bool HeadersInData(DelimiterInfo info, string[] headerValues, string[] rows)
{
var duplicate = 0;
var first = true;
foreach (var row in rows)
{
if (first)
{
first = false;
continue;
}
var values = row.Split(new char[] { info.Delimiter });
if (values.Length != headerValues.Length)
{
continue;
}
for (int i = 0; i < values.Length; i++)
{
if (values [i] == headerValues [i])
{
duplicate++;
}
}
}
return(duplicate >= rows.Length * 0.25);
}
private void AdjustConfidence(RecordFormatInfo format, DelimiterInfo info)
{
switch (info.Delimiter)
{
case '"': // Avoid the quote identifier
case '\'': // Avoid the quote identifier
format.mConfidence = (int)(format.Confidence * 0.2);
break;
case '/': // Avoid the date delimiters and url to be selected
case '.': // Avoid the decimal separator to be selected
format.mConfidence = (int)(format.Confidence * 0.4);
break;
case '@': // Avoid the mails separator to be selected
case '&': // Avoid this is near a letter and URLS
case '=': // Avoid because URLS contains it
format.mConfidence = (int)(format.Confidence * 0.6);
break;
case '-': // Avoid this other date separator
format.mConfidence = (int)(format.Confidence * 0.7);
break;
case ',': // Help the , ; tab ~ to be confident
case ';':
case '\t':
case '~':
format.mConfidence = (int)Math.Min(100, format.Confidence * 1.15);
break;
}
}
bool SameFirstLine(DelimiterInfo info, string[][] sampleData)
{
for (int i = 1; i < sampleData.Length; i++)
{
if (!SameHeaders(info, sampleData [0][0], sampleData [i][0]))
{
return(false);
}
}
return(true);
}
bool DetectIfContainsHeaders(DelimiterInfo info, string[][] sampleData)
{
if (sampleData.Length >= 2)
{
return(SameFirstLine(info, sampleData));
}
if (sampleData.Length >= 1)
{
var firstLine = sampleData [0] [0].Split(new char[] { info.Delimiter });
var res = AreAllHeaders(firstLine);
if (res == false)
{
return(false); // if has headers that starts with numbers so near sure are data and no header is present
}
if (HeadersInData(info, firstLine, sampleData[0]))
{
return(false);
}
return(true);
}
return(false);
}
private void AdjustConfidence(RecordFormatInfo format, DelimiterInfo info)
{
switch (info.Delimiter)
{
case '"': // Avoid the quote identifier
case '\'': // Avoid the quote identifier
format.mConfidence = (int)(format.Confidence * 0.2);
break;
case '/': // Avoid the date delimiters and url to be selected
case '.': // Avoid the decimal separator to be selected
format.mConfidence = (int)(format.Confidence * 0.4);
break;
case '@': // Avoid the mails separator to be selected
case '&': // Avoid this is near a letter and URLS
case '=': // Avoid because URLS contains it
format.mConfidence = (int)(format.Confidence * 0.6);
break;
case '-': // Avoid this other date separator
format.mConfidence = (int)(format.Confidence * 0.7);
break;
case ',': // Help the , ; tab to be confident
case ';':
case '\t':
format.mConfidence = (int)Math.Min(100, format.Confidence * 1.15);
break;
}
}
bool SameHeaders(DelimiterInfo info, string line1, string line2)
{
return(line1.Replace(info.Delimiter.ToString(), "").Trim()
== line2.Replace(info.Delimiter.ToString(), "").Trim());
}
bool SameHeaders(DelimiterInfo info, string line1, string line2)
{
return line1.Replace (info.Delimiter.ToString (), "").Trim ()
== line2.Replace (info.Delimiter.ToString (), "").Trim ();
}
bool SameFirstLine(DelimiterInfo info, string[][] sampleData)
{
for (int i = 1; i < sampleData.Length; i++) {
if (!SameHeaders (info, sampleData [0][0], sampleData [i][0]))
return false;
}
return true;
}
bool HeadersInData(DelimiterInfo info, string[] headerValues, string[] rows)
{
var duplicate = 0;
var first = true;
foreach (var row in rows) {
if (first) {
first = false;
continue;
}
var values = row.Split (new char[]{ info.Delimiter });
if (values.Length != headerValues.Length)
continue;
for (int i = 0; i < values.Length; i++) {
if (values [i] == headerValues [i])
duplicate++;
}
}
return duplicate >= rows.Length * 0.25;
}
bool DetectIfContainsHeaders(DelimiterInfo info, string[][] sampleData)
{
if (sampleData.Length >= 2) {
return SameFirstLine (info, sampleData);
}
if (sampleData.Length >= 1) {
var firstLine = sampleData [0] [0].Split (new char[]{ info.Delimiter });
var res = AreAllHeaders (firstLine);
if (res == false)
return false; // if has headers that starts with numbers so near sure are data and no header is present
if (HeadersInData(info, firstLine, sampleData[0]))
return false;
return true;
}
return false;
}
