void ProcessText(int pageNum, Element element, List <TextAndAxisValue> values, double bottomAxisValue, double topAxisValue, bool isMatchNum)
{
Rect bbox = new Rect();
element.GetBBox(bbox);
string text = element.GetTextString();
if (element.GetType() == Element.Type.e_text &&
!string.IsNullOrEmpty(element.GetTextString().RemoveBlankSpace()))
{
Point leftBottomPoint = GetCooridinateValueOfLeftBottomCorner(pageNum, bbox);
double horAxisValue = leftBottomPoint.y;
double verAxisValue = leftBottomPoint.x;
if (
IsBelow(leftBottomPoint, topAxisValue) &&
!IsBelow(leftBottomPoint, bottomAxisValue) &&
(!isMatchNum || Regex.IsMatch(text, @"\d")))
{
TextAndAxisValue textAndAxisValue = new TextAndAxisValue();
textAndAxisValue.text = text;
textAndAxisValue.yValue = horAxisValue;
textAndAxisValue.xValue = verAxisValue;
values.Add(textAndAxisValue);
}
}
}
void Merge(List <TextAndAxisValue> TextAndAxisValues)
{
for (int i = 0; i < TextAndAxisValues.Count; i++)
{
TextAndAxisValue current = TextAndAxisValues[i];
for (int j = i + 1; j < TextAndAxisValues.Count;)
{
TextAndAxisValue next = TextAndAxisValues[j];
if (Math.Abs(current.yValue - next.yValue) < 3)
{
if (current.xValue > next.xValue)
{
current.text = next.text + current.text;
}
else
{
current.text += next.text;
}
TextAndAxisValues.RemoveAt(j);
}
else
{
j++;
}
}
}
}
void Filtrate(List <TextAndAxisValue> TextAndAxisValues)
{
string regex = @"^(\d+|第\d+页(共\d+页)?|-\d+-)$";
for (int i = 0; i < TextAndAxisValues.Count;)
{
TextAndAxisValue current = TextAndAxisValues[i];
if (Regex.IsMatch(current.text.RemoveBlankSpace(), regex))
{
TextAndAxisValues.RemoveAt(i);
}
else
{
i++;
}
}
}