//////////////////////////////////////////////////////////////////////////////////////////////////
// GetPageObjectTextState
// get text object's of specified mcid the text state
//////////////////////////////////////////////////////////////////////////////////////////////////
static PdfTextState GetPageObjectTextState(PdsPageObject page_object, int mcid)
{
if (page_object.GetObjectType() == PdfPageObjectType.kPdsPageText)
{
PdsText text = (PdsText)page_object;
// check if this text page object has the same mcid
PdsContentMark content_mark = page_object.GetContentMark();
if (content_mark != null && content_mark.GetTagMcid() == mcid)
{
PdfTextState ts = text.GetTextState();
if (ts.font != null)
{
return(ts);
}
}
}
else if (page_object.GetObjectType() == PdfPageObjectType.kPdsPageForm)
{
// search for the text object inside of the form XObject
PdsForm form = (PdsForm)page_object;
var content = form.GetContent();
for (int i = 0; i < content.GetNumObjects(); i++)
{
var ts = GetPageObjectTextState(content.GetObject(i), mcid);
if (ts.font != null)
{
return(ts);
}
}
}
return(new PdfTextState());
}
private static void ProcessPageObject(PdfPage page, PdsPageObject obj, string savePath)
{
if (obj == null)
{
PdfixEngine.ThrowException();
}
switch (obj.GetObjectType())
{
case PdfPageObjectType.kPdsPageImage:
ExtractImage(page, (PdsImage)obj, savePath);
break;
case PdfPageObjectType.kPdsPageForm:
{
var form = (PdsForm)obj;
var content = form.GetContent();
for (int i = 0; i < content.GetNumObjects(); i++)
{
ProcessPageObject(page, content.GetObject(i), savePath);
}
}
break;
}
}
// check page object if it fits to
internal static void CheckPageObject(PdsPageObject page_object, KeyValuePair <string, int> layer)
{
switch (page_object.GetObjectType())
{
case PdfPageObjectType.kPdsPageText:
{
var text_obj = (PdsText)page_object;
string text = text_obj.GetText();
var layers = GetPageObjectLayers(page_object);
foreach (KeyValuePair <string, int> l in layers)
{
if (layer.Key == l.Key && layer.Value == l.Value)
{
Console.WriteLine(text);
break;
}
}
}
break;
}
}
// collect all bounding boxes of the page object with specified mcid
static List <PdfRect> GetMcidBBoxes(PdsPageObject obj, int mcid)
{
var bboxes = new List <PdfRect>();
// check object mcid
var content_mark = obj.GetContentMark();
if (content_mark != null && content_mark.GetTagMcid() == mcid)
{
bboxes.Add(obj.GetBBox());
}
else
{
if (obj.GetObjectType() == PdfPageObjectType.kPdsPageForm)
{
var form_obj = (PdsForm)obj;
for (var i = 0; i < form_obj.GetNumPageObjects(); i++)
{
bboxes.AddRange(GetMcidBBoxes(form_obj.GetPageObject(i), mcid));
}
}
}
return(bboxes);
}
private static void ProcessPageObject(PdfPage page, PdsPageObject obj, string savePath)
{
if (obj == null)
{
throw new Exception(_pdfix.GetError());
}
switch (obj.GetObjectType())
{
case PdfPageObjectType.kPdsPageImage:
ExtractImage(page, (PdsImage)obj, savePath);
break;
case PdfPageObjectType.kPdsPageForm:
{
var form = (PdsForm)obj;
for (int i = 0; i < form.GetNumPageObjects(); i++)
{
ProcessPageObject(page, form.GetPageObject(i), savePath);
}
}
break;
}
}