public static Lib.Data.Firma FirmaInText(string text)
{
if (string.IsNullOrEmpty(text))
{
return(null);
}
string value = Devmasters.TextUtil.RemoveDiacritics(TextUtil.NormalizeToBlockText(text).ToLower());
foreach (var k in Firma.Koncovky.Select(m => m.ToLower()).OrderByDescending(m => m.Length))
{
if (value.Contains(k))
{
value = value.Replace(k, k.Replace(' ', (char)160)); //nahrad mezery char(160) - non breaking space, aby to tvorilo 1 slovo
}
else if (k.EndsWith(".") && value.EndsWith(k.Substring(0, k.Length - 1)))
{
value = value.Replace(k.Substring(0, k.Length - 1), k.Replace(' ', (char)160)); //nahrad mezery char(160) - non breaking space, aby to tvorilo 1 slovo
}
}
//find company name
string[] words = value.Split(new char[] { ' ', ',', ';' }, StringSplitOptions.RemoveEmptyEntries);
//get back space instead of #160
words = words.Select(m => m.Replace((char)160, ' ')).ToArray();
for (int firstWord = 0; firstWord < words.Length; firstWord++)
{
for (int skipWord = 0; skipWord < words.Length - firstWord; skipWord++)
{
string[] cutWords = words.Skip(firstWord) //preskoc slovo na zacatku
.Reverse().Skip(skipWord).Reverse() // a ubirej od konce
.ToArray();
string wordCombination = cutWords.Aggregate((f, s) => f + " " + s);
string koncovka;
string firmaBezKoncovky = Lib.Data.Firma.JmenoBezKoncovkyFull(wordCombination, out koncovka);
string simpleName = Devmasters.TextUtil.RemoveDiacritics(firmaBezKoncovky).ToLower().Trim();
//+ "|" + koncovka;
if (firmaBezKoncovky.Length > 3 &&
StaticData.FirmyNazvyOnlyAscii.Get().ContainsKey(simpleName)
)
{
//nasel jsem ico?
foreach (var ico in StaticData.FirmyNazvyOnlyAscii.Get()[simpleName])
{
Firma f = Firmy.Get(ico); //TODO StaticData.FirmyNazvyAscii.Get()[simpleName]);
if (f.Valid)
{
var firmaFromText = TextUtil.ReplaceDuplicates(System.Text.RegularExpressions.Regex.Replace(wordCombination, @"[,;_""']", " ", defaultRegexOptions), ' ');
var firmaFromDB = TextUtil.ReplaceDuplicates(System.Text.RegularExpressions.Regex.Replace(f.Jmeno, @"[,;_""']", " ", defaultRegexOptions), ' ');
var rozdil = LevenshteinDistanceCompute(
TextUtil.RemoveDiacritics(firmaFromDB).ToLower(),
firmaFromText.ToLower()
);
var fKoncovka = f.KoncovkaFirmy();
var nextWord = "";
if (firstWord + cutWords.Length < words.Length - 1)
{
nextWord = words[firstWord + cutWords.Length];
}
if (string.IsNullOrEmpty(fKoncovka))
{
return(f);
}
if (!string.IsNullOrEmpty(fKoncovka) && LevenshteinDistanceCompute(cutWords.Last(), fKoncovka) < 2)
{
return(f);
}
if (!string.IsNullOrEmpty(fKoncovka) && LevenshteinDistanceCompute(nextWord, fKoncovka) < 2)
{
return(f);
}
}
}
//looking for next
//return null;
}
}
}
return(null);
}
