Implement a to_title_case text utility function

.github/workflows/website-build.yml

@@ -23,6 +23,7 @@ jobs:
           -DSOURCEMETA_CORE_LANG_NUMERIC:BOOL=OFF
           -DSOURCEMETA_CORE_LANG_ERROR:BOOL=OFF
           -DSOURCEMETA_CORE_LANG_OPTIONS:BOOL=OFF
+          -DSOURCEMETA_CORE_LANG_TEXT:BOOL=OFF
           -DSOURCEMETA_CORE_UNICODE:BOOL=OFF
           -DSOURCEMETA_CORE_PUNYCODE:BOOL=OFF
           -DSOURCEMETA_CORE_TIME:BOOL=OFF

.github/workflows/website-deploy.yml

@@ -33,6 +33,7 @@ jobs:
           -DSOURCEMETA_CORE_LANG_NUMERIC:BOOL=OFF
           -DSOURCEMETA_CORE_LANG_ERROR:BOOL=OFF
           -DSOURCEMETA_CORE_LANG_OPTIONS:BOOL=OFF
+          -DSOURCEMETA_CORE_LANG_TEXT:BOOL=OFF
           -DSOURCEMETA_CORE_UNICODE:BOOL=OFF
           -DSOURCEMETA_CORE_PUNYCODE:BOOL=OFF
           -DSOURCEMETA_CORE_TIME:BOOL=OFF

CMakeLists.txt

@@ -10,6 +10,7 @@ option(SOURCEMETA_CORE_LANG_PARALLEL "Build the Sourcemeta Core language paralle
 option(SOURCEMETA_CORE_LANG_NUMERIC "Build the Sourcemeta Core language numeric library" ON)
 option(SOURCEMETA_CORE_LANG_ERROR "Build the Sourcemeta Core language error library" ON)
 option(SOURCEMETA_CORE_LANG_OPTIONS "Build the Sourcemeta Core Options library" ON)
+option(SOURCEMETA_CORE_LANG_TEXT "Build the Sourcemeta Core language text library" ON)
 option(SOURCEMETA_CORE_UNICODE "Build the Sourcemeta Core Unicode library" ON)
 option(SOURCEMETA_CORE_PUNYCODE "Build the Sourcemeta Core Punycode library" ON)
 option(SOURCEMETA_CORE_TIME "Build the Sourcemeta Core time library" ON)
@@ -93,6 +94,10 @@ if(SOURCEMETA_CORE_LANG_OPTIONS)
   add_subdirectory(src/lang/options)
 endif()
 
+if(SOURCEMETA_CORE_LANG_TEXT)
+  add_subdirectory(src/lang/text)
+endif()
+
 if(SOURCEMETA_CORE_UNICODE)
   add_subdirectory(src/core/unicode)
 endif()
@@ -231,6 +236,10 @@ if(SOURCEMETA_CORE_TESTS)
     add_subdirectory(test/options)
   endif()
 
+  if(SOURCEMETA_CORE_LANG_TEXT)
+    add_subdirectory(test/text)
+  endif()
+
   if(SOURCEMETA_CORE_UNICODE)
     add_subdirectory(test/unicode)
   endif()

config.cmake.in

@@ -30,6 +30,7 @@ if(NOT SOURCEMETA_CORE_COMPONENTS)
   list(APPEND SOURCEMETA_CORE_COMPONENTS markdown)
   list(APPEND SOURCEMETA_CORE_COMPONENTS error)
   list(APPEND SOURCEMETA_CORE_COMPONENTS options)
+  list(APPEND SOURCEMETA_CORE_COMPONENTS text)
 endif()
 
 include(CMakeFindDependencyMacro)
@@ -136,6 +137,8 @@ foreach(component ${SOURCEMETA_CORE_COMPONENTS})
     include("${CMAKE_CURRENT_LIST_DIR}/sourcemeta_core_error.cmake")
   elseif(component STREQUAL "options")
     include("${CMAKE_CURRENT_LIST_DIR}/sourcemeta_core_options.cmake")
+  elseif(component STREQUAL "text")
+    include("${CMAKE_CURRENT_LIST_DIR}/sourcemeta_core_text.cmake")
   else()
     message(FATAL_ERROR "Unknown Sourcemeta Core component: ${component}")
   endif()

src/lang/text/CMakeLists.txt

@@ -0,0 +1,6 @@
+sourcemeta_library(NAMESPACE sourcemeta PROJECT core NAME text
+  SOURCES text.cc)
+
+if(SOURCEMETA_CORE_INSTALL)
+  sourcemeta_library_install(NAMESPACE sourcemeta PROJECT core NAME text)
+endif()

src/lang/text/include/sourcemeta/core/text.h

@@ -0,0 +1,39 @@
+#ifndef SOURCEMETA_CORE_TEXT_H_
+#define SOURCEMETA_CORE_TEXT_H_
+
+#ifndef SOURCEMETA_CORE_TEXT_EXPORT
+#include <sourcemeta/core/text_export.h>
+#endif
+
+#include <string> // std::string
+
+/// @defgroup text Text
+/// @brief A collection of general-purpose text manipulation utilities
+///
+/// This functionality is included as follows:
+///
+/// ```cpp
+/// #include <sourcemeta/core/text.h>
+/// ```
+
+namespace sourcemeta::core {
+
+/// @ingroup text
+///
+/// Convert a string to Title Case in place. For example:
+///
+/// ```cpp
+/// #include <sourcemeta/core/text.h>
+/// #include <cassert>
+/// #include <string>
+///
+/// std::string value{"hello_world"};
+/// sourcemeta::core::to_title_case(value);
+/// assert(value == "Hello World");
+/// ```
+SOURCEMETA_CORE_TEXT_EXPORT
+auto to_title_case(std::string &value) -> void;
+
+} // namespace sourcemeta::core
+
+#endif

src/lang/text/text.cc

@@ -0,0 +1,37 @@
+#include <sourcemeta/core/text.h>
+
+#include <cctype>  // std::isalpha, std::toupper
+#include <cstddef> // std::size_t
+
+namespace sourcemeta::core {
+
+auto to_title_case(std::string &value) -> void {
+  std::size_t write{0};
+  bool capitalize_next{true};
+  bool pending_separator{false};
+  for (const char character : value) {
+    if (character == '_' || character == '-') {
+      if (write > 0) {
+        pending_separator = true;
+      }
+    } else {
+      if (pending_separator) {
+        value[write++] = ' ';
+        pending_separator = false;
+        capitalize_next = true;
+      }
+      if (capitalize_next) {
+        value[write++] = static_cast<char>(
+            std::toupper(static_cast<unsigned char>(character)));
+        if (std::isalpha(static_cast<unsigned char>(character))) {
+          capitalize_next = false;
+        }
+      } else {
+        value[write++] = character;
+      }
+    }
+  }
+  value.resize(write);
+}
+
+} // namespace sourcemeta::core

test/packaging/find_package/CMakeLists.txt

@@ -25,3 +25,4 @@ target_link_libraries(core_hello PRIVATE sourcemeta::core::markdown)
 target_link_libraries(core_hello PRIVATE sourcemeta::core::options)
 target_link_libraries(core_hello PRIVATE sourcemeta::core::preprocessor)
 target_link_libraries(core_hello PRIVATE sourcemeta::core::jsonrpc)
+target_link_libraries(core_hello PRIVATE sourcemeta::core::text)

test/packaging/find_package/hello.cc

@@ -13,6 +13,7 @@
 #include <sourcemeta/core/parallel.h>
 #include <sourcemeta/core/preprocessor.h>
 #include <sourcemeta/core/punycode.h>
+#include <sourcemeta/core/text.h>
 #include <sourcemeta/core/time.h>
 #include <sourcemeta/core/unicode.h>
 #include <sourcemeta/core/uri.h>

test/text/CMakeLists.txt

@@ -0,0 +1,5 @@
+sourcemeta_googletest(NAMESPACE sourcemeta PROJECT core NAME text
+  SOURCES text_to_title_case_test.cc)
+
+target_link_libraries(sourcemeta_core_text_unit
+  PRIVATE sourcemeta::core::text)

test/text/text_to_title_case_test.cc

@@ -0,0 +1,179 @@
+#include <gtest/gtest.h>
+
+#include <sourcemeta/core/text.h>
+
+#include <string> // std::string
+
+TEST(Text_to_title_case, empty_string) {
+  std::string value{""};
+  sourcemeta::core::to_title_case(value);
+  EXPECT_EQ(value, "");
+}
+
+TEST(Text_to_title_case, single_lowercase_character) {
+  std::string value{"a"};
+  sourcemeta::core::to_title_case(value);
+  EXPECT_EQ(value, "A");
+}
+
+TEST(Text_to_title_case, single_uppercase_character) {
+  std::string value{"A"};
+  sourcemeta::core::to_title_case(value);
+  EXPECT_EQ(value, "A");
+}
+
+TEST(Text_to_title_case, single_underscore_is_empty) {
+  std::string value{"_"};
+  sourcemeta::core::to_title_case(value);
+  EXPECT_EQ(value, "");
+}
+
+TEST(Text_to_title_case, single_dash_is_empty) {
+  std::string value{"-"};
+  sourcemeta::core::to_title_case(value);
+  EXPECT_EQ(value, "");
+}
+
+TEST(Text_to_title_case, only_separators_is_empty) {
+  std::string value{"___"};
+  sourcemeta::core::to_title_case(value);
+  EXPECT_EQ(value, "");
+}
+
+TEST(Text_to_title_case, single_lowercase_word) {
+  std::string value{"hello"};
+  sourcemeta::core::to_title_case(value);
+  EXPECT_EQ(value, "Hello");
+}
+
+TEST(Text_to_title_case, already_title_cased_word) {
+  std::string value{"Hello"};
+  sourcemeta::core::to_title_case(value);
+  EXPECT_EQ(value, "Hello");
+}
+
+TEST(Text_to_title_case, all_uppercase_word) {
+  std::string value{"HELLO"};
+  sourcemeta::core::to_title_case(value);
+  EXPECT_EQ(value, "HELLO");
+}
+
+TEST(Text_to_title_case, snake_case_two_words) {
+  std::string value{"hello_world"};
+  sourcemeta::core::to_title_case(value);
+  EXPECT_EQ(value, "Hello World");
+}
+
+TEST(Text_to_title_case, kebab_case_two_words) {
+  std::string value{"hello-world"};
+  sourcemeta::core::to_title_case(value);
+  EXPECT_EQ(value, "Hello World");
+}
+
+TEST(Text_to_title_case, mixed_snake_and_kebab_separators) {
+  std::string value{"hello_world-test"};
+  sourcemeta::core::to_title_case(value);
+  EXPECT_EQ(value, "Hello World Test");
+}
+
+TEST(Text_to_title_case, snake_case_three_words) {
+  std::string value{"abc_def_ghi"};
+  sourcemeta::core::to_title_case(value);
+  EXPECT_EQ(value, "Abc Def Ghi");
+}
+
+TEST(Text_to_title_case, preserves_existing_uppercase_after_separator) {
+  std::string value{"Hello_World"};
+  sourcemeta::core::to_title_case(value);
+  EXPECT_EQ(value, "Hello World");
+}
+
+TEST(Text_to_title_case, leading_underscore_is_stripped) {
+  std::string value{"_hello"};
+  sourcemeta::core::to_title_case(value);
+  EXPECT_EQ(value, "Hello");
+}
+
+TEST(Text_to_title_case, leading_dash_is_stripped) {
+  std::string value{"-hello"};
+  sourcemeta::core::to_title_case(value);
+  EXPECT_EQ(value, "Hello");
+}
+
+TEST(Text_to_title_case, trailing_underscore_is_stripped) {
+  std::string value{"hello_"};
+  sourcemeta::core::to_title_case(value);
+  EXPECT_EQ(value, "Hello");
+}
+
+TEST(Text_to_title_case, trailing_dash_is_stripped) {
+  std::string value{"hello-"};
+  sourcemeta::core::to_title_case(value);
+  EXPECT_EQ(value, "Hello");
+}
+
+TEST(Text_to_title_case, separators_around_word_are_stripped) {
+  std::string value{"_hello_"};
+  sourcemeta::core::to_title_case(value);
+  EXPECT_EQ(value, "Hello");
+}
+
+TEST(Text_to_title_case, multiple_leading_separators_are_stripped) {
+  std::string value{"__hello"};
+  sourcemeta::core::to_title_case(value);
+  EXPECT_EQ(value, "Hello");
+}
+
+TEST(Text_to_title_case, multiple_trailing_separators_are_stripped) {
+  std::string value{"hello__"};
+  sourcemeta::core::to_title_case(value);
+  EXPECT_EQ(value, "Hello");
+}
+
+TEST(Text_to_title_case, consecutive_separators_collapse_to_single_space) {
+  std::string value{"hello__world"};
+  sourcemeta::core::to_title_case(value);
+  EXPECT_EQ(value, "Hello World");
+}
+
+TEST(Text_to_title_case, mixed_consecutive_separators_collapse) {
+  std::string value{"hello_-world"};
+  sourcemeta::core::to_title_case(value);
+  EXPECT_EQ(value, "Hello World");
+}
+
+TEST(Text_to_title_case, single_letter_words) {
+  std::string value{"a_b_c"};
+  sourcemeta::core::to_title_case(value);
+  EXPECT_EQ(value, "A B C");
+}
+
+TEST(Text_to_title_case, digits_pass_through) {
+  std::string value{"abc123def"};
+  sourcemeta::core::to_title_case(value);
+  EXPECT_EQ(value, "Abc123def");
+}
+
+TEST(Text_to_title_case, digit_after_separator) {
+  std::string value{"abc_123"};
+  sourcemeta::core::to_title_case(value);
+  EXPECT_EQ(value, "Abc 123");
+}
+
+TEST(Text_to_title_case, letter_after_leading_digits_in_segment) {
+  std::string value{"abc_123def"};
+  sourcemeta::core::to_title_case(value);
+  EXPECT_EQ(value, "Abc 123Def");
+}
+
+TEST(Text_to_title_case, leading_digits_then_letter_at_start) {
+  std::string value{"123abc"};
+  sourcemeta::core::to_title_case(value);
+  EXPECT_EQ(value, "123Abc");
+}
+
+TEST(Text_to_title_case, space_in_input_is_not_a_separator) {
+  std::string value{"hello world"};
+  sourcemeta::core::to_title_case(value);
+  EXPECT_EQ(value, "Hello world");
+}