the serial letter normalization added to the code to normalize the Heh , Ya , Kaf and etc to the correct character

2026-07-27 10:38:35 +03:30
parent 778544c321
commit 182db56e15
13 changed files with 337 additions and 11 deletions
--- a/scripts/verify-plate-normalization.mjs
+++ b/scripts/verify-plate-normalization.mjs
@@ -0,0 +1,86 @@
+#!/usr/bin/env node
+/**
+ * Quick manual verification for plate normalization.
+ *
+ * Usage:  node scripts/verify-plate-normalization.mjs
+ *
+ * Demonstrates that:
+ *   339ه‍77  (ه + U+200D)  ->  339ه77  ->  letter ه -> code 15
+ *   339ي77   (Arabic ي)    ->  339ی77  ->  letter ی -> code 16
+ */
+
+// ---------- paste the pure function here so no build needed ----------
+function normalizePlateText(text) {
+  if (!text) return text;
+  let result = String(text);
+  result = result.normalize("NFKC");
+  result = result.replace(/[\u200C\u200D\u200E\u200F\uFEFF]/g, "");
+  result = result.replace(/\u064A/g, "\u06CC"); // Arabic ي → Persian ی
+  result = result.trim();
+  return result;
+}
+
+// ---------- the canonical plate-letter → code mapping ----------
+const PLATE_LETTER_MAP = {
+  الف: 1, ب: 2, پ: 3, ج: 4, د: 5,
+  س: 6, ص: 7, ط: 8, ع: 9, ق: 10,
+  ل: 11, م: 12, ن: 13, و: 14, ه: 15,
+  ی: 16, ک: 17, ژ: 18, ت: 19, ث: 20,
+  ز: 21, ش: 22, ف: 23, گ: 24,
+};
+
+function extractAndMap(plateString) {
+  const normalized = normalizePlateText(plateString);
+
+  // Simple parser: digits | letter | digits  (left+right digits, 1 letter)
+  const match = normalized.match(/^(\d+)([^\d]+)(\d+)$/);
+  if (!match) return { input: plateString, normalized, error: "no match" };
+
+  const [, leftDigits, letterRaw, rightDigits] = match;
+  const letter = normalizePlateText(letterRaw);
+  const code = PLATE_LETTER_MAP[letter];
+
+  return {
+    input: plateString,
+    normalized,
+    leftDigits,
+    letter,
+    rightDigits,
+    code,
+    ok: code !== undefined,
+  };
+}
+
+// ---------- test cases ----------
+const cases = [
+  { desc: "ه + ZWJ (the original bug)", plate: "339ه\u200D77" },
+  { desc: "Arabic ي → Persian ی",        plate: "339ي77" },
+  { desc: "ه + ZWNJ",                    plate: "339ه\u200C77" },
+  { desc: "RTL mark around letter",      plate: "339\u200Fه\u200E77" },
+  { desc: "BOM + ZWJ combined",          plate: "\uFEFF339ه\u200D77" },
+  { desc: "clean letter (no junk)",       plate: "339ه77" },
+  { desc: "full plate: الف",             plate: "11الف22" },
+  { desc: "full plate: ی",               plate: "55ی99" },
+];
+
+console.log("Plate Normalization Verification\n");
+console.log("=".repeat(72));
+
+let allPassed = true;
+for (const { desc, plate } of cases) {
+  const r = extractAndMap(plate);
+  const status = r.ok ? "PASS" : "FAIL";
+  if (!r.ok) allPassed = false;
+
+  console.log(`\n[${status}] ${desc}`);
+  console.log(`  Input:     ${JSON.stringify(plate)}`);
+  console.log(`  Normalized: ${JSON.stringify(r.normalized)}`);
+  if (r.letter) {
+    console.log(`  Letter:    ${r.letter}  (U+${r.letter.charCodeAt(0).toString(16).toUpperCase().padStart(4, "0")})`);
+  }
+  console.log(`  Code:      ${r.code ?? "undefined"}`);
+}
+
+console.log("\n" + "=".repeat(72));
+console.log(allPassed ? "\n  All tests PASSED ✓" : "\n  Some tests FAILED ✗");
+process.exit(allPassed ? 0 : 1);