private void OpenDocument(string fileName)
{
using (LoadDocument.LoadDocumentDialog dlg = new LoadDocument.LoadDocumentDialog(fileName))
{
if (dlg.ShowDialog(this) == DialogResult.OK)
{
try
{
using (WaitCursor wait = new WaitCursor())
{
PDFDocument document = dlg.PDFDocument;
// Parse the document text
Dictionary <int, MyWord[]> documentText = MyWord.BuildWord(document);
// Initialize the document with annotations
SetDocument(document, documentText, fileName);
}
}
catch (Exception ex)
{
Messager.ShowError(this, ex.Message);
}
finally
{
UpdateUIState();
}
}
}
}
private void CopySelectedText()
{
// Convert the selected words to a single string
MyWord[] words = _selectedText[_currentPageNumber];
if (words != null)
{
StringBuilder sb = new StringBuilder();
for (int i = 0; i < words.Length; i++)
{
MyWord word = words[i];
sb.Append(word.Value);
if (word.IsEndOfLine)
{
sb.AppendLine();
}
else if (i != (words.Length - 1))
{
sb.Append(" ");
}
}
Clipboard.SetText(sb.ToString());
}
}
private string BuildPageText(MyWord[] words)
{
if (words == null || words.Length == 0)
{
return(string.Empty);
}
StringBuilder sb = new StringBuilder();
for (int i = 0; i < words.Length; i++)
{
MyWord word = words[i];
sb.Append(word.Value);
if (word.IsEndOfLine)
{
sb.AppendLine();
}
else if (i != (words.Length - 1))
{
sb.Append(" ");
}
}
return(sb.ToString());
}
private void SelectAllText()
{
if (_document == null || _documentText[_currentPageNumber] == null)
{
return;
}
// Get the text for the current page
_selectedText[_currentPageNumber] = new MyWord[_documentText[_currentPageNumber].Length];
_documentText[_currentPageNumber].CopyTo(_selectedText[_currentPageNumber], 0);
// Select it all (by replacing the selected text for the page for all text in the page)
// Re-paint the viewer
_viewerControl.RasterImageViewer.Invalidate();
}
private void BuildWordItems(Dictionary <int, MyWord[]> documentText, int pageNumber)
{
List <FindWordItem> findWordItems = new List <FindWordItem>();
MyWord[] words = null;
if (documentText[pageNumber] != null)
{
words = documentText[pageNumber];
}
if (words != null)
{
int beginIndex = 0;
StringBuilder sb = new StringBuilder();
for (int wordIndex = 0; wordIndex < words.Length; wordIndex++)
{
MyWord word = words[wordIndex];
sb.Append(word.Value);
// Add it as a FindWordItem
FindWordItem item = new FindWordItem();
item.BeginTextIndex = beginIndex;
item.EndTextIndex = beginIndex + word.Value.Length - 1;
item.WordIndex = wordIndex;
findWordItems.Add(item);
sb.Append(" ");
beginIndex = item.EndTextIndex + 2;
}
_currentPageText = sb.ToString();
}
else
{
_currentPageText = string.Empty;
}
_findWordItems = findWordItems.ToArray();
}
public static Dictionary <int, MyWord[]> BuildWord(PDFDocument document)
{
Dictionary <int, MyWord[]> pageWords = new Dictionary <int, MyWord[]>();
for (int pageNumber = 1; pageNumber <= document.Pages.Count; pageNumber++)
{
List <MyWord> words = new List <MyWord>();
PDFDocumentPage page = document.Pages[pageNumber - 1];
IList <PDFObject> objects = page.Objects;
if (objects != null && objects.Count > 0)
{
int objectIndex = 0;
int objectCount = objects.Count;
// Loop through all the objects
while (objectIndex < objectCount)
{
// Find the total bounding rectangle, begin and end index of the next word
LeadRect wordBounds = LeadRect.Empty;
int firstObjectIndex = objectIndex;
// Loop till we reach EndOfWord or reach the end of the objects
bool more = true;
while (more)
{
PDFObject obj = objects[objectIndex];
// Must be text and not a white character
if (obj.ObjectType == PDFObjectType.Text && !Char.IsWhiteSpace(obj.Code))
{
// Add the bounding rectangle of this object
PDFRect temp = page.ConvertRect(PDFCoordinateType.Pdf, PDFCoordinateType.Pixel, obj.Bounds);
LeadRect objectBounds = LeadRect.FromLTRB((int)temp.Left, (int)temp.Top, (int)temp.Right, (int)temp.Bottom);
if (wordBounds.IsEmpty)
{
wordBounds = objectBounds;
}
else
{
wordBounds = LeadRect.Union(wordBounds, objectBounds);
}
}
else
{
firstObjectIndex = objectIndex + 1;
}
objectIndex++;
more = (objectIndex < objectCount) && !obj.TextProperties.IsEndOfWord && !obj.TextProperties.IsEndOfLine;
}
if (firstObjectIndex == objectIndex)
{
continue;
}
// From the begin and end index, collect the characters into a string
StringBuilder sb = new StringBuilder();
for (int i = firstObjectIndex; i < objectIndex; i++)
{
if (objects[i].ObjectType == PDFObjectType.Text)
{
sb.Append(objects[i].Code);
}
}
// Add this word to the list
PDFObject lastObject = objects[objectIndex - 1];
MyWord word = new MyWord();
word.Value = sb.ToString();
word.Bounds = wordBounds;
word.IsEndOfLine = lastObject.TextProperties.IsEndOfLine;
words.Add(word);
}
}
// Add "IsEndOfLine" to the last word in the page, just in case it does not have it
if (words.Count > 0)
{
MyWord word = words[words.Count - 1];
word.IsEndOfLine = true;
words[words.Count - 1] = word;
}
pageWords.Add(pageNumber, words.ToArray());
}
return(pageWords);
}
