[vcpkg] Optimize string split slightly. (#11433)

11 files changed, 40 insertions, 70 deletions

diff --git a/toolsrc/src/vcpkg-test/strings.cpp b/toolsrc/src/vcpkg-test/strings.cpp
index d58d1b172..eb1f5697f 100644
--- a/toolsrc/src/vcpkg-test/strings.cpp
+++ b/toolsrc/src/vcpkg-test/strings.cpp
@@ -2,15 +2,16 @@
 
 #include <vcpkg/base/strings.h>
 
-#include <cstdint>
+#include <stdint.h>
 #include <utility>
 #include <vector>
+#include <string>
 
 TEST_CASE ("b32 encoding", "[strings]")
 {
-    using u64 = std::uint64_t;
+    using u64 = uint64_t;
 
-    std::vector<std::pair<std::uint64_t, std::string>> map;
+    std::vector<std::pair<u64, std::string>> map;
 
     map.emplace_back(0, "AAAAAAAAAAAAA");
     map.emplace_back(1, "BAAAAAAAAAAAA");
@@ -24,10 +25,19 @@ TEST_CASE ("b32 encoding", "[strings]")
     map.emplace_back(0x1405'64E7'FE7E'A88C, "MEK5H774ELBIB");
     map.emplace_back(0xFFFF'FFFF'FFFF'FFFF, "777777777777P");
 
-    std::string result;
     for (const auto& pr : map)
     {
-        result = vcpkg::Strings::b32_encode(pr.first);
         REQUIRE(vcpkg::Strings::b32_encode(pr.first) == pr.second);
     }
 }
+
+TEST_CASE ("split by char", "[strings]")
+{
+    using vcpkg::Strings::split;
+    using result_t = std::vector<std::string>;
+    REQUIRE(split(",,,,,,", ',').empty());
+    REQUIRE(split(",,a,,b,,", ',') == result_t{"a", "b"});
+    REQUIRE(split("hello world", ' ') == result_t{"hello", "world"});
+    REQUIRE(split("    hello  world    ", ' ') == result_t{"hello", "world"});
+    REQUIRE(split("no delimiters", ',') == result_t{"no delimiters"});
+}
diff --git a/toolsrc/src/vcpkg.cpp b/toolsrc/src/vcpkg.cpp
index a249a3d9d..7f9decce1 100644
--- a/toolsrc/src/vcpkg.cpp
+++ b/toolsrc/src/vcpkg.cpp
@@ -341,7 +341,7 @@ int main(const int argc, const char* const* const argv)
     const auto vcpkg_feature_flags_env = System::get_environment_variable("VCPKG_FEATURE_FLAGS");
     if (const auto v = vcpkg_feature_flags_env.get())
     {
-        auto flags = Strings::split(*v, ",");
+        auto flags = Strings::split(*v, ',');
         if (std::find(flags.begin(), flags.end(), "binarycaching") != flags.end()) GlobalState::g_binary_caching = true;
     }
 
diff --git a/toolsrc/src/vcpkg/base/files.cpp b/toolsrc/src/vcpkg/base/files.cpp
index 24e2038d9..7f4d18616 100644
--- a/toolsrc/src/vcpkg/base/files.cpp
+++ b/toolsrc/src/vcpkg/base/files.cpp
@@ -834,10 +834,10 @@ namespace vcpkg::Files
         {
 #if defined(_WIN32)
             static constexpr StringLiteral EXTS[] = {".cmd", ".exe", ".bat"};
-            auto paths = Strings::split(System::get_environment_variable("PATH").value_or_exit(VCPKG_LINE_INFO), ";");
+            auto paths = Strings::split(System::get_environment_variable("PATH").value_or_exit(VCPKG_LINE_INFO), ';');
 #else // ^^^ defined(_WIN32) // !defined(_WIN32) vvv
             static constexpr StringLiteral EXTS[] = {""};
-            auto paths = Strings::split(System::get_environment_variable("PATH").value_or_exit(VCPKG_LINE_INFO), ":");
+            auto paths = Strings::split(System::get_environment_variable("PATH").value_or_exit(VCPKG_LINE_INFO), ':');
 #endif // ^^^ !defined(_WIN32)
 
             std::vector<fs::path> ret;
diff --git a/toolsrc/src/vcpkg/base/strings.cpp b/toolsrc/src/vcpkg/base/strings.cpp
index d9dc4c088..167053d70 100644
--- a/toolsrc/src/vcpkg/base/strings.cpp
+++ b/toolsrc/src/vcpkg/base/strings.cpp
@@ -157,63 +157,23 @@ void Strings::trim_all_and_remove_whitespace_strings(std::vector<std::string>* s
     Util::erase_remove_if(*strings, [](const std::string& s) { return s.empty(); });
 }
 
-std::vector<std::string> Strings::split(const std::string& s, const std::string& delimiter)
+std::vector<std::string> Strings::split(const std::string& s, const char delimiter)
 {
     std::vector<std::string> output;
-
-    if (delimiter.empty())
-    {
-        output.push_back(s);
-        return output;
-    }
-
-    const size_t delimiter_length = delimiter.length();
-    size_t i = 0;
-    for (size_t pos = s.find(delimiter); pos != std::string::npos; pos = s.find(delimiter, pos))
-    {
-        output.push_back(s.substr(i, pos - i));
-        pos += delimiter_length;
-        i = pos;
-    }
-
-    // Add the rest of the string after the last delimiter, unless there is nothing after it
-    if (i != s.length())
-    {
-        output.push_back(s.substr(i, s.length()));
-    }
-
-    return output;
-}
-
-std::vector<std::string> Strings::split(const std::string& s, const std::string& delimiter, size_t max_count)
-{
-    std::vector<std::string> output;
-
-    Checks::check_exit(VCPKG_LINE_INFO, max_count >= 1);
-
-    if (delimiter.empty())
-    {
-        output.push_back(s);
-        return output;
-    }
-
-    const size_t delimiter_length = delimiter.length();
-    size_t i = 0;
-    for (size_t pos = s.find(delimiter); pos != std::string::npos; pos = s.find(delimiter, pos))
+    auto first = s.begin();
+    const auto last = s.end();
+    for (;;)
     {
-        if (output.size() == max_count - 1) break;
-        output.push_back(s.substr(i, pos - i));
-        pos += delimiter_length;
-        i = pos;
-    }
+        first = std::find_if(first, last, [=](const char c) { return c != delimiter; });
+        if (first == last)
+        {
+            return output;
+        }
 
-    // Add the rest of the string after the last delimiter, unless there is nothing after it
-    if (i != s.length())
-    {
-        output.push_back(s.substr(i, s.length()));
+        auto next = std::find(first, last, delimiter);
+        output.emplace_back(first, next);
+        first = next;
     }
-
-    return output;
 }
 
 std::vector<StringView> Strings::find_all_enclosed(StringView input, StringView left_delim, StringView right_delim)
@@ -223,7 +183,7 @@ std::vector<StringView> Strings::find_all_enclosed(StringView input, StringView
 
     std::vector<StringView> output;
 
-    while (true)
+    for (;;)
     {
         it_left = std::search(it_right, input.end(), left_delim.begin(), left_delim.end());
         if (it_left == input.end()) break;
diff --git a/toolsrc/src/vcpkg/base/system.process.cpp b/toolsrc/src/vcpkg/base/system.process.cpp
index 74c4390cc..e6e195010 100644
--- a/toolsrc/src/vcpkg/base/system.process.cpp
+++ b/toolsrc/src/vcpkg/base/system.process.cpp
@@ -267,7 +267,7 @@ namespace vcpkg
 
         if (k && !k->empty())
         {
-            auto vars = Strings::split(*k, ";");
+            auto vars = Strings::split(*k, ';');
 
             for (auto&& var : vars)
             {
diff --git a/toolsrc/src/vcpkg/build.cpp b/toolsrc/src/vcpkg/build.cpp
index a1a520970..675c75ea5 100644
--- a/toolsrc/src/vcpkg/build.cpp
+++ b/toolsrc/src/vcpkg/build.cpp
@@ -1057,7 +1057,7 @@ namespace vcpkg::Build
                             VCPKG_LINE_INFO, "Unknown setting for VCPKG_BUILD_TYPE: %s", variable_value);

                     break;

                 case VcpkgTripletVar::ENV_PASSTHROUGH:

-                    passthrough_env_vars = Strings::split(variable_value, ";");

+                    passthrough_env_vars = Strings::split(variable_value, ';');

                     break;

                 case VcpkgTripletVar::PUBLIC_ABI_OVERRIDE:

                     public_abi_override = variable_value.empty() ? nullopt : Optional<std::string>{variable_value};

diff --git a/toolsrc/src/vcpkg/cmakevars.cpp b/toolsrc/src/vcpkg/cmakevars.cpp
index 98b2546ed..114335e2c 100644
--- a/toolsrc/src/vcpkg/cmakevars.cpp
+++ b/toolsrc/src/vcpkg/cmakevars.cpp
@@ -174,7 +174,7 @@ namespace vcpkg::CMakeVars
         const auto ec_data = System::cmd_execute_and_capture_output(cmd_launch_cmake);
         Checks::check_exit(VCPKG_LINE_INFO, ec_data.exit_code == 0, ec_data.output);
 
-        const std::vector<std::string> lines = Strings::split(ec_data.output, "\n");
+        const std::vector<std::string> lines = Strings::split(ec_data.output, '\n');
 
         const auto end = lines.cend();
 
@@ -192,7 +192,7 @@ namespace vcpkg::CMakeVars
                 {
                     const std::string& line = *block_start;
 
-                    std::vector<std::string> s = Strings::split(line, "=");
+                    std::vector<std::string> s = Strings::split(line, '=');
                     Checks::check_exit(VCPKG_LINE_INFO,
                                        s.size() == 1 || s.size() == 2,
                                        "Expected format is [VARIABLE_NAME=VARIABLE_VALUE], but was [%s]",
diff --git a/toolsrc/src/vcpkg/commands.autocomplete.cpp b/toolsrc/src/vcpkg/commands.autocomplete.cpp
index 2857988ba..6b09e36f3 100644
--- a/toolsrc/src/vcpkg/commands.autocomplete.cpp
+++ b/toolsrc/src/vcpkg/commands.autocomplete.cpp
@@ -30,7 +30,7 @@ namespace vcpkg::Commands::Autocomplete
     {
         Metrics::g_metrics.lock()->set_send_metrics(false);
         const std::string to_autocomplete = Strings::join(" ", args.command_arguments);
-        const std::vector<std::string> tokens = Strings::split(to_autocomplete, " ");
+        const std::vector<std::string> tokens = Strings::split(to_autocomplete, ' ');
 
         std::smatch match;
 
diff --git a/toolsrc/src/vcpkg/commands.ci.cpp b/toolsrc/src/vcpkg/commands.ci.cpp
index eadfd8bdd..1e06be37d 100644
--- a/toolsrc/src/vcpkg/commands.ci.cpp
+++ b/toolsrc/src/vcpkg/commands.ci.cpp
@@ -407,7 +407,7 @@ namespace vcpkg::Commands::CI
         auto it_exclusions = options.settings.find(OPTION_EXCLUDE);
         if (it_exclusions != options.settings.end())
         {
-            auto exclusions = Strings::split(it_exclusions->second, ",");
+            auto exclusions = Strings::split(it_exclusions->second, ',');
             exclusions_set.insert(exclusions.begin(), exclusions.end());
         }
 
diff --git a/toolsrc/src/vcpkg/commands.porthistory.cpp b/toolsrc/src/vcpkg/commands.porthistory.cpp
index 44fccbb7f..e60e7dc63 100644
--- a/toolsrc/src/vcpkg/commands.porthistory.cpp
+++ b/toolsrc/src/vcpkg/commands.porthistory.cpp
@@ -48,8 +48,8 @@ namespace vcpkg::Commands::PortHistory
         auto output = run_git_command(paths, cmd);
 
         auto commits = Util::fmap(
-            Strings::split(output.output, "\n"), [](const std::string& line) -> auto {
-                auto parts = Strings::split(line, " ");
+            Strings::split(output.output, '\n'), [](const std::string& line) -> auto {
+                auto parts = Strings::split(line, ' ');
                 return std::make_pair(parts[0], parts[1]);
             });
 
diff --git a/toolsrc/src/vcpkg/logicexpression.cpp b/toolsrc/src/vcpkg/logicexpression.cpp
index c4365194d..06d949e06 100644
--- a/toolsrc/src/vcpkg/logicexpression.cpp
+++ b/toolsrc/src/vcpkg/logicexpression.cpp
@@ -55,7 +55,7 @@ namespace vcpkg
                 auto override_vars = evaluation_context.cmake_context.find("VCPKG_DEP_INFO_OVERRIDE_VARS");
                 if (override_vars != evaluation_context.cmake_context.end())
                 {
-                    auto cmake_list = Strings::split(override_vars->second, ";");
+                    auto cmake_list = Strings::split(override_vars->second, ';');
                     for (auto& override_id : cmake_list)
                     {
                         if (!override_id.empty())