// The algorithm for this method comes from the implementer notes in [MS-OI29500].pdf
// section 2.1.87
// The implementer notes are at:
// http://msdn.microsoft.com/en-us/library/ee908652.aspx
public static FontType DetermineFontTypeFromCharacter(char ch, CharStyleAttributes csa)
{
// If the run has the cs element ("[ISO/IEC-29500-1] §17.3.2.7; cs") or the rtl element ("[ISO/IEC-29500-1] §17.3.2.30; rtl"),
// then the cs (or cstheme if defined) font is used, regardless of the Unicode character values of the run’s content.
if (csa.Rtl)
{
return FontType.CS;
}
// A large percentage of characters will fall in the following rule.
// Unicode Block: Basic Latin
if (ch >= 0x00 && ch <= 0x7f)
{
return FontType.Ascii;
}
// If the eastAsia (or eastAsiaTheme if defined) attribute’s value is “Times New Roman” and the ascii (or asciiTheme if defined)
// and hAnsi (or hAnsiTheme if defined) attributes are equal, then the ascii (or asciiTheme if defined) font is used.
if (csa.EastAsiaFont == "Times New Roman" &&
csa.AsciiFont == csa.HAnsiFont)
{
return FontType.Ascii;
}
// Unicode BLock: Latin-1 Supplement
if (ch >= 0xA0 && ch <= 0xFF)
{
if (csa.Hint == "eastAsia")
{
if (ch == 0xA1 ||
ch == 0xA4 ||
ch == 0xA7 ||
ch == 0xA8 ||
ch == 0xAA ||
ch == 0xAD ||
ch == 0xAF ||
(ch >= 0xB0 && ch <= 0xB4) ||
(ch >= 0xB6 && ch <= 0xBA) ||
(ch >= 0xBC && ch <= 0xBF) ||
ch == 0xD7 ||
ch == 0xF7)
{
return FontType.EastAsia;
}
if (csa.EastAsiaLang == "zh-hant" ||
csa.EastAsiaLang == "zh-hans")
{
if (ch == 0xE0 ||
ch == 0xE1 ||
(ch >= 0xE8 && ch <= 0xEA) ||
(ch >= 0xEC && ch <= 0xED) ||
(ch >= 0xF2 && ch <= 0xF3) ||
(ch >= 0xF9 && ch <= 0xFA) ||
ch == 0xFC)
{
return FontType.EastAsia;
}
}
}
return FontType.HAnsi;
}
// Unicode Block: Latin Extended-A
if (ch >= 0x0100 && ch <= 0x017F)
{
if (csa.Hint == "eastAsia")
{
if (csa.EastAsiaLang == "zh-hant" ||
csa.EastAsiaLang == "zh-hans"
/* || the character set of the east Asia (or east Asia theme) font is Chinese5 || GB2312 todo */)
{
return FontType.EastAsia;
}
}
return FontType.HAnsi;
}
// Unicode Block: Latin Extended-B
if (ch >= 0x0180 && ch <= 0x024F)
{
if (csa.Hint == "eastAsia")
{
if (csa.EastAsiaLang == "zh-hant" ||
csa.EastAsiaLang == "zh-hans"
/* || the character set of the east Asia (or east Asia theme) font is Chinese5 || GB2312 todo */)
{
return FontType.EastAsia;
}
}
return FontType.HAnsi;
}
// Unicode Block: IPA Extensions
if (ch >= 0x0250 && ch <= 0x02AF)
{
if (csa.Hint == "eastAsia")
{
if (csa.EastAsiaLang == "zh-hant" ||
csa.EastAsiaLang == "zh-hans"
/* || the character set of the east Asia (or east Asia theme) font is Chinese5 || GB2312 todo */)
{
return FontType.EastAsia;
}
}
return FontType.HAnsi;
}
// Unicode Block: Spacing Modifier Letters
if (ch >= 0x02B0 && ch <= 0x02FF)
{
if (csa.Hint == "eastAsia")
{
return FontType.EastAsia;
}
return FontType.HAnsi;
}
// Unicode Block: Combining Diacritic Marks
if (ch >= 0x0300 && ch <= 0x036F)
{
if (csa.Hint == "eastAsia")
{
return FontType.EastAsia;
}
return FontType.HAnsi;
}
// Unicode Block: Greek
if (ch >= 0x0370 && ch <= 0x03CF)
{
if (csa.Hint == "eastAsia")
{
return FontType.EastAsia;
}
return FontType.HAnsi;
}
// Unicode Block: Cyrillic
if (ch >= 0x0400 && ch <= 0x04FF)
{
if (csa.Hint == "eastAsia")
{
return FontType.EastAsia;
}
return FontType.HAnsi;
}
// Unicode Block: Hebrew
if (ch >= 0x0590 && ch <= 0x05FF)
{
return FontType.Ascii;
}
// Unicode Block: Arabic
if (ch >= 0x0600 && ch <= 0x06FF)
{
return FontType.Ascii;
}
// Unicode Block: Syriac
if (ch >= 0x0700 && ch <= 0x074F)
{
return FontType.Ascii;
}
// Unicode Block: Arabic Supplement
if (ch >= 0x0750 && ch <= 0x077F)
{
return FontType.Ascii;
}
// Unicode Block: Thanna
if (ch >= 0x0780 && ch <= 0x07BF)
{
return FontType.Ascii;
}
// Unicode Block: Hangul Jamo
if (ch >= 0x1100 && ch <= 0x11FF)
{
return FontType.EastAsia;
}
// Unicode Block: Latin Extended Additional
if (ch >= 0x1E00 && ch <= 0x1EFF)
{
if (csa.Hint == "eastAsia" &&
(csa.EastAsiaLang == "zh-hant" ||
csa.EastAsiaLang == "zh-hans"))
{
return FontType.EastAsia;
}
return FontType.HAnsi;
}
// Unicode Block: General Punctuation
if (ch >= 0x2000 && ch <= 0x206F)
{
if (csa.Hint == "eastAsia")
{
return FontType.EastAsia;
}
return FontType.HAnsi;
}
// Unicode Block: Superscripts and Subscripts
if (ch >= 0x2070 && ch <= 0x209F)
{
if (csa.Hint == "eastAsia")
{
return FontType.EastAsia;
}
return FontType.HAnsi;
}
// Unicode Block: Currency Symbols
if (ch >= 0x20A0 && ch <= 0x20CF)
{
if (csa.Hint == "eastAsia")
{
return FontType.EastAsia;
}
return FontType.HAnsi;
}
// Unicode Block: Combining Diacritical Marks for Symbols
if (ch >= 0x20D0 && ch <= 0x20FF)
{
if (csa.Hint == "eastAsia")
{
return FontType.EastAsia;
}
return FontType.HAnsi;
}
// Unicode Block: Letter-like Symbols
if (ch >= 0x2100 && ch <= 0x214F)
{
if (csa.Hint == "eastAsia")
{
return FontType.EastAsia;
}
return FontType.HAnsi;
}
// Unicode Block: Number Forms
if (ch >= 0x2150 && ch <= 0x218F)
{
if (csa.Hint == "eastAsia")
{
return FontType.EastAsia;
}
return FontType.HAnsi;
}
// Unicode Block: Arrows
if (ch >= 0x2190 && ch <= 0x21FF)
{
if (csa.Hint == "eastAsia")
{
return FontType.EastAsia;
}
return FontType.HAnsi;
}
// Unicode Block: Mathematical Operators
if (ch >= 0x2200 && ch <= 0x22FF)
{
if (csa.Hint == "eastAsia")
{
return FontType.EastAsia;
}
return FontType.HAnsi;
}
// Unicode Block: Miscellaneous Technical
if (ch >= 0x2300 && ch <= 0x23FF)
{
if (csa.Hint == "eastAsia")
{
return FontType.EastAsia;
}
return FontType.HAnsi;
}
// Unicode Block: Control Pictures
if (ch >= 0x2400 && ch <= 0x243F)
{
if (csa.Hint == "eastAsia")
{
return FontType.EastAsia;
}
return FontType.HAnsi;
}
// Unicode Block: Optical Character Recognition
if (ch >= 0x2440 && ch <= 0x245F)
{
if (csa.Hint == "eastAsia")
{
return FontType.EastAsia;
}
return FontType.HAnsi;
}
// Unicode Block: Enclosed Alphanumerics
if (ch >= 0x2460 && ch <= 0x24FF)
{
if (csa.Hint == "eastAsia")
{
return FontType.EastAsia;
}
return FontType.HAnsi;
}
// Unicode Block: Box Drawing
if (ch >= 0x2500 && ch <= 0x257F)
{
if (csa.Hint == "eastAsia")
{
return FontType.EastAsia;
}
return FontType.HAnsi;
}
// Unicode Block: Block Elements
if (ch >= 0x2580 && ch <= 0x259F)
{
if (csa.Hint == "eastAsia")
{
return FontType.EastAsia;
}
return FontType.HAnsi;
}
// Unicode Block: Geometric Shapes
if (ch >= 0x25A0 && ch <= 0x25FF)
{
if (csa.Hint == "eastAsia")
{
return FontType.EastAsia;
}
return FontType.HAnsi;
}
// Unicode Block: Miscellaneous Symbols
if (ch >= 0x2600 && ch <= 0x26FF)
{
if (csa.Hint == "eastAsia")
{
return FontType.EastAsia;
}
return FontType.HAnsi;
}
// Unicode Block: Dingbats
if (ch >= 0x2700 && ch <= 0x27BF)
{
if (csa.Hint == "eastAsia")
{
return FontType.EastAsia;
}
return FontType.HAnsi;
}
// Unicode Block: CJK Radicals Supplement
if (ch >= 0x2E80 && ch <= 0x2EFF)
{
if (csa.Hint == "eastAsia")
{
return FontType.EastAsia;
}
return FontType.HAnsi;
}
// Unicode Block: Kangxi Radicals
if (ch >= 0x2F00 && ch <= 0x2FDF)
{
return FontType.EastAsia;
}
// Unicode Block: Ideographic Description Characters
if (ch >= 0x2FF0 && ch <= 0x2FFF)
{
return FontType.EastAsia;
}
// Unicode Block: CJK Symbols and Punctuation
if (ch >= 0x3000 && ch <= 0x303F)
{
return FontType.EastAsia;
}
// Unicode Block: Hiragana
if (ch >= 0x3040 && ch <= 0x309F)
{
return FontType.EastAsia;
}
// Unicode Block: Katakana
if (ch >= 0x30A0 && ch <= 0x30FF)
{
return FontType.EastAsia;
}
// Unicode Block: Bopomofo
if (ch >= 0x3100 && ch <= 0x312F)
{
return FontType.EastAsia;
}
// Unicode Block: Hangul Compatibility Jamo
if (ch >= 0x3130 && ch <= 0x318F)
{
return FontType.EastAsia;
}
// Unicode Block: Kanbun
if (ch >= 0x3190 && ch <= 0x319F)
{
return FontType.EastAsia;
}
// Unicode Block: Enclosed CJK Letters and Months
if (ch >= 0x3200 && ch <= 0x32FF)
{
return FontType.EastAsia;
}
// Unicode Block: CJK Compatibility
if (ch >= 0x3300 && ch <= 0x33FF)
{
return FontType.EastAsia;
}
// Unicode Block: CJK Unified Ideographs Extension A
if (ch >= 0x3400 && ch <= 0x4DBF)
{
return FontType.EastAsia;
}
// Unicode Block: CJK Unified Ideographs
if (ch >= 0x4E00 && ch <= 0x9FAF)
{
return FontType.EastAsia;
}
// Unicode Block: Yi Syllables
if (ch >= 0xA000 && ch <= 0xA48F)
{
return FontType.EastAsia;
}
// Unicode Block: Yi Radicals
if (ch >= 0xA490 && ch <= 0xA4CF)
{
return FontType.EastAsia;
}
// Unicode Block: Hangul Syllables
if (ch >= 0xAC00 && ch <= 0xD7AF)
{
return FontType.EastAsia;
}
// Unicode Block: High Surrogates
if (ch >= 0xD800 && ch <= 0xDB7F)
{
return FontType.EastAsia;
}
// Unicode Block: High Private Use Surrogates
if (ch >= 0xDB80 && ch <= 0xDBFF)
{
return FontType.EastAsia;
}
// Unicode Block: Low Surrogates
if (ch >= 0xDC00 && ch <= 0xDFFF)
{
return FontType.EastAsia;
}
// Unicode Block: Private Use Area
if (ch >= 0xE000 && ch <= 0xF8FF)
{
if (csa.Hint == "eastAsia")
{
return FontType.EastAsia;
}
return FontType.HAnsi;
}
// Unicode Block: CJK Compatibility Ideographs
if (ch >= 0xF900 && ch <= 0xFAFF)
{
return FontType.EastAsia;
}
// Unicode Block: Alphabetic Presentation Forms
if (ch >= 0xFB00 && ch <= 0xFB4F)
{
if (csa.Hint == "eastAsia")
{
if (ch >= 0xFB00 && ch <= 0xFB1C)
return FontType.EastAsia;
if (ch >= 0xFB1D && ch <= 0xFB4F)
return FontType.Ascii;
}
return FontType.HAnsi;
}
// Unicode Block: Arabic Presentation Forms-A
if (ch >= 0xFB50 && ch <= 0xFDFF)
{
return FontType.Ascii;
}
// Unicode Block: CJK Compatibility Forms
if (ch >= 0xFE30 && ch <= 0xFE4F)
{
return FontType.EastAsia;
}
// Unicode Block: Small Form Variants
if (ch >= 0xFE50 && ch <= 0xFE6F)
{
return FontType.EastAsia;
}
// Unicode Block: Arabic Presentation Forms-B
if (ch >= 0xFE70 && ch <= 0xFEFE)
{
return FontType.Ascii;
}
// Unicode Block: Halfwidth and Fullwidth Forms
if (ch >= 0xFF00 && ch <= 0xFFEF)
{
return FontType.EastAsia;
}
return FontType.HAnsi;
}
private static void AdjustFontAttributes(WordprocessingDocument wDoc, XElement paraOrRun, XElement pPr,
XElement rPr, FormattingAssemblerSettings settings)
{
XDocument themeXDoc = null;
if (wDoc.MainDocumentPart.ThemePart != null)
themeXDoc = wDoc.MainDocumentPart.ThemePart.GetXDocument();
XElement fontScheme = null;
XElement majorFont = null;
XElement minorFont = null;
if (themeXDoc != null)
{
fontScheme = themeXDoc.Root.Element(A.themeElements).Element(A.fontScheme);
majorFont = fontScheme.Element(A.majorFont);
minorFont = fontScheme.Element(A.minorFont);
}
var rFonts = rPr.Element(W.rFonts);
if (rFonts == null)
{
return;
}
var asciiTheme = (string)rFonts.Attribute(W.asciiTheme);
var hAnsiTheme = (string)rFonts.Attribute(W.hAnsiTheme);
var eastAsiaTheme = (string)rFonts.Attribute(W.eastAsiaTheme);
var cstheme = (string)rFonts.Attribute(W.cstheme);
string ascii = null;
string hAnsi = null;
string eastAsia = null;
string cs = null;
XElement minorLatin = null;
string minorLatinTypeface = null;
XElement majorLatin = null;
string majorLatinTypeface = null;
if (minorFont != null)
{
minorLatin = minorFont.Element(A.latin);
minorLatinTypeface = (string)minorLatin.Attribute("typeface");
}
if (majorFont != null)
{
majorLatin = majorFont.Element(A.latin);
majorLatinTypeface = (string)majorLatin.Attribute("typeface");
}
if (asciiTheme != null)
{
if (asciiTheme.StartsWith("minor") && minorLatinTypeface != null)
{
ascii = minorLatinTypeface;
}
else if (asciiTheme.StartsWith("major") && majorLatinTypeface != null)
{
ascii = majorLatinTypeface;
}
}
if (hAnsiTheme != null)
{
if (hAnsiTheme.StartsWith("minor") && minorLatinTypeface != null)
{
hAnsi = minorLatinTypeface;
}
else if (hAnsiTheme.StartsWith("major") && majorLatinTypeface != null)
{
hAnsi = majorLatinTypeface;
}
}
if (eastAsiaTheme != null)
{
if (eastAsiaTheme.StartsWith("minor") && minorLatinTypeface != null)
{
eastAsia = minorLatinTypeface;
}
else if (eastAsiaTheme.StartsWith("major") && majorLatinTypeface != null)
{
eastAsia = majorLatinTypeface;
}
}
if (cstheme != null)
{
if (cstheme.StartsWith("minor") && minorFont != null)
{
cs = (string)minorFont.Element(A.cs).Attribute("typeface");
}
else if (cstheme.StartsWith("major") && majorFont != null)
{
cs = (string)majorFont.Element(A.cs).Attribute("typeface");
}
}
if (ascii != null)
{
rFonts.SetAttributeValue(W.ascii, ascii);
}
if (hAnsi != null)
{
rFonts.SetAttributeValue(W.hAnsi, hAnsi);
}
if (eastAsia != null)
{
rFonts.SetAttributeValue(W.eastAsia, eastAsia);
}
if (cs != null)
{
rFonts.SetAttributeValue(W.cs, cs);
}
var firstTextNode = paraOrRun.Descendants(W.t).FirstOrDefault(t => t.Value.Length > 0);
string str = " ";
// if there is a run with no text in it, then no need to do any of the rest of this method.
if (firstTextNode == null && paraOrRun.Name == W.r)
return;
if (firstTextNode != null)
str = firstTextNode.Value;
var csa = new CharStyleAttributes(pPr, rPr);
// This module determines the font based on just the first character.
// Technically, a run can contain characters from different Unicode code blocks, and hence should be rendered with different fonts.
// However, Word breaks up runs that use more than one font into multiple runs. Other producers of WordprocessingML may not, so in
// that case, this routine may need to be augmented to look at all characters in a run.
/*
old code
var fontFamilies = str.select(function (c) {
var ft = Pav.DetermineFontTypeFromCharacter(c, csa);
switch (ft) {
case Pav.FontType.Ascii:
return cast(rFonts.attribute(W.ascii));
case Pav.FontType.HAnsi:
return cast(rFonts.attribute(W.hAnsi));
case Pav.FontType.EastAsia:
return cast(rFonts.attribute(W.eastAsia));
case Pav.FontType.CS:
return cast(rFonts.attribute(W.cs));
default:
return null;
}
})
.where(function (f) { return f != null && f != ""; })
.distinct()
.select(function (f) { return new Pav.FontFamily(f); })
.toArray();
*/
var charToExamine = str.FirstOrDefault(c => ! WeakAndNeutralDirectionalCharacters.Contains(c));
if (charToExamine == '\0')
charToExamine = str[0];
var ft = DetermineFontTypeFromCharacter(charToExamine, csa);
string fontType = null;
string languageType = null;
switch (ft)
{
case FontType.Ascii:
fontType = (string)rFonts.Attribute(W.ascii);
languageType = "western";
break;
case FontType.HAnsi:
fontType = (string)rFonts.Attribute(W.hAnsi);
languageType = "western";
break;
case FontType.EastAsia:
if (settings.RestrictToSupportedLanguages)
throw new UnsupportedLanguageException("EastAsia languages are not supported");
fontType = (string)rFonts.Attribute(W.eastAsia);
languageType = "eastAsia";
break;
case FontType.CS:
if (settings.RestrictToSupportedLanguages)
throw new UnsupportedLanguageException("Complex script (RTL) languages are not supported");
fontType = (string)rFonts.Attribute(W.cs);
languageType = "bidi";
break;
}
if (fontType != null)
{
if (paraOrRun.Attribute(PtOpenXml.FontName) == null)
{
XAttribute fta = new XAttribute(PtOpenXml.FontName, fontType.ToString());
paraOrRun.Add(fta);
}
else
{
paraOrRun.Attribute(PtOpenXml.FontName).Value = fontType.ToString();
}
}
if (languageType != null)
{
if (paraOrRun.Attribute(PtOpenXml.LanguageType) == null)
{
XAttribute lta = new XAttribute(PtOpenXml.LanguageType, languageType);
paraOrRun.Add(lta);
}
else
{
paraOrRun.Attribute(PtOpenXml.LanguageType).Value = languageType;
}
}
}
