summaryrefslogtreecommitdiff
path: root/ext/bg/js/japanese.js
diff options
context:
space:
mode:
authortoasted-nutbread <toasted-nutbread@users.noreply.github.com>2020-04-10 11:56:18 -0400
committertoasted-nutbread <toasted-nutbread@users.noreply.github.com>2020-04-10 11:56:18 -0400
commit42a2917bf7aa3ab424ada2fc3acf224b74020a7f (patch)
tree647438015b7b526ddc2969d2c00e9e43b8bb792f /ext/bg/js/japanese.js
parentf177e3699ae85bafa3a648f5c03fbb8e303a6be3 (diff)
Add support for collapsing emphatic character sequences
Diffstat (limited to 'ext/bg/js/japanese.js')
-rw-r--r--ext/bg/js/japanese.js38
1 files changed, 37 insertions, 1 deletions
diff --git a/ext/bg/js/japanese.js b/ext/bg/js/japanese.js
index 2a2b39fd..e8b258cb 100644
--- a/ext/bg/js/japanese.js
+++ b/ext/bg/js/japanese.js
@@ -83,6 +83,8 @@
const ITERATION_MARK_CODE_POINT = 0x3005;
+ const HIRAGANA_SMALL_TSU_CODE_POINT = 0x3063;
+ const KATAKANA_SMALL_TSU_CODE_POINT = 0x30c3;
// Existing functions
@@ -373,6 +375,39 @@
}
+ // Miscellaneous
+
+ function collapseEmphaticSequences(sourceText, fullCollapse, sourceMap=null) {
+ let result = '';
+ let collapseCodePoint = -1;
+ const hasSourceMap = (sourceMap !== null);
+ for (const char of sourceText) {
+ const c = char.codePointAt(0);
+ if (c === HIRAGANA_SMALL_TSU_CODE_POINT || c === KATAKANA_SMALL_TSU_CODE_POINT) {
+ if (collapseCodePoint !== c) {
+ collapseCodePoint = c;
+ if (!fullCollapse) {
+ result += char;
+ continue;
+ }
+ }
+ } else {
+ collapseCodePoint = -1;
+ result += char;
+ continue;
+ }
+
+ if (hasSourceMap) {
+ const index = result.length;
+ if (index > 0) {
+ sourceMap.combine(index - 1, 1);
+ }
+ }
+ }
+ return result;
+ }
+
+
// Exports
Object.assign(jp, {
@@ -384,6 +419,7 @@
convertHalfWidthKanaToFullWidth,
convertAlphabeticToKana,
distributeFurigana,
- distributeFuriganaInflected
+ distributeFuriganaInflected,
+ collapseEmphaticSequences
});
})();