public ToxyDocument Parse()
{
if (!File.Exists(Context.Path))
{
throw new FileNotFoundException("File " + Context.Path + " is not found");
}
bool extractHeader = false;
if (Context.Properties.ContainsKey("ExtractHeader"))
{
extractHeader = Utility.IsTrue(Context.Properties["ExtractHeader"]);
}
bool extractFooter = false;
if (Context.Properties.ContainsKey("ExtractFooter"))
{
extractFooter = Utility.IsTrue(Context.Properties["ExtractFooter"]);
}
ToxyDocument rdoc = new ToxyDocument();
using (FileStream stream = File.OpenRead(Context.Path))
{
HWPFDocument worddoc = new HWPFDocument(stream);
if (extractHeader && worddoc.GetHeaderStoryRange() != null)
{
StringBuilder sb = new StringBuilder();
rdoc.Header = worddoc.GetHeaderStoryRange().Text;
}
if (extractFooter && worddoc.GetFootnoteRange() != null)
{
StringBuilder sb = new StringBuilder();
rdoc.Footer = worddoc.GetFootnoteRange().Text;
}
for (int i = 0; i < worddoc.GetRange().NumParagraphs; i++)
{
Paragraph para = worddoc.GetRange().GetParagraph(i);
string text = para.Text;
ToxyParagraph p = new ToxyParagraph();
p.Text = text;
//var runs = para.Runs;
p.StyleID = para.GetStyleIndex().ToString();
//for (int i = 0; i < runs.Count; i++)
//{
// var run = runs[i];
//}
rdoc.Paragraphs.Add(p);
}
}
return(rdoc);
}
public void TestContents()
{
HWPFDocument doc = docAscii;
for (int run = 0; run < 3; run++)
{
Range r;
// Now check the real ranges
r = doc.GetRange();
Assert.AreEqual(
a_page_1 +
page_break + "\r" +
a_page_2,
r.Text
);
r = doc.GetHeaderStoryRange();
Assert.AreEqual(
headerDef +
a_header +
footerDef +
a_footer +
endHeaderFooter,
r.Text
);
r = doc.GetOverallRange();
Assert.AreEqual(
a_page_1 +
page_break + "\r" +
a_page_2 +
headerDef +
a_header +
footerDef +
a_footer +
endHeaderFooter +
"\r",
r.Text
);
// Write out and read back in again, Ready for
// the next run of the Test
// TODO run more than once
if (run < 1)
{
doc = HWPFTestDataSamples.WriteOutAndReadBack(doc);
}
}
}
public HeaderStories(HWPFDocument doc)
{
this.headerStories = doc.GetHeaderStoryRange();
FileInformationBlock fib = doc.GetFileInformationBlock();
// If there's no PlcfHdd, nothing to do
if (fib.GetCcpHdd() == 0)
{
return;
}
if (fib.GetPlcfHddSize() == 0)
{
return;
}
// Handle the PlcfHdd
plcfHdd = new PlexOfCps(
doc.GetTableStream(), fib.GetPlcfHddOffset(),
fib.GetPlcfHddSize(), 0
);
}
public HeaderStories(HWPFDocument doc)
{
this.headerStories = doc.GetHeaderStoryRange();
FileInformationBlock fib = doc.GetFileInformationBlock();
// If there's no PlcfHdd, nothing to do
if (fib.GetCcpHdd() == 0)
{
return;
}
if (fib.GetPlcfHddSize() == 0)
{
return;
}
// Handle the PlcfHdd
plcfHdd = new PlexOfCps(
doc.GetTableStream(), fib.GetPlcfHddOffset(),
fib.GetPlcfHddSize(), 0
);
}
public void TestContentsUnicode()
{
Range r;
// Now check the real ranges
r = docUnicode.GetRange();
Assert.AreEqual(
u_page_1 +
page_break + "\r" +
u_page_2,
r.Text
);
r = docUnicode.GetHeaderStoryRange();
Assert.AreEqual(
headerDef +
u_header +
footerDef +
u_footer +
endHeaderFooter,
r.Text
);
r = docUnicode.GetOverallRange();
Assert.AreEqual(
u_page_1 +
page_break + "\r" +
u_page_2 +
headerDef +
u_header +
footerDef +
u_footer +
endHeaderFooter +
"\r",
r.Text
);
}