public void Rejects_Short_Script_Short_Label()
{
string scriptLabel = "C";
string realLabel = "D";
Assert.IsFalse(FuzzyTextComparer.FuzzyEquals(scriptLabel, realLabel));
}
public void Can_Match_1()
{
string scriptLabel = "Invoice";
string realLabel = "Invoice";
Assert.IsTrue(FuzzyTextComparer.FuzzyEquals(scriptLabel, realLabel));
}
public void Can_Match_Partial_Label_2()
{
string scriptLabel = "Customer Number";
string realLabel = "stomer Number";
Assert.IsTrue(FuzzyTextComparer.FuzzyEquals(scriptLabel, realLabel));
}
public void Rejects_2()
{
string scriptLabel = "To";
string realLabel = "From";
Assert.IsFalse(FuzzyTextComparer.FuzzyEquals(scriptLabel, realLabel));
}
public void Rejects_3()
{
string scriptLabel = "Banana Factory";
string realLabel = "Apple Factory";
Assert.IsFalse(FuzzyTextComparer.FuzzyEquals(scriptLabel, realLabel));
}
public void Can_Match_3()
{
string scriptLabel = "Banana Factory";
string realLabel = "Banana Factory";
Assert.IsTrue(FuzzyTextComparer.FuzzyEquals(scriptLabel, realLabel));
}
public void Rejects_1()
{
string scriptLabel = "Invoice";
string realLabel = "Address";
Assert.IsFalse(FuzzyTextComparer.FuzzyEquals(scriptLabel, realLabel));
}
public void Rejects_Different()
{
string scriptLabel = "Banana";
string realLabel = "Apple";
Assert.IsFalse(FuzzyTextComparer.FuzzyEquals(scriptLabel, realLabel));
}
public void Rejects_Short()
{
string scriptLabel = "BA";
string realLabel = "AB";
Assert.IsFalse(FuzzyTextComparer.FuzzyEquals(scriptLabel, realLabel));
}
public void Matches_Short()
{
string scriptLabel = "AB";
string realLabel = "AB";
Assert.IsTrue(FuzzyTextComparer.FuzzyEquals(scriptLabel, realLabel));
}
private AnalyzedTableColumns[] GetAnalyzedTableColumns(Page page, TableColumn[] tableColumns)
{
// Experimentation shows that trimming the Y index more can help normalise table headers
var tableAnalysisPage = new PageTrimmer().TrimPage(page, 20, 20, true);
// This finds the row in the page that can best match the TextType and label definitions along with the indices where they match.
var bestMatch = tableColumns
.SelectMany((tableColumn, TableIndex) => tableAnalysisPage
.GetIndexWhere(t =>
tableColumn.LabelParameters.Any(v => FuzzyTextComparer.FuzzyEquals(v, t.Value))
)
.Select(Index => new
{
TableIndex,
Index,
tableColumn.TextType
}))
.GroupBy(x => x.Index.Coordinate.Y)
.OrderByDescending(x => x.Count())
.FirstOrDefault();
if (bestMatch == null)
{
return null;
}
var columns = bestMatch
.GroupBy(x => x.TableIndex)
.Select(x => x.First())
.Select(x => new AnalyzedTableColumns()
{
Rows = GetTableColumn(page, (PageIndex)x.Index.Clone(), x.TextType),
ColumnIndex = x.TableIndex
})
.ToArray();
return columns;
}
private bool ValidateTextTypeMatch(PageUnit pageUnit, string textType, string[] matchContent) => pageUnit != null && pageUnit?.TextType == textType && (matchContent?.Any(x => FuzzyTextComparer.FuzzyEquals(x, pageUnit.Value)) ?? true);
