public bool HasEscherPicture(CharacterRun run)
{
if (run.IsSpecialCharacter() && !run.IsObj() && !run.IsOle2() && !run.IsData() && run.Text.StartsWith("\u0008"))
{
return(true);
}
return(false);
}
/**
* determines whether specified CharacterRun Contains reference to a picture
* @param run
*/
public bool HasPicture(CharacterRun run)
{
if (run.IsSpecialCharacter() && !run.IsObj() && !run.IsOle2() && !run.IsData())
{
// Image should be in it's own run, or in a run with the end-of-special marker
if ("\u0001".Equals(run.Text) || "\u0001\u0015".Equals(run.Text))
{
return(IsBlockContainsImage(run.GetPicOffset()));
}
}
return(false);
}
protected void ProcessField(HWPFDocument wordDocument, Range parentRange,
int currentTableLevel, Field field, XmlElement currentBlock)
{
switch (field.Type)
{
case 37: // page reference
{
Range firstSubrange = field.FirstSubrange(parentRange);
if (firstSubrange != null)
{
String formula = firstSubrange.Text;
Regex pagerefPattern = new Regex("[ \\t\\r\\n]*PAGEREF ([^ ]*)[ \\t\\r\\n]*\\\\h[ \\t\\r\\n]*");
Match match = pagerefPattern.Match(formula);
if (match.Success)
{
String pageref = match.Groups[1].Value;
ProcessPageref(wordDocument, currentBlock,
field.SecondSubrange(parentRange),
currentTableLevel, pageref);
return;
}
//Pattern pagerefPattern = Pattern
// .compile("[ \\t\\r\\n]*PAGEREF ([^ ]*)[ \\t\\r\\n]*\\\\h[ \\t\\r\\n]*");
//Matcher matcher = pagerefPattern.matcher(formula);
//if (matcher.find())
//{
// String pageref = matcher.group(1);
// processPageref(wordDocument, currentBlock,
// field.secondSubrange(parentRange),
// currentTableLevel, pageref);
// return;
//}
}
break;
}
case 58: // Embedded Object
{
if (!field.HasSeparator())
{
logger.Log(POILogger.WARN, parentRange + " contains " + field
+ " with 'Embedded Object' but without separator mark");
return;
}
CharacterRun separator = field.GetMarkSeparatorCharacterRun(parentRange);
if (separator.IsOle2())
{
// the only supported so far
bool processed = ProcessOle2(wordDocument, separator,
currentBlock);
// if we didn't output OLE - output field value
if (!processed)
{
ProcessCharacters(wordDocument, currentTableLevel,
field.SecondSubrange(parentRange), currentBlock);
}
return;
}
break;
}
case 88: // hyperlink
{
Range firstSubrange = field.FirstSubrange(parentRange);
if (firstSubrange != null)
{
String formula = firstSubrange.Text;
Regex hyperlinkPattern = new Regex("[ \\t\\r\\n]*HYPERLINK \"(.*)\"[ \\t\\r\\n]*");
Match match = hyperlinkPattern.Match(formula);
if (match.Success)
{
String hyperlink = match.Groups[1].Value;
ProcessHyperlink(wordDocument, currentBlock,
field.SecondSubrange(parentRange),
currentTableLevel, hyperlink);
return;
}
//Pattern hyperlinkPattern = Pattern
// .compile("[ \\t\\r\\n]*HYPERLINK \"(.*)\"[ \\t\\r\\n]*");
//Matcher matcher = hyperlinkPattern.matcher(formula);
//if (matcher.find())
//{
// String hyperlink = matcher.group(1);
// processHyperlink(wordDocument, currentBlock,
// field.secondSubrange(parentRange),
// currentTableLevel, hyperlink);
// return;
//}
}
break;
}
}
logger.Log(POILogger.WARN, parentRange + " contains " + field
+ " with unsupported type or format");
ProcessCharacters(wordDocument, currentTableLevel,
field.SecondSubrange(parentRange), currentBlock);
}
protected bool ProcessCharacters(HWPFDocumentCore wordDocument, int currentTableLevel, Range range, XmlElement block)
{
if (range == null)
{
return(false);
}
bool haveAnyText = false;
/*
* In text there can be fields, bookmarks, may be other structures (code
* below allows extension). Those structures can overlaps, so either we
* should process char-by-char (slow) or find a correct way to
* reconstruct the structure of range -- sergey
*/
IList <Structure> structures = new List <Structure>();
if (wordDocument is HWPFDocument)
{
HWPFDocument doc = (HWPFDocument)wordDocument;
Dictionary <int, List <Bookmark> > rangeBookmarks = doc.GetBookmarks()
.GetBookmarksStartedBetween(range.StartOffset, range.EndOffset);
if (rangeBookmarks != null)
{
foreach (KeyValuePair <int, List <Bookmark> > kv in rangeBookmarks)
{
List <Bookmark> lists = kv.Value;
foreach (Bookmark bookmark in lists)
{
if (!bookmarkStack.Contains(bookmark))
{
AddToStructures(structures, new Structure(bookmark));
}
}
}
}
// TODO: dead fields?
for (int c = 0; c < range.NumCharacterRuns; c++)
{
CharacterRun characterRun = range.GetCharacterRun(c);
if (characterRun == null)
{
throw new NullReferenceException();
}
Field aliveField = ((HWPFDocument)wordDocument).GetFields()
.GetFieldByStartOffset(FieldsDocumentPart.MAIN,
characterRun.StartOffset);
if (aliveField != null)
{
AddToStructures(structures, new Structure(aliveField));
}
}
}
//structures = new ArrayList<Structure>( structures );
//Collections.sort( structures );
SortedList <Structure, Structure> sl = new SortedList <Structure, Structure>();
foreach (Structure s in structures)
{
sl.Add(s, s);
}
structures.Clear();
((List <Structure>)structures).AddRange(sl.Values);
int previous = range.StartOffset;
foreach (Structure structure in structures)
{
if (structure.Start != previous)
{
Range subrange = new Range(previous, structure.Start, range);
//{
// public String toString()
// {
// return "BetweenStructuresSubrange " + super.ToString();
// }
//};
ProcessCharacters(wordDocument, currentTableLevel, subrange, block);
}
if (structure.StructureObject is Bookmark)
{
// other bookmarks with same boundaries
IList <Bookmark> bookmarks = new List <Bookmark>();
IEnumerator <List <Bookmark> > iterator = ((HWPFDocument)wordDocument).GetBookmarks().GetBookmarksStartedBetween(structure.Start, structure.Start + 1).Values.GetEnumerator();
iterator.MoveNext();
foreach (Bookmark bookmark in iterator.Current)
{
if (bookmark.Start == structure.Start &&
bookmark.End == structure.End)
{
bookmarks.Add(bookmark);
}
}
bookmarkStack.AddRange(bookmarks);
try
{
int end = Math.Min(range.EndOffset, structure.End);
Range subrange = new Range(structure.Start, end, range);
/*{
* public String toString()
* {
* return "BookmarksSubrange " + super.ToString();
* }
* };*/
ProcessBookmarks(wordDocument, block, subrange,
currentTableLevel, bookmarks);
}
finally
{
bookmarkStack.RemoveAll((e) => { return(bookmarks.Contains(e)); });
}
}
else if (structure.StructureObject is Field)
{
Field field = (Field)structure.StructureObject;
ProcessField((HWPFDocument)wordDocument, range, currentTableLevel, field, block);
}
else
{
throw new NotSupportedException("NYI: " + structure.StructureObject.GetType().ToString());
}
previous = Math.Min(range.EndOffset, structure.End);
}
if (previous != range.StartOffset)
{
if (previous > range.EndOffset)
{
logger.Log(POILogger.WARN, "Latest structure in ", range,
" ended at #" + previous, " after range boundaries [",
range.StartOffset + "; " + range.EndOffset,
")");
return(true);
}
if (previous < range.EndOffset)
{
Range subrange = new Range(previous, range.EndOffset, range);
/*{
* @Override
* public String toString()
* {
* return "AfterStructureSubrange " + super.ToString();
* }
* };*/
ProcessCharacters(wordDocument, currentTableLevel, subrange,
block);
}
return(true);
}
for (int c = 0; c < range.NumCharacterRuns; c++)
{
CharacterRun characterRun = range.GetCharacterRun(c);
if (characterRun == null)
{
throw new NullReferenceException();
}
if (wordDocument is HWPFDocument && ((HWPFDocument)wordDocument).GetPicturesTable().HasPicture(characterRun))
{
HWPFDocument newFormat = (HWPFDocument)wordDocument;
Picture picture = newFormat.GetPicturesTable().ExtractPicture(characterRun, true);
ProcessImage(block, characterRun.Text[0] == 0x01, picture);
continue;
}
string text = characterRun.Text;
byte[] textByte = System.Text.Encoding.GetEncoding("iso-8859-1").GetBytes(text);
//if ( text.getBytes().length == 0 )
if (textByte.Length == 0)
{
continue;
}
if (characterRun.IsSpecialCharacter())
{
if (text[0] == SPECCHAR_AUTONUMBERED_FOOTNOTE_REFERENCE && (wordDocument is HWPFDocument))
{
HWPFDocument doc = (HWPFDocument)wordDocument;
ProcessNoteAnchor(doc, characterRun, block);
continue;
}
if (text[0] == SPECCHAR_DRAWN_OBJECT &&
(wordDocument is HWPFDocument))
{
HWPFDocument doc = (HWPFDocument)wordDocument;
ProcessDrawnObject(doc, characterRun, block);
continue;
}
if (characterRun.IsOle2() && (wordDocument is HWPFDocument))
{
HWPFDocument doc = (HWPFDocument)wordDocument;
ProcessOle2(doc, characterRun, block);
continue;
}
}
if (textByte[0] == FIELD_BEGIN_MARK)
//if ( text.getBytes()[0] == FIELD_BEGIN_MARK )
{
if (wordDocument is HWPFDocument)
{
Field aliveField = ((HWPFDocument)wordDocument).GetFields().GetFieldByStartOffset(
FieldsDocumentPart.MAIN, characterRun.StartOffset);
if (aliveField != null)
{
ProcessField(((HWPFDocument)wordDocument), range,
currentTableLevel, aliveField, block);
int continueAfter = aliveField.GetFieldEndOffset();
while (c < range.NumCharacterRuns &&
range.GetCharacterRun(c).EndOffset <= continueAfter)
{
c++;
}
if (c < range.NumCharacterRuns)
{
c--;
}
continue;
}
}
int skipTo = TryDeadField(wordDocument, range,
currentTableLevel, c, block);
if (skipTo != c)
{
c = skipTo;
continue;
}
continue;
}
if (textByte[0] == FIELD_SEPARATOR_MARK)
{
// shall not appear without FIELD_BEGIN_MARK
continue;
}
if (textByte[0] == FIELD_END_MARK)
{
// shall not appear without FIELD_BEGIN_MARK
continue;
}
if (characterRun.IsSpecialCharacter() || characterRun.IsObj() ||
characterRun.IsOle2())
{
continue;
}
if (text.EndsWith("\r") ||
(text[text.Length - 1] == BEL_MARK && currentTableLevel != int.MinValue))
{
text = text.Substring(0, text.Length - 1);
}
{
// line breaks
StringBuilder stringBuilder = new StringBuilder();
foreach (char charChar in text.ToCharArray())
{
if (charChar == 11)
{
if (stringBuilder.Length > 0)
{
OutputCharacters(block, characterRun,
stringBuilder.ToString());
stringBuilder.Length = 0;
}
ProcessLineBreak(block, characterRun);
}
else if (charChar == 30)
{
// Non-breaking hyphens are stored as ASCII 30
stringBuilder.Append(UNICODECHAR_NONBREAKING_HYPHEN);
}
else if (charChar == 31)
{
// Non-required hyphens to zero-width space
stringBuilder.Append(UNICODECHAR_ZERO_WIDTH_SPACE);
}
else if (charChar >= 0x20 || charChar == 0x09 ||
charChar == 0x0A || charChar == 0x0D)
{
stringBuilder.Append(charChar);
}
}
if (stringBuilder.Length > 0)
{
OutputCharacters(block, characterRun,
stringBuilder.ToString());
stringBuilder.Length = 0;
}
}
haveAnyText |= text.Trim().Length != 0;
}
return(haveAnyText);
}
