fix BiDi to affect RandALcat labels only

Author: SorinGFS

#/tests/0/50.json

@@ -1,6 +1,6 @@
 {
-    "description": "KATAKANA MIDDLE DOT with Hiragana valid for RFC 5890-5892, but fails Bidi rule 1",
+    "description": "KATAKANA MIDDLE DOT with Hiragana",
     "comment": "https://tools.ietf.org/html/rfc5891#section-4.2.3.3 https://tools.ietf.org/html/rfc5892#appendix-A.7",
     "data": "・ぁ",
-    "valid": false
+    "valid": true
 }

#/tests/0/51.json

@@ -1,6 +1,6 @@
 {
-    "description": "KATAKANA MIDDLE DOT with Katakana valid for RFC 5890-5892, but fails Bidi rule 1",
+    "description": "KATAKANA MIDDLE DOT with Katakana",
     "comment": "https://tools.ietf.org/html/rfc5891#section-4.2.3.3 https://tools.ietf.org/html/rfc5892#appendix-A.7",
     "data": "・ァ",
-    "valid": false
+    "valid": true
 }

#/tests/0/52.json

@@ -1,6 +1,6 @@
 {
-    "description": "KATAKANA MIDDLE DOT with Han valid for RFC 5890-5892, but fails Bidi rule 1",
+    "description": "KATAKANA MIDDLE DOT with Han",
     "comment": "https://tools.ietf.org/html/rfc5891#section-4.2.3.3 https://tools.ietf.org/html/rfc5892#appendix-A.7",
     "data": "・丈",
-    "valid": false
+    "valid": true
 }

#/tests/0/55.json

@@ -1,6 +1,6 @@
 {
-    "description": "Extended Arabic-Indic digits not mixed with Arabic-Indic digits, but fail Bidi rule 1",
+    "description": "Extended Arabic-Indic digits mixed with ASCII digits",
     "comment": "https://tools.ietf.org/html/rfc5891#section-4.2.3.3 https://tools.ietf.org/html/rfc5892#appendix-A.9",
     "data": "۰0",
-    "valid": false
+    "valid": true
 }

#/tests/0/66.json

@@ -1,5 +1,5 @@
 {
-    "description": "Bidi Rule 1 — L-branch fail: contains L but doesn't start with L",
-    "data": "1g",
-    "valid": false
+    "description": "BiDi rules not affecting label that does not contain RTL chars",
+    "data": "1host",
+    "valid": true
 }

#/tests/0/76.json

@@ -1,5 +1,5 @@
 {
-    "description": "Bidi Rule 4 violation: AN only",
+    "description": "AN only in label not affected by BiDi rules (no R or AL char present)",
     "data": "١٢",
-    "valid": false
+    "valid": true
 }

#/tests/0/77.json

@@ -1,5 +1,5 @@
 {
-    "description": "Bidi Rule 4 — fail: mixed EN and AN",
+    "description": "mixed EN and AN not affected by BiDi rules (no R or AL char present)",
     "data": "1١",
-    "valid": false
+    "valid": true
 }

#/tests/0/78.json

@@ -1,5 +1,5 @@
 {
-    "description": "Bidi Rule 4 — fail: mixed AN and EN",
+    "description": "mixed AN and EN not affected by BiDi rules (no R or AL char present)",
     "data": "١2",
-    "valid": false
+    "valid": true
 }

#/tests/0/82.json

@@ -1,5 +1,5 @@
 {
-    "description": "Bidi Rule 5 — fail: starts L but contains AN (disallowed)",
+    "description": "mixed L with AN not affected by BiDi rules (no R or AL char present)",
     "data": "g١",
-    "valid": false
+    "valid": true
 }

index.js

@@ -60,13 +60,6 @@ function uts46map(label) {
 function cpHex(cp) {
     return `char '${String.fromCodePoint(cp)}' ` + JSON.stringify('(U+' + cp.toString(16).toUpperCase().padStart(4, '0') + ')');
 }
-// digits: Arabic-Indic sets distinction (1 and 2), ASCII→0
-function digitSet(cp) {
-    if (cp >= 0x30 && cp <= 0x39) return 0; // ASCII → EN (0)
-    if (cp >= 0x0660 && cp <= 0x0669) return 1; // Arabic-Indic
-    if (cp >= 0x06f0 && cp <= 0x06f9) return 2; // Extended Arabic-Indic
-    return 0;
-}
 // main validator
 function isIdnHostname(hostname) {
     // basic hostname checks
@@ -102,13 +95,14 @@ function isIdnHostname(hostname) {
         if (aceLabel.length > 63) throwIdnaLengthError('Final ASCII Compatible Encoding (ACE) label cannot exceed 63 bytes (RFC 5890 §2.3.2.1).');
         aceHostnameLength += aceLabel.length + 1;
         // hyphen rules (the other one is covered by bidi)
+        if (/^-|-$/.test(label)) throwIdnaSyntaxError('Label cannot begin or end with hyphen-minus (RFC 5891 §4.2.3.1).');
         if (label.indexOf('--') === 2) throwIdnaSyntaxError('Label cannot contain consecutive hyphen-minus in the 3rd and 4th positions (RFC 5891 §4.2.3.1).');
         // leading combining marks check (some are not covered by bidi)
         if (/^\p{M}$/u.test(String.fromCodePoint(label.codePointAt(0)))) throwIdnaSyntaxError(`Label cannot begin with combining/enclosing mark ${cpHex(label.codePointAt(0))} (RFC 5891 §4.2.3.2).`);
         // spread cps for context and bidi checks
         const cps = Array.from(label).map((char) => char.codePointAt(0));
         let joinTypes = '';
-        let digitSeen = 0;
+        let digits = '';
         let bidiClasses = [];
         // per-codepoint contextual checks
         for (let j = 0; j < cps.length; j++) {
@@ -143,14 +137,11 @@ function isIdnHostname(hostname) {
                 }
             }
             // check mixed digit sets
-            const ds = digitSet(cp);
-            if (ds) {
-                if (digitSeen && digitSeen !== ds) throwIdnaContextOError('Mixed digit sets in hostname not allowed (RFC 5892 Appendix A.8/A.9).');
-                digitSeen = ds;
-            }
+            if ((cp >= 0x0660 && cp <= 0x0669) || (cp >= 0x06f0 && cp <= 0x06f9)) digits += (cp < 0x06f0 ? 'a' : 'e' );
+            if (j === cps.length - 1 && /^(?=.*a)(?=.*e).*$/.test(digits)) throwIdnaContextOError('Arabic-Indic digits cannot be mixed with Extended Arabic-Indic digits (RFC 5892 Appendix A.8/A.9).');
             // validate bidi
             bidiClasses.push(getRange(bidi_ranges, cp));
-            if (j === cps.length - 1) {
+            if (j === cps.length - 1 && (bidiClasses.includes('R') || bidiClasses.includes('AL'))) {
                 if (bidiClasses[0] === 'R' || bidiClasses[0] === 'AL') {
                     for (let cls of bidiClasses) if (!['R', 'AL', 'AN', 'EN', 'ET', 'ES', 'CS', 'ON', 'BN', 'NSM'].includes(cls)) throwIdnaBidiError(`'${label}' breaks rule #2: Only R, AL, AN, EN, ET, ES, CS, ON, BN, NSM allowed in label (RFC 5893 §2.2)`);
                     if (!/(R|AL|EN|AN)(NSM)*$/.test(bidiClasses.join(''))) throwIdnaBidiError(`'${label}' breaks rule #3: label must end with R, AL, EN, or AN, followed by zero or more NSM (RFC 5893 §2.3)`);