From a2027fb609835a3f95557b39611e3d11b89fca03 Mon Sep 17 00:00:00 2001
From: brettw <brettw@chromium.org>
Date: Mon, 13 Jul 2015 19:24:50 -0700
Subject: Remove CaseInsensitiveCompare from string_util.h

There were a number of callers in net using this for HTTP headers. I think
these callers actually just need ASCII case-insensitive comparisons so these
were changed.

The omnibox code used this functor. I added a new omnibox-specific one which
does not have the locale issues of the old string_util one, but which still
has the UTF-16 and combining accent issues (described in great detail in
the comment for this).

The Windows installer code can't depend on ICU so it calls the Win32 function
to do case-insensitive comparisons. This should match the system comparison
for registry keys better anyway.

I also changed a caller of StartsWith to use this version. I wrote this
StartsWith call using ToLower in a previous patch, but it turns out that the
lengths of case-mapped strings do change in practice, making the offset
computations of the suyrrounding code incorrect. This new version will be
like the old version (will miss some cases of case-insensitive equality) but
will handle 0x80-0xFF properly.

BUG=24917

Review URL: https://codereview.chromium.org/1230583014

Cr-Commit-Position: refs/heads/master@{#338624}
---
 base/strings/string_util.cc | 18 +++++++++++++++---
 base/strings/string_util.h  | 22 ++++++++--------------
 2 files changed, 23 insertions(+), 17 deletions(-)

(limited to 'base/strings')

diff --git a/base/strings/string_util.cc b/base/strings/string_util.cc
index fcd8ddf..ae5fb80 100644
--- a/base/strings/string_util.cc
+++ b/base/strings/string_util.cc
@@ -101,6 +101,18 @@ template<> struct NonASCIIMask<8, wchar_t> {
 };
 #endif  // WCHAR_T_IS_UTF32
 
+// DO NOT USE. http://crbug.com/24917
+//
+// tolower() will given incorrect results for non-ASCII characters. Use the
+// ASCII version, base::i18n::ToLower, or base::i18n::FoldCase. This is here
+// for backwards-compat for StartsWith until such calls can be updated.
+struct CaseInsensitiveCompareDeprecated {
+ public:
+  bool operator()(char16 x, char16 y) const {
+    return tolower(x) == tolower(y);
+  }
+};
+
 }  // namespace
 
 namespace base {
@@ -611,7 +623,7 @@ bool StartsWith(const string16& str,
     if (search.size() > str.size())
       return false;
     return std::equal(search.begin(), search.end(), str.begin(),
-                      CaseInsensitiveCompare<char16>());
+                      CaseInsensitiveCompareDeprecated());
   }
   return StartsWith(StringPiece16(str), StringPiece16(search),
                     CompareCase::SENSITIVE);
@@ -667,10 +679,10 @@ bool EndsWith(const string16& str,
       return false;
     return std::equal(search.begin(), search.end(),
                       str.begin() + (str.size() - search.size()),
-                      CaseInsensitiveCompare<char16>());
+                      CaseInsensitiveCompareDeprecated());
   }
   return EndsWith(StringPiece16(str), StringPiece16(search),
-                    CompareCase::SENSITIVE);
+                  CompareCase::SENSITIVE);
 }
 
 char HexDigitToInt(wchar_t c) {
diff --git a/base/strings/string_util.h b/base/strings/string_util.h
index 5d26f1c5..e4abce2 100644
--- a/base/strings/string_util.h
+++ b/base/strings/string_util.h
@@ -103,20 +103,14 @@ template <class Char> inline Char ToUpperASCII(Char c) {
   return (c >= 'a' && c <= 'z') ? (c + ('A' - 'a')) : c;
 }
 
-// Function objects to aid in comparing/searching strings.
-
-// DO NOT USE. tolower() will given incorrect results for non-ASCII characters.
-// Use the ASCII version, base::i18n::ToLower, or base::i18n::FoldCase.
-template<typename Char> struct CaseInsensitiveCompare {
- public:
-  bool operator()(Char x, Char y) const {
-    // TODO(darin): Do we really want to do locale sensitive comparisons here?
-    // ANSWER(brettw): No.
-    // See http://crbug.com/24917
-    return tolower(x) == tolower(y);
-  }
-};
-
+// Functor for case-insensitive ASCII comparisons for STL algorithms like
+// std::search.
+//
+// Note that a full Unicode version of this functor is not possible to write
+// because case mappings might change the number of characters, depend on
+// context (combining accents), and require handling UTF-16. If you need
+// proper Unicode support, use base::i18n::ToLower/FoldCase and then just
+// use a normal operator== on the result.
 template<typename Char> struct CaseInsensitiveCompareASCII {
  public:
   bool operator()(Char x, Char y) const {
-- 
cgit v1.1