public WordHistogram Clone()
{
WordHistogram clone = new WordHistogram(_language);
foreach (KeyValuePair<string, int> kv in _frequencies)
{
clone._frequencies[kv.Key] = kv.Value;
}
return clone;
}
private static void CompressColumns(Dictionary<string, AggregationPolicy> aggregationPolicies, PoiService aggregate)
{
// Compress all keyword columns to a single column (this may be a setting later).
var labelsWithKeywordAggregation = aggregationPolicies.Where(kv => kv.Value.IsKeywordAggregation).Select(kv => kv.Key);
foreach (BaseContent poi in aggregate.PoIs)
{
WordHistogram sumHistogram = null;
foreach (string label in labelsWithKeywordAggregation)
{
string labelValue;
if (poi.Labels.TryGetValue(label, out labelValue))
{
WordHistogram wordHistogram = new WordHistogram();
wordHistogram.Append(labelValue);
if (sumHistogram == null)
{
sumHistogram = wordHistogram.Clone();
}
else
{
sumHistogram.Merge(wordHistogram);
}
}
}
if (sumHistogram != null)
{
poi.Keywords = sumHistogram;
}
}
// Remove all columns that have been compressed to keywords.
foreach (BaseContent poI in aggregate.PoIs)
{
foreach (KeyValuePair<string, AggregationPolicy> kv in aggregationPolicies)
{
if (!kv.Value.DataIsNumeric &&
kv.Value.NonNumericAggregationPolicy == AggregationPolicy.NonNumericAggregation.Keywords)
{
poI.Labels.Remove(kv.Key);
}
}
}
// Remove obsolete labels; only in the PoiTypes; they are already gone from the Pois.
foreach (BaseContent baseContent in aggregate.PoITypes)
{
List<string> toOmit = new List<string>();
foreach (string label in baseContent.Labels.Keys)
{
AggregationPolicy policy;
bool remove = true;
if (aggregationPolicies.TryGetValue(label, out policy))
{
if (!policy.IsOmit &&
!(policy.DataIsNumeric &&
policy.NonNumericAggregationPolicy == AggregationPolicy.NonNumericAggregation.Keywords))
{
remove = false;
}
}
if (remove)
{
toOmit.Add(label);
}
}
foreach (string label in toOmit)
{
baseContent.Labels.Remove(label);
}
}
}
private string AggregateNonNumeric(IEnumerable<string> data)
{
switch (NonNumericAggregationPolicy)
{
case NonNumericAggregation.Concatenate:
StringBuilder sb = new StringBuilder();
foreach (string s in data)
{
sb.Append(s);
}
return sb.ToString();
case NonNumericAggregation.KeepFirst:
return data.FirstOrDefault();
case NonNumericAggregation.Keywords:
StringBuilder allText = new StringBuilder();
foreach (string text in data)
{
allText.Append(text).Append(" ");
}
WordHistogram histogram = new WordHistogram();
return histogram.ToGeoJson(); // Pity, we only support aggregation to strings.
case NonNumericAggregation.Omit:
default:
return null;
}
}
public void FromXmlBase(ref XElement res, string directoryName)
{
//var xmlSerializer = new XmlSerializer(typeof (PoI));
//var p = xmlSerializer.Deserialize(res.CreateReader());
try
{
Id = res.GetGuid("Id");
if (Id == Guid.Empty) Id = Guid.NewGuid();
var n = res.GetString("Name");
ContentId = res.GetString("PoiId", "");
if (String.IsNullOrEmpty(ContentId)) ContentId = n;
Priority = res.GetInt("Priority", 2);
UserId = res.GetString("UserId");
DateLong = res.GetLong("Date", DateTime.Now.ToEpoch());
UpdatedLong = res.GetLong("Updated", DateTime.Now.ToEpoch());
Layer = res.GetString("Layer");
MaxItems = res.GetNullInt("MaxItems");
var xMid = res.Element("MetaInfoData");
PoiTypeId = res.GetString("PoiTypeId", "");
IsVisibleInMenu = res.GetBool("IsVisibleInMenu", true);
Orientation = res.GetDouble("Orientation", 0.0);
//if (!string.IsNullOrEmpty(PoiTypeId))
//{
//}
if (xMid != null)
{
var metaInfo = new MetaInfoCollection();
foreach (var xMi in xMid.Elements())
{
var mi = new MetaInfo();
mi.FromXml(xMi);
metaInfo.Add(mi);
}
MetaInfo = metaInfo;
}
if (res.Element("WKT") != null)
{
var xElement = res.Element("WKT");
if (xElement != null) WktText = xElement.Value;
}
var xlabels = res.Element("Labels");
if (xlabels != null)
{
Labels = new Dictionary<string, string>();
foreach (var xk in xlabels.Elements())
{
var k = xk.Name.LocalName;
// Restore keys starting with numbers or having % or '.
k = k.Replace(LabelPercentSubst, "%");
k = k.Replace(LabelQuoteSubst, "'");
if (k.StartsWith(LabelNumPrefix))
{
k = k.Substring(LabelNumPrefix.Length);
}
var s = xk.InnerXml();
Labels[k] = s.RestoreInvalidCharacters();
Labels[k] = Labels[k].Replace("<", "<").Replace(">", ">");
}
}
var xkeywords = res.Element("Keywords");
if (xkeywords != null)
{
Keywords = new WordHistogram();
Keywords.FromXml(xkeywords);
}
if (res.Element("Style") != null)
{
try
{
var newStyle = new PoIStyle();
newStyle.FromXml(res.Element("Style"), directoryName, false); //, Service.Settings); // TODO REVIEW: Settings were ignored.
Style = newStyle;
}
catch (Exception)
{
// OK, keep the old style.
}
}
var media = res.Element("AllMedia");
if (media != null)
{
AllMedia = new BindableCollection<Media>();
foreach (var m in media.Elements())
{
var me = new Media { Content = this };
me.FromXml(m);
AllMedia.Add(me);
}
}
var xpos = res.Element("Position");
if (xpos != null)
Position = new Position(xpos.GetDouble(Position.LONG_LABEL), xpos.GetDouble(Position.LAT_LABEL), xpos.GetDouble(Position.ALT_LABEL)); // TODO Remember other Position attributes.
var px = res.Element("Points");
var mo = res.Element("Models");
if (mo != null)
{
Models = new List<Model>();
foreach (var xm in mo.Elements())
{
var m = new Model();
m.FromXml(xm);
Models.Add(m);
}
}
if (px == null) return;
var pp = px.Value;
Points = new ObservableCollection<Point>();
var ppo = pp.Split(' ');
foreach (var poss in ppo)
{
var split = poss.Split(',');
var pt = new Point(
Double.Parse(split[0], CultureInfo.InvariantCulture),
Double.Parse(split[1], CultureInfo.InvariantCulture));
Points.Add(pt);
}
}
catch (SystemException e)
{
Logger.Log("DataServer.BaseContent", "Error reading XML " + res + " from " + directoryName, e.Message, Logger.Level.Error, true);
}
}
public IConvertibleGeoJson FromGeoJson(JObject geoJsonObject, bool newObject = true)
{
var poi = (newObject) ? new PoI() : this;
//var lockId = f["properties"]["lock"].Value<string>();
//var pos = f["properties"]["pos"].Value<string>();
//var angle = f["properties"]["angle"].Value<double>();
foreach (var prp in geoJsonObject["properties"].OfType<JProperty>())
{
// TODO REVIEW There are more properties of PoIs that should not end up in the labels, but should be set to a property. Mentioned under ToGeoJson:
// Orientation, ContentId, Layer, Date, MaxItems.
if (Equals(prp.Name, "Id"))
{
poi.PoiId = prp.Value.ToString();
}
else if (Equals(prp.Name, "FeatureTypeId"))
{
poi.PoiTypeId = prp.Value.ToString();
}
else
{
poi.Labels[prp.Name] = prp.Value.ToString().RestoreInvalidCharacters();
}
}
JToken keywordToken;
if (geoJsonObject.TryGetValue("keywords", out keywordToken))
{
var histogram = new WordHistogram();
histogram.FromGeoJson(keywordToken.ToString(Formatting.None), false); // Not very efficient.
Keywords = histogram;
}
var wkt = (new WellKnownTextIO().FromGeoJson(geoJsonObject["geometry"].ToString(Formatting.None), false)).ToString(); // TODO More efficiency if we allow JObjects as input.
if (!string.IsNullOrEmpty(wkt))
{
poi.WktText = wkt;
}
return poi;
}
public void Merge(WordHistogram other)
{
foreach (KeyValuePair<string, int> kv in other._frequencies)
{
int currentFrequency;
if (_frequencies.TryGetValue(kv.Key, out currentFrequency))
{
_frequencies[kv.Key] = currentFrequency + other._frequencies[kv.Key];
}
else
{
_frequencies[kv.Key] = other._frequencies[kv.Key];
}
}
}
public static string DistinctWordsInText(this string text, string language = null)
{
if (language == null) language = text.Language(); // Slow approximate language detector.
IEnumerable<string> distinctWords = new WordHistogram(language, text).DistinctWords;
return String.Join(" ", distinctWords);
}
public static WordHistogram Histogram(this string text)
{
WordHistogram wordHistogram = new WordHistogram(text.Language(), text); // Slow approximate language detector.
return wordHistogram;
}
