private void generateFinalRect(List <TextWithRect> rectAndTextList)
{
int listSize = rectAndTextList.Count;
if (listSize == 1)
{
this.myPoints.Add(rectAndTextList[0]);
}
else
{
TextWithRect tmpObj = rectAndTextList[0];
for (int i = 1; i < listSize; i++)
{
TextWithRect tmp = rectAndTextList[i];
if (tmp.Rect.Bottom == tmpObj.Rect.Bottom)//if in the same line
{
tmpObj.Text += " " + tmp.Text;
tmpObj.Rect.Top = tmp.Rect.Top;
tmpObj.Rect.Right = tmp.Rect.Right;
}
else
{
this.myPoints.Add(tmpObj);
tmpObj = rectAndTextList[i];
}
}
this.myPoints.Add(tmpObj);
}
}
private void mergeAndStoreChunk(List <TextWithRect> tmpList)
{
TextWithRect mergedChunk = tmpList[0];
int tmpListCount = tmpList.Count();
for (int i = 1; i < tmpListCount; i++)
{
TextWithRect nowChunk = tmpList[i];
mergedChunk.Rect.Right = nowChunk.Rect.Right;
mergedChunk.Rect.Top = nowChunk.Rect.Top;
mergedChunk.Text += nowChunk.Text;
}
string chunkText = mergedChunk.Text.ToLower().Trim();
if (chunkText.Length == 0 || chunkText.Equals("table") || chunkText.Equals("figure") || chunkText.Equals(".rev.") || chunkText.Equals("inc."))
{
return;
}
docText.Add(index, mergedChunk);
docTextString += chunkText + " ";
index += chunkText.Length + 1;
List <string> tmpTerms = new List <string>(mergedChunk.Text.Split(' '));
}
private void matchTopicTerms()
{
int docTextCount = docText.Count;
foreach (string term in topicTerms)
{
List <TextWithRect> rectAndTextList = new List <TextWithRect>();
List <string> subTerms = new List <string>();
if (term.Contains(" "))
{
subTerms = term.Split(' ').ToList();
}
else
{
subTerms.Add(term);
}
int subTermCount = subTerms.Count;
int startPlace = 0;
int tmpIndex = docTextString.IndexOf(term, startPlace);
while (tmpIndex >= 0)
{
char[] sps = docTextString.Substring(tmpIndex + term.Length, 1).ToCharArray();
if (sps.Length > 0 && ((sps[0] == 's') || !((sps[0] > 'a' && sps[0] < 'z') || (sps[0] > 'A' && sps[0] < 'Z'))))
{
List <int> keys = docText.Keys.ToList();
for (int i = 0; i < docTextCount; i++)
{
int curKey = keys[i];
TextWithRect textRect = docText[curKey];
string curText = textRect.Text.ToLower();
int textLength = curText.Length;
if (curKey <= tmpIndex && curKey + textLength > tmpIndex && !rectAndTextList.Contains(textRect))
{
if (subTermCount == 1)
{
rectAndTextList.Add(textRect);
break;
}
else if (subTermCount > 1)
{
string startTerm = subTerms[0];
int tmpLength = startTerm.Length;
List <TextWithRect> rectList = new List <TextWithRect>();
rectList.Add(textRect);
string tmpTextAmount = curText + " ";
if (curText.Contains(startTerm))
{
for (int j = 1; j < subTermCount; j++)
{
int tmpKey = keys[i + j];
TextWithRect tmpRect = docText[tmpKey];
rectList.Add(tmpRect);
string tmpText = tmpRect.Text.ToLower();
tmpTextAmount += tmpText + " ";
}
if (tmpTextAmount.Contains(term + " ") || tmpTextAmount.Contains(term + "s") || tmpTextAmount.Contains(term + ".") || tmpTextAmount.Contains(term + "]") || tmpTextAmount.Contains(term + "es"))
{
foreach (TextWithRect tmpRect in rectList)
{
rectAndTextList.Add(tmpRect);
}
}
}
}
}
}
}
startPlace = tmpIndex + term.Length;
tmpIndex = docTextString.IndexOf(term, startPlace);
}
if (rectAndTextList.Count > 0)
{
generateFinalRect(rectAndTextList);
}
}
}