore_1.store.nationalFlags[fId]) {
objectProp[constants_1.consts.ONTOLOGY.HAS_FLAG] = globalStore_1.store.nationalFlags[fId];
}
else {
objectProp = funktologies_1.entityMaker(constants_1.consts.ONTOLOGY.HAS_FLAG, constants_1.consts.ONTOLOGY.ONT_FLAG, fId, "National Flag of " + country);
globalStore_1.store.nationalFlags[fId] = objectProp[constants_1.consts.ONTOLOGY.HAS_FLAG];
}
flag = objectProp[constants_1.consts.ONTOLOGY.HAS_FLAG];
globalStore_1.store.countries[countryId].objectProperties.push(funktologies_1.entityRefMaker(constants_1.consts.ONTOLOGY.HAS_FLAG, objectProp));
}
bailOut = false;
});
if (bailOut) {
return;
}
cheerioElem('div.flagBox').each(function (index, element) {
var a = cheerioElem(element).find('img').attr('src');
var flagImgUrl;
if (a && a.replace('../', '')) {
flagImgUrl = constants_1.consts.BASE.URL_BASE + a.replace('../', '');
}
if (flagImgUrl) {
flag.datatypeProperties[constants_1.consts.ONTOLOGY.DT_LOCATOR_URI] = flagImgUrl;
}
// TODO: scrape physical image from url and store it.
});
cheerioElem('div.modalFlagDesc').each(function (index, element) {
var b = cheerioElem(element).find('div.photogallery_captiontext').text().trim();
if (!b) {
return;
}
if (flag) {
flag.datatypeProperties[constants_1.consts.ONTOLOGY.DT_CONTENT_DESCRIPTION] = b.replace(/\\n/g, '').trim();
}
else {
flag.datatypeProperties[constants_1.consts.ONTOLOGY.DT_CONTENT_DESCRIPTION] = b.trim();
}
});
}
exports.getFlag = getFlag;
;