override protected void ExtractMetaData (Gsf.Input sumStream, Gsf.Input docSumStream)
{
int count = 0;
DocProp prop = null;
if (sumMeta != null) {
prop = sumMeta.Lookup ("gsf:word-count");
if (prop != null)
count = (int) prop.Val;
if (count > 0)
AddProperty (Beagle.Property.NewUnsearched ("fixme:word-count", count));
count = 0;
prop = sumMeta.Lookup ("gsf:page-count");
if (prop != null)
count = (int) prop.Val;
if (count > 0)
AddProperty (Beagle.Property.NewUnsearched ("fixme:page-count", count));
}
}
private static int ParseElement(Gsf.Input stream)
{
byte [] data = stream.Read(8);
if (data == null)
return 0;
RecordType.TypeCode opcode = (RecordType.TypeCode) (data[2] + (data[3] << 8));
int length = data[4] + (data[5] << 8) + (data[6] << 16) + (data[7] << 24);
RecordType type = RecordType.Find (opcode);
if (type.is_container) {
int length_remaining = length;
while (length_remaining > 0) {
int elem_length = ParseElement(stream);
if (elem_length == 0)
return 0;
length_remaining -= elem_length;
}
} else {
if (length != 0) {
data = stream.Read(length);
if (data == null)
return 0;
string val = null;
if (opcode == RecordType.TypeCode.TextBytesAtom) {
val = System.Text.Encoding.GetEncoding (28591).GetString (data);
} else if (opcode == RecordType.TypeCode.TextCharsAtom) {
val = System.Text.Encoding.Unicode.GetString (data);
}
if (val != null)
System.Console.WriteLine(val);
}
}
return length + 8;
}
void PullMetaData (Gsf.Input sum_stream, Gsf.Input doc_stream)
{
DocProp prop = null;
sumMeta = new DocMetaData ();
if (sum_stream != null)
Msole.MetadataRead (sum_stream, sumMeta);
else
Logger.Log.Warn ("SummaryInformationStream not found in {0}", FileName);
docSumMeta = new DocMetaData ();
if (doc_stream != null)
Msole.MetadataRead (doc_stream, docSumMeta);
else
Logger.Log.Warn ("DocumentSummaryInformationStream not found in {0}", FileName);
if (sumMeta != null) {
prop = sumMeta.Lookup ("dc:title");
if (prop != null)
AddProperty (Beagle.Property.New ("dc:title", prop.Val as string));
prop = sumMeta.Lookup ("dc:subject");
if (prop != null)
AddProperty (Beagle.Property.New ("dc:subject", prop.Val as string));
prop = sumMeta.Lookup ("dc:description");
if (prop != null)
AddProperty (Beagle.Property.New ("dc:description", prop.Val as string));
prop = sumMeta.Lookup ("gsf:keywords");
if (prop != null)
AddProperty (Beagle.Property.New ("fixme:keywords", prop.Val as string));
prop = sumMeta.Lookup ("gsf:creator");
if (prop != null)
AddProperty (Beagle.Property.New ("fixme:author", prop.Val as string));
prop = sumMeta.Lookup ("gsf:last-saved-by");
if (prop != null)
AddProperty (Beagle.Property.New ("fixme:last-saved-by", prop.Val as string));
prop = sumMeta.Lookup ("gsf:generator");
if (prop != null)
AddProperty (Beagle.Property.New ("fixme:generator", prop.Val as string));
prop = sumMeta.Lookup ("gsf:template");
if (prop != null)
AddProperty (Beagle.Property.New ("fixme:template", prop.Val as string));
}
if (docSumMeta != null) {
prop = docSumMeta.Lookup ("gsf:company");
if (prop != null)
AddProperty (Beagle.Property.New ("fixme:company", prop.Val as string));
prop = docSumMeta.Lookup ("gsf:category");
if (prop != null)
AddProperty (Beagle.Property.New ("fixme:category", prop.Val as string));
prop = docSumMeta.Lookup ("CreativeCommons_LicenseURL");
if (prop != null)
AddProperty (Beagle.Property.New ("fixme:license", prop.Val as string));
}
ExtractMetaData (sum_stream, doc_stream);
if (sumMeta != null)
sumMeta.Dispose ();
if (docSumMeta != null)
docSumMeta.Dispose ();
}
protected virtual void ExtractMetaData (Gsf.Input sum_stream,
Gsf.Input doc_stream)
{
}
