Auto merge of #26039 - SimonSapin:case-mapping, r=alexcrichton

* Add “complex” mappings to `char::to_lowercase` and `char::to_uppercase`, making them yield sometimes more than on `char`: #25800. `str::to_lowercase` and `str::to_uppercase` are affected as well.
* Add `char::to_titlecase`, since it’s the same algorithm (just different data). However this does **not** add `str::to_titlecase`, as that would require UAX#29 Unicode Text Segmentation which we decided not to include in of `std`: rust-lang/rfcs#1054 I made `char::to_titlecase` immediately `#[stable]`, since it’s so similar to `char::to_uppercase` that’s already stable. Let me know if it should be `#[unstable]` for a while.
* Add a special case for upper-case Sigma in word-final position in `str::to_lowercase`: #26035. This is the only language-independent conditional mapping currently in `SpecialCasing.txt`.
* Stabilize `str::to_lowercase` and `str::to_uppercase`. The `&self -> String` on `str` signature seems straightforward enough, and the only relevant issue I’ve found is #24536 about naming. But `char` already has stable methods with the same name, and deprecating them for a rename doesn’t seem worth it.

r? @alexcrichton

Author: bors

src/etc/unicode.py

@@ -72,8 +72,9 @@ def is_surrogate(n):
 def load_unicode_data(f):
     fetch(f)
     gencats = {}
-    upperlower = {}
-    lowerupper = {}
+    to_lower = {}
+    to_upper = {}
+    to_title = {}
     combines = {}
     canon_decomp = {}
     compat_decomp = {}
@@ -103,12 +104,16 @@ def load_unicode_data(f):
 
         # generate char to char direct common and simple conversions
         # uppercase to lowercase
-        if gencat == "Lu" and lowcase != "" and code_org != lowcase:
-            upperlower[code] = int(lowcase, 16)
+        if lowcase != "" and code_org != lowcase:
+            to_lower[code] = (int(lowcase, 16), 0, 0)
 
         # lowercase to uppercase
-        if gencat == "Ll" and upcase != "" and code_org != upcase:
-            lowerupper[code] = int(upcase, 16)
+        if upcase != "" and code_org != upcase:
+            to_upper[code] = (int(upcase, 16), 0, 0)
+
+        # title case
+        if titlecase.strip() != "" and code_org != titlecase:
+            to_title[code] = (int(titlecase, 16), 0, 0)
 
         # store decomposition, if given
         if decomp != "":
@@ -144,7 +149,32 @@ def load_unicode_data(f):
     gencats = group_cats(gencats)
     combines = to_combines(group_cats(combines))
 
-    return (canon_decomp, compat_decomp, gencats, combines, lowerupper, upperlower)
+    return (canon_decomp, compat_decomp, gencats, combines, to_upper, to_lower, to_title)
+
+def load_special_casing(f, to_upper, to_lower, to_title):
+    fetch(f)
+    for line in fileinput.input(f):
+        data = line.split('#')[0].split(';')
+        if len(data) == 5:
+            code, lower, title, upper, _comment = data
+        elif len(data) == 6:
+            code, lower, title, upper, condition, _comment = data
+            if condition.strip():  # Only keep unconditional mappins
+                continue
+        else:
+            continue
+        code = code.strip()
+        lower = lower.strip()
+        title = title.strip()
+        upper = upper.strip()
+        key = int(code, 16)
+        for (map_, values) in [(to_lower, lower), (to_upper, upper), (to_title, title)]:
+            if values != code:
+                values = [int(i, 16) for i in values.split()]
+                for _ in range(len(values), 3):
+                    values.append(0)
+                assert len(values) == 3
+                map_[key] = values
 
 def group_cats(cats):
     cats_out = {}
@@ -279,7 +309,7 @@ def load_east_asian_width(want_widths, except_cats):
     return widths
 
 def escape_char(c):
-    return "'\\u{%x}'" % c
+    return "'\\u{%x}'" % c if c != 0 else "'\\0'"
 
 def emit_bsearch_range_table(f):
     f.write("""
@@ -319,7 +349,7 @@ def emit_property_module(f, mod, tbl, emit):
         f.write("    }\n\n")
     f.write("}\n\n")
 
-def emit_conversions_module(f, lowerupper, upperlower):
+def emit_conversions_module(f, to_upper, to_lower, to_title):
     f.write("pub mod conversions {")
     f.write("""
     use core::cmp::Ordering::{Equal, Less, Greater};
@@ -328,21 +358,28 @@ def emit_conversions_module(f, lowerupper, upperlower):
     use core::option::Option::{Some, None};
     use core::result::Result::{Ok, Err};
 
-    pub fn to_lower(c: char) -> char {
-        match bsearch_case_table(c, LuLl_table) {
-          None        => c,
-          Some(index) => LuLl_table[index].1
+    pub fn to_lower(c: char) -> [char; 3] {
+        match bsearch_case_table(c, to_lowercase_table) {
+          None        => [c, '\\0', '\\0'],
+          Some(index) => to_lowercase_table[index].1
+        }
+    }
+
+    pub fn to_upper(c: char) -> [char; 3] {
+        match bsearch_case_table(c, to_uppercase_table) {
+            None        => [c, '\\0', '\\0'],
+            Some(index) => to_uppercase_table[index].1
         }
     }
 
-    pub fn to_upper(c: char) -> char {
-        match bsearch_case_table(c, LlLu_table) {
-            None        => c,
-            Some(index) => LlLu_table[index].1
+    pub fn to_title(c: char) -> [char; 3] {
+        match bsearch_case_table(c, to_titlecase_table) {
+            None        => [c, '\\0', '\\0'],
+            Some(index) => to_titlecase_table[index].1
         }
     }
 
-    fn bsearch_case_table(c: char, table: &'static [(char, char)]) -> Option<usize> {
+    fn bsearch_case_table(c: char, table: &'static [(char, [char; 3])]) -> Option<usize> {
         match table.binary_search_by(|&(key, _)| {
             if c == key { Equal }
             else if key < c { Less }
@@ -354,10 +391,18 @@ def emit_conversions_module(f, lowerupper, upperlower):
     }
 
 """)
-    emit_table(f, "LuLl_table",
-        sorted(upperlower.iteritems(), key=operator.itemgetter(0)), is_pub=False)
-    emit_table(f, "LlLu_table",
-        sorted(lowerupper.iteritems(), key=operator.itemgetter(0)), is_pub=False)
+    t_type = "&'static [(char, [char; 3])]"
+    pfun = lambda x: "(%s,[%s,%s,%s])" % (
+        escape_char(x[0]), escape_char(x[1][0]), escape_char(x[1][1]), escape_char(x[1][2]))
+    emit_table(f, "to_lowercase_table",
+        sorted(to_lower.iteritems(), key=operator.itemgetter(0)),
+        is_pub=False, t_type = t_type, pfun=pfun)
+    emit_table(f, "to_uppercase_table",
+        sorted(to_upper.iteritems(), key=operator.itemgetter(0)),
+        is_pub=False, t_type = t_type, pfun=pfun)
+    emit_table(f, "to_titlecase_table",
+        sorted(to_title.iteritems(), key=operator.itemgetter(0)),
+        is_pub=False, t_type = t_type, pfun=pfun)
     f.write("}\n\n")
 
 def emit_grapheme_module(f, grapheme_table, grapheme_cats):
@@ -591,8 +636,10 @@ def optimize_width_table(wtable):
 pub const UNICODE_VERSION: (u64, u64, u64) = (%s, %s, %s);
 """ % unicode_version)
         (canon_decomp, compat_decomp, gencats, combines,
-                lowerupper, upperlower) = load_unicode_data("UnicodeData.txt")
-        want_derived = ["XID_Start", "XID_Continue", "Alphabetic", "Lowercase", "Uppercase"]
+                to_upper, to_lower, to_title) = load_unicode_data("UnicodeData.txt")
+        load_special_casing("SpecialCasing.txt", to_upper, to_lower, to_title)
+        want_derived = ["XID_Start", "XID_Continue", "Alphabetic", "Lowercase", "Uppercase",
+                        "Cased", "Case_Ignorable"]
         derived = load_properties("DerivedCoreProperties.txt", want_derived)
         scripts = load_properties("Scripts.txt", [])
         props = load_properties("PropList.txt",
@@ -611,7 +658,7 @@ def optimize_width_table(wtable):
 
         # normalizations and conversions module
         emit_norm_module(rf, canon_decomp, compat_decomp, combines, norm_props)
-        emit_conversions_module(rf, lowerupper, upperlower)
+        emit_conversions_module(rf, to_upper, to_lower, to_title)
 
         ### character width module
         width_table = []

src/libcollections/str.rs

@@ -1816,11 +1816,40 @@ impl str {
     /// let s = "HELLO";
     /// assert_eq!(s.to_lowercase(), "hello");
     /// ```
-    #[unstable(feature = "collections")]
+    #[stable(feature = "unicode_case_mapping", since = "1.2.0")]
     pub fn to_lowercase(&self) -> String {
         let mut s = String::with_capacity(self.len());
-        s.extend(self[..].chars().flat_map(|c| c.to_lowercase()));
+        for (i, c) in self[..].char_indices() {
+            if c == 'Σ' {
+                // Σ maps to σ, except at the end of a word where it maps to ς.
+                // This is the only conditional (contextual) but language-independent mapping
+                // in `SpecialCasing.txt`,
+                // so hard-code it rather than have a generic "condition" mechanim.
+                // See https://github.com/rust-lang/rust/issues/26035
+                map_uppercase_sigma(self, i, &mut s)
+            } else {
+                s.extend(c.to_lowercase());
+            }
+        }
         return s;
+
+        fn map_uppercase_sigma(from: &str, i: usize, to: &mut String) {
+            // See http://www.unicode.org/versions/Unicode7.0.0/ch03.pdf#G33992
+            // for the definition of `Final_Sigma`.
+            debug_assert!('Σ'.len_utf8() == 2);
+            let is_word_final =
+                case_ignoreable_then_cased(from[..i].chars().rev()) &&
+                !case_ignoreable_then_cased(from[i + 2..].chars());
+            to.push_str(if is_word_final { "ς" } else { "σ" });
+        }
+
+        fn case_ignoreable_then_cased<I: Iterator<Item=char>>(iter: I) -> bool {
+            use rustc_unicode::derived_property::{Cased, Case_Ignorable};
+            match iter.skip_while(|&c| Case_Ignorable(c)).next() {
+                Some(c) => Cased(c),
+                None => false,
+            }
+        }
     }
 
     /// Returns the uppercase equivalent of this string.
@@ -1833,7 +1862,7 @@ impl str {
     /// let s = "hello";
     /// assert_eq!(s.to_uppercase(), "HELLO");
     /// ```
-    #[unstable(feature = "collections")]
+    #[stable(feature = "unicode_case_mapping", since = "1.2.0")]
     pub fn to_uppercase(&self) -> String {
         let mut s = String::with_capacity(self.len());
         s.extend(self[..].chars().flat_map(|c| c.to_uppercase()));

src/libcollectionstest/str.rs

@@ -1687,6 +1687,45 @@ fn trim_ws() {
                "");
 }
 
+#[test]
+fn to_lowercase() {
+    assert_eq!("".to_lowercase(), "");
+    assert_eq!("AÉDžaé ".to_lowercase(), "aédžaé ");
+
+    // https://github.com/rust-lang/rust/issues/26035
+    assert_eq!("ΑΣ".to_lowercase(), "ας");
+    assert_eq!("Α'Σ".to_lowercase(), "α'ς");
+    assert_eq!("Α''Σ".to_lowercase(), "α''ς");
+
+    assert_eq!("ΑΣ Α".to_lowercase(), "ας α");
+    assert_eq!("Α'Σ Α".to_lowercase(), "α'ς α");
+    assert_eq!("Α''Σ Α".to_lowercase(), "α''ς α");
+
+    assert_eq!("ΑΣ' Α".to_lowercase(), "ας' α");
+    assert_eq!("ΑΣ'' Α".to_lowercase(), "ας'' α");
+
+    assert_eq!("Α'Σ' Α".to_lowercase(), "α'ς' α");
+    assert_eq!("Α''Σ'' Α".to_lowercase(), "α''ς'' α");
+
+    assert_eq!("Α Σ".to_lowercase(), "α σ");
+    assert_eq!("Α 'Σ".to_lowercase(), "α 'σ");
+    assert_eq!("Α ''Σ".to_lowercase(), "α ''σ");
+
+    assert_eq!("Σ".to_lowercase(), "σ");
+    assert_eq!("'Σ".to_lowercase(), "'σ");
+    assert_eq!("''Σ".to_lowercase(), "''σ");
+
+    assert_eq!("ΑΣΑ".to_lowercase(), "ασα");
+    assert_eq!("ΑΣ'Α".to_lowercase(), "ασ'α");
+    assert_eq!("ΑΣ''Α".to_lowercase(), "ασ''α");
+}
+
+#[test]
+fn to_uppercase() {
+    assert_eq!("".to_uppercase(), "");
+    assert_eq!("aéDžßfiᾀ".to_uppercase(), "AÉDŽSSFIἈΙ");
+}
+
 mod pattern {
     use std::str::pattern::Pattern;
     use std::str::pattern::{Searcher, ReverseSearcher};

src/libcoretest/char.rs

@@ -58,6 +58,8 @@ fn test_to_lowercase() {
     fn lower(c: char) -> char {
         let mut it = c.to_lowercase();
         let c = it.next().unwrap();
+        // As of Unicode version 7.0.0, `SpecialCasing.txt` has no lower-case mapping
+        // to multiple code points.
         assert!(it.next().is_none());
         c
     }
@@ -73,29 +75,54 @@ fn test_to_lowercase() {
     assert_eq!(lower('Μ'), 'μ');
     assert_eq!(lower('Α'), 'α');
     assert_eq!(lower('Σ'), 'σ');
+    assert_eq!(lower('Dž'), 'dž');
+    assert_eq!(lower('fi'), 'fi');
 }
 
 #[test]
 fn test_to_uppercase() {
-    fn upper(c: char) -> char {
-        let mut it = c.to_uppercase();
-        let c = it.next().unwrap();
-        assert!(it.next().is_none());
-        c
+    fn upper(c: char) -> Vec<char> {
+        c.to_uppercase().collect()
     }
-    assert_eq!(upper('a'), 'A');
-    assert_eq!(upper('ö'), 'Ö');
-    assert_eq!(upper('ß'), 'ß'); // not ẞ: Latin capital letter sharp s
-    assert_eq!(upper('ü'), 'Ü');
-    assert_eq!(upper('💩'), '💩');
-
-    assert_eq!(upper('σ'), 'Σ');
-    assert_eq!(upper('τ'), 'Τ');
-    assert_eq!(upper('ι'), 'Ι');
-    assert_eq!(upper('γ'), 'Γ');
-    assert_eq!(upper('μ'), 'Μ');
-    assert_eq!(upper('α'), 'Α');
-    assert_eq!(upper('ς'), 'Σ');
+    assert_eq!(upper('a'), ['A']);
+    assert_eq!(upper('ö'), ['Ö']);
+    assert_eq!(upper('ß'), ['S', 'S']); // not ẞ: Latin capital letter sharp s
+    assert_eq!(upper('ü'), ['Ü']);
+    assert_eq!(upper('💩'), ['💩']);
+
+    assert_eq!(upper('σ'), ['Σ']);
+    assert_eq!(upper('τ'), ['Τ']);
+    assert_eq!(upper('ι'), ['Ι']);
+    assert_eq!(upper('γ'), ['Γ']);
+    assert_eq!(upper('μ'), ['Μ']);
+    assert_eq!(upper('α'), ['Α']);
+    assert_eq!(upper('ς'), ['Σ']);
+    assert_eq!(upper('Dž'), ['DŽ']);
+    assert_eq!(upper('fi'), ['F', 'I']);
+    assert_eq!(upper('ᾀ'), ['Ἀ', 'Ι']);
+}
+
+#[test]
+fn test_to_titlecase() {
+    fn title(c: char) -> Vec<char> {
+        c.to_titlecase().collect()
+    }
+    assert_eq!(title('a'), ['A']);
+    assert_eq!(title('ö'), ['Ö']);
+    assert_eq!(title('ß'), ['S', 's']); // not ẞ: Latin capital letter sharp s
+    assert_eq!(title('ü'), ['Ü']);
+    assert_eq!(title('💩'), ['💩']);
+
+    assert_eq!(title('σ'), ['Σ']);
+    assert_eq!(title('τ'), ['Τ']);
+    assert_eq!(title('ι'), ['Ι']);
+    assert_eq!(title('γ'), ['Γ']);
+    assert_eq!(title('μ'), ['Μ']);
+    assert_eq!(title('α'), ['Α']);
+    assert_eq!(title('ς'), ['Σ']);
+    assert_eq!(title('DŽ'), ['Dž']);
+    assert_eq!(title('fi'), ['F', 'i']);
+    assert_eq!(title('ᾀ'), ['ᾈ']);
 }
 
 #[test]

src/librustc_lint/lib.rs

@@ -32,7 +32,7 @@
 
 #![feature(box_patterns)]
 #![feature(box_syntax)]
-#![feature(collections)]
+#![cfg_attr(stage0, feature(collections))]
 #![feature(core)]
 #![feature(quote)]
 #![feature(rustc_diagnostic_macros)]