idna: implement support for reporting errors on invalid IDNA2008 characters

idna/src/make_uts46_mapping_table.py

@@ -78,6 +78,12 @@ for line in txt:
             unicode_str = u''.join(char(c) for c in fields[2].strip().split(' '))
         elif mapping == "Deviation":
             unicode_str = u''
+
+    if len(fields) > 3:
+        assert fields[3].strip() in ('NV8', 'XV8'), fields[3]
+        assert mapping == 'Valid', mapping
+        mapping = 'DisallowedIdna2008'
+
     ranges.append((first, last, mapping, unicode_str))
 
 def mergeable_key(r):
@@ -86,7 +92,7 @@ def mergeable_key(r):
     # These types have associated data, so we should not merge them.
     if mapping in ('Mapped', 'Deviation', 'DisallowedStd3Mapped'):
         return r
-    assert mapping in ('Valid', 'Ignored', 'Disallowed', 'DisallowedStd3Valid')
+    assert mapping in ('Valid', 'Ignored', 'Disallowed', 'DisallowedStd3Valid', 'DisallowedIdna2008')
     return mapping
 
 grouped_ranges = itertools.groupby(ranges, key=mergeable_key)

idna/src/uts46.rs

@@ -48,6 +48,7 @@ enum Mapping {
     Disallowed,
     DisallowedStd3Valid,
     DisallowedStd3Mapped(StringTableSlice),
+    DisallowedIdna2008,
 }
 
 struct Range {
@@ -140,6 +141,12 @@ impl<'a> Iterator for Mapper<'a> {
                     self.slice = Some(decode_slice(slice).chars());
                     continue;
                 }
+                Mapping::DisallowedIdna2008 => {
+                    if self.config.use_idna_2008_rules {
+                        self.errors.disallowed_in_idna_2008 = true;
+                    }
+                    codepoint
+                }
             });
         }
     }
@@ -310,7 +317,7 @@ fn check_validity(label: &str, config: Config, errors: &mut Errors) {
 
     // V6: Check against Mapping Table
     if label.chars().any(|c| match *find_char(c) {
-        Mapping::Valid => false,
+        Mapping::Valid | Mapping::DisallowedIdna2008 => false,
         Mapping::Deviation(_) => config.transitional_processing,
         Mapping::DisallowedStd3Valid => config.use_std3_ascii_rules,
         _ => true,
@@ -510,6 +517,7 @@ pub struct Config {
     transitional_processing: bool,
     verify_dns_length: bool,
     check_hyphens: bool,
+    use_idna_2008_rules: bool,
 }
 
 /// The defaults are that of https://url.spec.whatwg.org/#idna
@@ -524,6 +532,7 @@ impl Default for Config {
 
             // Only use for to_ascii, not to_unicode
             verify_dns_length: false,
+            use_idna_2008_rules: false,
         }
     }
 }
@@ -553,6 +562,12 @@ impl Config {
         self
     }
 
+    #[inline]
+    pub fn use_idna_2008_rules(mut self, value: bool) -> Self {
+        self.use_idna_2008_rules = value;
+        self
+    }
+
     /// http://www.unicode.org/reports/tr46/#ToASCII
     pub fn to_ascii(self, domain: &str) -> Result<String, Errors> {
         let mut result = String::new();
@@ -599,6 +614,7 @@ pub struct Errors {
     disallowed_character: bool,
     too_long_for_dns: bool,
     too_short_for_dns: bool,
+    disallowed_in_idna_2008: bool,
 }
 
 impl Errors {
@@ -615,6 +631,7 @@ impl Errors {
             disallowed_character,
             too_long_for_dns,
             too_short_for_dns,
+            disallowed_in_idna_2008,
         } = *self;
         punycode
             || check_hyphens
@@ -627,6 +644,7 @@ impl Errors {
             || disallowed_character
             || too_long_for_dns
             || too_short_for_dns
+            || disallowed_in_idna_2008
     }
 }
 
@@ -644,6 +662,7 @@ impl fmt::Debug for Errors {
             disallowed_character,
             too_long_for_dns,
             too_short_for_dns,
+            disallowed_in_idna_2008,
         } = *self;
 
         let fields = [
@@ -661,6 +680,7 @@ impl fmt::Debug for Errors {
             ("disallowed_character", disallowed_character),
             ("too_long_for_dns", too_long_for_dns),
             ("too_short_for_dns", too_short_for_dns),
+            ("disallowed_in_idna_2008", disallowed_in_idna_2008),
         ];
 
         let mut empty = true;

idna/tests/unit.rs

@@ -114,3 +114,20 @@ fn test_v8_bidi_rules() {
     // Bidi chars may be punycode-encoded
     assert!(config.to_ascii("xn--0ca24w").is_err());
 }
+
+#[test]
+fn emoji_domains() {
+    // HOT BEVERAGE is allowed here...
+    let config = idna::Config::default()
+        .verify_dns_length(true)
+        .use_std3_ascii_rules(true);
+    assert_eq!(config.to_ascii("☕.com").unwrap(), "xn--53h.com");
+
+    // ... but not here
+    let config = idna::Config::default()
+        .verify_dns_length(true)
+        .use_std3_ascii_rules(true)
+        .use_idna_2008_rules(true);
+    let error = format!("{:?}", config.to_ascii("☕.com").unwrap_err());
+    assert!(error.contains("disallowed_in_idna_2008"));
+}