Legacy Document

Inside Macintosh: Text /

Appendix B - International Resources / Transliteration Resource (Type 'trsl')

Rule-Based Format

In the rule-based version of the transliteration resource, the header is followed immediately by a 2-byte field containing a count of the number of rules in the resource; the rules immediately follow the count field and constitute the remainder of the resource. This is the definition of the rule-based resource header:

TYPE  RuleBasedTrslRecord = 
   RECORD
      sourceType:    Integer; {target type for left side of rule}
      targetType:    Integer; {target type for right side of rule}
      formatNumber:  Integer; {format of this resource}
      propertyFlag:  Integer; {transliteration property flags}
      numberOfRules: Integer; {number of rules that follow}
   END;

Figure B-19 shows the format of a rule.

Figure B-19 Format of a transliteration rule

• The length of the rule is a byte that specifies the actual number of bytes in the rule, excluding the length byte itself.
• The left side of the rule contains the source pattern, a sequence of one or more character codes that the TransliterateText function compares to the source string. If it finds a match, it returns the right side of the rule (the target pattern). The rules are organized to implement the longest match algorithm, meaning that the longest source pattern that matches a particular target pattern is the one that is converted. For instance, the rule

  abb --> hello

  takes precedence over the rule

  ab --> hello

Some rules in some versions of the transliteration resource incorporate a look-ahead feature, in which a particular source pattern is converted to its target pattern only if it is followed by other specific characters. For example, if "[" represents the look-ahead symbol, the characters preceding it in the left side of the rule are converted to the right side of the rule only if the characters following "[" in the left side of the rule match the subsequent characters in the input string.

If these are the matching rules:

Left side	Rignt side
a[bc	A
b	B
c	C
d[ef	D
f	F

Then if we have the input string

abcdf

the output string will be:

ABCdF

because, in the input string, the characters "bc" follow "a", but the characters "ef" do not follow "d".