feat(insights): fix DBSCAN, Persian embeddings crash, D3 physics layouts, and D3 node not found runtime error

memento-note/lib/clip/rtl-content.ts

@@ -0,0 +1,112 @@
+/** Détection RTL et enveloppe HTML pour contenus clippés (persan, arabe, hébreu). */
+
+const RTL_CHAR = /[\u0590-\u05FF\u0600-\u06FF\u0700-\u074F\u0750-\u077F\u08A0-\u08FF\uFB50-\uFDFF\uFE70-\uFEFF]/
+const LTR_CHAR = /[A-Za-z0-9]/
+
+export type ClipTextDirection = 'rtl' | 'ltr'
+
+export interface ClipLocaleHint {
+  direction: ClipTextDirection
+  lang?: 'fa' | 'ar' | 'he'
+}
+
+export function inferLangFromUrl(url: string): ClipLocaleHint['lang'] | undefined {
+  const lower = url.toLowerCase()
+  if (/\/persian\b|\/fa\b|lang=fa|bbc\.com\/persian/.test(lower)) return 'fa'
+  if (/\/arabic\b|\/ar\b|lang=ar/.test(lower)) return 'ar'
+  if (/\/hebrew\b|\/he\b|lang=he/.test(lower)) return 'he'
+  return undefined
+}
+
+export function detectTextDirection(text: string): ClipTextDirection {
+  const sample = text.replace(/\s+/g, '').slice(0, 4000)
+  if (!sample) return 'ltr'
+
+  let rtl = 0
+  let ltr = 0
+  for (const ch of sample) {
+    if (RTL_CHAR.test(ch)) rtl++
+    else if (LTR_CHAR.test(ch)) ltr++
+  }
+  if (rtl === 0) return 'ltr'
+  return rtl >= ltr ? 'rtl' : 'ltr'
+}
+
+/** Direction du titre de note (éviter dir="auto" qui casse les chiffres persans). */
+export function resolveTitleDirection(title: string, sourceUrl?: string | null): ClipTextDirection {
+  if (sourceUrl && inferLangFromUrl(sourceUrl)) return 'rtl'
+  return detectTextDirection(title)
+}
+
+export function resolveTitleLang(
+  title: string,
+  sourceUrl?: string | null,
+): ClipLocaleHint['lang'] | undefined {
+  const urlLang = sourceUrl ? inferLangFromUrl(sourceUrl) : undefined
+  if (urlLang) return urlLang
+  if (detectTextDirection(title) !== 'rtl') return undefined
+  return resolveClipLocale(sourceUrl || '', title).lang
+}
+
+export function resolveClipLocale(url: string, ...texts: string[]): ClipLocaleHint {
+  const combined = texts.filter(Boolean).join('\n')
+  const direction = detectTextDirection(combined)
+  const urlLang = inferLangFromUrl(url)
+  let lang = urlLang
+
+  if (!lang && direction === 'rtl') {
+    if (/[\u06AF\u06CC\u06A9\u067E\u0686\u0698\u200C]/.test(combined)) lang = 'fa'
+    else if (/[\u0590-\u05FF]/.test(combined)) lang = 'he'
+    else lang = 'ar'
+  }
+
+  return { direction, lang }
+}
+
+/** Applique dir/lang sur les blocs HTML extraits (Readability ne les conserve pas toujours). */
+export function applyRtlToHtmlBlocks(html: string, hint: ClipLocaleHint): string {
+  if (hint.direction !== 'rtl') return html
+
+  const langAttr = hint.lang ? ` lang="${hint.lang}"` : ''
+  const blockTags = ['p', 'h1', 'h2', 'h3', 'h4', 'li', 'ul', 'ol', 'blockquote', 'figcaption']
+
+  let out = html
+  for (const tag of blockTags) {
+    out = out.replace(new RegExp(`<${tag}(\\s[^>]*)?>`, 'gi'), (match, attrs = '') => {
+      if (/dir\s*=/.test(attrs)) return match
+      return `<${tag}${attrs} dir="rtl"${langAttr}>`
+    })
+  }
+
+  return out
+}
+
+export function wrapClipArticleHtml(innerHtml: string, hint: ClipLocaleHint): string {
+  if (hint.direction !== 'rtl') return innerHtml
+  const langAttr = hint.lang ? ` lang="${hint.lang}"` : ''
+  return `<div class="clip-article clip-article--rtl" dir="rtl"${langAttr}>${innerHtml}</div>`
+}
+
+export function wrapClipPlainParagraph(text: string, hint: ClipLocaleHint): string {
+  const escaped = text
+    .replace(/&/g, '&amp;')
+    .replace(/</g, '&lt;')
+    .replace(/>/g, '&gt;')
+  const langAttr = hint.lang ? ` lang="${hint.lang}"` : ''
+  const dirAttr = hint.direction === 'rtl' ? ' dir="rtl"' : ''
+  return `<p${dirAttr}${langAttr}>${escaped}</p>`
+}
+
+export function readPageLocaleFromHtml(html: string): Pick<ClipLocaleHint, 'direction' | 'lang'> {
+  const dirMatch =
+    html.match(/<html[^>]*\sdir=["'](rtl|ltr)["']/i) ||
+    html.match(/<body[^>]*\sdir=["'](rtl|ltr)["']/i)
+  const langMatch =
+    html.match(/<html[^>]*\slang=["']([^"']+)["']/i) ||
+    html.match(/<body[^>]*\slang=["']([^"']+)["']/i)
+  const direction: ClipTextDirection = dirMatch?.[1]?.toLowerCase() === 'rtl' ? 'rtl' : 'ltr'
+  const rawLang = langMatch?.[1]?.split('-')[0]?.toLowerCase()
+  const lang =
+    rawLang === 'fa' || rawLang === 'ar' || rawLang === 'he' ? (rawLang as ClipLocaleHint['lang']) : undefined
+  return { direction, lang }
+}