private void btnCountInterwiki_Click(object sender, EventArgs e)
{
var entityTypes = CurrentActiveEntityTypes();
var entitiesWithWikiData = allEntities.Where(x => x.wiki != null && !String.IsNullOrEmpty(x.wiki.wikidata));
var workItems = entitiesWithWikiData.Where(x => entityTypes.Contains(x.type));
StringBuilder collisions = new StringBuilder();
var siteLinkCount = _bot.CountSiteLinks(workItems, collisions);
StringBuilder builder = new StringBuilder();
builder.AppendFormat(CultureInfo.CurrentUICulture, "{0} entities on WikiData", workItems.Count());
builder.AppendLine();
foreach (var value in siteLinkCount)
{
builder.AppendFormat(CultureInfo.CurrentUICulture, " {0}: {1}", value.Key, value.Value);
builder.AppendLine();
}
var result = builder.ToString();
edtCollisions.Text = collisions.ToString();
var formWikiDataEntries = new StringDisplayForm(
"WikiData language coverage",
result);
formWikiDataEntries.Show();
}
private void btnConvert_Click(object sender, EventArgs e)
{
var entities = new List <Entity>();
UInt32 count = 0;
UInt32 baseGeocode = Convert.ToUInt32(edtGeocode.Value);
foreach (var line in edtText.Lines)
{
var name = line.Replace('\t', ' ').Trim();
if (!String.IsNullOrEmpty(name))
{
count++;
var entityType = EntityType.Muban;
if (chkStripBefore.Checked)
{
var startPosition = name.IndexOf(ThaiLanguageHelper.Ban);
if (startPosition >= 0)
{
name = name.Substring(startPosition);
}
if (chkStripAfter.Checked)
{
name = name.Split(' ').First();
}
}
var entity = new Entity();
entity.name = name;
entity.type = entityType;
entity.geocode = baseGeocode * 100 + count;
entities.Add(entity);
}
}
if (Romanizator != null)
{
List <RomanizationEntry> dummy;
var romanizations = Romanizator.FindRomanizationSuggestions(out dummy, entities);
foreach (var entry in romanizations)
{
var entity = entities.First(x => x.geocode == entry.Geocode);
entity.english = entry.English;
}
}
StringBuilder mubanListBuilder = new StringBuilder();
foreach (var entity in entities)
{
if (!String.IsNullOrEmpty(entity.english))
{
mubanListBuilder.AppendLine(String.Format("<entity type=\"{0}\" geocode=\"{1}\" name=\"{2}\" english=\"{3}\" />",
entity.type, entity.geocode, entity.name, entity.english));
}
else
{
mubanListBuilder.AppendLine(String.Format("<entity type=\"{0}\" geocode=\"{1}\" name=\"{2}\" />",
entity.type, entity.geocode, entity.name));
}
}
var form = new StringDisplayForm("Muban", mubanListBuilder.ToString());
form.Show();
}
private void btnStatistics_Click(object sender, EventArgs e)
{
var entitiesWithWikiData = allEntities.Where(x => x.wiki != null && !String.IsNullOrEmpty(x.wiki.wikidata));
var wikiDataLinks = new List <String>();
wikiDataLinks.AddRange(entitiesWithWikiData.Select(x => x.wiki.wikidata));
var allOffices = allEntities.SelectMany(x => x.office);
//var officesWithWikiData = allOffices.Where(y => y.wiki != null && !String.IsNullOrEmpty(y.wiki.wikidata));
//wikiDataLinks.AddRange(officesWithWikiData.Select(x => x.wiki.wikidata));
// write to CSV file?
var fittingEntitiesByType = entitiesWithWikiData.GroupBy(y => y.type).OrderBy(z => z.Count()).ToList();
var allEntitiesByType = allEntities.Where(x => !x.IsObsolete).GroupBy(y => y.type);
foreach (var expectedType in WikiBase.WikiDataItems)
{
if (expectedType.Key != EntityType.Country)
{
if (allEntitiesByType.Any(x => x.Key == expectedType.Key))
{
if (!fittingEntitiesByType.Any(x => x.Key == expectedType.Key))
{
var emptyEntry = new EntityTypeGrouping <EntityType, Entity>();
emptyEntry.Key = expectedType.Key;
fittingEntitiesByType.Add(emptyEntry);
}
}
}
}
StringBuilder builder = new StringBuilder();
foreach (var type in fittingEntitiesByType)
{
var fittingAllEntities = allEntitiesByType.First(x => x.Key == type.Key);
var expectedCount = fittingAllEntities.Count();
var actualCount = type.Count();
builder.AppendFormat(CultureInfo.CurrentUICulture, "{0}: {1} of {2}", type.Key, type.Count(), expectedCount);
if (actualCount != expectedCount && expectedCount - actualCount < 5)
{
builder.Append(" (");
foreach (var entry in fittingAllEntities)
{
if (!entitiesWithWikiData.Contains(entry))
{
builder.AppendFormat(CultureInfo.CurrentUICulture, "{0},", entry.geocode);
}
}
builder.Append(")");
}
builder.AppendLine();
}
builder.AppendLine();
//var officesWithWikiDataByType = officesWithWikiData.GroupBy(x => x.type).OrderBy(y => y.Count());
//foreach ( var type in officesWithWikiDataByType )
//{
// builder.AppendFormat(CultureInfo.CurrentUICulture,"{0}: {1}", type.Key, type.Count());
// builder.AppendLine();
//}
//builder.AppendLine();
var announcementsWithWikiData = GlobalData.AllGazetteAnnouncements.entry.Where(x => x.wiki != null && !String.IsNullOrEmpty(x.wiki.wikidata));
if (announcementsWithWikiData.Any())
{
builder.AppendFormat(CultureInfo.CurrentUICulture, "Announcements: {0}", announcementsWithWikiData.Count());
builder.AppendLine();
builder.AppendLine();
}
wikiDataLinks.AddRange(announcementsWithWikiData.Select(x => x.wiki.wikidata));
var duplicateWikiDataLinks = wikiDataLinks.GroupBy(x => x).Where(y => y.Count() > 1);
if (duplicateWikiDataLinks.Any())
{
builder.AppendLine("Duplicate links:");
foreach (var wikiDataLink in duplicateWikiDataLinks)
{
builder.AppendLine(wikiDataLink.Key);
}
}
var noUpgradeHistoryEntry = new List <Entity>();
foreach (var entity in allEntities.Where(x => x.type.IsCompatibleEntityType(EntityType.Thesaban) && x.tambonSpecified && !x.IsObsolete))
{
if (!entity.history.Items.Any(x => x is HistoryStatus))
{
noUpgradeHistoryEntry.Add(entity);
}
}
noUpgradeHistoryEntry.Sort((x, y) => x.geocode.CompareTo(y.geocode));
if (noUpgradeHistoryEntry.Any())
{
builder.AppendFormat(CultureInfo.CurrentUICulture, "No history ({0}):", noUpgradeHistoryEntry.Count);
builder.AppendLine();
foreach (var entity in noUpgradeHistoryEntry)
{
builder.AppendFormat(CultureInfo.CurrentUICulture, "{0}: {1}", entity.geocode, entity.english);
builder.AppendLine();
}
}
var result = builder.ToString();
var formWikiDataEntries = new StringDisplayForm(
String.Format(CultureInfo.CurrentUICulture, "Wikidata coverage ({0})", entitiesWithWikiData.Count()),
result);
formWikiDataEntries.Show();
}
private void btnConvert_Click(object sender, EventArgs e)
{
var entities = new List<Entity>();
UInt32 count = 0;
UInt32 baseGeocode = Convert.ToUInt32(edtGeocode.Value);
foreach ( var line in edtText.Lines )
{
var name = line.Replace('\t', ' ').Trim();
if ( !String.IsNullOrEmpty(name) )
{
count++;
var entityType = EntityType.Muban;
if ( chkStripBefore.Checked )
{
var startPosition = name.IndexOf(ThaiLanguageHelper.Ban);
if ( startPosition >= 0 )
{
name = name.Substring(startPosition);
}
if ( chkStripAfter.Checked )
{
name = name.Split(' ').First();
}
}
var entity = new Entity();
entity.name = name;
entity.type = entityType;
entity.geocode = baseGeocode * 100 + count;
entities.Add(entity);
}
}
if ( Romanizator != null )
{
List<RomanizationEntry> dummy;
var romanizations = Romanizator.FindRomanizationSuggestions(out dummy, entities);
foreach ( var entry in romanizations )
{
var entity = entities.First(x => x.geocode == entry.Geocode);
entity.english = entry.English;
}
}
StringBuilder mubanListBuilder = new StringBuilder();
foreach ( var entity in entities )
{
if ( !String.IsNullOrEmpty(entity.english) )
{
mubanListBuilder.AppendLine(String.Format("<entity type=\"{0}\" geocode=\"{1}\" name=\"{2}\" english=\"{3}\" />",
entity.type, entity.geocode, entity.name, entity.english));
}
else
{
mubanListBuilder.AppendLine(String.Format("<entity type=\"{0}\" geocode=\"{1}\" name=\"{2}\" />",
entity.type, entity.geocode, entity.name));
}
}
var form = new StringDisplayForm("Muban", mubanListBuilder.ToString());
form.Show();
}