Update digit-value to account for characters in the non-ASCII range

Signed-off-by: yamacir-kit <[email protected]>

.gitignore

@@ -1,3 +1,4 @@
 build
+include/meevax/kernel/digit_value.hpp
 src/kernel/basis.cpp
 src/kernel/version.cpp

CMakeLists.txt

@@ -47,41 +47,25 @@ else()
  set(${PROJECT_NAME}_BYTE_ORDER "little-endian")
 endif()
 
-set(${PROJECT_NAME}_RESOURCES ${CMAKE_CURRENT_SOURCE_DIR}/configure/UnicodeData.txt)
-set(${PROJECT_NAME}_OBJECTS "")
-foreach(EACH IN LISTS ${PROJECT_NAME}_RESOURCES)
- get_filename_component(FILENAME ${EACH} NAME)
- get_filename_component(FILEPATH ${EACH} PATH)
- message(STATUS "Copying resource ${EACH} as ")
- add_custom_command(
- OUTPUT ${CMAKE_CURRENT_BINARY_DIR}/${FILENAME}.o
- WORKING_DIRECTORY ${FILEPATH}
- COMMAND ${CMAKE_OBJCOPY}
- ARGS -I binary
- -O elf64-x86-64
- -B i386
- ${FILENAME}
- ${CMAKE_CURRENT_BINARY_DIR}/${FILENAME}.o)
- list(APPEND ${PROJECT_NAME}_OBJECTS ${CMAKE_CURRENT_BINARY_DIR}/${FILENAME}.o)
-endforeach()
+execute_process(
+ COMMAND ${CMAKE_CURRENT_SOURCE_DIR}/script/unicode.sh --digit-value
+ OUTPUT_VARIABLE ${PROJECT_NAME}_DIGIT_VALUE)
 
+file(GLOB ${PROJECT_NAME}_BASIS_SOURCES ${CMAKE_CURRENT_SOURCE_DIR}/basis/*.ss)
 file(READ ${CMAKE_CURRENT_SOURCE_DIR}/configure/help.txt ${PROJECT_NAME}_HELP_UNCONFIGURED)
+file(WRITE ${CMAKE_CURRENT_SOURCE_DIR}/include/${PROJECT_NAME}/kernel/digit_value.hpp ${${PROJECT_NAME}_DIGIT_VALUE})
 
 string(CONFIGURE ${${PROJECT_NAME}_HELP_UNCONFIGURED} ${PROJECT_NAME}_HELP)
-
 string(TOLOWER ${CMAKE_SYSTEM_NAME} ${PROJECT_NAME}_SYSTEM_NAME)
 
-configure_file(${CMAKE_CURRENT_SOURCE_DIR}/configure/version.cpp ${CMAKE_CURRENT_SOURCE_DIR}/src/kernel/version.cpp)
-configure_file(${CMAKE_CURRENT_SOURCE_DIR}/configure/README.md ${CMAKE_CURRENT_SOURCE_DIR}/README.md)
-
-file(GLOB ${PROJECT_NAME}_BASIS_SOURCES ${CMAKE_CURRENT_SOURCE_DIR}/basis/*.ss)
-
 foreach(EACH IN LISTS ${PROJECT_NAME}_BASIS_SOURCES)
  get_filename_component(FILENAME ${EACH} NAME)
  file(READ ${EACH} ${PROJECT_NAME}_BASIS_${FILENAME})
 endforeach()
 
-configure_file(${CMAKE_CURRENT_SOURCE_DIR}/configure/basis.cpp ${CMAKE_CURRENT_SOURCE_DIR}/src/kernel/basis.cpp)
+configure_file(${CMAKE_CURRENT_SOURCE_DIR}/configure/README.md ${CMAKE_CURRENT_SOURCE_DIR}/README.md)
+configure_file(${CMAKE_CURRENT_SOURCE_DIR}/configure/basis.cpp ${CMAKE_CURRENT_SOURCE_DIR}/src/kernel/basis.cpp)
+configure_file(${CMAKE_CURRENT_SOURCE_DIR}/configure/version.cpp ${CMAKE_CURRENT_SOURCE_DIR}/src/kernel/version.cpp)
 
 # ---- Target kernel -----------------------------------------------------------
 

README.md

@@ -74,7 +74,7 @@ Subset of R7RS-small.
 cmake -B build -DCMAKE_BUILD_TYPE=Release
 cd build
 make package
-sudo apt install build/meevax_0.4.796_amd64.deb
+sudo apt install build/meevax_0.4.797_amd64.deb
 ```
 
 or
@@ -106,15 +106,15 @@ sudo rm -rf /usr/local/share/meevax
 
 | Target Name | Description
 |-------------|-------------
-| `all` | Build shared-library `libmeevax.0.4.796.so` and executable `meevax`
+| `all` | Build shared-library `libmeevax.0.4.797.so` and executable `meevax`
 | `test` | Test executable `meevax`
-| `package` | Generate debian package `meevax_0.4.796_amd64.deb`
+| `package` | Generate debian package `meevax_0.4.797_amd64.deb`
 | `install` | Copy files into `/usr/local` directly
 
 ## Usage
 
 ```
-Meevax Lisp 0.4.796
+Meevax Lisp 0.4.797
 
 Usage:
  meevax [option...] [file...]

VERSION

@@ -1 +1 @@
-0.4.796
+0.4.797

include/meevax/kernel/character.hpp

@@ -47,11 +47,13 @@ inline namespace kernel
  return std::char_traits<char_type>::eq_int_type(c1, c2);
  }
 
- inline constexpr auto eq(int_type const& c) const
+ constexpr auto eq(int_type const& c) const
  {
  return std::char_traits<char_type>::eq_int_type(codepoint, c);
  }
 
+ auto digit_value() const -> object const&;
+
  static constexpr auto is_ascii(int_type c)
  {
  return 0x00 <= c and c <= 0x7F;
@@ -62,7 +64,7 @@ inline namespace kernel
  return eq(eof(), c);
  }
 
- inline constexpr operator int_type() const
+ constexpr operator int_type() const
  {
  return codepoint;
  }

script/unicode.sh

@@ -0,0 +1,38 @@
+#!/bin/sh -e
+
+# https://www.unicode.org/L2/L1999/UnicodeData.html
+#
+# \1 = Code value
+# \2 = Character name
+# \3 = General category
+# \4 = Canonical combining classes
+# \5 = Bidirectional category
+# \6 = Character decomposition mapping
+# \7 = Decimal digit value
+# \8 = Digit value
+# \9 = Numeric value
+# \10 = Mirrored
+# \11 = Unicode 1.0 Name
+# \12 = 10646 comment field
+# \13 = Uppercase mapping
+# \14 = Lowercase mapping
+# \15 = Titlecase mapping
+
+unicode_data='^([^;]*);([^;]*);([^;]*);([^;]*);([^;]*);([^;]*);([^;]*);([^;]*);([^;]*);([^;]*);([^;]*);([^;]*);([^;]*);([^;]*);(.*)$'
+
+substitute()
+{
+ sed -E "s/$unicode_data/$1/g" "$(git rev-parse --show-toplevel)/configure/UnicodeData.txt"
+}
+
+digit_value()
+{
+ substitute '{ 0x\1, make_number("\9") },' | grep -e '{ .\+, make_number(".\+") },'
+}
+
+for each in "$@"
+do
+ case "$each" in
+ --digit-value ) digit_value ;;
+ esac
+done

src/kernel/boot.cpp

@@ -72,14 +72,7 @@ inline namespace kernel
 
  library.define<function>("digit-value", [](let const& xs)
  {
- if (auto c = xs[0].as<character>(); std::isdigit(c.codepoint))
- {
- return make<exact_integer>(c.codepoint - '0');
- }
- else
- {
- return f;
- }
+ return xs[0].as<character>().digit_value();
  });
 
  library.define<function>("integer->char", [](let const& xs)

src/kernel/character.cpp

@@ -14,13 +14,31 @@
  limitations under the License.
 */
 
+#include <meevax/kernel/boolean.hpp>
 #include <meevax/kernel/eof.hpp>
 #include <meevax/kernel/error.hpp>
+#include <meevax/kernel/number.hpp>
 
 namespace meevax
 {
 inline namespace kernel
 {
+ auto character::digit_value() const -> object const&
+ {
+ static const std::unordered_map<character::int_type, object> table {
+ #include <meevax/kernel/digit_value.hpp>
+ };
+
+ if (auto iterator = table.find(codepoint); iterator != table.end())
+ {
+ return iterator->second;
+ }
+ else
+ {
+ return f;
+ }
+ }
+
  character::operator std::string() const
  {
  auto chars = std::array<char, 5>();

test/r7rs.ss

@@ -1299,11 +1299,11 @@
 
 (check (digit-value #\3) => 3)
 
-; (check (digit-value #\x0664) => 4)
+(check (digit-value #\x0664) => 4)
 
-; (check (digit-value #\x0AE6) => 0)
+(check (digit-value #\x0AE6) => 0)
 
-; (check (digit-value #\x0EA6) => #f) ; BUG: MEMORY-LEAK
+(check (digit-value #\x0EA6) => #f)
 
 ; ---- 6.7. --------------------------------------------------------------------
 
@@ -1619,4 +1619,4 @@
 
 (check-report)
 
-(exit (check-passed? 429))
+(exit (check-passed? 432))