ICU stuff

.gitignore

@@ -73,10 +73,12 @@
 /lttoolbox/lt-expand
 /python/Makefile
 /python/Makefile.in
+/python/lttoolbox.i
 /python/lttoolbox_wrap.cpp
 /python/lttoolbox.py
 /python/setup.py
 /python/build*
 *.egg-info/
 *.egg
 **/.mypy_cache/
+*~

.travis.yml

@@ -6,6 +6,11 @@ compiler:
   - clang
   - gcc
 
+addons:
+  homebrew:
+    packages:
+	- icu4c
+
 before_install:
   - if [ $TRAVIS_OS_NAME = linux ]; then sudo apt-get install -y swig; else brew install swig; fi
 script:

CMakeLists.txt

@@ -110,7 +110,7 @@ endif()
 # Unlocked I/O functions
 include(CheckSymbolExists)
 set(CMAKE_REQUIRED_DEFINITIONS -D_POSIX_C_SOURCE=200112 -D_GNU_SOURCE)
-foreach(func fread_unlocked fwrite_unlocked fgetc_unlocked fputc_unlocked fputs_unlocked fgetwc_unlocked fputwc_unlocked fputws_unlocked)
+foreach(func fread_unlocked fwrite_unlocked fgetc_unlocked fputc_unlocked fputs_unlocked)
 	string(TOUPPER ${func} _uc)
 	CHECK_SYMBOL_EXISTS(${func} "stdio.h" HAVE_DECL_${_uc})
 	if(HAVE_DECL_${_uc})

configure.ac

@@ -38,29 +38,8 @@ AC_ARG_ENABLE(profile,
               [CXXFLAGS="-pg -g -Wall"; CFLAGS="-pg -g -Wall"; LDFLAGS="-pg"])
 
 
-PKG_CHECK_MODULES(LTTOOLBOX, [libxml-2.0 >= 2.6.17])
-
-# Check for wide strings
-AC_DEFUN([AC_CXX_WSTRING],[
-  AC_CACHE_CHECK(whether the compiler supports wide strings,
-  ac_cv_cxx_wstring,
-  [AC_LANG_SAVE
-   AC_LANG_CPLUSPLUS
-   AC_COMPILE_IFELSE([AC_LANG_PROGRAM([[#include <string>]],[[
-std::wstring test = L"test";
-   ]])],
-   [ac_cv_cxx_wstring=yes], [ac_cv_cxx_wstring=no])
-   AC_LANG_RESTORE
-  ])
-])
-
-AC_CXX_WSTRING
-
-if test "$ac_cv_cxx_wstring" = no
-then
-  AC_MSG_ERROR([Missing wide string support])
-fi
-
+PKG_CHECK_MODULES(LIBXML, [libxml-2.0 >= 2.6.17])
+PKG_CHECK_MODULES(ICU, [icu-i18n, icu-io, icu-uc])
 
 # Checks for libraries.
 AC_CHECK_LIB(xml2, xmlReaderForFile)
@@ -78,7 +57,7 @@ AC_TYPE_SIZE_T
 AC_FUNC_ERROR_AT_LINE
 
 AC_CHECK_DECLS([fread_unlocked, fwrite_unlocked, fgetc_unlocked, \
-fputc_unlocked, fputs_unlocked, fgetwc_unlocked, fputwc_unlocked, fputws_unlocked, ungetwc_unlocked])
+fputc_unlocked, fputs_unlocked])
 
 AC_CHECK_FUNCS([setlocale strdup getopt_long])
 

lttoolbox/CMakeLists.txt

@@ -57,7 +57,6 @@ if(WIN32)
 		win32/regex.c
 		win32/regex.h
 		win32/unistd.h
-		${CMAKE_SOURCE_DIR}/utf8/utf8_fwrap.h
 		${LIBLTTOOLBOX_SOURCES}
 		)
 	if(NOT VCPKG_TOOLCHAIN)

lttoolbox/Makefile.am

@@ -1,15 +1,15 @@
 
 h_sources = alphabet.h att_compiler.h buffer.h compiler.h compression.h  \
-            deserialiser.h entry_token.h expander.h fst_processor.h lt_locale.h \
-            ltstr.h match_exe.h match_node.h match_state.h my_stdio.h node.h \
+            deserialiser.h entry_token.h expander.h fst_processor.h input_file.h lt_locale.h \
+            match_exe.h match_node.h match_state.h my_stdio.h node.h \
             pattern_list.h regexp_compiler.h serialiser.h sorted_vector.h state.h \
             transducer.h trans_exe.h xml_parse_util.h exception.h tmx_compiler.h \
-            string_to_wostream.h
+            ustring.h
 cc_sources = alphabet.cc att_compiler.cc compiler.cc compression.cc entry_token.cc \
-             expander.cc fst_processor.cc lt_locale.cc match_exe.cc \
+             expander.cc fst_processor.cc input_file.cc lt_locale.cc match_exe.cc \
              match_node.cc match_state.cc node.cc pattern_list.cc \
              regexp_compiler.cc sorted_vector.cc state.cc transducer.cc \
-             trans_exe.cc xml_parse_util.cc tmx_compiler.cc
+             trans_exe.cc xml_parse_util.cc tmx_compiler.cc ustring.cc
 
 library_includedir = $(includedir)/$(PACKAGE_NAME)-$(VERSION_API)/$(PACKAGE_NAME)
 library_include_HEADERS = $(h_sources)
@@ -27,33 +27,16 @@ lttoolboxlib = $(prefix)/lib
 
 lttoolbox_DATA = dix.dtd dix.rng dix.rnc acx.rng xsd/dix.xsd xsd/acx.xsd
 
-lt_print_SOURCES = lt_print.cc
-lt_print_LDADD = liblttoolbox$(VERSION_MAJOR).la
-lt_print_LDFLAGS = -llttoolbox$(VERSION_MAJOR) $(LTTOOLBOX_LIBS)
+LDADD = liblttoolbox$(VERSION_MAJOR).la
+AM_LDFLAGS = -llttoolbox$(VERSION_MAJOR) $(LIBXML_LIBS) $(ICU_LIBS)
 
+lt_print_SOURCES = lt_print.cc
 lt_trim_SOURCES = lt_trim.cc
-lt_trim_LDADD = liblttoolbox$(VERSION_MAJOR).la
-lt_trim_LDFLAGS = -llttoolbox$(VERSION_MAJOR) $(LTTOOLBOX_LIBS)
-
 lt_comp_SOURCES = lt_comp.cc
-lt_comp_LDADD = liblttoolbox$(VERSION_MAJOR).la
-lt_comp_LDFLAGS = -llttoolbox$(VERSION_MAJOR) $(LTTOOLBOX_LIBS)
-
 lt_proc_SOURCES = lt_proc.cc
-lt_proc_LDADD = liblttoolbox$(VERSION_MAJOR).la
-lt_proc_LDFLAGS = -llttoolbox$(VERSION_MAJOR) $(LTTOOLBOX_LIBS)
-
 lt_expand_SOURCES = lt_expand.cc
-lt_expand_LDADD = liblttoolbox$(VERSION_MAJOR).la
-lt_expand_LDFLAGS = -llttoolbox$(VERSION_MAJOR) $(LTTOOLBOX_LIBS)
-
 lt_tmxcomp_SOURCES = lt_tmxcomp.cc
-lt_tmxcomp_LDADD = liblttoolbox$(VERSION_MAJOR).la
-lt_tmxcomp_LDFLAGS = -llttoolbox$(VERSION_MAJOR) $(LTTOOLBOX_LIBS)
-
 lt_tmxproc_SOURCES = lt_tmxproc.cc
-lt_tmxproc_LDADD = liblttoolbox$(VERSION_MAJOR).la
-lt_tmxproc_LDFLAGS = -llttoolbox$(VERSION_MAJOR) $(LTTOOLBOX_LIBS)
 
 #lt-validate-dictionary: Makefile.am validate-header.sh
 #	@echo "Creating lt-validate-dictionary script"
@@ -67,10 +50,7 @@ lt_tmxproc_LDFLAGS = -llttoolbox$(VERSION_MAJOR) $(LTTOOLBOX_LIBS)
 
 man_MANS = lt-comp.1 lt-expand.1 lt-proc.1 lt-tmxcomp.1 lt-tmxproc.1 lt-print.1 lt-trim.1
 
-INCLUDES = -I$(top_srcdir) $(LTTOOLBOX_CFLAGS)
-if WINDOWS
-  INCLUDES += -I$(top_srcdir)/utf8
-endif
+INCLUDES = -I$(top_srcdir) -I$(top_srcdir)/utf8 $(LIBXML_CFLAGS) $(ICU_CFLAGS)
 CLEANFILES = *~
 
 EXTRA_DIST = dix.dtd dix.rng dix.rnc acx.rng xsd/dix.xsd xsd/acx.xsd $(man_MANS)