public static String GetText(string docFile)
{
HWPFDocumentCore wordDocument = AbstractWordUtils
.LoadDoc(docFile);
return(GetText(wordDocument));
}
public static XmlDocument Process(string docFile)
{
HWPFDocumentCore wordDocument = AbstractWordUtils
.LoadDoc(docFile);
WordToTextConverter wordToTextConverter = new WordToTextConverter(new XmlDocument());
wordToTextConverter.ProcessDocument(wordDocument);
return(wordToTextConverter.Document);
}
public Triplet Update(Triplet original)
{
if (!string.IsNullOrEmpty(original.fontName))
{
String fontName = original.fontName;
if (fontName.EndsWith(" Regular"))
{
fontName = AbstractWordUtils.SubstringBeforeLast(fontName,
" Regular");
}
if (fontName
.EndsWith(" \u041F\u043E\u043B\u0443\u0436\u0438\u0440\u043D\u044B\u0439"))
{
fontName = AbstractWordUtils
.SubstringBeforeLast(fontName,
" \u041F\u043E\u043B\u0443\u0436\u0438\u0440\u043D\u044B\u0439")
+ " Bold";
}
if (fontName
.EndsWith(" \u041F\u043E\u043B\u0443\u0436\u0438\u0440\u043D\u044B\u0439 \u041A\u0443\u0440\u0441\u0438\u0432"))
{
fontName = AbstractWordUtils
.SubstringBeforeLast(
fontName,
" \u041F\u043E\u043B\u0443\u0436\u0438\u0440\u043D\u044B\u0439 \u041A\u0443\u0440\u0441\u0438\u0432")
+ " Bold Italic";
}
if (fontName.EndsWith(" \u041A\u0443\u0440\u0441\u0438\u0432"))
{
fontName = AbstractWordUtils.SubstringBeforeLast(fontName,
" \u041A\u0443\u0440\u0441\u0438\u0432") + " Italic";
}
original.fontName = fontName;
}
if (!string.IsNullOrEmpty(original.fontName))
{
if ("Times Regular".Equals(original.fontName) ||
"Times-Regular".Equals(original.fontName))
{
original.fontName = "Times";
original.bold = false;
original.italic = false;
}
if ("Times Bold".Equals(original.fontName) ||
"Times-Bold".Equals(original.fontName))
{
original.fontName = "Times";
original.bold = true;
original.italic = false;
}
if ("Times Italic".Equals(original.fontName) ||
"Times-Italic".Equals(original.fontName))
{
original.fontName = "Times";
original.bold = false;
original.italic = true;
}
if ("Times Bold Italic".Equals(original.fontName) ||
"Times-BoldItalic".Equals(original.fontName))
{
original.fontName = "Times";
original.bold = true;
original.italic = true;
}
}
return(original);
}
protected void ProcessParagraphes(HWPFDocumentCore wordDocument,
XmlElement flow, Range range, int currentTableLevel)
{
ListTables listTables = wordDocument.GetListTables();
int currentListInfo = 0;
int paragraphs = range.NumParagraphs;
for (int p = 0; p < paragraphs; p++)
{
Paragraph paragraph = range.GetParagraph(p);
if (paragraph.IsInTable() && paragraph.GetTableLevel() != currentTableLevel)
{
if (paragraph.GetTableLevel() < currentTableLevel)
{
throw new InvalidOperationException(
"Trying to process table cell with higher level ("
+ paragraph.GetTableLevel()
+ ") than current table level ("
+ currentTableLevel
+ ") as inner table part");
}
Table table = range.GetTable(paragraph);
ProcessTable(wordDocument, flow, table);
p += table.NumParagraphs;
p--;
continue;
}
if (paragraph.Text.Equals("\u000c"))
{
ProcessPageBreak(wordDocument, flow);
}
if (paragraph.GetIlfo() != currentListInfo)
{
currentListInfo = paragraph.GetIlfo();
}
if (currentListInfo != 0)
{
if (listTables != null)
{
ListFormatOverride listFormatOverride = listTables.GetOverride(paragraph.GetIlfo());
String label = AbstractWordUtils.GetBulletText(listTables,
paragraph, listFormatOverride.GetLsid());
ProcessParagraph(wordDocument, flow, currentTableLevel,
paragraph, label);
}
else
{
logger.Log(POILogger.WARN,
"Paragraph #" + paragraph.StartOffset + "-"
+ paragraph.EndOffset
+ " has reference to list structure #"
+ currentListInfo
+ ", but listTables not defined in file");
ProcessParagraph(wordDocument, flow, currentTableLevel,
paragraph, string.Empty);
}
}
else
{
ProcessParagraph(wordDocument, flow, currentTableLevel,
paragraph, string.Empty);
}
}
}
public static String GetText(DirectoryNode root)
{
HWPFDocumentCore wordDocument = AbstractWordUtils.LoadDoc(root);
return(GetText(wordDocument));
}