public void Read10000Format(string filePath)
{
var sb = new StringBuilder();
int lineCounter = 0;
// read in the dictionary file in the wiktionary_frequency_list.txt format
foreach (var line in File.ReadLines(filePath, _isoLatin1Encoding)) {
lineCounter++;
// ignore blank lines
if (string.IsNullOrEmpty(line))
continue;
// parse
var elements = line.Split(new String[] {
" "
}, StringSplitOptions.RemoveEmptyEntries);
// ignore words with only one letter or digit
//if (elements[1].Length < 2) continue;
var word = new WordElement();
wordElements.Add(word);
word.Place = lineCounter;
word.Frequency = int.Parse(elements[0]);
word.Word = elements[1];
// also add to the string
sb.Append(word.Word);
sb.Append("\r\n");
}
words = sb.ToString();
}
public void Read10KFormat(string filePath)
{
var sb = new StringBuilder();
int lineCounter = 0;
// read in the dictionary file in the ord10k.csv format
foreach (var line in File.ReadLines(filePath, _isoLatin1Encoding))
{
lineCounter++;
// skip header
if (lineCounter == 1)
{
continue;
}
// ignore blank lines
if (string.IsNullOrEmpty(line))
{
continue;
}
// parse
var elements = line.Split(new String[] {
","
}, StringSplitOptions.RemoveEmptyEntries);
// ignore words with only one letter or digit
//if (elements[4].Length < 2) continue;
var word = new WordElement();
wordElements.Add(word);
word.Place = int.Parse(elements[0]);
word.Frequency = int.Parse(elements[1]);
word.Word = elements[4];
// also add to the string
sb.Append(word.Word);
sb.Append("\r\n");
}
words = sb.ToString();
}
