[base] Introduce wide string split utilities

This change implements wide string overloads to string_split.h
and string_util.h.
Since these functions are not supposed to be used by Cross-Platform
code it is guarded by correpsonding #ifdef guards. Furthermore,
this change performs some minor code clean-ups and performance
improvements.

Bug: 911896
Change-Id: I579b55a398307c890b30a027699b74a9afb1a77e
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/1823851
Commit-Queue: Jan Wilken Dörrie <jdoerrie@chromium.org>
Reviewed-by: Robert Liao <robliao@chromium.org>
Reviewed-by: Daniel Cheng <dcheng@chromium.org>
Cr-Commit-Position: refs/heads/master@{#703617}

base/strings/string_piece.cc

@@ -219,8 +219,11 @@ size_t find_first_of(const StringPiece& self,
 size_t find_first_of(const StringPiece16& self,
                      const StringPiece16& s,
                      size_t pos) {
+  // Use the faster std::find() if searching for a single character.
   StringPiece16::const_iterator found =
-      std::find_first_of(self.begin() + pos, self.end(), s.begin(), s.end());
+      s.size() == 1 ? std::find(self.begin() + pos, self.end(), s[0])
+                    : std::find_first_of(self.begin() + pos, self.end(),
+                                         s.begin(), s.end());
   if (found == self.end())
     return StringPiece16::npos;
   return found - self.begin();

base/strings/string_split.cc

@@ -14,27 +14,15 @@ namespace base {
 
 namespace {
 
-// PieceToOutputType converts a StringPiece as needed to a given output type,
-// which is either the same type of StringPiece (a NOP) or the corresponding
-// non-piece string type.
-//
-// The default converter is a NOP, it works when the OutputType is the
-// correct StringPiece.
-template<typename Str, typename OutputType>
-OutputType PieceToOutputType(BasicStringPiece<Str> piece) {
-  return piece;
-}
-template<>  // Convert StringPiece to std::string
-std::string PieceToOutputType<std::string, std::string>(StringPiece piece) {
-  return piece.as_string();
-}
-template<>  // Convert StringPiece16 to string16.
-string16 PieceToOutputType<string16, string16>(StringPiece16 piece) {
-  return piece.as_string();
-}
-
 // Returns either the ASCII or UTF-16 whitespace.
 template<typename Str> BasicStringPiece<Str> WhitespaceForType();
+#if defined(OS_WIN) && defined(BASE_STRING16_IS_STD_U16STRING)
+template <>
+WStringPiece WhitespaceForType<std::wstring>() {
+  return kWhitespaceWide;
+}
+#endif
+
 template<> StringPiece16 WhitespaceForType<string16>() {
   return kWhitespaceUTF16;
 }
@@ -42,37 +30,12 @@ template<> StringPiece WhitespaceForType<std::string>() {
   return kWhitespaceASCII;
 }
 
-// Optimize the single-character case to call find() on the string instead,
-// since this is the common case and can be made faster. This could have been
-// done with template specialization too, but would have been less clear.
-//
-// There is no corresponding FindFirstNotOf because StringPiece already
-// implements these different versions that do the optimized searching.
-size_t FindFirstOf(StringPiece piece, char c, size_t pos) {
-  return piece.find(c, pos);
-}
-size_t FindFirstOf(StringPiece16 piece, char16 c, size_t pos) {
-  return piece.find(c, pos);
-}
-size_t FindFirstOf(StringPiece piece, StringPiece one_of, size_t pos) {
-  return piece.find_first_of(one_of, pos);
-}
-size_t FindFirstOf(StringPiece16 piece, StringPiece16 one_of, size_t pos) {
-  return piece.find_first_of(one_of, pos);
-}
-
 // General string splitter template. Can take 8- or 16-bit input, can produce
-// the corresponding string or StringPiece output, and can take single- or
-// multiple-character delimiters.
-//
-// DelimiterType is either a character (Str::value_type) or a string piece of
-// multiple characters (BasicStringPiece<Str>). StringPiece has a version of
-// find for both of these cases, and the single-character version is the most
-// common and can be implemented faster, which is why this is a template.
-template<typename Str, typename OutputStringType, typename DelimiterType>
+// the corresponding string or StringPiece output.
+template <typename OutputStringType, typename Str>
 static std::vector<OutputStringType> SplitStringT(
     BasicStringPiece<Str> str,
-    DelimiterType delimiter,
+    BasicStringPiece<Str> delimiter,
     WhitespaceHandling whitespace,
     SplitResult result_type) {
   std::vector<OutputStringType> result;
@@ -81,7 +44,7 @@ static std::vector<OutputStringType> SplitStringT(
 
   size_t start = 0;
   while (start != Str::npos) {
-    size_t end = FindFirstOf(str, delimiter, start);
+    size_t end = str.find_first_of(delimiter, start);
 
     BasicStringPiece<Str> piece;
     if (end == Str::npos) {
@@ -96,7 +59,7 @@ static std::vector<OutputStringType> SplitStringT(
       piece = TrimString(piece, WhitespaceForType<Str>(), TRIM_ALL);
 
     if (result_type == SPLIT_WANT_ALL || !piece.empty())
-      result.push_back(PieceToOutputType<Str, OutputStringType>(piece));
+      result.emplace_back(piece);
   }
   return result;
 }
@@ -130,16 +93,16 @@ bool AppendStringKeyValue(StringPiece input,
   return true;
 }
 
-template <typename Str, typename OutputStringType>
-void SplitStringUsingSubstrT(BasicStringPiece<Str> input,
-                             BasicStringPiece<Str> delimiter,
-                             WhitespaceHandling whitespace,
-                             SplitResult result_type,
-                             std::vector<OutputStringType>* result) {
+template <typename OutputStringType, typename Str>
+std::vector<OutputStringType> SplitStringUsingSubstrT(
+    BasicStringPiece<Str> input,
+    BasicStringPiece<Str> delimiter,
+    WhitespaceHandling whitespace,
+    SplitResult result_type) {
   using Piece = BasicStringPiece<Str>;
   using size_type = typename Piece::size_type;
 
-  result->clear();
+  std::vector<OutputStringType> result;
   for (size_type begin_index = 0, end_index = 0; end_index != Piece::npos;
        begin_index = end_index + delimiter.size()) {
     end_index = input.find(delimiter, begin_index);
@@ -151,8 +114,10 @@ void SplitStringUsingSubstrT(BasicStringPiece<Str> input,
       term = TrimString(term, WhitespaceForType<Str>(), TRIM_ALL);
 
     if (result_type == SPLIT_WANT_ALL || !term.empty())
-      result->push_back(PieceToOutputType<Str, OutputStringType>(term));
+      result.emplace_back(term);
   }
+
+  return result;
 }
 
 }  // namespace
@@ -161,48 +126,29 @@ std::vector<std::string> SplitString(StringPiece input,
                                      StringPiece separators,
                                      WhitespaceHandling whitespace,
                                      SplitResult result_type) {
-  if (separators.size() == 1) {
-    return SplitStringT<std::string, std::string, char>(
-        input, separators[0], whitespace, result_type);
-  }
-  return SplitStringT<std::string, std::string, StringPiece>(
-      input, separators, whitespace, result_type);
+  return SplitStringT<std::string>(input, separators, whitespace, result_type);
 }
 
 std::vector<string16> SplitString(StringPiece16 input,
                                   StringPiece16 separators,
                                   WhitespaceHandling whitespace,
                                   SplitResult result_type) {
-  if (separators.size() == 1) {
-    return SplitStringT<string16, string16, char16>(
-        input, separators[0], whitespace, result_type);
-  }
-  return SplitStringT<string16, string16, StringPiece16>(
-      input, separators, whitespace, result_type);
+  return SplitStringT<string16>(input, separators, whitespace, result_type);
 }
 
 std::vector<StringPiece> SplitStringPiece(StringPiece input,
                                           StringPiece separators,
                                           WhitespaceHandling whitespace,
                                           SplitResult result_type) {
-  if (separators.size() == 1) {
-    return SplitStringT<std::string, StringPiece, char>(
-        input, separators[0], whitespace, result_type);
-  }
-  return SplitStringT<std::string, StringPiece, StringPiece>(
-      input, separators, whitespace, result_type);
+  return SplitStringT<StringPiece>(input, separators, whitespace, result_type);
 }
 
 std::vector<StringPiece16> SplitStringPiece(StringPiece16 input,
                                             StringPiece16 separators,
                                             WhitespaceHandling whitespace,
                                             SplitResult result_type) {
-  if (separators.size() == 1) {
-    return SplitStringT<string16, StringPiece16, char16>(
-        input, separators[0], whitespace, result_type);
-  }
-  return SplitStringT<string16, StringPiece16, StringPiece16>(
-      input, separators, whitespace, result_type);
+  return SplitStringT<StringPiece16>(input, separators, whitespace,
+                                     result_type);
 }
 
 bool SplitStringIntoKeyValuePairs(StringPiece input,
@@ -240,18 +186,16 @@ std::vector<string16> SplitStringUsingSubstr(StringPiece16 input,
                                              StringPiece16 delimiter,
                                              WhitespaceHandling whitespace,
                                              SplitResult result_type) {
-  std::vector<string16> result;
-  SplitStringUsingSubstrT(input, delimiter, whitespace, result_type, &result);
-  return result;
+  return SplitStringUsingSubstrT<string16>(input, delimiter, whitespace,
+                                           result_type);
 }
 
 std::vector<std::string> SplitStringUsingSubstr(StringPiece input,
                                                 StringPiece delimiter,
                                                 WhitespaceHandling whitespace,
                                                 SplitResult result_type) {
-  std::vector<std::string> result;
-  SplitStringUsingSubstrT(input, delimiter, whitespace, result_type, &result);
-  return result;
+  return SplitStringUsingSubstrT<std::string>(input, delimiter, whitespace,
+                                              result_type);
 }
 
 std::vector<StringPiece16> SplitStringPieceUsingSubstr(
@@ -260,8 +204,8 @@ std::vector<StringPiece16> SplitStringPieceUsingSubstr(
     WhitespaceHandling whitespace,
     SplitResult result_type) {
   std::vector<StringPiece16> result;
-  SplitStringUsingSubstrT(input, delimiter, whitespace, result_type, &result);
-  return result;
+  return SplitStringUsingSubstrT<StringPiece16>(input, delimiter, whitespace,
+                                                result_type);
 }
 
 std::vector<StringPiece> SplitStringPieceUsingSubstr(
@@ -269,9 +213,41 @@ std::vector<StringPiece> SplitStringPieceUsingSubstr(
     StringPiece delimiter,
     WhitespaceHandling whitespace,
     SplitResult result_type) {
-  std::vector<StringPiece> result;
-  SplitStringUsingSubstrT(input, delimiter, whitespace, result_type, &result);
-  return result;
+  return SplitStringUsingSubstrT<StringPiece>(input, delimiter, whitespace,
+                                              result_type);
+}
+
+#if defined(OS_WIN) && defined(BASE_STRING16_IS_STD_U16STRING)
+std::vector<std::wstring> SplitString(WStringPiece input,
+                                      WStringPiece separators,
+                                      WhitespaceHandling whitespace,
+                                      SplitResult result_type) {
+  return SplitStringT<std::wstring>(input, separators, whitespace, result_type);
+}
+
+std::vector<WStringPiece> SplitStringPiece(WStringPiece input,
+                                           WStringPiece separators,
+                                           WhitespaceHandling whitespace,
+                                           SplitResult result_type) {
+  return SplitStringT<WStringPiece>(input, separators, whitespace, result_type);
+}
+
+std::vector<std::wstring> SplitStringUsingSubstr(WStringPiece input,
+                                                 WStringPiece delimiter,
+                                                 WhitespaceHandling whitespace,
+                                                 SplitResult result_type) {
+  return SplitStringUsingSubstrT<std::wstring>(input, delimiter, whitespace,
+                                               result_type);
+}
+
+std::vector<WStringPiece> SplitStringPieceUsingSubstr(
+    WStringPiece input,
+    WStringPiece delimiter,
+    WhitespaceHandling whitespace,
+    SplitResult result_type) {
+  return SplitStringUsingSubstrT<WStringPiece>(input, delimiter, whitespace,
+                                               result_type);
 }
+#endif
 
 }  // namespace base

base/strings/string_split.h

@@ -12,6 +12,7 @@
 #include "base/base_export.h"
 #include "base/strings/string16.h"
 #include "base/strings/string_piece.h"
+#include "build/build_config.h"
 
 namespace base {
 
@@ -132,6 +133,31 @@ BASE_EXPORT std::vector<StringPiece> SplitStringPieceUsingSubstr(
     WhitespaceHandling whitespace,
     SplitResult result_type);
 
+#if defined(OS_WIN) && defined(BASE_STRING16_IS_STD_U16STRING)
+BASE_EXPORT std::vector<std::wstring> SplitString(WStringPiece input,
+                                                  WStringPiece separators,
+                                                  WhitespaceHandling whitespace,
+                                                  SplitResult result_type);
+
+BASE_EXPORT std::vector<WStringPiece> SplitStringPiece(
+    WStringPiece input,
+    WStringPiece separators,
+    WhitespaceHandling whitespace,
+    SplitResult result_type);
+
+BASE_EXPORT std::vector<std::wstring> SplitStringUsingSubstr(
+    WStringPiece input,
+    WStringPiece delimiter,
+    WhitespaceHandling whitespace,
+    SplitResult result_type);
+
+BASE_EXPORT std::vector<WStringPiece> SplitStringPieceUsingSubstr(
+    WStringPiece input,
+    WStringPiece delimiter,
+    WhitespaceHandling whitespace,
+    SplitResult result_type);
+#endif
+
 }  // namespace base
 
 #endif  // BASE_STRINGS_STRING_SPLIT_H_

base/strings/string_util.cc

@@ -262,8 +262,8 @@ bool RemoveChars(const std::string& input,
   return ReplaceCharsT(input, remove_chars, StringPiece(), output);
 }
 
-template<typename Str>
-TrimPositions TrimStringT(const Str& input,
+template <typename Str>
+TrimPositions TrimStringT(BasicStringPiece<Str> input,
                           BasicStringPiece<Str> trim_chars,
                           TrimPositions positions,
                           Str* output) {
@@ -271,40 +271,40 @@ TrimPositions TrimStringT(const Str& input,
   // a StringPiece version of input to be able to call find* on it with the
   // StringPiece version of trim_chars (normally the trim_chars will be a
   // constant so avoid making a copy).
-  BasicStringPiece<Str> input_piece(input);
   const size_t last_char = input.length() - 1;
-  const size_t first_good_char = (positions & TRIM_LEADING) ?
-      input_piece.find_first_not_of(trim_chars) : 0;
-  const size_t last_good_char = (positions & TRIM_TRAILING) ?
-      input_piece.find_last_not_of(trim_chars) : last_char;
+  const size_t first_good_char =
+      (positions & TRIM_LEADING) ? input.find_first_not_of(trim_chars) : 0;
+  const size_t last_good_char = (positions & TRIM_TRAILING)
+                                    ? input.find_last_not_of(trim_chars)
+                                    : last_char;
 
   // When the string was all trimmed, report that we stripped off characters
   // from whichever position the caller was interested in. For empty input, we
   // stripped no characters, but we still need to clear |output|.
-  if (input.empty() ||
-      (first_good_char == Str::npos) || (last_good_char == Str::npos)) {
+  if (input.empty() || first_good_char == Str::npos ||
+      last_good_char == Str::npos) {
     bool input_was_empty = input.empty();  // in case output == &input
     output->clear();
     return input_was_empty ? TRIM_NONE : positions;
   }
 
   // Trim.
-  *output =
-      input.substr(first_good_char, last_good_char - first_good_char + 1);
+  output->assign(input.data() + first_good_char,
+                 last_good_char - first_good_char + 1);
 
   // Return where we trimmed from.
   return static_cast<TrimPositions>(
-      ((first_good_char == 0) ? TRIM_NONE : TRIM_LEADING) |
-      ((last_good_char == last_char) ? TRIM_NONE : TRIM_TRAILING));
+      (first_good_char == 0 ? TRIM_NONE : TRIM_LEADING) |
+      (last_good_char == last_char ? TRIM_NONE : TRIM_TRAILING));
 }
 
-bool TrimString(const string16& input,
+bool TrimString(StringPiece16 input,
                 StringPiece16 trim_chars,
                 string16* output) {
   return TrimStringT(input, trim_chars, TRIM_ALL, output) != TRIM_NONE;
 }
 
-bool TrimString(const std::string& input,
+bool TrimString(StringPiece input,
                 StringPiece trim_chars,
                 std::string* output) {
   return TrimStringT(input, trim_chars, TRIM_ALL, output) != TRIM_NONE;
@@ -370,7 +370,7 @@ void TruncateUTF8ToByteSize(const std::string& input,
     output->clear();
 }
 
-TrimPositions TrimWhitespace(const string16& input,
+TrimPositions TrimWhitespace(StringPiece16 input,
                              TrimPositions positions,
                              string16* output) {
   return TrimStringT(input, StringPiece16(kWhitespaceUTF16), positions, output);
@@ -381,7 +381,7 @@ StringPiece16 TrimWhitespace(StringPiece16 input,
   return TrimStringPieceT(input, StringPiece16(kWhitespaceUTF16), positions);
 }
 
-TrimPositions TrimWhitespaceASCII(const std::string& input,
+TrimPositions TrimWhitespaceASCII(StringPiece input,
                                   TrimPositions positions,
                                   std::string* output) {
   return TrimStringT(input, StringPiece(kWhitespaceASCII), positions, output);
@@ -1085,6 +1085,32 @@ string16 ReplaceStringPlaceholders(const string16& format_string,
   return result;
 }
 
+#if defined(OS_WIN) && defined(BASE_STRING16_IS_STD_U16STRING)
+
+TrimPositions TrimWhitespace(WStringPiece input,
+                             TrimPositions positions,
+                             std::wstring* output) {
+  return TrimStringT(input, WStringPiece(kWhitespaceWide), positions, output);
+}
+
+WStringPiece TrimWhitespace(WStringPiece input, TrimPositions positions) {
+  return TrimStringPieceT(input, WStringPiece(kWhitespaceWide), positions);
+}
+
+bool TrimString(WStringPiece input,
+                WStringPiece trim_chars,
+                std::wstring* output) {
+  return TrimStringT(input, trim_chars, TRIM_ALL, output) != TRIM_NONE;
+}
+
+WStringPiece TrimString(WStringPiece input,
+                        WStringPiece trim_chars,
+                        TrimPositions positions) {
+  return TrimStringPieceT(input, trim_chars, positions);
+}
+
+#endif
+
 // The following code is compatible with the OpenBSD lcpy interface.  See:
 //   http://www.gratisoft.us/todd/papers/strlcpy.html
 //   ftp://ftp.openbsd.org/pub/OpenBSD/src/lib/libc/string/{wcs,str}lcpy.c

base/strings/string_util.h

@@ -204,10 +204,10 @@ enum TrimPositions {
 //
 // It is safe to use the same variable for both |input| and |output| (this is
 // the normal usage to trim in-place).
-BASE_EXPORT bool TrimString(const string16& input,
+BASE_EXPORT bool TrimString(StringPiece16 input,
                             StringPiece16 trim_chars,
                             string16* output);
-BASE_EXPORT bool TrimString(const std::string& input,
+BASE_EXPORT bool TrimString(StringPiece input,
                             StringPiece trim_chars,
                             std::string* output);
 
@@ -296,12 +296,12 @@ inline string16 AsString16(WStringPiece str) {
 //
 // The std::string versions return where whitespace was found.
 // NOTE: Safe to use the same variable for both input and output.
-BASE_EXPORT TrimPositions TrimWhitespace(const string16& input,
+BASE_EXPORT TrimPositions TrimWhitespace(StringPiece16 input,
                                          TrimPositions positions,
                                          string16* output);
 BASE_EXPORT StringPiece16 TrimWhitespace(StringPiece16 input,
                                          TrimPositions positions);
-BASE_EXPORT TrimPositions TrimWhitespaceASCII(const std::string& input,
+BASE_EXPORT TrimPositions TrimWhitespaceASCII(StringPiece input,
                                               TrimPositions positions,
                                               std::string* output);
 BASE_EXPORT StringPiece TrimWhitespaceASCII(StringPiece input,
@@ -532,6 +532,23 @@ BASE_EXPORT string16 ReplaceStringPlaceholders(const string16& format_string,
                                                const string16& a,
                                                size_t* offset);
 
+#if defined(OS_WIN) && defined(BASE_STRING16_IS_STD_U16STRING)
+BASE_EXPORT TrimPositions TrimWhitespace(WStringPiece input,
+                                         TrimPositions positions,
+                                         std::wstring* output);
+
+BASE_EXPORT WStringPiece TrimWhitespace(WStringPiece input,
+                                        TrimPositions positions);
+
+BASE_EXPORT bool TrimString(WStringPiece input,
+                            WStringPiece trim_chars,
+                            std::wstring* output);
+
+BASE_EXPORT WStringPiece TrimString(WStringPiece input,
+                                    WStringPiece trim_chars,
+                                    TrimPositions positions);
+#endif
+
 }  // namespace base
 
 #if defined(OS_WIN)