Upload src/viNumberFix.js

src/viNumberFix.js

@@ -0,0 +1,161 @@
+/**
+ * Vietnamese text processing for both display and TTS.
+ *
+ * DESIGN: 
+ * - Display text (chat & subtitles): numbers stay as digits for clean visual display
+ * - TTS audio: numbers are converted to Vietnamese words for proper pronunciation
+ * - Source links in greetings: rendered as clickable popup links
+ */
+
+// Vietnamese number words
+const DIGITS = [
+  "không", "một", "hai", "ba", "bốn", "năm", "sáu", "bảy", "tám", "chín",
+];
+
+function numberToVietnamese(n) {
+  if (n === 0) return "không";
+  if (n < 0) return "âm " + numberToVietnamese(-n);
+
+  const units = ["", "nghìn", "triệu", "tỷ"];
+  const groups = [];
+  let temp = n;
+  while (temp > 0) {
+    groups.push(temp % 1000);
+    temp = Math.floor(temp / 1000);
+  }
+  if (groups.length === 0) groups.push(0);
+
+  const readGroup = (g) => {
+    if (g === 0) return "";
+    const h = Math.floor(g / 100);
+    const r = g % 100;
+    let s = "";
+    if (h > 0) s += DIGITS[h] + " trăm ";
+    else if (groups.length > 1) s += "không trăm ";
+    if (r === 0) return s.trim();
+    if (r < 10) {
+      s += (h > 0 && r === 5) ? "lẻ năm" : "lẻ " + DIGITS[r];
+    } else if (r < 20) {
+      s += "mười" + (r === 10 ? "" : (r === 15 ? " lăm" : " " + DIGITS[r % 10]));
+    } else {
+      const t = Math.floor(r / 10);
+      const o = r % 10;
+      s += ["", "", "hai mươi", "ba mươi", "bốn mươi", "năm mươi",
+            "sáu mươi", "bảy mươi", "tám mươi", "chín mươi"][t];
+      if (o === 1) s += " mốt";
+      else if (o === 5) s += " lăm";
+      else if (o > 0) s += " " + DIGITS[o];
+    }
+    return s.replace(/\s+/g, " ").trim();
+  };
+
+  let result = "";
+  for (let i = groups.length - 1; i >= 0; i--) {
+    const g = groups[i];
+    if (g === 0) continue;
+    result += readGroup(g);
+    if (i > 0) result += " " + units[i] + " ";
+  }
+  return result.replace(/\s+/g, " ").trim();
+}
+
+/**
+ * Normalize Vietnamese text for display.
+ * Numbers remain as digits for clean visual display.
+ */
+export function normalizeVietnameseText(text) {
+  if (!text || typeof text !== "string") return text;
+  return text.replace(/\s+/g, " ").trim();
+}
+
+/**
+ * Check if text contains Vietnamese characters.
+ */
+export function containsVietnamese(text) {
+  return /[àáạảãâầấậẩẫăằắặẳẵèéẹẻẽêềếệểễìíịỉĩòóọỏõôồốộổỗơờớợởỡùúụủũưừứựửữỳýỵỷỹđ]/i.test(text);
+}
+
+/**
+ * Convert Arabic numerals to Vietnamese words for TTS.
+ * Display texts keep the original digits.
+ */
+export function numbersToVietnameseWords(text) {
+  if (!text || typeof text !== "string") return text;
+  
+  const units = ["", "nghìn", "triệu", "tỷ"];
+  
+  const numberToVietnameseLocal = (n) => {
+    if (n === 0) return "không";
+    if (n < 0) return "âm " + numberToVietnameseLocal(-n);
+    
+    const groups = [];
+    let temp = n;
+    while (temp > 0) {
+      groups.push(temp % 1000);
+      temp = Math.floor(temp / 1000);
+    }
+    if (groups.length === 0) groups.push(0);
+    
+    const readGroup = (g) => {
+      if (g === 0) return "";
+      const h = Math.floor(g / 100);
+      const r = g % 100;
+      let s = "";
+      if (h > 0) s += DIGITS[h] + " trăm ";
+      else if (groups.length > 1) s += "không trăm ";
+      if (r === 0) return s.trim();
+      if (r < 10) {
+        s += (h > 0 && r === 5) ? "lẻ năm" : "lẻ " + DIGITS[r];
+      } else if (r < 20) {
+        s += "mười" + (r === 10 ? "" : (r === 15 ? " lăm" : " " + DIGITS[r % 10]));
+      } else {
+        const t = Math.floor(r / 10);
+        const o = r % 10;
+        s += ["", "", "hai mươi", "ba mươi", "bốn mươi", "năm mươi",
+              "sáu mươi", "bảy mươi", "tám mươi", "chín mươi"][t];
+        if (o === 1) s += " mốt";
+        else if (o === 5) s += " lăm";
+        else if (o > 0) s += " " + DIGITS[o];
+      }
+      return s.replace(/\s+/g, " ").trim();
+    };
+    
+    let result = "";
+    for (let i = groups.length - 1; i >= 0; i--) {
+      const g = groups[i];
+      if (g === 0) continue;
+      result += readGroup(g);
+      if (i > 0) result += " " + units[i] + " ";
+    }
+    return result.replace(/\s+/g, " ").trim();
+  };
+  
+  // Match integers (including negative) - convert to Vietnamese words
+  return text.replace(/\b-?\d+\b/g, (match) => {
+    const num = parseInt(match);
+    return numberToVietnameseLocal(num);
+  });
+}
+
+/**
+ * Smart normalizer: only run on Vietnamese text, only on final transcripts.
+ */
+export function smartNormalize(text, partial = false) {
+  if (!text || partial) return text;
+  if (containsVietnamese(text)) {
+    return normalizeVietnameseText(text);
+  }
+  return text;
+}
+
+/**
+ * Convert text for TTS: Vietnamese numbers to words for proper pronunciation.
+ * Call this before sending text to TTS synthesis.
+ */
+export function prepareForTTS(text) {
+  if (!text || typeof text !== "string") return text;
+  if (containsVietnamese(text)) {
+    return numbersToVietnameseWords(text);
+  }
+  return text;
+}