14 lines
651 B
Text
14 lines
651 B
Text
# Writing SCRIPT(S) per language — DECLARED, not sniffed (D39.60 §5 G3): the echo detector, the repair
|
|
# script-guard and the CJK-language predicate must know a language's script WITHOUT guessing it from the text
|
|
# (a sniffed verdict is not reproducible on a resume). `lang<TAB>scriptname`, ONE script per row; a language
|
|
# written in several scripts (ja: kanji + both kana) lists each. Script NAMES resolve to Unicode ranges via
|
|
# scriptRanges (internal/lang/script.go). Add a language = add its row(s); add a script = add a scriptRanges
|
|
# entry + the row that names it.
|
|
zh han
|
|
ja han
|
|
ja hiragana
|
|
ja katakana
|
|
ko hangul
|
|
ko han
|
|
en latin
|
|
ru cyrillic
|