private int GetStartIndex(GMWord word)
{
int startIndex = (word.BoundingBox.YMin - yEpsilon) / segmentSize;
if (startIndex < 0)
{
startIndex = 0;
}
return(startIndex);
}
private int GetEndIndex(GMWord word)
{
int endIndex = (word.BoundingBox.YMax + yEpsilon) / segmentSize;
if (endIndex >= hash.Count())
{
endIndex = hash.Count() - 1;
}
return(endIndex);
}
private void AddWord(GMWord word)
{
int startIndex = GetStartIndex(word);
int endIndex = GetEndIndex(word);
for (int index = startIndex; index <= endIndex; index++)
{
hash[index].Add(word);
}
}
//return HashSet coincidered words
private static IEnumerable <GMWord> EnumerateWordsFromFineReader(TextGeometryModel fineReaderModel, TextGeometryModel sampleModel,
LinesHashInGeometryModel tesseractCache)
{
foreach (var block in fineReaderModel.TextBlocks())
{
var blockBox = block.BoundingBox;
var thisBlock =
sampleModel.AddTextBlock(new BoundingBox(blockBox.XMin, blockBox.YMin, blockBox.XMax, blockBox.YMax));
foreach (var paragraph in block.Paragraphs())
{
var paragraphBox = paragraph.BoundingBox;
var thisParagraph =
thisBlock.AddParagraph(new BoundingBox(paragraphBox.XMin, paragraphBox.YMin, paragraphBox.XMax,
paragraphBox.YMax));
foreach (var line in paragraph.Lines())
{
var lineBox = line.BoundingBox;
var thisLine =
thisParagraph.AddLine(new BoundingBox(lineBox.XMin, lineBox.YMin, lineBox.XMax, lineBox.YMax));
foreach (var word in line.Words())
{
var wordBox = word.BoundingBox;
var accuracy = 50;
if (tesseractCache.Contains(word, String.Compare))
{
yield return(word);
accuracy = 100;
}
var thisWord = new GMWord(wordBox, word.Text, accuracy);
thisLine.AddWord(thisWord);
}
}
}
foreach (var word in block.StandaloneWords())
{
var wordBox = word.BoundingBox;
var accuracy = 50;
if (tesseractCache.Contains(word, String.Compare))
{
yield return(word);
accuracy = 100;
}
var thisWord = new GMWord(wordBox, word.Text, accuracy);
thisBlock.AddStandaloneWord(thisWord);
}
}
}
public bool Contains(GMWord word, Func <string, string, int> comparator = null)
{
if (comparator == null)
{
comparator = (w1, w2) => String.Compare(w1, w2, StringComparison.OrdinalIgnoreCase);
}
var startIndex = GetStartIndex(word);
var endIndex = GetEndIndex(word);
for (int i = startIndex; i <= endIndex; i++)
{
if (startIndex >= hash.Length || endIndex >= hash.Length)
{
continue;
}
if (hash[i].Any(w => EqualsBoxes(w.BoundingBox, word.BoundingBox) && comparator(w.Text, word.Text) == 0))//String.Compare(w.Text, word.Text, StringComparison.OrdinalIgnoreCase) == 0))
{
return(true);
}
}
return(false);
}