private void AppendNameInfo(StringBuilder builder)
{
builder.AppendLine(String.Format("Name equal: {0} times", SuffixFrequency(String.Empty)));
List <String> standardSuffices = new List <String>()
{
"เหนือ", "ใต้", "พัฒนา", "ใหม่", "ทอง", "น้อย", "ใน"
};
foreach (String suffix in standardSuffices)
{
builder.AppendLine(String.Format("Suffix {0}: {1} times", suffix, SuffixFrequency(suffix)));
}
builder.AppendLine("Suffix with number:" + SuffixFrequencyNumbers().ToString() + " times");
List <String> standardPrefixes = new List <String>()
{
"ใหม่"
};
foreach (String prefix in standardPrefixes)
{
builder.AppendLine(String.Format("Prefix {0}: {1} times", prefix, PrefixFrequency(prefix)));
}
builder.AppendLine();
builder.Append("Other suffices: ");
var sortedSuffices = new List <KeyValuePair <String, Int32> >();
foreach (var keyValuePair in _newNameSuffix)
{
String name = ThaiNumeralHelper.ReplaceThaiNumerals(keyValuePair.Key);
if (standardSuffices.Contains(name))
{
}
else if (String.IsNullOrEmpty(keyValuePair.Key))
{
}
else if (name.IsNumeric())
{
}
else
{
sortedSuffices.Add(keyValuePair);
}
}
sortedSuffices.Sort(delegate(KeyValuePair <String, Int32> x, KeyValuePair <String, Int32> y)
{
return(y.Value.CompareTo(x.Value));
});
foreach (var keyValuePair in sortedSuffices)
{
builder.Append(keyValuePair.Key + " (" + keyValuePair.Value.ToString() + ") ");
}
builder.AppendLine();
}
protected Int32 SuffixFrequencyNumbers()
{
Int32 result = 0;
foreach (var keyValue in _newNameSuffix)
{
String name = ThaiNumeralHelper.ReplaceThaiNumerals(keyValue.Key);
if ((!String.IsNullOrEmpty(name)) && (name.IsNumeric()))
{
result = result + keyValue.Value;
}
}
return(result);
}
private GazetteEntry ParseSingeItem(String value)
{
value = value.Replace("\t", "");
GazetteEntry retval = null;
Int32 position = value.IndexOf(EntryURL);
if (position >= 0)
{
retval = new GazetteEntry();
position = position + EntryURL.Length;
Int32 position2 = value.IndexOf(EntryURLend);
retval.uri = value.Substring(position, position2 - position);
value = value.Substring(position2, value.Length - position2);
position = value.IndexOf(EntryTitle) + EntryTitle.Length;
position2 = value.IndexOf(EntryTitleEnd);
retval.title = value.Substring(position, position2 - position).Trim();
value = value.Substring(position2, value.Length - position2);
position = value.IndexOf(EntryVolumeorPage) + EntryVolumeorPage.Length;
position2 = value.IndexOf(ColumnEnd, position);
string volume = value.Substring(position, position2 - position);
retval.volume = Convert.ToByte(ThaiNumeralHelper.ReplaceThaiNumerals(volume));
value = value.Substring(position2, value.Length - position2);
position = value.IndexOf(EntryIssue) + EntryIssue.Length;
position2 = value.IndexOf(ColumnEnd, position);
retval.issue = ThaiNumeralHelper.ReplaceThaiNumerals(value.Substring(position, position2 - position).Trim());
value = value.Substring(position2, value.Length - position2);
position = value.IndexOf(EntryDate) + EntryDate.Length;
position2 = value.IndexOf(ColumnEnd, position);
string Date = value.Substring(position, position2 - position);
retval.publication = ThaiDateHelper.ParseThaiDate(Date);
value = value.Substring(position2, value.Length - position2);
position = value.IndexOf(EntryVolumeorPage) + EntryVolumeorPage.Length;
position2 = value.IndexOf(ColumnEnd, position);
string page = value.Substring(position, position2 - position);
retval.page = ThaiNumeralHelper.ReplaceThaiNumerals(page);
if (retval.title.Contains('[') && retval.title.EndsWith("]"))
{
var beginSubTitle = retval.title.LastIndexOf('[');
retval.subtitle = retval.title.Substring(beginSubTitle + 1, retval.title.Length - beginSubTitle - 2).Trim();
retval.title = retval.title.Substring(0, beginSubTitle - 1).Trim();
}
}
return(retval);
}
public static DateTime ParseThaiDate(String value)
{
String monthString = String.Empty;
Int32 month = 0;
String yearString = String.Empty;
Int32 year = 0;
Int32 day = 0;
Int32 position = 0;
String date = ThaiNumeralHelper.ReplaceThaiNumerals(value);
position = date.IndexOf(' ');
day = Convert.ToInt32(date.Substring(0, position), CultureInfo.InvariantCulture);
date = date.Substring(position + 1, date.Length - position - 1);
position = date.IndexOf(' ');
monthString = date.Substring(0, position).Trim();
month = ThaiMonthNames[monthString];
// TODO: Weren't there some very old ones with KhoSo as well?
position = date.IndexOf(BuddhistEra, StringComparison.Ordinal) + BuddhistEra.Length;
yearString = date.Substring(position, date.Length - position);
year = Convert.ToInt32(yearString, CultureInfo.InvariantCulture);
if (year < 2100)
{
year = year + 543; // there are entries in KhoSo but with "พ.ศ." in the returned info
}
if ((year < 2484) & (month < 4))
{
year = year - 542;
}
else
{
year = year - 543;
}
return(new DateTime(year, month, day));
}
public List <RomanizationEntry> FindRomanizationSuggestions(out List <RomanizationEntry> romanizationMissing, IEnumerable <Entity> entitiesToCheck)
{
var result = new List <RomanizationEntry>();
romanizationMissing = new List <RomanizationEntry>();
foreach (var entityToCheck in entitiesToCheck)
{
if (String.IsNullOrEmpty(entityToCheck.name))
{
continue;
}
if (String.IsNullOrEmpty(entityToCheck.english))
{
String foundEnglishName = String.Empty;
if (Romanizations.Keys.Contains(entityToCheck.name))
{
foundEnglishName = entityToCheck.name;
}
else
{
var searchName = entityToCheck.name.StripBanOrChumchon();
if (Romanizations.Keys.Contains(searchName))
{
foundEnglishName = searchName;
}
else
{
// Chumchon may have the name "Chumchon Ban ..."
searchName = searchName.StripBanOrChumchon();
if (Romanizations.Keys.Contains(searchName))
{
foundEnglishName = searchName;
}
}
}
if (!String.IsNullOrEmpty(foundEnglishName))
{
result.Add(CreateRomanizationEntry(entityToCheck, Romanizations[foundEnglishName]));
}
else
{
Boolean found = false;
String name = entityToCheck.name.StripBanOrChumchon();
name = ThaiNumeralHelper.ReplaceThaiNumerals(name);
List <Char> numerals = new List <Char>()
{
'0', '1', '2', '3', '4', '5', '6', '7', '8', '9'
};
foreach (Char c in numerals)
{
name = name.Replace(c, ' ');
}
name = name.Trim();
foreach (var keyValuePair in ThaiLanguageHelper.NameSuffixRomanizations)
{
if (entityToCheck.name.EndsWith(keyValuePair.Key))
{
String searchString = name.Substring(0, name.Length - keyValuePair.Key.Length).StripBanOrChumchon();
if (String.IsNullOrEmpty(searchString))
{
result.Add(CreateRomanizationEntry(entityToCheck, keyValuePair.Value));
found = true;
}
else if (Romanizations.Keys.Contains(searchString))
{
result.Add(CreateRomanizationEntry(entityToCheck, Romanizations[searchString] + " " + keyValuePair.Value));
found = true;
}
}
}
if (!found)
{
var prefixes = ThaiLanguageHelper.NamePrefixRomanizations.Union(ThaiLanguageHelper.NameSuffixRomanizations);
foreach (var keyValuePair in prefixes)
{
if (name.StartsWith(keyValuePair.Key))
{
String searchString = name.Substring(keyValuePair.Key.Length);
if (String.IsNullOrEmpty(searchString))
{
result.Add(CreateRomanizationEntry(entityToCheck, keyValuePair.Value));
found = true;
}
else if (Romanizations.Keys.Contains(searchString))
{
result.Add(CreateRomanizationEntry(entityToCheck, keyValuePair.Value + " " + Romanizations[searchString]));
found = true;
}
}
}
}
if (!found)
{
romanizationMissing.Add(new RomanizationEntry(entityToCheck.geocode, entityToCheck.name));
}
}
}
}
return(result);
}