Separate out file_util functions that use ICU into their own file.

This prevents ~300K of ICU being linked into chrome.exe due to a false
dependency.


git-svn-id: svn://svn.chromium.org/chrome/trunk/src@13281 0039d316-1c4b-4281-b951-d872f2087c98

3 files changed, 92 insertions, 72 deletions

diff --git a/base/base.gyp b/base/base.gyp
index ee56c96..325ab65 100644
--- a/base/base.gyp
+++ b/base/base.gyp
@@ -92,6 +92,7 @@
         'file_path.h',
         'file_util.cc',
         'file_util.h',
+        'file_util_icu.cc',
         'file_util_linux.cc',
         'file_util_mac.mm',
         'file_util_posix.cc',
diff --git a/base/file_util.cc b/base/file_util.cc
index cb6c7e01..9ba6eed 100644
--- a/base/file_util.cc
+++ b/base/file_util.cc
@@ -1,4 +1,4 @@
-// Copyright (c) 2006-2008 The Chromium Authors. All rights reserved.
+// Copyright (c) 2006-2009 The Chromium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
@@ -14,7 +14,6 @@
 #include "base/file_path.h"
 #include "base/logging.h"
 #include "base/string_util.h"
-#include "unicode/uniset.h"
 
 #include "base/string_piece.h"
 #include "base/sys_string_conversions.h"
@@ -144,76 +143,6 @@ void ReplaceExtension(FilePath* path, const FilePath::StringType& extension) {
   value.append(clean_extension);
 }
 
-void ReplaceIllegalCharacters(std::wstring* file_name, int replace_char) {
-  DCHECK(file_name);
-
-  // Control characters, formatting characters, non-characters, and
-  // some printable ASCII characters regarded as dangerous ('"*/:<>?\\').
-  // See  http://blogs.msdn.com/michkap/archive/2006/11/03/941420.aspx
-  // and http://msdn2.microsoft.com/en-us/library/Aa365247.aspx
-  // TODO(jungshik): Revisit the set. ZWJ and ZWNJ are excluded because they
-  // are legitimate in Arabic and some S/SE Asian scripts. However, when used
-  // elsewhere, they can be confusing/problematic.
-  // Also, consider wrapping the set with our Singleton class to create and
-  // freeze it only once. Note that there's a trade-off between memory and
-  // speed.
-
-  UErrorCode status = U_ZERO_ERROR;
-#if defined(WCHAR_T_IS_UTF16)
-  UnicodeSet illegal_characters(UnicodeString(
-      L"[[\"*/:<>?\\\\|][:Cc:][:Cf:] - [\u200c\u200d]]"), status);
-#else
-  UnicodeSet illegal_characters(UNICODE_STRING_SIMPLE(
-      "[[\"*/:<>?\\\\|][:Cc:][:Cf:] - [\\u200c\\u200d]]").unescape(), status);
-#endif
-  DCHECK(U_SUCCESS(status));
-  // Add non-characters. If this becomes a performance bottleneck by
-  // any chance, check |ucs4 & 0xFFFEu == 0xFFFEu|, instead.
-  illegal_characters.add(0xFDD0, 0xFDEF);
-  for (int i = 0; i <= 0x10; ++i) {
-    int plane_base = 0x10000 * i;
-    illegal_characters.add(plane_base + 0xFFFE, plane_base + 0xFFFF);
-  }
-  illegal_characters.freeze();
-  DCHECK(!illegal_characters.contains(replace_char) && replace_char < 0x10000);
-
-  // Remove leading and trailing whitespace.
-  TrimWhitespace(*file_name, TRIM_ALL, file_name);
-
-  std::wstring::size_type i = 0;
-  std::wstring::size_type length = file_name->size();
-  const wchar_t* wstr = file_name->data();
-#if defined(WCHAR_T_IS_UTF16)
-  // Using |span| method of UnicodeSet might speed things up a bit, but
-  // it's not likely to matter here.
-  std::wstring temp;
-  temp.reserve(length);
-  while (i < length) {
-    UChar32 ucs4;
-    std::wstring::size_type prev = i;
-    U16_NEXT(wstr, i, length, ucs4);
-    if (illegal_characters.contains(ucs4)) {
-      temp.push_back(replace_char);
-    } else if (ucs4 < 0x10000) {
-      temp.push_back(ucs4);
-    } else {
-      temp.push_back(wstr[prev]);
-      temp.push_back(wstr[prev + 1]);
-    }
-  }
-  file_name->swap(temp);
-#elif defined(WCHAR_T_IS_UTF32)
-  while (i < length) {
-    if (illegal_characters.contains(wstr[i])) {
-      (*file_name)[i] = replace_char;
-    }
-    ++i;
-  }
-#else
-#error wchar_t* should be either UTF-16 or UTF-32
-#endif
-}
-
 bool ContentsEqual(const FilePath& filename1, const FilePath& filename2) {
   // We open the file in binary format even if they are text files because
   // we are just comparing that bytes are exactly same in both files and not
diff --git a/base/file_util_icu.cc b/base/file_util_icu.cc
new file mode 100644
index 0000000..0b9830d
--- /dev/null
+++ b/base/file_util_icu.cc
@@ -0,0 +1,90 @@
+// Copyright (c) 2009 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// File utilities that use the ICU library go in this file.  Functions using ICU
+// are separated from the other functions to prevent ICU being pulled in by the
+// linker if there is a false dependency.
+//
+// (The VS2005 linker finds such a false dependency and adds ~300K of ICU to
+// chrome.exe if this code lives in file_util.cc, even though none of this code
+// is called.)
+
+#include "base/file_util.h"
+
+#include "base/string_util.h"
+#include "unicode/uniset.h"
+
+namespace file_util {
+
+void ReplaceIllegalCharacters(std::wstring* file_name, int replace_char) {
+  DCHECK(file_name);
+
+  // Control characters, formatting characters, non-characters, and
+  // some printable ASCII characters regarded as dangerous ('"*/:<>?\\').
+  // See  http://blogs.msdn.com/michkap/archive/2006/11/03/941420.aspx
+  // and http://msdn2.microsoft.com/en-us/library/Aa365247.aspx
+  // TODO(jungshik): Revisit the set. ZWJ and ZWNJ are excluded because they
+  // are legitimate in Arabic and some S/SE Asian scripts. However, when used
+  // elsewhere, they can be confusing/problematic.
+  // Also, consider wrapping the set with our Singleton class to create and
+  // freeze it only once. Note that there's a trade-off between memory and
+  // speed.
+
+  UErrorCode status = U_ZERO_ERROR;
+#if defined(WCHAR_T_IS_UTF16)
+  UnicodeSet illegal_characters(UnicodeString(
+      L"[[\"*/:<>?\\\\|][:Cc:][:Cf:] - [\u200c\u200d]]"), status);
+#else
+  UnicodeSet illegal_characters(UNICODE_STRING_SIMPLE(
+      "[[\"*/:<>?\\\\|][:Cc:][:Cf:] - [\\u200c\\u200d]]").unescape(), status);
+#endif
+  DCHECK(U_SUCCESS(status));
+  // Add non-characters. If this becomes a performance bottleneck by
+  // any chance, check |ucs4 & 0xFFFEu == 0xFFFEu|, instead.
+  illegal_characters.add(0xFDD0, 0xFDEF);
+  for (int i = 0; i <= 0x10; ++i) {
+    int plane_base = 0x10000 * i;
+    illegal_characters.add(plane_base + 0xFFFE, plane_base + 0xFFFF);
+  }
+  illegal_characters.freeze();
+  DCHECK(!illegal_characters.contains(replace_char) && replace_char < 0x10000);
+
+  // Remove leading and trailing whitespace.
+  TrimWhitespace(*file_name, TRIM_ALL, file_name);
+
+  std::wstring::size_type i = 0;
+  std::wstring::size_type length = file_name->size();
+  const wchar_t* wstr = file_name->data();
+#if defined(WCHAR_T_IS_UTF16)
+  // Using |span| method of UnicodeSet might speed things up a bit, but
+  // it's not likely to matter here.
+  std::wstring temp;
+  temp.reserve(length);
+  while (i < length) {
+    UChar32 ucs4;
+    std::wstring::size_type prev = i;
+    U16_NEXT(wstr, i, length, ucs4);
+    if (illegal_characters.contains(ucs4)) {
+      temp.push_back(replace_char);
+    } else if (ucs4 < 0x10000) {
+      temp.push_back(ucs4);
+    } else {
+      temp.push_back(wstr[prev]);
+      temp.push_back(wstr[prev + 1]);
+    }
+  }
+  file_name->swap(temp);
+#elif defined(WCHAR_T_IS_UTF32)
+  while (i < length) {
+    if (illegal_characters.contains(wstr[i])) {
+      (*file_name)[i] = replace_char;
+    }
+    ++i;
+  }
+#else
+#error wchar_t* should be either UTF-16 or UTF-32
+#endif
+}
+
+}  // namespace