Geo taxonomy: download the taxonomy data from geonames.org

There are a number of web servies we could use. All have their drawbacks.
This one is free with free data. It's daily limits are reasonably high.
For many coordinates I tested the results were good, for others at least
not terrible.

We can always consider supporting multiple such services. But for now this
seems like a reasonable choice.

Signed-off-by: Dirk Hohndel <dirk@hohndel.org>

2 files changed, 102 insertions, 20 deletions

diff --git a/divesitehelpers.cpp b/divesitehelpers.cpp
index 62b48ba94..4cc40f203 100644
--- a/divesitehelpers.cpp
+++ b/divesitehelpers.cpp
@@ -41,54 +41,135 @@ void ReverseGeoLookupThread::run() {
 
 	QNetworkRequest request;
 	QNetworkAccessManager *rgl = new QNetworkAccessManager();
+	QEventLoop loop;
+	QString mapquestURL("http://open.mapquestapi.com/nominatim/v1/reverse.php?format=json&accept-language=%1&lat=%2&lon=%3");
+	QString geonamesURL("http://api.geonames.org/findNearbyPlaceNameJSON?language=%1&lat=%2&lng=%3&radius=50&username=dirkhh");
+	QString geonamesOceanURL("http://api.geonames.org/oceanJSON?language=%1&lat=%2&lng=%3&radius=50&username=dirkhh");
+	QString divelogsURL("https://www.divelogs.de/mapsearch_divespotnames.php?lat=%1&lng=%2&radius=50");
+	QTimer timer;
+
 	request.setRawHeader("Accept", "text/json");
 	request.setRawHeader("User-Agent", getUserAgent().toUtf8());
-	QEventLoop loop;
-	QString apiCall("http://open.mapquestapi.com/nominatim/v1/reverse.php?format=json&accept-language=%1&lat=%2&lon=%3");
+	connect(&timer, SIGNAL(timeout()), &loop, SLOT(quit()));
+
 	Q_FOREACH (const GeoLookupInfo& info, geo_lookup_data ) {
-		request.setUrl(apiCall.arg(uiLanguage(NULL)).arg(info.lat.udeg / 1000000.0).arg(info.lon.udeg / 1000000.0));
+		struct dive_site *ds = get_dive_site_by_uuid(info.uuid);
+
+		// first check the findNearbyPlaces API from geonames - that should give us country, state, city
+		request.setUrl(geonamesURL.arg(uiLanguage(NULL)).arg(info.lat.udeg / 1000000.0).arg(info.lon.udeg / 1000000.0));
 
 		QNetworkReply *reply = rgl->get(request);
-		QTimer timer;
 		timer.setSingleShot(true);
-
-		QEventLoop loop;
-		connect(&timer, SIGNAL(timeout()), &loop, SLOT(quit()));
 		connect(reply, SIGNAL(finished()), &loop, SLOT(quit()));
-		timer.start(500);   // 30 secs. timeout
+		timer.start(5000);   // 5 secs. timeout
 		loop.exec();
 
 		if(timer.isActive()) {
 			timer.stop();
-			if(reply->error() > 0)
+			if(reply->error() > 0) {
+				report_error("got error accessing geonames.org: %s", reply->errorString());
 				goto clear_reply;
-
+			}
 			int v = reply->attribute(QNetworkRequest::HttpStatusCodeAttribute).toInt();
 			if (v < 200 || v >= 300)
 				goto clear_reply;
-
+			QByteArray fullReply = reply->readAll();
 			QJsonParseError errorObject;
-			QJsonDocument jsonDoc = QJsonDocument::fromJson(reply->readAll(), &errorObject);
-			if (errorObject.error != QJsonParseError::NoError)
+			QJsonDocument jsonDoc = QJsonDocument::fromJson(fullReply, &errorObject);
+			if (errorObject.error != QJsonParseError::NoError) {
+				report_error("error parsing geonames.org response: %s", errorObject.errorString());
 				goto clear_reply;
-
+			}
 			QJsonObject obj = jsonDoc.object();
-			QJsonObject address = obj.value("address").toObject();
-
-			struct dive_site *ds = get_dive_site_by_uuid(info.uuid);
-			ds->notes = add_to_string(ds->notes, "countrytag: %s", address.value("country").toString().toUtf8().data());
-
+			QVariant geoNamesObject = obj.value("geonames").toVariant();
+			QVariantList geoNames = geoNamesObject.toList();
+			if (geoNames.count() > 0) {
+				QVariantMap firstData = geoNames.at(0).toMap();
+				int ri = 0;
+				if (ds->taxonomy.category == NULL)
+					ds->taxonomy.category = alloc_taxonomy();
+				// get all the data - OCEAN is special, so start at COUNTRY
+				for (int j = COUNTRY; j < NR_CATEGORIES; j++) {
+					if (firstData[taxonomy_api_names[j]].isValid()) {
+						ds->taxonomy.category[ri].category = j;
+						ds->taxonomy.category[ri].origin = taxonomy::GEOCODED;
+						free((void*)ds->taxonomy.category[ri].value);
+						ds->taxonomy.category[ri].value = copy_string(qPrintable(firstData[taxonomy_api_names[j]].toString()));
+						ri++;
+					}
+				}
+				ds->taxonomy.nr = ri;
+				mark_divelist_changed(true);
+			} else {
+				report_error("geonames.org did not provide reverse lookup information");
+				qDebug() << "no reverse geo lookup; geonames returned\n" << fullReply;
+			}
 		} else {
+			report_error("timeout accessing geonames.org");
+			disconnect(reply, SIGNAL(finished()), &loop, SLOT(quit()));
+			reply->abort();
+		}
+		// next check the oceans API to figure out the body of water
+		request.setUrl(geonamesOceanURL.arg(uiLanguage(NULL)).arg(info.lat.udeg / 1000000.0).arg(info.lon.udeg / 1000000.0));
+		reply = rgl->get(request);
+		connect(reply, SIGNAL(finished()), &loop, SLOT(quit()));
+		timer.start(5000);   // 5 secs. timeout
+		loop.exec();
+		if(timer.isActive()) {
+			timer.stop();
+			if(reply->error() > 0) {
+				report_error("got error accessing oceans API of geonames.org: %s", reply->errorString());
+				goto clear_reply;
+			}
+			int v = reply->attribute(QNetworkRequest::HttpStatusCodeAttribute).toInt();
+			if (v < 200 || v >= 300)
+				goto clear_reply;
+			QByteArray fullReply = reply->readAll();
+			QJsonParseError errorObject;
+			QJsonDocument jsonDoc = QJsonDocument::fromJson(fullReply, &errorObject);
+			if (errorObject.error != QJsonParseError::NoError) {
+				report_error("error parsing geonames.org response: %s", errorObject.errorString());
+				goto clear_reply;
+			}
+			QJsonObject obj = jsonDoc.object();
+			QVariant oceanObject = obj.value("ocean").toVariant();
+			QVariantMap oceanName = oceanObject.toMap();
+			if (oceanName["name"].isValid()) {
+				if (ds->taxonomy.category == NULL)
+					ds->taxonomy.category = alloc_taxonomy();
+				ds->taxonomy.category[ds->taxonomy.nr].category = OCEAN;
+				qDebug() << "set category of slot" << ds->taxonomy.nr << "to OCEAN(1)";
+				ds->taxonomy.category[ds->taxonomy.nr].origin = taxonomy::GEOCODED;
+				ds->taxonomy.category[ds->taxonomy.nr].value = copy_string(qPrintable(oceanName["name"].toString()));
+				ds->taxonomy.nr++;
+				mark_divelist_changed(true);
+			}
+		} else {
+			report_error("timeout accessing geonames.org");
 			disconnect(reply, SIGNAL(finished()), &loop, SLOT(quit()));
 			reply->abort();
 		}
 
-		clear_reply:
+clear_reply:
 		reply->deleteLater();
 	}
 	rgl->deleteLater();
 }
 
+void ReverseGeoLookupThread::lookup(dive_site *ds)
+{
+	if (!ds)
+		return;
+	GeoLookupInfo info;
+	info.lat = ds->latitude;
+	info.lon = ds->longitude;
+	info.uuid = ds->uuid;
+
+	geo_lookup_data.clear();
+	geo_lookup_data.append(info);
+	run();
+}
+
 extern "C" void add_geo_information_for_lookup(degrees_t latitude, degrees_t longitude, uint32_t uuid) {
 	GeoLookupInfo info;
 	info.lat = latitude;
diff --git a/divesitehelpers.h b/divesitehelpers.h
index dad60be36..a08069bc0 100644
--- a/divesitehelpers.h
+++ b/divesitehelpers.h
@@ -8,6 +8,7 @@ class ReverseGeoLookupThread : public QThread {
 Q_OBJECT
 public:
 	static ReverseGeoLookupThread *instance();
+	void lookup(struct dive_site *ds);
 	void run() Q_DECL_OVERRIDE;
 
 private: