/// <summary>
/// Parses all glyphs from the page xml file.
/// </summary>
/// <returns>List of glyph objects with ID and Unicode propagated.</returns>
public List <Glyph> GetGlyphs()
{
/*
* For some odd reason we have to specify our own xml prefix for the namespace,
* because .NET didn't accept a URI for the default namespace xmlns.
*
*/
XmlNamespaceManager namespaceManager = new XmlNamespaceManager(reader.NameTable);
namespaceManager.AddNamespace("aletheia", pageXmlNamespace.NamespaceName);
var Extracts = from REC in aletheiaDoc.Descendants(pageXmlNamespace + "TextLine")
select new Glyph
{
ID = (string)(REC.Attribute("id") ?? new XAttribute("id", string.Empty)),
Unicode = REC.XPathSelectElement("./aletheia:TextEquiv/aletheia:Unicode", namespaceManager) != null
? REC.XPathSelectElement("./aletheia:TextEquiv/aletheia:Unicode", namespaceManager).Value
: string.Empty,
PointsString = REC.XPathSelectElement("./aletheia:Coords", namespaceManager) != null
? REC.XPathSelectElement("./aletheia:Coords", namespaceManager).Attribute("points").Value
: string.Empty
};
return(Extracts.ToList());
}
public void IngestImages(string LangName, string FontName, string BaseName, string XMLFile, string PathToImages, bool UseSubList)
{
XDocument AletheiaDoc = XDocument.Load(XMLFile);
XNamespace ns = "http://schema.primaresearch.org/PAGE/gts/pagecontent/2010-03-19";
var Extracts = from REC in AletheiaDoc.Descendants(ns + "Glyph")
select new
{
ID = (string)(REC.Attribute("id") ?? new XAttribute("id", string.Empty)),
Char = (string)((REC.Element(ns + "TextEquiv") != null) ? (REC.Element(ns + "TextEquiv").Element(ns + "Unicode") != null)
? REC.Element(ns + "TextEquiv").Element(ns + "Unicode").Value : string.Empty
: string.Empty)
};
string[] GlyphFiles = System.IO.Directory.GetFiles(PathToImages);
foreach (var E in Extracts)
{
if (E.ID != "" && E.Char.Trim() != "")
{
string ImagePath = "";
string FixedChar = E.Char.Trim();
if (UseSubList)
{
FixedChar = FixGlyphChar(FixedChar);
}
for (int x = 0; x < GlyphFiles.Length; x++)
{
if (GlyphFiles[x].Replace(PathToImages + "\\", "").Replace(BaseName + "_", "").Replace(".tif", "") == E.ID)
{
ImagePath = GlyphFiles[x].Replace(db.DataDirectory, "");
break;
}
}
if (ImagePath != "")
{
int gIndex = this.FindGlyph(FixedChar);
if (gIndex < 0)
{
this.AddGlyph(FixedChar, ImagePath);
}
else
{
this.Glyphs[gIndex].AddImage(ImagePath);
}
}
}
}
}