public static void TestParser(string[] srcNameList, List <string> trgNameList)
{
var sbp = new StatsBasedParser(trgNameList);
//int z = 0; foreach (var srcName in srcNameList) { Debug.WriteLine("{0}", ++z); foreach (var m in sbp.ClosestMatches(srcName)) Debug.WriteLine("{0,64} => {1}", srcName, m); }
if (Debugger.IsAttached)
{
Debugger.Break();
}
}
public List <string> ClosestMatches(string src)
{
//wp>Debug.WriteLine(src, "\n::");
var matches = new List <string>();
//0: if there is an exact match:
if (_trgList.Contains(src, StringComparer.OrdinalIgnoreCase)) //tu: !!!!!!!!! //todo: remove ToLower from word stats view !!!!!!!
{
var exactMatch = _trgList.FirstOrDefault(r => string.Compare(r, src, true) == 0);
MatchBaseCsv = exactMatch;
matches.Add(exactMatch);
return(matches);
}
//1a: order by FoU:
var wa = src.Split(OlpPresets.DelimAll, StringSplitOptions.RemoveEmptyEntries);
var srcWordsByFoU = GetWordStats(wa, _wordStatCollection);
//1: if there is a SINGLE match by a single word with Fou==1:
if (srcWordsByFoU.Count(r => r.Usage == 1) == 1)
{
var singleUseWord = srcWordsByFoU.First(r => r.Usage == 1);
var trgMatch = _trgList.First(r => r.ToLower().Contains(singleUseWord.Word.ToLower()));
matches.Add(trgMatch);
MatchBaseCsv = string.Join("·", srcWordsByFoU.Select(r => r.Word));
return(matches);
}
//1: if there is more than 1 match by a single word with Fou==1:
if (srcWordsByFoU.Count(r => r.Usage == 1) > 1)
{
foreach (var matchByFoU1 in srcWordsByFoU.Where(r => r.Usage == 1))
{
if (!matches.Contains(matchByFoU1.Word, StringComparer.OrdinalIgnoreCase))
{
matches.Add(matchByFoU1.Word);
}
}
Debug.Assert(matches.Count == 1, "Need more wits to find a better way to tell which one matches");
MatchBaseCsv = string.Join("·", srcWordsByFoU.Select(r => r.Word));
return(matches);
}
//3: if contains all the words from src name, ignoring the org order
var matchByAllWords = _trgList.Where(r => StatsBasedParser.ContainsAllWordsOpt(r, wa));
if (matchByAllWords.Any())
{
MatchBaseCsv = string.Join("·", wa);
return(matchByAllWords.ToList());
}
//3: if contains the words from src name and stats array
if (srcWordsByFoU.Count() > 1)
{
for (int i = srcWordsByFoU.Count(); i > 0; i--)
{
var notAllOrdered = srcWordsByFoU.OrderBy(r => r.Usage).Take(i).Select(r => r.Word).ToArray();
var matchBy1stWords = _trgList.Where(r => StatsBasedParser.ContainsAllWordsOpt(r, notAllOrdered));
if (matchBy1stWords.Any())
{
MatchBaseCsv = string.Join("·", notAllOrdered);
return(matchBy1stWords.ToList());
}
}
}
return(matches);
}