.RTF File Format – Characters and languages

Character formatting

The specification describes \cgrid, \g, \gcw and \gridtbl only as above. To read negative \expnd values from Word for the Macintosh, which writes −7 to −1 as 57 to 63, use only the low six bits of the value.

Underlines

Character borders, shading and highlighting

<shading> = (\chshdng | <pat>) \chcfpat? \chcbpat?

Tracked changes

Associated character properties

A run of text can carry two sets of properties: one for left-to-right text and one for right-to-left text, or one each for low ASCII, high single-byte and double-byte characters. The word that comes last before the text (\ltrch or \rtlch; \loch, \hich or \dbch) says which kind the run is; the associated (\a…) properties written after the other words apply to the other kinds.

<atext>   = <ltrrun> | <rtlrun>
<ltrrun>  = \rtlch \af & <aprops>* \ltrch <ptext>
<rtlrun>  = \ltrch \af & <aprops>* \rtlch <ptext>
<atext>   = <losbrun> | <hisbrun> | <dbrun>
<losbrun> = \hich \af & <aprops> \dbch \af & <aprops> \loch <ptext>
<hisbrun> = \loch \af & <aprops> \dbch \af & <aprops> \hich <ptext>
<dbrun>   = \loch \af & <aprops> \hich \af & <aprops> \dbch <ptext>

The specification’s examples:

• \ltrch\af2\ab\au\rtlch\ul Sample Text – a right-to-left run in the default bidirectional font, underlined; its left-to-right font is font 2, bold and underlined (\au is printed so in the specification).
• \plain\rtlch\ltrch Sample Text – a left-to-right run; both fonts are the default font.
• \rtlch\af5\ab\ai\ltrch\ul Sample Text – a left-to-right run in the default font, underlined; its right-to-left font is font 5, bold and italic. A reader that does not support underline in the associated font underlines both.

Far East text

Double-byte text (Japanese, Chinese, Korean) is written in the code page of its font, one \'hh per byte. Writers should escape every byte below 0x20 or above 0x7F, and the three special characters \, { and } (0x5C, 0x7B, 0x7D). A reader that meets a lead byte of a double-byte character takes the next byte as its trail byte:

Japanese Word sets the character set in the font table with \fcharset. It reads \cpg437 as \fcharset0 and \cpg932 as \fcharset128, and ignores \cpg when both appear. With neither, it uses the font’s metrics, and assumes Shift JIS for an unknown font. Single-byte characters are mapped between character sets (Macintosh to ANSI, for example); double-byte characters are not.

When a run has no \loch, \hich or \dbch, Japanese Word treats bytes below 0x80 as \loch; a lead byte followed by a trail byte as one \dbch character; a lead byte followed by anything else, or at the end of the run, as \hich; and any other byte from 0x80 as \hich. In a style, the <dbrun> form should be used. If a style lacks one of the three, Word uses \f for \loch, any ANSI font for \hich and any Shift JIS font for \dbch. The specification’s example:

{\fonttbl{\f5\fswiss\fcharset0\fprq2 Arial;}{\f27\froman\fcharset128\fprq1 Mincho;}}
{\stylesheet{\loch\af5\hich\af5\dbch\f27\fs20\snext0 Normal;}}
\pard\plain
{\dbch\f27\fs20 \'82\'b1\'82\'ea\'82\'cd}
{\loch\f5 Test }
{\dbch\f27\'82\'c5\'82\'b7\'81B}
\par}

RTF-J words

RTF-J, a Japanese dialect of RTF, has three words of its own, which Word reads as follows:

Language codes

The values of \lang, \langfe, \langnp, \langfenp, \alang, \deflang, \deflangfe and \ksulang are Windows language identifiers. The 1.6 specification’s table, in its order; 1024 (0x0400) means no language, used to skip proofing.

A language identifier is a 16-bit number: the low ten bits are the primary language (0x09 English, 0x0C French) and the high six bits the sublanguage (1 the main country, 2 the next, and so on), so 0x0809 is English, sublanguage 2, British.