ID_Compat_Math_Start and ID_Compat_Math_Continue

* Adds two new functions to check if a character is part of the Mathematical Compatibility Notation Profile as defined by Unicode. * Small cleanup of tests and improved error message for failed tests
unicode-rs · Danvil · Jul 18, 2025 · Jul 18, 2025 · d7f2543d53b6f31d00b109c2a86eb34faa84819f
commit d7f2543d53b6f31d00b109c2a86eb34faa84819f
diff --git a/scripts/unicode.py b/scripts/unicode.py
@@ -158,14 +158,12 @@ def emit_table(f, name, t_data, t_type = "&[(char, char)]", is_pub=True,
     format_table_content(f, data, 8)
     f.write("\n    ];\n\n")
 
-def emit_property_module(f, mod, tbl, emit):
-    f.write("pub mod %s {\n" % mod)
+def emit_property_module_contents(f, tbl, emit):
     for cat in sorted(emit):
         emit_table(f, "%s_table" % cat, tbl[cat])
         f.write("    pub fn %s(c: char) -> bool {\n" % cat)
         f.write("        super::bsearch_range_table(c, %s_table)\n" % cat)
         f.write("    }\n\n")
-    f.write("}\n\n")
 
 if __name__ == "__main__":
     r = "tables.rs"
@@ -187,6 +185,16 @@ def emit_property_module(f, mod, tbl, emit):
 """ % unicode_version)
         emit_bsearch_range_table(rf)
 
+        rf.write("pub mod derived_property {\n")
+
+        # XID_Start and XID_Continue
         want_derived = ["XID_Start", "XID_Continue"]
         derived = load_properties("DerivedCoreProperties.txt", want_derived)
-        emit_property_module(rf, "derived_property", derived, want_derived)
+        emit_property_module_contents(rf, derived, want_derived)
+
+        # ID_Compat_Math_Start and ID_Compat_Math_Continue
+        want_derived = ["ID_Compat_Math_Start", "ID_Compat_Math_Continue"]
+        derived = load_properties("PropList.txt", want_derived)
+        emit_property_module_contents(rf, derived, want_derived)
+
+        rf.write("}\n\n")
diff --git a/src/lib.rs b/src/lib.rs
@@ -67,6 +67,22 @@ pub trait UnicodeXID {
     /// [UAX #31](http://unicode.org/reports/tr31/#NFKC_Modifications),
     /// mostly similar to 'ID_Continue' but modified for closure under NFKx.
     fn is_xid_continue(self) -> bool;
+
+    /// Returns whether the specified `char` satisfies the
+    /// `ID_Compat_Math_Start` Unicode property.
+    ///
+    /// `ID_Compat_Math_Start` is a Unicode Property specified in
+    /// [UAX #31](http://unicode.org/reports/tr31/#NFKC_Modifications)
+    /// as part of the Mathematical Compatibility Notation Profile.
+    fn is_id_compat_math_start(self) -> bool;
+
+    /// Returns whether the specified `char` satisfies the
+    /// `ID_Compat_Math_Continue` Unicode property.
+    ///
+    /// `ID_Compat_Math_Continue` is a Unicode Property specified in
+    /// [UAX #31](http://unicode.org/reports/tr31/#NFKC_Modifications)
+    /// as part of the Mathematical Compatibility Notation Profile.
+    fn is_id_compat_math_continue(self) -> bool;
 }
 
 impl UnicodeXID for char {
@@ -87,4 +103,14 @@ impl UnicodeXID for char {
             || self == '_'
             || (self > '\x7f' && derived_property::XID_Continue(self))
     }
+
+    #[inline]
+    fn is_id_compat_math_start(self) -> bool {
+        derived_property::ID_Compat_Math_Start(self)
+    }
+
+    #[inline]
+    fn is_id_compat_math_continue(self) -> bool {
+        derived_property::ID_Compat_Math_Continue(self)
+    }
 }
diff --git a/src/tables.rs b/src/tables.rs
@@ -1534,4 +1534,49 @@ pub mod derived_property {
     pub fn XID_Start(c: char) -> bool {
         super::bsearch_range_table(c, XID_Start_table)
     }
+
+    static ID_Compat_Math_Continue_table: &[(char, char)] = &[
+        ('\u{b2}', '\u{b3}'),
+        ('\u{b9}', '\u{b9}'),
+        ('\u{2070}', '\u{2070}'),
+        ('\u{2074}', '\u{207e}'),
+        ('\u{2080}', '\u{208e}'),
+        ('\u{2202}', '\u{2202}'),
+        ('\u{2207}', '\u{2207}'),
+        ('\u{221e}', '\u{221e}'),
+        ('\u{1d6c1}', '\u{1d6c1}'),
+        ('\u{1d6db}', '\u{1d6db}'),
+        ('\u{1d6fb}', '\u{1d6fb}'),
+        ('\u{1d715}', '\u{1d715}'),
+        ('\u{1d735}', '\u{1d735}'),
+        ('\u{1d74f}', '\u{1d74f}'),
+        ('\u{1d76f}', '\u{1d76f}'),
+        ('\u{1d789}', '\u{1d789}'),
+        ('\u{1d7a9}', '\u{1d7a9}'),
+        ('\u{1d7c3}', '\u{1d7c3}'),
+    ];
+
+    pub fn ID_Compat_Math_Continue(c: char) -> bool {
+        super::bsearch_range_table(c, ID_Compat_Math_Continue_table)
+    }
+
+    static ID_Compat_Math_Start_table: &[(char, char)] = &[
+        ('\u{2202}', '\u{2202}'),
+        ('\u{2207}', '\u{2207}'),
+        ('\u{221e}', '\u{221e}'),
+        ('\u{1d6c1}', '\u{1d6c1}'),
+        ('\u{1d6db}', '\u{1d6db}'),
+        ('\u{1d6fb}', '\u{1d6fb}'),
+        ('\u{1d715}', '\u{1d715}'),
+        ('\u{1d735}', '\u{1d735}'),
+        ('\u{1d74f}', '\u{1d74f}'),
+        ('\u{1d76f}', '\u{1d76f}'),
+        ('\u{1d789}', '\u{1d789}'),
+        ('\u{1d7a9}', '\u{1d7a9}'),
+        ('\u{1d7c3}', '\u{1d7c3}'),
+    ];
+
+    pub fn ID_Compat_Math_Start(c: char) -> bool {
+        super::bsearch_range_table(c, ID_Compat_Math_Start_table)
+    }
 }
diff --git a/src/tests.rs b/src/tests.rs
@@ -15,7 +15,6 @@ use std::prelude::v1::*;
 #[cfg(feature = "bench")]
 use test::Bencher;
 
-#[cfg(feature = "bench")]
 use UnicodeXID;
 
 #[cfg(feature = "bench")]
@@ -24,7 +23,7 @@ fn cargo_is_xid_start(b: &mut Bencher) {
     let string = iter::repeat('a').take(4096).collect::<String>();
 
     b.bytes = string.len() as u64;
-    b.iter(|| string.chars().all(super::UnicodeXID::is_xid_start));
+    b.iter(|| string.chars().all(UnicodeXID::is_xid_start));
 }
 
 #[cfg(feature = "bench")]
@@ -42,7 +41,7 @@ fn cargo_xid_continue(b: &mut Bencher) {
     let string = iter::repeat('a').take(4096).collect::<String>();
 
     b.bytes = string.len() as u64;
-    b.iter(|| string.chars().all(super::UnicodeXID::is_xid_continue));
+    b.iter(|| string.chars().all(UnicodeXID::is_xid_continue));
 }
 
 #[cfg(feature = "bench")]
@@ -54,12 +53,20 @@ fn stdlib_xid_continue(b: &mut Bencher) {
     b.iter(|| string.chars().all(char::is_xid_continue));
 }
 
+fn assert_ch_pretty(ok: bool, ch: char) {
+    assert!(
+        ok,
+        "test failed for unicode character <U+{:X}>: {}",
+        ch as u32, ch
+    );
+}
+
 #[test]
 fn test_is_xid_start() {
     let chars = ['A', 'Z', 'a', 'z', '\u{1000d}', '\u{10026}'];
 
-    for ch in &chars {
-        assert!(super::UnicodeXID::is_xid_start(*ch), "{}", ch);
+    for &ch in &chars {
+        assert_ch_pretty(UnicodeXID::is_xid_start(ch), ch);
     }
 }
 
@@ -69,17 +76,17 @@ fn test_is_not_xid_start() {
         '\x00', '\x01', '0', '9', ' ', '[', '<', '{', '(', '\u{02c2}', '\u{ffff}',
     ];
 
-    for ch in &chars {
-        assert!(!super::UnicodeXID::is_xid_start(*ch), "{}", ch);
+    for &ch in &chars {
+        assert_ch_pretty(!UnicodeXID::is_xid_start(ch), ch);
     }
 }
 
 #[test]
 fn test_is_xid_continue() {
     let chars = ['0', '9', 'A', 'Z', 'a', 'z', '_', '\u{1000d}', '\u{10026}'];
 
-    for ch in &chars {
-        assert!(super::UnicodeXID::is_xid_continue(*ch), "{}", ch);
+    for &ch in &chars {
+        assert_ch_pretty(UnicodeXID::is_xid_continue(ch), ch);
     }
 }
 
@@ -90,6 +97,33 @@ fn test_is_not_xid_continue() {
     ];
 
     for &ch in &chars {
-        assert!(!super::UnicodeXID::is_xid_continue(ch), "{}", ch);
+        assert_ch_pretty(!UnicodeXID::is_xid_continue(ch), ch);
+    }
+}
+
+#[test]
+fn test_is_id_compat_math_start() {
+    let chars = ['∇', '∂', '∞'];
+
+    for &ch in &chars {
+        assert_ch_pretty(UnicodeXID::is_id_compat_math_start(ch), ch);
+    }
+}
+
+#[test]
+fn test_is_not_id_compat_math_start() {
+    let chars = ['²', '³', '₂', '₃'];
+
+    for &ch in &chars {
+        assert_ch_pretty(!UnicodeXID::is_xid_start(ch), ch);
+    }
+}
+
+#[test]
+fn test_is_id_compat_math_continue() {
+    let chars = ['∇', '∂', '∞', '²', '³', '₂', '₃'];
+
+    for &ch in &chars {
+        assert_ch_pretty(UnicodeXID::is_id_compat_math_continue(ch), ch);
     }
 }