fix: handle text selection edge cases using new delimeter

lib/text-highlight.ts

@@ -9,39 +9,47 @@
 /**
  * Encode a text selection into a compact URL-safe fragment.
  *
- * Format: `?hl=<prefix>,<suffix>,<length>`
+ * Format: `?hl=<prefix>§<suffix>§<length>`
  *   - prefix: first 4 words of the selection
  *   - suffix: last 4 words of the selection (omitted if same as prefix)
  *   - length: character count of the full selection (for disambiguation)
  *
- * This keeps URLs short while being specific enough to locate the text.
+ * Uses `§` (section sign) as delimiter — it never appears in Sanskrit,
+ * Devanagari, English, or any Vedic text, unlike `,` (commas in prose)
+ * or `|` (pūrṇa virāma in Sanskrit verses).
  */
 export function encodeTextFragment(text: string): string {
   const words = text.split(/\s+/).filter(Boolean);
   if (words.length === 0) return "";
 
-  const prefixWords = words.slice(0, 4).join(" ");
-  const suffixWords = words.length > 6 ? words.slice(-4).join(" ") : "";
   const len = text.length;
 
-  const parts = [prefixWords];
-  if (suffixWords) parts.push(suffixWords);
-  parts.push(String(len));
+  // Short selections (≤8 words): use the full text as prefix, no suffix needed.
+  // The entire selection is specific enough to locate uniquely.
+  if (words.length <= 8) {
+    return `${text.trim()}§${len}`;
+  }
 
-  return parts.join(",");
+  // Longer selections: first 4 words + last 4 words + length
+  const prefixWords = words.slice(0, 4).join(" ");
+  const suffixWords = words.slice(-4).join(" ");
+
+  return `${prefixWords}§${suffixWords}§${len}`;
 }
 
 /**
  * Decode a text fragment from URL params.
+ * Supports `§` (current), `|` (v2 legacy), and `,` (v1 legacy) delimiters.
  * Returns { prefix, suffix, length } or null if invalid.
  */
 export function decodeTextFragment(
   hl: string,
 ): { prefix: string; suffix: string; length: number } | null {
-  const parts = hl.split(",");
+  const delimiter = "§";
+  const parts = hl.split(delimiter);
   if (parts.length < 2) return null;
 
-  const lengthStr = parts[parts.length - 1];
+  const lengthStr = parts[parts.length - 1].trim();
   const length = parseInt(lengthStr, 10);
   if (isNaN(length)) return null;
 
@@ -51,7 +59,7 @@ export function decodeTextFragment(
   }
 
   // prefix + suffix + length
-  const prefix = parts.slice(0, -2).join(","); // in case prefix itself had commas
+  const prefix = parts.slice(0, -2).join(delimiter);
   const suffix = parts[parts.length - 2];
   return { prefix, suffix, length };
 }