} // XML
/// <summary>
/// Load from an XML file
/// </summary>
/// <param name="file"></param>
/// <returns></returns>
public static XDict Load(String file)
{
XDict d = Utils.LoadXml <XDict>(file);
d.Reindex();
return(d);
}
public XDict Parse(string file)
{
XDict dict = new XDict();
FillHeader(dict);
// Read
using (Stream s = new FileStream(file, FileMode.Open, FileAccess.Read))
{
byte[] buf = new byte[6];
int pos = s.Read(buf, 0, buf.Length);
// First four bytes: BGL signature 0x12340001 or 0x12340002 (big-endian)
if (pos < 6 ||
(buf[0] == 0x12 && buf[1] == 0x34 && buf[2] == 0x00 &&
(buf[4] == 0x01 || buf[4] == 0x02)))
{
throw new FileLoadException("Invalid file: no BGL signature: " + file);
}
int gzipHeaderPos = buf[4] << 8 | buf[5];
if (gzipHeaderPos < 6)
{
throw new FileLoadException("No gzip ptr");
}
s.Seek(gzipHeaderPos, SeekOrigin.Begin);
parseMetaData(new GZipInputStream(s));
if (null == this.SrcEncoding && string.IsNullOrWhiteSpace(this.SrcEnc))
{
throw new InvalidDataException("Failed to detect source encoding in BGL file. Please provide encoding via proper constructor.");
}
else if (null == this.SrcEncoding)
{
// if SrcEncoding not set at command line
this.SrcEncoding = Encoding.GetEncoding(this.SrcEnc);
}
if (null == this.DstEncoding && string.IsNullOrWhiteSpace(this.DstEnc))
{
throw new InvalidDataException("Failed to detect destination encoding in BGL file. Please provide encoding via proper constructor.");
}
else if (null == this.DstEncoding)
{
// if DstEncoding not set at command line
this.DstEncoding = Encoding.GetEncoding(this.DstEnc);
}
s.Seek(gzipHeaderPos, SeekOrigin.Begin);
ParseUnzipped(new GZipInputStream(s), dict);
}
return(dict);
}
public XDict Parse(string file)
{
XDict dict = new XDict();
FillHeader(dict);
// Read
using (Stream s = new FileStream(file, FileMode.Open, FileAccess.Read))
{
byte[] buf = new byte[6];
int pos = s.Read(buf, 0, buf.Length);
// First four bytes: BGL signature 0x12340001 or 0x12340002 (big-endian)
if (pos < 6 ||
(buf[0] == 0x12 && buf[1] == 0x34 && buf[2] == 0x00 &&
(buf[4] == 0x01 || buf[4] == 0x02)))
{
throw new FileLoadException("Invalid file: no BGL signature: " + file);
}
int gzipHeaderPos = buf[4] << 8 | buf[5];
if (gzipHeaderPos < 6) { throw new FileLoadException("No gzip ptr"); }
s.Seek(gzipHeaderPos, SeekOrigin.Begin);
parseMetaData(new GZipInputStream(s));
if (null == this.SrcEncoding && string.IsNullOrWhiteSpace(this.SrcEnc))
{
throw new InvalidDataException("Failed to detect source encoding in BGL file. Please provide encoding via proper constructor.");
}
else if (null == this.SrcEncoding)
{
// if SrcEncoding not set at command line
this.SrcEncoding = Encoding.GetEncoding(this.SrcEnc);
}
if (null == this.DstEncoding && string.IsNullOrWhiteSpace(this.DstEnc))
{
throw new InvalidDataException("Failed to detect destination encoding in BGL file. Please provide encoding via proper constructor.");
}
else if (null == this.DstEncoding)
{
// if DstEncoding not set at command line
this.DstEncoding = Encoding.GetEncoding(this.DstEnc);
}
s.Seek(gzipHeaderPos, SeekOrigin.Begin);
ParseUnzipped(new GZipInputStream(s), dict);
}
return dict;
}
internal static void GenerateHtml(XDict dict, string outputPath)
{
StreamWriter outFileStream = new StreamWriter(outputPath, false, System.Text.Encoding.UTF8);
int numOfEntries = dict.Entries.Count;
int curEntry = 1;
int top = Console.CursorTop;
int left = Console.CursorLeft;
//header
outFileStream.WriteLine("<html>\n\t<head>\n\t\t<meta http-equiv=\"Content-Type\" content=\"text/html;charset=utf-8\" />\n\t\t<title>Babylon Dictionary</title>\n\t</head>\n\t<body>");
foreach(XDictEntry entry in dict.Entries)
{
Console.SetCursorPosition(left, top);
Console.Write("Entry #{0}/{1} ({2})...", curEntry.ToString(), numOfEntries.ToString(), (double.Parse(curEntry.ToString()) / double.Parse(numOfEntries.ToString())).ToString("P"));
curEntry++;
outFileStream.WriteLine("\t\t<idx:entry name=\"dic\">");
outFileStream.WriteLine("\t\t\t\t<idx:orth>" + entry.Word.Text);
if (entry.Comments.Count > 0)
{
outFileStream.WriteLine("\t\t\t\t<idx:infl>");
foreach (XWordInfo comment in entry.Comments)
{
if (!string.IsNullOrWhiteSpace(comment.Text))
outFileStream.WriteLine("\t\t\t\t\t<idx:iform value=\"" + comment.Text + "\" />");
}
outFileStream.WriteLine("\t\t\t\t</idx:infl>");
}
outFileStream.WriteLine("\t\t\t</idx:orth>");
foreach (XWordInfo definition in entry.Definitions)
{
outFileStream.WriteLine("\t\t\t<p><ul><li><blockquote>" + definition.Text + "</blockquote></li></ul></p>");
}
outFileStream.WriteLine("\t\t</idx:entry>");
}
outFileStream.WriteLine("\t</body>\n</html>");
outFileStream.Close();
Console.ForegroundColor = ConsoleColor.Magenta;
Console.WriteLine("\n\nGenerated HTML file \"{0}\".", Path.GetFileName(outputPath), numOfEntries);
}
void ParseUnzipped(Stream s, XDict dict)
{
while (true)
{
BglBlock block = BglBlock.Read(s);
if (block == null)
{
break;
}
if (block.Type == 1 || block.Type == 10)
{
XDictEntry e = ParseEntry(block);
dict.AddEntry(e);
}
}
}
protected void FillHeader(XDict dict)
{
//dict.Source = "BCL: " + _bclName;
//dict.Url = @"http://www.babylon.com/define/122/English-Thai-Dictionary.html";
//dict.Comment = "License unknown. Content is user created, and the website screams 'free dictionary', but BCL format is closed, no public spec.";
}
protected void FillHeader(XDict dict)
{
//dict.Source = "BCL: " + _bclName;
//dict.Url = @"http://www.babylon.com/define/122/English-Thai-Dictionary.html";
//dict.Comment = "License unknown. Content is user created, and the website screams 'free dictionary', but BCL format is closed, no public spec.";
}
void ParseUnzipped(Stream s, XDict dict)
{
while (true)
{
BglBlock block = BglBlock.Read(s);
if (block == null) { break; }
if (block.Type == 1 || block.Type == 10)
{
XDictEntry e = ParseEntry(block);
dict.AddEntry(e);
}
}
}
