/**
* Parse search results from a search result site
*
* @param pUrl
*/
private void parseSearchResults(String pUrl) {
LOGGER.info("Started parsing: " + pUrl);
Document doc = null;
doc = ParserUtils.connectGetUrl(ParserUtils.getUri(pUrl).toASCIIString());
doc.setBaseUri(DEFAULT_VSP_URL);
Elements results = doc.select("div[class*=map-list-item]");
for (Element result : results) {
PersistentEntity ent = new PersistentEntity();
Elements infoElement = result.select("div[class*=info-content]");
LOGGER.debug(infoElement.select("p[class*=establishment-category]").first().ownText());
String tmp =
result
.select("div[class*=info-content]")
.select("p[class*=establishment-category]")
.first()
.ownText();
ent.setIndustry(new Utf8(tmp.split("/")[0]));
ent.setLabel(new Utf8(tmp));
// getting same as value to where it is
EylloLink link =
ParserUtils.detectUrl(
infoElement.select("p[class*=establishment-name]").select("a").first());
if (link != null) {
LOGGER.debug(DEFAULT_VSP_URL + link.getLinkHref());
ent.putToSameAs(
new Utf8(DEFAULT_VSP_URL + link.getLinkHref()), new Utf8(link.getLinkText()));
ent.setName(new Utf8(link.getLinkText()));
}
// getting its address and phone
PersistentPoint point = new PersistentPoint();
infoElement = result.select("div[class*=establishment-details]").select("p");
ent.addToTelephones(new Utf8(infoElement.get(0).ownText()));
point.setAddress(new Utf8(infoElement.get(0).text()));
if (!result.attr("data-lng").toString().equals("")
&& !result.attr("data-lat").toString().equals("")) {
// Format in [lon, lat], note, the order of lon/lat here in order to conform with GeoJSON.
point.addToCoordinates(Double.parseDouble(result.attr("data-lng")));
point.addToCoordinates(Double.parseDouble(result.attr("data-lat")));
point.setAccuracy(EylloLocation.GEOCODER_VERIF_ACC_HIGH);
}
ent.setPersistentpoint(point);
ent.addToScenarioId(getScenarioId());
this.pEntities.add(ent);
}
LOGGER.info("Completed getting basic information from entities.");
}
