compat(epub): detect missing writing direction for some EPUBs (#2803)

2025-12-28 19:43:49 +01:00
parent 988fbc8c85
commit b76a2adf61
5 changed files with 42 additions and 5 deletions
@@ -132,7 +132,8 @@ const FoliateViewer: React.FC<{
          const bookData = getBookData(bookKey);
          if (viewSettings && detail.type === 'text/css')
            return transformStylesheet(data, width, height, viewSettings.vertical);
-          if (viewSettings && bookData && detail.type === 'application/xhtml+xml') {
+          const isHtml = detail.type === 'application/xhtml+xml' || detail.type === 'text/html';
+          if (viewSettings && bookData && isHtml) {
            const ctx: TransformContext = {
              bookKey,
              viewSettings,
@@ -1,4 +1,4 @@
-import { detectLanguage, isSameLang, isValidLang } from '@/utils/lang';
+import { detectLanguage, getLanguageInfo, isSameLang, isValidLang } from '@/utils/lang';
 import type { Transformer } from './types';

 export const languageTransformer: Transformer = {
@@ -21,12 +21,25 @@ export const languageTransformer: Transformer = {
          isValidLang(primaryLanguage) && primaryLanguage !== 'en'
            ? primaryLanguage
            : detectLanguage(mainContent);
+        const languageInfo = getLanguageInfo(lang || '');
        const newLangAttr = ` lang="${lang}"`;
        const newXmlLangAttr = ` xml:lang="${lang}"`;
-        attrs = langMatch ? attrs.replace(langRegex, newLangAttr) : attrs + newLangAttr;
-        attrs = xmlLangMatch ? attrs.replace(xmlLangRegex, newXmlLangAttr) : attrs + newXmlLangAttr;
+        const dirAttr = languageInfo?.direction === 'rtl' ? ' dir="rtl"' : '';
+        attrs = langMatch ? attrs.replace(langRegex, newLangAttr) : attrs + newLangAttr + dirAttr;
+        attrs = xmlLangMatch
+          ? attrs.replace(xmlLangRegex, newXmlLangAttr)
+          : attrs + newXmlLangAttr + dirAttr;
        result = result.replace(attrsMatch[0], `<html${attrs}>`);
      }
+    } else {
+      const lang =
+        isValidLang(primaryLanguage) && primaryLanguage !== 'en'
+          ? primaryLanguage
+          : detectLanguage(result.replace(/<[^>]+>/g, ' '));
+      const languageInfo = getLanguageInfo(lang || '');
+      const dirAttr = languageInfo?.direction === 'rtl' ? ' dir="rtl"' : '';
+      const newAttrs = ` lang="${lang}" xml:lang="${lang}" ${dirAttr}`;
+      result = result.replace(/<html>/i, `<html${newAttrs}>`);
    }
    return result;
  },
@@ -4,3 +4,8 @@ export type Insets = {
  left: number;
  right: number;
 };
+
+export interface LocaleWithTextInfo extends Intl.Locale {
+  getTextInfo?: () => { direction: string };
+  textInfo?: { direction: string };
+}
@@ -2,6 +2,7 @@ import { BookDoc } from '@/libs/document';
 import { BookNote, BookSearchConfig, BookSearchResult } from '@/types/book';
 import { TTSGranularity } from '@/services/tts';
 import { TTS } from 'foliate-js/tts.js';
+import { LocaleWithTextInfo } from './misc';

 export const NOTE_PREFIX = 'foliate-note:';

@@ -33,8 +34,10 @@ export interface FoliateView extends HTMLElement {
  book: BookDoc;
  tts: TTS | null;
  language: {
-    locale?: string;
+    locale?: LocaleWithTextInfo;
    isCJK?: boolean;
+    canonical?: string;
+    direction?: string;
  };
  history: {
    canGoBack: boolean;
@@ -1,3 +1,4 @@
+import { LocaleWithTextInfo } from '@/types/misc';
 import { franc } from 'franc-min';
 import { iso6392 } from 'iso-639-2';
 import { iso6393To1 } from 'iso-639-3';
@@ -116,3 +117,17 @@ export const detectLanguage = (content: string): string => {
    return 'en';
  }
 };
+
+export const getLanguageInfo = (lang: string) => {
+  if (!lang) return {};
+  try {
+    const canonical = Intl.getCanonicalLocales(lang)[0]!;
+    const locale = new Intl.Locale(canonical) as LocaleWithTextInfo;
+    const isCJK = ['zh', 'ja', 'kr'].includes(locale.language);
+    const direction = (locale.getTextInfo?.() ?? locale.textInfo)?.direction;
+    return { canonical, locale, isCJK, direction };
+  } catch (e) {
+    console.warn(e);
+    return {};
+  }
+};