/**
* 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.");
}
/**
* Parsing existing entities
*
* @param pEntity
*/
public void parseIndividualEnt(PersistentEntity pEntity) {
Document doc = null;
Iterator<Entry<Utf8, Utf8>> it = pEntity.getSameAs().entrySet().iterator();
while (it.hasNext()) {
Map.Entry<Utf8, Utf8> pairs = (Map.Entry<Utf8, Utf8>) it.next();
// Reading individual URLs
LOGGER.info(
"Parsing entity from: " + ParserUtils.getUri(pairs.getKey().toString()).toASCIIString());
doc =
ParserUtils.connectGetUrl(ParserUtils.getUri(pairs.getKey().toString()).toASCIIString());
if (doc == null && !validateSite(doc)) {
break;
} else {
doc.setBaseUri(VejaSaoPauloParser.DEFAULT_VSP_URL);
StringBuilder strBuilder = new StringBuilder();
// getting working hours
Elements workElems =
doc.select("div[class*=information-unwanted]").select("div[class*=working-hours]");
if (workElems != null && workElems.size() > 0) {
for (Element info : workElems.select("div[class*=hours]").select("p"))
strBuilder.append(info.text().replace("-", "_")).append(ParserProperties.INFO_SEP);
pEntity.setSchedule(new Utf8(strBuilder.toString()));
}
// getting price range
workElems =
doc.select("div[class*=information-unwanted]")
.select("div[class*=price]")
.select("p[class*=price-range]");
strBuilder.delete(0, strBuilder.length());
if (workElems != null && workElems.size() > 0) {
strBuilder.append(
doc.select("div[class*=price]").select("h3").first().text()
+ ParserProperties.DESC_SEP);
strBuilder.append(workElems.text());
pEntity.addToExtraInfo(new Utf8(strBuilder.toString()));
// LOGGER.debug(strBuilder.toString());
}
// getting payment information
workElems =
doc.select("div[class*=information-unwanted]")
.select("div[class*=payment]")
.select("p");
strBuilder.delete(0, strBuilder.length());
if (workElems != null && workElems.size() > 0) {
strBuilder.append(
doc.select("div[class*=payment]").select("h3").first().text()
+ ParserProperties.DESC_SEP);
for (Element infoElem : workElems)
if (!infoElem.text().trim().equals("")) {
strBuilder.append(infoElem.text().trim() + ParserProperties.INFO_SEP);
}
} // END-IF_PAYMENT
// getting services provided information
workElems =
doc.select("div[class*=information-unwanted]")
.select("div[class*=services]")
.select("div[class*=information-services]")
.select("p");
strBuilder.delete(0, strBuilder.length());
if (workElems != null && workElems.size() > 0) {
for (Element infoElem : workElems) {
if (infoElem.hasClass("observation")) {
pEntity.addToExtraInfo(new Utf8("Observation :" + infoElem.text()));
} else if (!infoElem.text().equals("")) {
pEntity.addToServices(new Utf8(infoElem.text()));
}
// LOGGER.debug(infoElem.text());
}
} // END-IF_SERVICES
// getting home url
workElems = doc.select("div[class*=information-unwanted]").select("div[class*=website]");
if (workElems != null && workElems.size() > 0) {
EylloLink homeLink =
ParserUtils.detectUrl(
workElems
.select("div[class*=information-website]")
.select("p")
.select("a")
.first());
if (homeLink != null) {
pEntity.setHomepage(new Utf8(homeLink.getLinkHref()));
pEntity.putToSameAs(new Utf8(homeLink.getLinkHref()), new Utf8(homeLink.getLinkText()));
}
} // END-IF_URL
pEntity.setDescription(new Utf8(""));
} // END-IF_VALID_URL
} // END-WHILE
}
