From 372673b58c09355c6ca963a05e2cbc142b832eac Mon Sep 17 00:00:00 2001
From: Tim Chevalier <chevalier@alum.wellesley.edu>
Date: Fri, 23 Mar 2012 14:41:02 -0700
Subject: [PATCH] Remove char::to_lower, char::to_upper, use libc versions
 instead

As per Graydon's comments on #1985: remove char::to_lower and
char::to_upper. The str versions of these functions now call
libc::tolower and libc::toupper (using wrappers that cast between
char and c_char). Not sure how much better that is, but it at least
makes it clearer that these functions are Unicode-unsafe.
---
 src/libcore/char.rs | 44 +-------------------------------------------
 src/libcore/libc.rs |  4 ++--
 src/libcore/str.rs  | 25 ++++++++++++++-----------
 3 files changed, 17 insertions(+), 56 deletions(-)

diff --git a/src/libcore/char.rs b/src/libcore/char.rs
index ad9b7388d6ec..910c28409a1e 100644
--- a/src/libcore/char.rs
+++ b/src/libcore/char.rs
@@ -38,7 +38,7 @@ export is_alphabetic,
        is_lowercase, is_uppercase,
        is_whitespace, is_alphanumeric,
        is_ascii, is_digit,
-       to_digit, to_lower, to_upper, cmp;
+       to_digit, cmp;
 
 import is_alphabetic = unicode::derived_property::Alphabetic;
 import is_XID_start = unicode::derived_property::XID_Start;
@@ -122,30 +122,6 @@ pure fn to_digit(c: char, radix: uint) -> option<uint> {
     else { none }
 }
 
-/*
- FIXME: works only on ASCII (Issue #1985)
-*/
-#[doc = "Convert a char to the corresponding lower case."]
-pure fn to_lower(c: char) -> char {
-    assert is_ascii(c);
-    alt c {
-      'A' to 'Z' { ((c as u8) + 32u8) as char }
-      _ { c }
-    }
-}
-
-/*
- FIXME: works only on ASCII (Issue 1985)
-*/
-#[doc = "Convert a char to the corresponding upper case."]
-pure fn to_upper(c: char) -> char {
-    assert is_ascii(c);
-    alt c {
-      'a' to 'z' { ((c as u8) - 32u8) as char }
-      _ { c }
-    }
-}
-
 #[doc = "
 Compare two chars
 
@@ -206,24 +182,6 @@ fn test_to_digit() {
     assert to_digit('$', 36u) == none;
 }
 
-#[test]
-fn test_to_lower() {
-    assert (to_lower('H') == 'h');
-    assert (to_lower('e') == 'e');
-    // non-ASCII, shouldn't work (see earlier FIXME)
-    //assert (to_lower('Ö') == 'ö');
-    //assert (to_lower('ß') == 'ß');
-}
-
-#[test]
-fn test_to_upper() {
-    assert (to_upper('l') == 'L');
-    assert (to_upper('Q') == 'Q');
-    // non-ASCII, shouldn't work (see earlier FIXME)
-    //assert (to_upper('ü') == 'Ü');
-    //assert (to_upper('ß') == 'ß');
-}
-
 #[test]
 fn test_is_ascii() unsafe {
    assert str::all("banana", char::is_ascii);
diff --git a/src/libcore/libc.rs b/src/libcore/libc.rs
index b46b9cb7d5d0..912ab3a7429d 100644
--- a/src/libcore/libc.rs
+++ b/src/libcore/libc.rs
@@ -736,8 +736,8 @@ mod funcs {
             fn isspace(c: c_int) -> c_int;
             fn isupper(c: c_int) -> c_int;
             fn isxdigit(c: c_int) -> c_int;
-            fn tolower(c: c_int) -> c_int;
-            fn toupper(c: c_int) -> c_int;
+            fn tolower(c: c_char) -> c_char;
+            fn toupper(c: c_char) -> c_char;
         }
 
         #[nolink]
diff --git a/src/libcore/str.rs b/src/libcore/str.rs
index 95dc2e130339..057c0b0d187f 100644
--- a/src/libcore/str.rs
+++ b/src/libcore/str.rs
@@ -95,7 +95,6 @@ export
 
    unsafe;
 
-
 #[abi = "cdecl"]
 native mod rustrt {
     fn rust_str_push(&s: str, ch: u8);
@@ -492,14 +491,14 @@ fn words(s: str) -> [str] {
     split_nonempty(s, {|c| char::is_whitespace(c)})
 }
 
-#[doc = "Convert a string to lowercase"]
+#[doc = "Convert a string to lowercase. ASCII only"]
 fn to_lower(s: str) -> str {
-    map(s, char::to_lower)
+    map(s, {|c| (libc::tolower(c as libc::c_char)) as char})
 }
 
-#[doc = "Convert a string to uppercase"]
+#[doc = "Convert a string to uppercase. ASCII only"]
 fn to_upper(s: str) -> str {
-    map(s, char::to_upper)
+    map(s, {|c| (libc::toupper(c as libc::c_char)) as char})
 }
 
 #[doc = "
@@ -1629,6 +1628,8 @@ mod unsafe {
 #[cfg(test)]
 mod tests {
 
+    import libc::c_char;
+
     #[test]
     fn test_eq() {
         assert (eq("", ""));
@@ -1937,9 +1938,9 @@ mod tests {
 
     #[test]
     fn test_to_upper() {
-        // char::to_upper, and hence str::to_upper
+        // libc::toupper, and hence str::to_upper
         // are culturally insensitive: they only work for ASCII
-        // (see Issue #1985)
+        // (see Issue #1347)
         let unicode = ""; //"\u65e5\u672c"; // uncomment once non-ASCII works
         let input = "abcDEF" + unicode + "xyz:.;";
         let expected = "ABCDEF" + unicode + "XYZ:.;";
@@ -1949,8 +1950,9 @@ mod tests {
 
     #[test]
     fn test_to_lower() {
-        assert "" == map("", char::to_lower);
-        assert "ymca" == map("YMCA", char::to_lower);
+        assert "" == map("", {|c| libc::tolower(c as c_char) as char});
+        assert "ymca" == map("YMCA",
+                             {|c| libc::tolower(c as c_char) as char});
     }
 
     #[test]
@@ -2401,8 +2403,9 @@ mod tests {
 
     #[test]
     fn test_map() {
-        assert "" == map("", char::to_upper);
-        assert "YMCA" == map("ymca", char::to_upper);
+        assert "" == map("", {|c| libc::toupper(c as c_char) as char});
+        assert "YMCA" == map("ymca", {|c| libc::toupper(c as c_char)
+              as char});
     }
 
     #[test]