add alternate label separator tests

Author: SorinGFS

#/tests/0/87.json

@@ -0,0 +1,5 @@
+{
+    "description": "single dot",
+    "data": ".",
+    "valid": false
+}

#/tests/0/88.json

@@ -0,0 +1,5 @@
+{
+    "description": "single ideographic full stop",
+    "data": "。",
+    "valid": false
+}

#/tests/0/89.json

@@ -0,0 +1,5 @@
+{
+    "description": "single fullwidth full stop",
+    "data": "．",
+    "valid": false
+}

#/tests/0/90.json

@@ -0,0 +1,5 @@
+{
+    "description": "single halfwidth ideographic full stop",
+    "data": "｡",
+    "valid": false
+}

#/tests/0/91.json

@@ -0,0 +1,5 @@
+{
+    "description": "dot as label separator",
+    "data": "a.b",
+    "valid": true
+}

#/tests/0/92.json

@@ -0,0 +1,5 @@
+{
+    "description": "ideographic full stop as label separator",
+    "data": "a。b",
+    "valid": true
+}

#/tests/0/93.json

@@ -0,0 +1,5 @@
+{
+    "description": "fullwidth full stop as label separator",
+    "data": "a．b",
+    "valid": true
+}

#/tests/0/94.json

@@ -0,0 +1,5 @@
+{
+    "description": "halfwidth ideographic full stop as label separator",
+    "data": "a｡b",
+    "valid": true
+}

index.js

@@ -142,6 +142,7 @@ function isIdnHostname(hostname) {
             // validate bidi
             bidiClasses.push(getRange(bidi_ranges, cp));
             if (j === cps.length - 1 && (bidiClasses.includes('R') || bidiClasses.includes('AL'))) {
+                // order of chars in label (RFC 5890 §2.3.3)
                 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)`);

readme.md

@@ -105,7 +105,7 @@ Each release will have its `major` and `minor` version identical with the relate
 6. **Mapping phase (at validation time)**:
 
     - For each input label the validator:
-        1. Splits the hostname into labels (by `.`). Empty labels are rejected.
+        1. Splits the hostname into labels (by `.` or alternate label separators). Empty labels are rejected.
         2. For each label, maps codepoints according to `mappings` (`valid` and `deviation` are passed as they are, `mapped` are replaced with the corresponding codepoints, `ignored` are ignored, any other chars are triggering `IdnaUnicodeError`).
         3. Normalizes the resulting mapped label with NFC.
         4. Checks length limits (label ≤ 63, full name ≤ 253 octets after ASCII punycode conversion).
@@ -155,6 +155,7 @@ Each release will have its `major` and `minor` version identical with the relate
 
 ## What does _not_ do
 
+-   This validator does not support `context` or `locale` specific [Special Casing](https://www.unicode.org/Public/16.0.0/ucd/SpecialCasing.txt) mappings. For such needs some sort of `mapping` must be done before using this validator.
 -   This validator does not support `UTS#46 useTransitional` backward compatibility flag.
 -   This validator does not support `UTS#46 STD3 ASCII rules`, when required they can be enforced on separate layer.
 -   This validator does not attempt to resolve or query DNS — it only validates label syntax/mapping/contextual/bidi rules.