static void Main(string[] args)
{
// Load first document
TextExtractor document1 = new TextExtractor();
document1.RegistrationName = "demo";
document1.RegistrationKey = "demo";
document1.LoadDocumentFromFile(@".\comparison1.pdf");
// Load second document
TextExtractor document2 = new TextExtractor();
document2.RegistrationName = "demo";
document2.RegistrationKey = "demo";
document2.LoadDocumentFromFile(@".\comparison2.pdf");
// Compare documents
TextComparer comparer = new TextComparer();
comparer.RegistrationName = "demo";
comparer.RegistrationKey = "demo";
comparer.Compare(document1, document2);
// Generate report
comparer.GenerateHtmlReport(@".\report.html");
document1.Dispose();
document2.Dispose();
// Open the report in default browser
ProcessStartInfo processStartInfo = new ProcessStartInfo(@".\report.html");
processStartInfo.UseShellExecute = true;
Process.Start(processStartInfo);
}
public void Compare_StringAndValue_True()
{
var comparer = new TextComparer();
var result = comparer.Compare("string", "(value)");
Assert.That(result.AreEqual, Is.True);
}
public void Compare_NullAndAny_True()
{
var comparer = new TextComparer();
var result = comparer.Compare(null, "(any)");
Assert.That(result.AreEqual, Is.True);
}
public void Compare_NullAndString_False()
{
var comparer = new TextComparer();
var result = comparer.Compare(null, "string");
Assert.That(result.AreEqual, Is.False);
}
static void Main(string[] args)
{
// Load first document
TextExtractor document1 = new TextExtractor();
document1.RegistrationName = "demo";
document1.RegistrationKey = "demo";
document1.LoadDocumentFromFile("comparison1.pdf");
// Load second document
TextExtractor document2 = new TextExtractor();
document2.RegistrationName = "demo";
document2.RegistrationKey = "demo";
document2.LoadDocumentFromFile("comparison2.pdf");
// Compare documents
TextComparer comparer = new TextComparer();
comparer.RegistrationName = "demo";
comparer.RegistrationKey = "demo";
comparer.Compare(document1, document2);
// Generate report
comparer.GenerateHtmlReport("report.html");
// Open the report in browser
System.Diagnostics.Process.Start("report.html");
}
public void Compare_NullAndValue_False()
{
var comparer = new TextComparer();
var result = comparer.Compare(null, "(value)");
Assert.That(result.AreEqual, Is.False);
}
public void Compare_TwoSpacesAndBlankPlaceHolder_True()
{
var comparer = new TextComparer();
var result = comparer.Compare(" ", "(blank)");
Assert.That(result.AreEqual, Is.True);
}
public void Compare_StringAndSubstring_False()
{
var comparer = new TextComparer();
var result = comparer.Compare("string", "str");
Assert.That(result.AreEqual, Is.False);
}
public void Compare_StringAndSameStringUppercase_False()
{
var comparer = new TextComparer();
var result = comparer.Compare("string", "STRING");
Assert.That(result.AreEqual, Is.False);
}
public void Compare_1AndEmptyPlaceHolder_False()
{
var comparer = new TextComparer();
var result = comparer.Compare(1, "(empty)");
Assert.That(result.AreEqual, Is.False);
}
public void Compare_EmptyAndEmptyPlaceHolder_True()
{
var comparer = new TextComparer();
var result = comparer.Compare(string.Empty, "(empty)");
Assert.That(result.AreEqual, Is.True);
}
public void Compare_NullAndNullPlaceHolder_True()
{
var comparer = new TextComparer();
var result = comparer.Compare(null, "(null)");
Assert.That(result.AreEqual, Is.True);
}
public void Compare_IgnoreCase_true(string def)
{
var tolerance = new TextToleranceFactory().Instantiate(def);
var comparer = new TextComparer();
var result = comparer.Compare("Seddryck", "seddryck", tolerance);
Assert.That(result.AreEqual, Is.True);
}
public void Compare_LargeDifference_False(string def)
{
var tolerance = new TextToleranceFactory().Instantiate(def);
var comparer = new TextComparer();
var result = comparer.Compare("Seddryck", "Undefined", tolerance);
Assert.That(result.AreEqual, Is.False);
}
protected override bool ApplyWithReference(object reference, object x)
{
var cpr = new TextComparer();
return(cpr.Compare(x, reference).AreEqual);
}
public bool Compare(object x, object y)
{
var cpr = new TextComparer();
return(cpr.Compare(x, y).AreEqual);
}
private void CompareValues(List <CompareHelper> keyMatchingRows, List <DataRow> nonMatchingValueRows)
{
if (Settings.KeysDef != ResultSetComparisonSettings.KeysChoice.All)
{
foreach (var rxHelper in keyMatchingRows)
{
var ryHelper = yDict[rxHelper.Keys];
//if (ryHelper.ValuesHashed == rxHelper.ValuesHashed)
//{
// // quick shortcut. If the hash of the values matches, then there is no further need to test
// continue;
//}
var rx = rxHelper.DataRowObj;
var ry = ryHelper.DataRowObj;
var isRowOnError = false;
for (int i = 0; i < rx.Table.Columns.Count; i++)
{
if (Settings.GetColumnRole(i) == ColumnRole.Value)
{
//Any management
if (rx[i].ToString() != "(any)" && ry[i].ToString() != "(any)")
{
//Null management
if (rx.IsNull(i) || ry.IsNull(i))
{
if ((!rx.IsNull(i) && rx[i].ToString() != "(blank)") || (!ry.IsNull(i) && ry[i].ToString() != "(blank)"))
{
ry.SetColumnError(i, ry.IsNull(i) ? rx[i].ToString() : "(null)");
if (!isRowOnError)
{
isRowOnError = true;
nonMatchingValueRows.Add(ry);
}
}
}
//(value) management
else if (rx[i].ToString() == "(value)" || ry[i].ToString() == "(value)")
{
if (rx.IsNull(i) || ry.IsNull(i))
{
ry.SetColumnError(i, rx[i].ToString());
if (!isRowOnError)
{
isRowOnError = true;
nonMatchingValueRows.Add(ry);
}
}
}
//Not Null management
else
{
ComparerResult result = null;
//Numeric
if (Settings.GetColumnType(i) == ColumnType.Numeric)
{
//Convert to decimal
if (Settings.IsRounding(i))
{
result = numericComparer.Compare(rx[i], ry[i], Settings.GetRounding(i));
}
else
{
result = numericComparer.Compare(rx[i], ry[i], Settings.GetTolerance(i));
}
}
//Date and Time
else if (Settings.GetColumnType(i) == ColumnType.DateTime)
{
//Convert to dateTime
if (Settings.IsRounding(i))
{
result = dateTimeComparer.Compare(rx[i], ry[i], Settings.GetRounding(i));
}
else
{
result = dateTimeComparer.Compare(rx[i], ry[i], Settings.GetTolerance(i));
}
}
//Boolean
else if (Settings.GetColumnType(i) == ColumnType.Boolean)
{
//Convert to bool
result = booleanComparer.Compare(rx[i], ry[i]);
}
//Text
else
{
result = textComparer.Compare(rx[i], ry[i]);
}
//If are not equal then we need to set the message in the ColumnError.
if (!result.AreEqual)
{
ry.SetColumnError(i, result.Message);
if (!isRowOnError)
{
isRowOnError = true;
nonMatchingValueRows.Add(ry);
}
}
}
}
}
}
}
}
}
public override bool Apply(object x)
{
var cpr = new TextComparer();
return(cpr.Compare(x, Reference).AreEqual);
}
protected ResultSetCompareResult doCompare(DataTable x, DataTable y)
{
var chrono = DateTime.Now;
var columnsCount = Math.Max(y.Columns.Count, x.Columns.Count);
if (Settings == null)
{
BuildDefaultSettings(columnsCount);
}
else
{
Settings.ApplyTo(columnsCount);
}
Settings.ConsoleDisplay();
WriteSettingsToDataTableProperties(y, Settings);
WriteSettingsToDataTableProperties(x, Settings);
CheckSettingsAndDataTable(y, Settings);
CheckSettingsAndDataTable(x, Settings);
CheckSettingsAndFirstRow(y, Settings);
CheckSettingsAndFirstRow(x, Settings);
var keyComparer = new DataRowKeysComparer(Settings, x.Columns.Count);
CalculateHashValues(x, xDict, keyComparer, false);
CalculateHashValues(y, yDict, keyComparer, true);
chrono = DateTime.Now;
List <CompareHelper> missingRows;
{
var missingRowKeys = xDict.Keys.Except(yDict.Keys);
missingRows = new List <CompareHelper>(missingRowKeys.Count());
foreach (Int64 i in missingRowKeys)
{
missingRows.Add(xDict[i]);
}
}
Trace.WriteLineIf(NBiTraceSwitch.TraceInfo, string.Format("Missing rows: {0} [{1}]", missingRows.Count(), DateTime.Now.Subtract(chrono).ToString(@"d\d\.hh\h\:mm\m\:ss\s\ \+fff\m\s")));
chrono = DateTime.Now;
List <CompareHelper> unexpectedRows;
{
var unexpectedRowKeys = yDict.Keys.Except(xDict.Keys);
unexpectedRows = new List <CompareHelper>(unexpectedRowKeys.Count());
foreach (Int64 i in unexpectedRowKeys)
{
unexpectedRows.Add(yDict[i]);
}
}
Trace.WriteLineIf(NBiTraceSwitch.TraceInfo, string.Format("Unexpected rows: {0} [{1}]", unexpectedRows.Count(), DateTime.Now.Subtract(chrono).ToString(@"d\d\.hh\h\:mm\m\:ss\s\ \+fff\m\s")));
chrono = DateTime.Now;
List <CompareHelper> keyMatchingRows;
{
var keyMatchingRowKeys = xDict.Keys.Intersect(yDict.Keys);
keyMatchingRows = new List <CompareHelper>(keyMatchingRowKeys.Count());
foreach (Int64 i in keyMatchingRowKeys)
{
keyMatchingRows.Add(xDict[i]);
}
}
Trace.WriteLineIf(NBiTraceSwitch.TraceInfo, string.Format("Rows with a matching key and not duplicated: {0} [{1}]", keyMatchingRows.Count(), DateTime.Now.Subtract(chrono).ToString(@"d\d\.hh\h\:mm\m\:ss\s\ \+fff\m\s")));
chrono = DateTime.Now;
var nonMatchingValueRows = new List <DataRow>();
// If all of the columns make up the key, then we already know which rows match and which don't.
// So there is no need to continue testing
if (Settings.KeysDef != ResultSetComparisonSettings.KeysChoice.All)
{
foreach (var rxHelper in keyMatchingRows)
{
var ryHelper = yDict[rxHelper.KeysHashed];
//if (ryHelper.ValuesHashed == rxHelper.ValuesHashed)
//{
// // quick shortcut. If the hash of the values matches, then there is no further need to test
// continue;
//}
var rx = rxHelper.DataRowObj;
var ry = ryHelper.DataRowObj;
for (int i = 0; i < rx.Table.Columns.Count; i++)
{
if (Settings.IsValue(i))
{
//Null management
if (rx.IsNull(i) || ry.IsNull(i))
{
if (!rx.IsNull(i) || !ry.IsNull(i))
{
ry.SetColumnError(i, ry.IsNull(i) ? rx[i].ToString() : "(null)");
if (!nonMatchingValueRows.Contains(ry))
{
nonMatchingValueRows.Add(ry);
}
}
}
//(value) management
else if (rx[i].ToString() == "(value)" || ry[i].ToString() == "(value)")
{
if (rx.IsNull(i) || ry.IsNull(i))
{
ry.SetColumnError(i, rx[i].ToString());
if (!nonMatchingValueRows.Contains(ry))
{
nonMatchingValueRows.Add(ry);
}
}
}
//Not Null management
else
{
ComparerResult result = null;
//Numeric
if (Settings.IsNumeric(i))
{
//Convert to decimal
if (Settings.IsRounding(i))
{
result = numericComparer.Compare(rx[i], ry[i], Settings.GetRounding(i));
}
else
{
result = numericComparer.Compare(rx[i], ry[i], Settings.GetTolerance(i));
}
}
//Date and Time
else if (Settings.IsDateTime(i))
{
//Convert to dateTime
if (Settings.IsRounding(i))
{
result = dateTimeComparer.Compare(rx[i], ry[i], Settings.GetRounding(i));
}
else
{
result = dateTimeComparer.Compare(rx[i], ry[i], Settings.GetTolerance(i));
}
}
//Boolean
else if (Settings.IsBoolean(i))
{
//Convert to bool
result = booleanComparer.Compare(rx[i], ry[i]);
}
//Text
else
{
result = textComparer.Compare(rx[i], ry[i]);
}
//If are not equal then we need to set the message in the ColumnError.
if (!result.AreEqual)
{
ry.SetColumnError(i, result.Message);
if (!nonMatchingValueRows.Contains(ry))
{
nonMatchingValueRows.Add(ry);
}
}
}
}
}
}
}
Trace.WriteLineIf(NBiTraceSwitch.TraceInfo, string.Format("Rows with a matching key but without matching value: {0} [{1}]", nonMatchingValueRows.Count(), DateTime.Now.Subtract(chrono).ToString(@"d\d\.hh\h\:mm\m\:ss\s\ \+fff\m\s")));
var duplicatedRows = new List <DataRow>(); // Dummy place holder
return(ResultSetCompareResult.Build(
missingRows.Select(a => a.DataRowObj).ToList(),
unexpectedRows.Select(a => a.DataRowObj).ToList(),
duplicatedRows,
keyMatchingRows.Select(a => a.DataRowObj).ToList(),
nonMatchingValueRows
));
}
private IEnumerable <DataRow> CompareValues(DataRow rx, DataRow ry)
{
var nonMatchingValueRows = new List <DataRow>();
var isRowOnError = false;
for (int i = 0; i < rx.Table.Columns.Count; i++)
{
if (Settings.GetColumnRole(i) == ColumnRole.Value)
{
//Null management
if (rx.IsNull(i) || ry.IsNull(i))
{
if (!rx.IsNull(i) || !ry.IsNull(i))
{
ry.SetColumnError(i, ry.IsNull(i) ? rx[i].ToString() : "(null)");
if (!isRowOnError)
{
isRowOnError = true;
nonMatchingValueRows.Add(ry);
}
}
}
//(value) management
else if (rx[i].ToString() == "(value)" || ry[i].ToString() == "(value)")
{
if (rx.IsNull(i) || ry.IsNull(i))
{
ry.SetColumnError(i, rx[i].ToString());
if (!isRowOnError)
{
isRowOnError = true;
nonMatchingValueRows.Add(ry);
}
}
}
//Not Null management
else
{
ComparerResult result = null;
//Numeric
if (Settings.GetColumnType(i) == ColumnType.Numeric)
{
//Convert to decimal
if (Settings.IsRounding(i))
{
result = numericComparer.Compare(rx[i], ry[i], Settings.GetRounding(i));
}
else
{
result = numericComparer.Compare(rx[i], ry[i], Settings.GetTolerance(i));
}
}
//Date and Time
else if (Settings.GetColumnType(i) == ColumnType.DateTime)
{
//Convert to dateTime
if (Settings.IsRounding(i))
{
result = dateTimeComparer.Compare(rx[i], ry[i], Settings.GetRounding(i));
}
else
{
result = dateTimeComparer.Compare(rx[i], ry[i], Settings.GetTolerance(i));
}
}
//Boolean
else if (Settings.GetColumnType(i) == ColumnType.Boolean)
{
//Convert to bool
result = booleanComparer.Compare(rx[i], ry[i]);
}
//Text
else
{
result = textComparer.Compare(rx[i], ry[i]);
}
//If are not equal then we need to set the message in the ColumnError.
if (!result.AreEqual)
{
ry.SetColumnError(i, result.Message);
if (!isRowOnError)
{
isRowOnError = true;
nonMatchingValueRows.Add(ry);
}
}
}
}
}
return(nonMatchingValueRows);
}