diff options
| author | Boris Kolpackov <boris@codesynthesis.com> | 2010-01-07 13:50:11 +0200 |
|---|---|---|
| committer | Boris Kolpackov <boris@codesynthesis.com> | 2010-01-07 13:50:11 +0200 |
| commit | 76d23e639004517db8f9469d64ac1789f8449365 (patch) | |
| tree | dbafc8c4e31a97f74046c98af19d0fe76f360001 | |
| parent | c30caae30bc64974eeaa1e81aa2abdc203f5120d (diff) | |
Add support for ISO-8859-1 as application encoding
New runtime configuration parameter, XSDE_ENCODING. New option,
--char-encoding. New test, tests/cxx/hybrid/iso8859-1.
133 files changed, 2744 insertions, 318 deletions
@@ -7,6 +7,17 @@ Version 3.2.0 * Automatic mapping for the urn-style XML namespaces. The last component in the urn name is used to derive the C++ namespace name. + * Support for ISO-8859-1 in addition to UTF-8 as application encoding. + Note that this encoding is not the same as the XML document encoding + that is being parsed or serialized. Rather, it is the encoding that + is used inside the application. When an XML document is parsed the + character data is automatically converted to the application encoding. + Similarly, when an XML document is serialized, the data in the + application encoding is automatically converted to the resulting + document encoding. To select a particular encoding, configure the + XSD/e runtime library accordingly and pass the --char-encoding option + to the XSD/e compiler when translating your schemas. + C++/Hybrid * String-based types that use XML Schema restriction by enumeration are diff --git a/build/configuration.make b/build/configuration.make index 05318c9..49624e5 100644 --- a/build/configuration.make +++ b/build/configuration.make @@ -10,6 +10,7 @@ $(call include-once,$(scf_root)/configuration-rules.make,$(dcf_root)) # xsde_arch_width := xsde_byteorder := +xsde_encoding := xsde_stl := xsde_stl_iterator := xsde_iostream := @@ -35,6 +36,7 @@ ifdef xsde_stl $(out_root)/%: xsde_arch_width := $(xsde_arch_width) $(out_root)/%: xsde_byteorder := $(xsde_byteorder) +$(out_root)/%: xsde_encoding := $(xsde_encoding) $(out_root)/%: xsde_stl := $(xsde_stl) $(out_root)/%: xsde_stl_iterator := $(xsde_stl_iterator) $(out_root)/%: xsde_iostream := $(xsde_iostream) diff --git a/build/configure b/build/configure index e49a7c5..8295e76 100755 --- a/build/configure +++ b/build/configure @@ -39,6 +39,15 @@ else fi $echo +$echo "Please select the application character encoding:" +$echo +$echo "(1) UTF-8" +$echo "(2) ISO-8859-1" +$echo + +encoding=`read_option "utf8 iso8859-1" "utf8"` + +$echo $echo "Would you like the generated code and runtime to use STL?" $echo @@ -166,6 +175,7 @@ fi echo "xsde_arch_width := $arch_width" >$1 echo "xsde_byteorder := $byteorder" >>$1 +echo "xsde_encoding := $encoding" >>$1 echo "xsde_stl := $stl" >>$1 echo "xsde_stl_iterator := $stl_iter" >>$1 echo "xsde_iostream := $iostream" >>$1 @@ -208,6 +208,8 @@ if [ "$lib" != "" ]; then gen fi +gen 'EXTRA_XSDFLAGS += --char-encoding $(XSDE_ENCODING)' +gen gen 'ifeq ($(XSDE_STL),n)' gen 'EXTRA_XSDFLAGS += --no-stl' gen 'endif' @@ -357,6 +359,8 @@ if [ "$lib" != "" ]; then gen fi +gen 'EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding $(XSDE_ENCODING)' +gen gen '!if "$(XSDE_STL)" == "n"' gen 'EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-stl' gen '!endif' diff --git a/build/xsde/hybrid/xsd-cxx.make b/build/xsde/hybrid/xsd-cxx.make index c026291..540fa1f 100644 --- a/build/xsde/hybrid/xsd-cxx.make +++ b/build/xsde/hybrid/xsd-cxx.make @@ -56,6 +56,7 @@ ops := \ --sskel-file-suffix $(xsd_sskel_suffix) \ --simpl-file-suffix $(xsd_simpl_suffix) +ops += --char-encoding $(xsde_encoding) ifeq ($(xsde_stl),n) ops += --no-stl diff --git a/build/xsde/parser/xsd-cxx.make b/build/xsde/parser/xsd-cxx.make index cd657d6..f9c87e9 100644 --- a/build/xsde/parser/xsd-cxx.make +++ b/build/xsde/parser/xsd-cxx.make @@ -34,6 +34,8 @@ ifneq ($(xsd_pimpl_suffix),) ops += --impl-file-suffix $(xsd_pimpl_suffix) endif +ops += --char-encoding $(xsde_encoding) + ifeq ($(xsde_stl),n) ops += --no-stl endif diff --git a/build/xsde/serializer/xsd-cxx.make b/build/xsde/serializer/xsd-cxx.make index 2f57782..a042ef9 100644 --- a/build/xsde/serializer/xsd-cxx.make +++ b/build/xsde/serializer/xsd-cxx.make @@ -34,6 +34,8 @@ ifneq ($(xsd_pimpl_suffix),) ops += --impl-file-suffix $(xsd_simpl_suffix) endif +ops += --char-encoding $(xsde_encoding) + ifeq ($(xsde_stl),n) ops += --no-stl endif diff --git a/dist/build/config.make b/dist/build/config.make index 87a39c6..b05df1a 100644 --- a/dist/build/config.make +++ b/dist/build/config.make @@ -14,6 +14,7 @@ include $(root)/config/config.make XSDE_PLATFORM := $(strip $(XSDE_PLATFORM)) XSDE_ARCH_WIDTH := $(strip $(XSDE_ARCH_WIDTH)) XSDE_BYTEORDER := $(strip $(XSDE_BYTEORDER)) +XSDE_ENCODING := $(strip $(XSDE_ENCODING)) XSDE_STL := $(strip $(XSDE_STL)) XSDE_IOSTREAM := $(strip $(XSDE_IOSTREAM)) XSDE_EXCEPTIONS := $(strip $(XSDE_EXCEPTIONS)) diff --git a/dist/config/config.make b/dist/config/config.make index ac3518c..fccb68f 100644 --- a/dist/config/config.make +++ b/dist/config/config.make @@ -37,12 +37,26 @@ XSDE_PLATFORM := posix XSDE_ARCH_WIDTH := 32 -# Platform byte order. Valid values are 'b' for big-endian -# and 'l' for little-endian. +# Platform byte order. Valid values are 'b' for big-endian and 'l' +# for little-endian. # XSDE_BYTEORDER := l +# Application character encoding. Valid values are 'utf8' for UTF-8 +# and 'iso8859-1' for ISO-8859-1. Note that this encoding is not +# the same as the XML document encoding that is being parsed or +# serialized. Rather, it is the encoding that is used inside the +# application. When an XML document is parsed, the character data +# is automatically converted to the application encoding. Similarly, +# when an XML document is serialized, the data in the application +# encoding is automatically converted to the resulting document +# encoding. Also don't forget to use the --char-encoding option +# when compiling your schemas if using an encoding other than UTF-8. +# +XSDE_ENCODING := utf8 + + # Set to 'n' if you don't have STL (std::string, etc.). Also don't # forget to use the --no-stl option when compiling your schemas. # diff --git a/dist/config/config.nmake b/dist/config/config.nmake index 599d660..0e4008f 100644 --- a/dist/config/config.nmake +++ b/dist/config/config.nmake @@ -42,6 +42,20 @@ XSDE_ARCH_WIDTH = 32 XSDE_BYTEORDER = l +# Application character encoding. Valid values are 'utf8' for UTF-8 +# and 'iso8859-1' for ISO-8859-1. Note that this encoding is not +# the same as the XML document encoding that is being parsed or +# serialized. Rather, it is the encoding that is used inside the +# application. When an XML document is parsed, the character data +# is automatically converted to the application encoding. Similarly, +# when an XML document is serialized, the data in the application +# encoding is automatically converted to the resulting document +# encoding. Also don't forget to use the --char-encoding option +# when compiling your schemas if using an encoding other than UTF-8. +# +XSDE_ENCODING = utf8 + + # Set to 'n' if you don't have STL (std::string, etc.). Also don't # forget to use the --no-stl option when compiling your schemas. # diff --git a/dist/etc/evc-4.0/config.nmake b/dist/etc/evc-4.0/config.nmake index cb4777f..136f281 100644 --- a/dist/etc/evc-4.0/config.nmake +++ b/dist/etc/evc-4.0/config.nmake @@ -45,6 +45,20 @@ XSDE_ARCH_WIDTH = 32 XSDE_BYTEORDER = l +# Application character encoding. Valid values are 'utf8' for UTF-8 +# and 'iso8859-1' for ISO-8859-1. Note that this encoding is not +# the same as the XML document encoding that is being parsed or +# serialized. Rather, it is the encoding that is used inside the +# application. When an XML document is parsed, the character data +# is automatically converted to the application encoding. Similarly, +# when an XML document is serialized, the data in the application +# encoding is automatically converted to the resulting document +# encoding. Also don't forget to use the --char-encoding option +# when compiling your schemas if using an encoding other than UTF-8. +# +XSDE_ENCODING = utf8 + + # Set to 'n' if you don't have STL (std::string, etc.). Also don't # forget to use the --no-stl option when compiling your schemas. # diff --git a/dist/etc/iphone/config-device.make b/dist/etc/iphone/config-device.make index a52c484..c6e8634 100644 --- a/dist/etc/iphone/config-device.make +++ b/dist/etc/iphone/config-device.make @@ -50,6 +50,20 @@ XSDE_ARCH_WIDTH := 32 XSDE_BYTEORDER := l +# Application character encoding. Valid values are 'utf8' for UTF-8 +# and 'iso8859-1' for ISO-8859-1. Note that this encoding is not +# the same as the XML document encoding that is being parsed or +# serialized. Rather, it is the encoding that is used inside the +# application. When an XML document is parsed, the character data +# is automatically converted to the application encoding. Similarly, +# when an XML document is serialized, the data in the application +# encoding is automatically converted to the resulting document +# encoding. Also don't forget to use the --char-encoding option +# when compiling your schemas if using an encoding other than UTF-8. +# +XSDE_ENCODING := utf8 + + # Set to 'n' if you don't have STL (std::string, etc.). Also don't # forget to use the --no-stl option when compiling your schemas. # diff --git a/dist/etc/iphone/config-simulator.make b/dist/etc/iphone/config-simulator.make index 3ab24df..afff0a6 100644 --- a/dist/etc/iphone/config-simulator.make +++ b/dist/etc/iphone/config-simulator.make @@ -50,6 +50,20 @@ XSDE_ARCH_WIDTH := 32 XSDE_BYTEORDER := l +# Application character encoding. Valid values are 'utf8' for UTF-8 +# and 'iso8859-1' for ISO-8859-1. Note that this encoding is not +# the same as the XML document encoding that is being parsed or +# serialized. Rather, it is the encoding that is used inside the +# application. When an XML document is parsed, the character data +# is automatically converted to the application encoding. Similarly, +# when an XML document is serialized, the data in the application +# encoding is automatically converted to the resulting document +# encoding. Also don't forget to use the --char-encoding option +# when compiling your schemas if using an encoding other than UTF-8. +# +XSDE_ENCODING := utf8 + + # Set to 'n' if you don't have STL (std::string, etc.). Also don't # forget to use the --no-stl option when compiling your schemas. # diff --git a/dist/etc/lynxos/config-4.2.make b/dist/etc/lynxos/config-4.2.make index beb401e..5be4262 100644 --- a/dist/etc/lynxos/config-4.2.make +++ b/dist/etc/lynxos/config-4.2.make @@ -49,6 +49,20 @@ XSDE_ARCH_WIDTH := 32 XSDE_BYTEORDER := l +# Application character encoding. Valid values are 'utf8' for UTF-8 +# and 'iso8859-1' for ISO-8859-1. Note that this encoding is not +# the same as the XML document encoding that is being parsed or +# serialized. Rather, it is the encoding that is used inside the +# application. When an XML document is parsed, the character data +# is automatically converted to the application encoding. Similarly, +# when an XML document is serialized, the data in the application +# encoding is automatically converted to the resulting document +# encoding. Also don't forget to use the --char-encoding option +# when compiling your schemas if using an encoding other than UTF-8. +# +XSDE_ENCODING := utf8 + + # Set to 'n' if you don't have STL (std::string, etc.). Also don't # forget to use the --no-stl option when compiling your schemas. # diff --git a/dist/etc/lynxos/config-5.0.make b/dist/etc/lynxos/config-5.0.make index 3c64855..25a15b2 100644 --- a/dist/etc/lynxos/config-5.0.make +++ b/dist/etc/lynxos/config-5.0.make @@ -49,6 +49,20 @@ XSDE_ARCH_WIDTH := 32 XSDE_BYTEORDER := l +# Application character encoding. Valid values are 'utf8' for UTF-8 +# and 'iso8859-1' for ISO-8859-1. Note that this encoding is not +# the same as the XML document encoding that is being parsed or +# serialized. Rather, it is the encoding that is used inside the +# application. When an XML document is parsed, the character data +# is automatically converted to the application encoding. Similarly, +# when an XML document is serialized, the data in the application +# encoding is automatically converted to the resulting document +# encoding. Also don't forget to use the --char-encoding option +# when compiling your schemas if using an encoding other than UTF-8. +# +XSDE_ENCODING := utf8 + + # Set to 'n' if you don't have STL (std::string, etc.). Also don't # forget to use the --no-stl option when compiling your schemas. # diff --git a/dist/etc/qnx/config-6.3-gcc-2.95.make b/dist/etc/qnx/config-6.3-gcc-2.95.make index 4a6bb1b..3211450 100644 --- a/dist/etc/qnx/config-6.3-gcc-2.95.make +++ b/dist/etc/qnx/config-6.3-gcc-2.95.make @@ -48,6 +48,20 @@ XSDE_ARCH_WIDTH := 32 XSDE_BYTEORDER := l +# Application character encoding. Valid values are 'utf8' for UTF-8 +# and 'iso8859-1' for ISO-8859-1. Note that this encoding is not +# the same as the XML document encoding that is being parsed or +# serialized. Rather, it is the encoding that is used inside the +# application. When an XML document is parsed, the character data +# is automatically converted to the application encoding. Similarly, +# when an XML document is serialized, the data in the application +# encoding is automatically converted to the resulting document +# encoding. Also don't forget to use the --char-encoding option +# when compiling your schemas if using an encoding other than UTF-8. +# +XSDE_ENCODING := utf8 + + # Set to 'n' if you don't have STL (std::string, etc.). Also don't # forget to use the --no-stl option when compiling your schemas. # diff --git a/dist/etc/qnx/config-6.3-gcc-3.3.make b/dist/etc/qnx/config-6.3-gcc-3.3.make index 770693b..e0a84db 100644 --- a/dist/etc/qnx/config-6.3-gcc-3.3.make +++ b/dist/etc/qnx/config-6.3-gcc-3.3.make @@ -48,6 +48,20 @@ XSDE_ARCH_WIDTH := 32 XSDE_BYTEORDER := l +# Application character encoding. Valid values are 'utf8' for UTF-8 +# and 'iso8859-1' for ISO-8859-1. Note that this encoding is not +# the same as the XML document encoding that is being parsed or +# serialized. Rather, it is the encoding that is used inside the +# application. When an XML document is parsed, the character data +# is automatically converted to the application encoding. Similarly, +# when an XML document is serialized, the data in the application +# encoding is automatically converted to the resulting document +# encoding. Also don't forget to use the --char-encoding option +# when compiling your schemas if using an encoding other than UTF-8. +# +XSDE_ENCODING := utf8 + + # Set to 'n' if you don't have STL (std::string, etc.). Also don't # forget to use the --no-stl option when compiling your schemas. # diff --git a/dist/etc/qnx/config-6.4.make b/dist/etc/qnx/config-6.4.make index a302d0d..71f40db 100644 --- a/dist/etc/qnx/config-6.4.make +++ b/dist/etc/qnx/config-6.4.make @@ -48,6 +48,20 @@ XSDE_ARCH_WIDTH := 32 XSDE_BYTEORDER := l +# Application character encoding. Valid values are 'utf8' for UTF-8 +# and 'iso8859-1' for ISO-8859-1. Note that this encoding is not +# the same as the XML document encoding that is being parsed or +# serialized. Rather, it is the encoding that is used inside the +# application. When an XML document is parsed, the character data +# is automatically converted to the application encoding. Similarly, +# when an XML document is serialized, the data in the application +# encoding is automatically converted to the resulting document +# encoding. Also don't forget to use the --char-encoding option +# when compiling your schemas if using an encoding other than UTF-8. +# +XSDE_ENCODING := utf8 + + # Set to 'n' if you don't have STL (std::string, etc.). Also don't # forget to use the --no-stl option when compiling your schemas. # diff --git a/dist/etc/vc-8.0/config-max.nmake b/dist/etc/vc-8.0/config-max.nmake index 07b46bb..59d1a1c 100644 --- a/dist/etc/vc-8.0/config-max.nmake +++ b/dist/etc/vc-8.0/config-max.nmake @@ -46,6 +46,20 @@ XSDE_ARCH_WIDTH = 32 XSDE_BYTEORDER = l +# Application character encoding. Valid values are 'utf8' for UTF-8 +# and 'iso8859-1' for ISO-8859-1. Note that this encoding is not +# the same as the XML document encoding that is being parsed or +# serialized. Rather, it is the encoding that is used inside the +# application. When an XML document is parsed, the character data +# is automatically converted to the application encoding. Similarly, +# when an XML document is serialized, the data in the application +# encoding is automatically converted to the resulting document +# encoding. Also don't forget to use the --char-encoding option +# when compiling your schemas if using an encoding other than UTF-8. +# +XSDE_ENCODING = utf8 + + # Set to 'n' if you don't have STL (std::string, etc.). Also don't # forget to use the --no-stl option when compiling your schemas. # diff --git a/dist/etc/vc-8.0/config-min.nmake b/dist/etc/vc-8.0/config-min.nmake index 286e349..bdb9863 100644 --- a/dist/etc/vc-8.0/config-min.nmake +++ b/dist/etc/vc-8.0/config-min.nmake @@ -46,6 +46,20 @@ XSDE_ARCH_WIDTH = 32 XSDE_BYTEORDER = l +# Application character encoding. Valid values are 'utf8' for UTF-8 +# and 'iso8859-1' for ISO-8859-1. Note that this encoding is not +# the same as the XML document encoding that is being parsed or +# serialized. Rather, it is the encoding that is used inside the +# application. When an XML document is parsed, the character data +# is automatically converted to the application encoding. Similarly, +# when an XML document is serialized, the data in the application +# encoding is automatically converted to the resulting document +# encoding. Also don't forget to use the --char-encoding option +# when compiling your schemas if using an encoding other than UTF-8. +# +XSDE_ENCODING = utf8 + + # Set to 'n' if you don't have STL (std::string, etc.). Also don't # forget to use the --no-stl option when compiling your schemas. # diff --git a/dist/etc/vc-9.0/config-max.nmake b/dist/etc/vc-9.0/config-max.nmake index 938bc3d..3a11fe7 100644 --- a/dist/etc/vc-9.0/config-max.nmake +++ b/dist/etc/vc-9.0/config-max.nmake @@ -46,6 +46,20 @@ XSDE_ARCH_WIDTH = 32 XSDE_BYTEORDER = l +# Application character encoding. Valid values are 'utf8' for UTF-8 +# and 'iso8859-1' for ISO-8859-1. Note that this encoding is not +# the same as the XML document encoding that is being parsed or +# serialized. Rather, it is the encoding that is used inside the +# application. When an XML document is parsed, the character data +# is automatically converted to the application encoding. Similarly, +# when an XML document is serialized, the data in the application +# encoding is automatically converted to the resulting document +# encoding. Also don't forget to use the --char-encoding option +# when compiling your schemas if using an encoding other than UTF-8. +# +XSDE_ENCODING = utf8 + + # Set to 'n' if you don't have STL (std::string, etc.). Also don't # forget to use the --no-stl option when compiling your schemas. # diff --git a/dist/etc/vc-9.0/config-min.nmake b/dist/etc/vc-9.0/config-min.nmake index b98e9a9..9df85b5 100644 --- a/dist/etc/vc-9.0/config-min.nmake +++ b/dist/etc/vc-9.0/config-min.nmake @@ -46,6 +46,20 @@ XSDE_ARCH_WIDTH = 32 XSDE_BYTEORDER = l +# Application character encoding. Valid values are 'utf8' for UTF-8 +# and 'iso8859-1' for ISO-8859-1. Note that this encoding is not +# the same as the XML document encoding that is being parsed or +# serialized. Rather, it is the encoding that is used inside the +# application. When an XML document is parsed, the character data +# is automatically converted to the application encoding. Similarly, +# when an XML document is serialized, the data in the application +# encoding is automatically converted to the resulting document +# encoding. Also don't forget to use the --char-encoding option +# when compiling your schemas if using an encoding other than UTF-8. +# +XSDE_ENCODING = utf8 + + # Set to 'n' if you don't have STL (std::string, etc.). Also don't # forget to use the --no-stl option when compiling your schemas. # diff --git a/dist/etc/vxworks/config-5.5.1.make b/dist/etc/vxworks/config-5.5.1.make index 9d6d8c5..89bf164 100644 --- a/dist/etc/vxworks/config-5.5.1.make +++ b/dist/etc/vxworks/config-5.5.1.make @@ -49,6 +49,20 @@ XSDE_ARCH_WIDTH := 32 XSDE_BYTEORDER := b +# Application character encoding. Valid values are 'utf8' for UTF-8 +# and 'iso8859-1' for ISO-8859-1. Note that this encoding is not +# the same as the XML document encoding that is being parsed or +# serialized. Rather, it is the encoding that is used inside the +# application. When an XML document is parsed, the character data +# is automatically converted to the application encoding. Similarly, +# when an XML document is serialized, the data in the application +# encoding is automatically converted to the resulting document +# encoding. Also don't forget to use the --char-encoding option +# when compiling your schemas if using an encoding other than UTF-8. +# +XSDE_ENCODING := utf8 + + # Set to 'n' if you don't have STL (std::string, etc.). Also don't # forget to use the --no-stl option when compiling your schemas. # diff --git a/dist/etc/vxworks/config-6.4-max.make b/dist/etc/vxworks/config-6.4-max.make index 0d7ba41..b1b08fd 100644 --- a/dist/etc/vxworks/config-6.4-max.make +++ b/dist/etc/vxworks/config-6.4-max.make @@ -61,6 +61,20 @@ XSDE_ARCH_WIDTH := 32 XSDE_BYTEORDER := l +# Application character encoding. Valid values are 'utf8' for UTF-8 +# and 'iso8859-1' for ISO-8859-1. Note that this encoding is not +# the same as the XML document encoding that is being parsed or +# serialized. Rather, it is the encoding that is used inside the +# application. When an XML document is parsed, the character data +# is automatically converted to the application encoding. Similarly, +# when an XML document is serialized, the data in the application +# encoding is automatically converted to the resulting document +# encoding. Also don't forget to use the --char-encoding option +# when compiling your schemas if using an encoding other than UTF-8. +# +XSDE_ENCODING := utf8 + + # Set to 'n' if you don't have STL (std::string, etc.). Also don't # forget to use the --no-stl option when compiling your schemas. # diff --git a/dist/etc/vxworks/config-6.4-min.make b/dist/etc/vxworks/config-6.4-min.make index 3efd917..3435c70 100644 --- a/dist/etc/vxworks/config-6.4-min.make +++ b/dist/etc/vxworks/config-6.4-min.make @@ -48,6 +48,20 @@ XSDE_ARCH_WIDTH := 32 XSDE_BYTEORDER := l +# Application character encoding. Valid values are 'utf8' for UTF-8 +# and 'iso8859-1' for ISO-8859-1. Note that this encoding is not +# the same as the XML document encoding that is being parsed or +# serialized. Rather, it is the encoding that is used inside the +# application. When an XML document is parsed, the character data +# is automatically converted to the application encoding. Similarly, +# when an XML document is serialized, the data in the application +# encoding is automatically converted to the resulting document +# encoding. Also don't forget to use the --char-encoding option +# when compiling your schemas if using an encoding other than UTF-8. +# +XSDE_ENCODING := utf8 + + # Set to 'n' if you don't have STL (std::string, etc.). Also don't # forget to use the --no-stl option when compiling your schemas. # diff --git a/dist/etc/vxworks/config-6.7-max.make b/dist/etc/vxworks/config-6.7-max.make index f6e8674..43731df 100644 --- a/dist/etc/vxworks/config-6.7-max.make +++ b/dist/etc/vxworks/config-6.7-max.make @@ -61,6 +61,20 @@ XSDE_ARCH_WIDTH := 32 XSDE_BYTEORDER := l +# Application character encoding. Valid values are 'utf8' for UTF-8 +# and 'iso8859-1' for ISO-8859-1. Note that this encoding is not +# the same as the XML document encoding that is being parsed or +# serialized. Rather, it is the encoding that is used inside the +# application. When an XML document is parsed, the character data +# is automatically converted to the application encoding. Similarly, +# when an XML document is serialized, the data in the application +# encoding is automatically converted to the resulting document +# encoding. Also don't forget to use the --char-encoding option +# when compiling your schemas if using an encoding other than UTF-8. +# +XSDE_ENCODING := utf8 + + # Set to 'n' if you don't have STL (std::string, etc.). Also don't # forget to use the --no-stl option when compiling your schemas. # diff --git a/dist/etc/vxworks/config-6.7-min.make b/dist/etc/vxworks/config-6.7-min.make index 42861a3..cf8d622 100644 --- a/dist/etc/vxworks/config-6.7-min.make +++ b/dist/etc/vxworks/config-6.7-min.make @@ -48,6 +48,20 @@ XSDE_ARCH_WIDTH := 32 XSDE_BYTEORDER := l +# Application character encoding. Valid values are 'utf8' for UTF-8 +# and 'iso8859-1' for ISO-8859-1. Note that this encoding is not +# the same as the XML document encoding that is being parsed or +# serialized. Rather, it is the encoding that is used inside the +# application. When an XML document is parsed, the character data +# is automatically converted to the application encoding. Similarly, +# when an XML document is serialized, the data in the application +# encoding is automatically converted to the resulting document +# encoding. Also don't forget to use the --char-encoding option +# when compiling your schemas if using an encoding other than UTF-8. +# +XSDE_ENCODING := utf8 + + # Set to 'n' if you don't have STL (std::string, etc.). Also don't # forget to use the --no-stl option when compiling your schemas. # diff --git a/dist/examples/cxx/hybrid/binary/cdr/makefile b/dist/examples/cxx/hybrid/binary/cdr/makefile index 1453f78..9e0cd7d 100644 --- a/dist/examples/cxx/hybrid/binary/cdr/makefile +++ b/dist/examples/cxx/hybrid/binary/cdr/makefile @@ -8,6 +8,10 @@ override LIBS += -lACE # EXTRA_CPPFLAGS := -I$(root)/libxsde +ifeq ($(XSDE_ENCODING),iso8859-1) +EXTRA_XSDFLAGS += --char-encoding iso8859-1 +endif + ifeq ($(XSDE_STL),n) EXTRA_XSDFLAGS += --no-stl endif diff --git a/dist/examples/cxx/hybrid/binary/cdr/nmakefile b/dist/examples/cxx/hybrid/binary/cdr/nmakefile index ce9210b..ffa65b6 100644 --- a/dist/examples/cxx/hybrid/binary/cdr/nmakefile +++ b/dist/examples/cxx/hybrid/binary/cdr/nmakefile @@ -8,6 +8,10 @@ LIBS = $(LIBS) ACE.lib # EXTRA_CPPFLAGS = /I$(root)\libxsde +!if "$(XSDE_ENCODING)" == "iso8859-1" +EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding iso8859-1 +!endif + !if "$(XSDE_STL)" == "n" EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-stl !endif diff --git a/dist/examples/cxx/hybrid/binary/custom/makefile b/dist/examples/cxx/hybrid/binary/custom/makefile index 9d02861..55f1944 100644 --- a/dist/examples/cxx/hybrid/binary/custom/makefile +++ b/dist/examples/cxx/hybrid/binary/custom/makefile @@ -6,6 +6,10 @@ include $(root)/build/cxx/rules.make # EXTRA_CPPFLAGS := -I$(root)/libxsde +ifeq ($(XSDE_ENCODING),iso8859-1) +EXTRA_XSDFLAGS += --char-encoding iso8859-1 +endif + ifeq ($(XSDE_STL),n) EXTRA_XSDFLAGS += --no-stl endif diff --git a/dist/examples/cxx/hybrid/binary/custom/nmakefile b/dist/examples/cxx/hybrid/binary/custom/nmakefile index 00a0964..02e193f 100644 --- a/dist/examples/cxx/hybrid/binary/custom/nmakefile +++ b/dist/examples/cxx/hybrid/binary/custom/nmakefile @@ -6,6 +6,10 @@ root = ..\..\..\..\.. # EXTRA_CPPFLAGS = /I$(root)\libxsde +!if "$(XSDE_ENCODING)" == "iso8859-1" +EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding iso8859-1 +!endif + !if "$(XSDE_STL)" == "n" EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-stl !endif diff --git a/dist/examples/cxx/hybrid/binary/xdr/makefile b/dist/examples/cxx/hybrid/binary/xdr/makefile index 257536d..ebd0241 100644 --- a/dist/examples/cxx/hybrid/binary/xdr/makefile +++ b/dist/examples/cxx/hybrid/binary/xdr/makefile @@ -6,6 +6,10 @@ include $(root)/build/cxx/rules.make # EXTRA_CPPFLAGS := -I$(root)/libxsde +ifeq ($(XSDE_ENCODING),iso8859-1) +EXTRA_XSDFLAGS += --char-encoding iso8859-1 +endif + ifeq ($(XSDE_STL),n) EXTRA_XSDFLAGS += --no-stl endif diff --git a/dist/examples/cxx/hybrid/binary/xdr/nmakefile b/dist/examples/cxx/hybrid/binary/xdr/nmakefile index 332bbeb..49cbbb2 100644 --- a/dist/examples/cxx/hybrid/binary/xdr/nmakefile +++ b/dist/examples/cxx/hybrid/binary/xdr/nmakefile @@ -6,6 +6,10 @@ root = ..\..\..\..\.. # EXTRA_CPPFLAGS = /I$(root)\libxsde +!if "$(XSDE_ENCODING)" == "iso8859-1" +EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding iso8859-1 +!endif + !if "$(XSDE_STL)" == "n" EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-stl !endif diff --git a/dist/examples/cxx/hybrid/compositors/makefile b/dist/examples/cxx/hybrid/compositors/makefile index dd36b46..beda699 100644 --- a/dist/examples/cxx/hybrid/compositors/makefile +++ b/dist/examples/cxx/hybrid/compositors/makefile @@ -6,6 +6,10 @@ include $(root)/build/cxx/rules.make # EXTRA_CPPFLAGS := -I$(root)/libxsde +ifeq ($(XSDE_ENCODING),iso8859-1) +EXTRA_XSDFLAGS += --char-encoding iso8859-1 +endif + ifeq ($(XSDE_STL),n) EXTRA_XSDFLAGS += --no-stl endif diff --git a/dist/examples/cxx/hybrid/compositors/nmakefile b/dist/examples/cxx/hybrid/compositors/nmakefile index 14039ac..3178b0a 100644 --- a/dist/examples/cxx/hybrid/compositors/nmakefile +++ b/dist/examples/cxx/hybrid/compositors/nmakefile @@ -6,6 +6,10 @@ root = ..\..\..\.. # EXTRA_CPPFLAGS = /I$(root)\libxsde +!if "$(XSDE_ENCODING)" == "iso8859-1" +EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding iso8859-1 +!endif + !if "$(XSDE_STL)" == "n" EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-stl !endif diff --git a/dist/examples/cxx/hybrid/custom/wildcard/makefile b/dist/examples/cxx/hybrid/custom/wildcard/makefile index fb85b48..f316afa 100644 --- a/dist/examples/cxx/hybrid/custom/wildcard/makefile +++ b/dist/examples/cxx/hybrid/custom/wildcard/makefile @@ -6,6 +6,10 @@ include $(root)/build/cxx/rules.make # EXTRA_CPPFLAGS := -I$(root)/libxsde +ifeq ($(XSDE_ENCODING),iso8859-1) +EXTRA_XSDFLAGS += --char-encoding iso8859-1 +endif + ifeq ($(XSDE_LONGLONG),n) EXTRA_XSDFLAGS += --no-long-long endif diff --git a/dist/examples/cxx/hybrid/custom/wildcard/nmakefile b/dist/examples/cxx/hybrid/custom/wildcard/nmakefile index 329c993..37f1293 100644 --- a/dist/examples/cxx/hybrid/custom/wildcard/nmakefile +++ b/dist/examples/cxx/hybrid/custom/wildcard/nmakefile @@ -6,6 +6,10 @@ root = ..\..\..\..\.. # EXTRA_CPPFLAGS = /I$(root)\libxsde +!if "$(XSDE_ENCODING)" == "iso8859-1" +EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding iso8859-1 +!endif + !if "$(XSDE_LONGLONG)" == "n" EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-long-long !endif diff --git a/dist/examples/cxx/hybrid/filter/makefile b/dist/examples/cxx/hybrid/filter/makefile index 586a166..baf45ce 100644 --- a/dist/examples/cxx/hybrid/filter/makefile +++ b/dist/examples/cxx/hybrid/filter/makefile @@ -6,6 +6,10 @@ include $(root)/build/cxx/rules.make # EXTRA_CPPFLAGS := -I$(root)/libxsde +ifeq ($(XSDE_ENCODING),iso8859-1) +EXTRA_XSDFLAGS += --char-encoding iso8859-1 +endif + ifeq ($(XSDE_LONGLONG),n) EXTRA_XSDFLAGS += --no-long-long endif diff --git a/dist/examples/cxx/hybrid/filter/nmakefile b/dist/examples/cxx/hybrid/filter/nmakefile index 6db5615..d535457 100644 --- a/dist/examples/cxx/hybrid/filter/nmakefile +++ b/dist/examples/cxx/hybrid/filter/nmakefile @@ -6,6 +6,10 @@ root = ..\..\..\.. # EXTRA_CPPFLAGS = /I$(root)\libxsde +!if "$(XSDE_ENCODING)" == "iso8859-1" +EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding iso8859-1 +!endif + !if "$(XSDE_LONGLONG)" == "n" EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-long-long !endif diff --git a/dist/examples/cxx/hybrid/hello/makefile b/dist/examples/cxx/hybrid/hello/makefile index 5499a21..9b2130d 100644 --- a/dist/examples/cxx/hybrid/hello/makefile +++ b/dist/examples/cxx/hybrid/hello/makefile @@ -6,6 +6,10 @@ include $(root)/build/cxx/rules.make # EXTRA_CPPFLAGS := -I$(root)/libxsde +ifeq ($(XSDE_ENCODING),iso8859-1) +EXTRA_XSDFLAGS += --char-encoding iso8859-1 +endif + ifeq ($(XSDE_STL),n) EXTRA_XSDFLAGS += --no-stl endif diff --git a/dist/examples/cxx/hybrid/hello/nmakefile b/dist/examples/cxx/hybrid/hello/nmakefile index ea80a40..59d7111 100644 --- a/dist/examples/cxx/hybrid/hello/nmakefile +++ b/dist/examples/cxx/hybrid/hello/nmakefile @@ -6,6 +6,10 @@ root = ..\..\..\.. # EXTRA_CPPFLAGS = /I$(root)\libxsde +!if "$(XSDE_ENCODING)" == "iso8859-1" +EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding iso8859-1 +!endif + !if "$(XSDE_STL)" == "n" EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-stl !endif diff --git a/dist/examples/cxx/hybrid/library/makefile b/dist/examples/cxx/hybrid/library/makefile index 512a0ba..944c724 100644 --- a/dist/examples/cxx/hybrid/library/makefile +++ b/dist/examples/cxx/hybrid/library/makefile @@ -6,6 +6,10 @@ include $(root)/build/cxx/rules.make # EXTRA_CPPFLAGS := -I$(root)/libxsde +ifeq ($(XSDE_ENCODING),iso8859-1) +EXTRA_XSDFLAGS += --char-encoding iso8859-1 +endif + ifeq ($(XSDE_LONGLONG),n) EXTRA_XSDFLAGS += --no-long-long endif diff --git a/dist/examples/cxx/hybrid/library/nmakefile b/dist/examples/cxx/hybrid/library/nmakefile index e1d1a72..17f3ce0 100644 --- a/dist/examples/cxx/hybrid/library/nmakefile +++ b/dist/examples/cxx/hybrid/library/nmakefile @@ -6,6 +6,10 @@ root = ..\..\..\.. # EXTRA_CPPFLAGS = /I$(root)\libxsde +!if "$(XSDE_ENCODING)" == "iso8859-1" +EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding iso8859-1 +!endif + !if "$(XSDE_LONGLONG)" == "n" EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-long-long !endif diff --git a/dist/examples/cxx/hybrid/minimal/makefile b/dist/examples/cxx/hybrid/minimal/makefile index 8224235..53a52d4 100644 --- a/dist/examples/cxx/hybrid/minimal/makefile +++ b/dist/examples/cxx/hybrid/minimal/makefile @@ -8,6 +8,10 @@ EXTRA_CPPFLAGS := -I$(root)/libxsde EXTRA_XSDFLAGS := --no-stl --no-exceptions +ifeq ($(XSDE_ENCODING),iso8859-1) +EXTRA_XSDFLAGS += --char-encoding iso8859-1 +endif + ifeq ($(XSDE_IOSTREAM),n) EXTRA_XSDFLAGS += --no-iostream endif diff --git a/dist/examples/cxx/hybrid/minimal/nmakefile b/dist/examples/cxx/hybrid/minimal/nmakefile index 060be2c..30d6805 100644 --- a/dist/examples/cxx/hybrid/minimal/nmakefile +++ b/dist/examples/cxx/hybrid/minimal/nmakefile @@ -8,6 +8,10 @@ EXTRA_CPPFLAGS = /I$(root)\libxsde EXTRA_XSDFLAGS = --no-stl --no-exceptions +!if "$(XSDE_ENCODING)" == "iso8859-1" +EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding iso8859-1 +!endif + !if "$(XSDE_IOSTREAM)" == "n" EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-iostream !endif diff --git a/dist/examples/cxx/hybrid/multiroot/makefile b/dist/examples/cxx/hybrid/multiroot/makefile index 7329f41..f4fda3f 100644 --- a/dist/examples/cxx/hybrid/multiroot/makefile +++ b/dist/examples/cxx/hybrid/multiroot/makefile @@ -6,6 +6,10 @@ include $(root)/build/cxx/rules.make # EXTRA_CPPFLAGS := -I$(root)/libxsde +ifeq ($(XSDE_ENCODING),iso8859-1) +EXTRA_XSDFLAGS += --char-encoding iso8859-1 +endif + ifeq ($(XSDE_STL),n) EXTRA_XSDFLAGS += --no-stl endif diff --git a/dist/examples/cxx/hybrid/multiroot/nmakefile b/dist/examples/cxx/hybrid/multiroot/nmakefile index d3c4e35..b550ddf 100644 --- a/dist/examples/cxx/hybrid/multiroot/nmakefile +++ b/dist/examples/cxx/hybrid/multiroot/nmakefile @@ -6,6 +6,10 @@ root = ..\..\..\.. # EXTRA_CPPFLAGS = /I$(root)\libxsde +!if "$(XSDE_ENCODING)" == "iso8859-1" +EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding iso8859-1 +!endif + !if "$(XSDE_STL)" == "n" EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-stl !endif diff --git a/dist/examples/cxx/hybrid/polymorphism/makefile b/dist/examples/cxx/hybrid/polymorphism/makefile index e1f40a1..cc7148c 100644 --- a/dist/examples/cxx/hybrid/polymorphism/makefile +++ b/dist/examples/cxx/hybrid/polymorphism/makefile @@ -6,6 +6,10 @@ include $(root)/build/cxx/rules.make # EXTRA_CPPFLAGS := -I$(root)/libxsde +ifeq ($(XSDE_ENCODING),iso8859-1) +EXTRA_XSDFLAGS += --char-encoding iso8859-1 +endif + ifeq ($(XSDE_STL),n) EXTRA_XSDFLAGS += --no-stl endif diff --git a/dist/examples/cxx/hybrid/polymorphism/nmakefile b/dist/examples/cxx/hybrid/polymorphism/nmakefile index bd86945..de9e356 100644 --- a/dist/examples/cxx/hybrid/polymorphism/nmakefile +++ b/dist/examples/cxx/hybrid/polymorphism/nmakefile @@ -6,6 +6,10 @@ root = ..\..\..\.. # EXTRA_CPPFLAGS = /I$(root)\libxsde +!if "$(XSDE_ENCODING)" == "iso8859-1" +EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding iso8859-1 +!endif + !if "$(XSDE_STL)" == "n" EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-stl !endif diff --git a/dist/examples/cxx/hybrid/polyroot/makefile b/dist/examples/cxx/hybrid/polyroot/makefile index a422aa3..752d936 100644 --- a/dist/examples/cxx/hybrid/polyroot/makefile +++ b/dist/examples/cxx/hybrid/polyroot/makefile @@ -6,6 +6,10 @@ include $(root)/build/cxx/rules.make # EXTRA_CPPFLAGS := -I$(root)/libxsde +ifeq ($(XSDE_ENCODING),iso8859-1) +EXTRA_XSDFLAGS += --char-encoding iso8859-1 +endif + ifeq ($(XSDE_STL),n) EXTRA_XSDFLAGS += --no-stl endif diff --git a/dist/examples/cxx/hybrid/polyroot/nmakefile b/dist/examples/cxx/hybrid/polyroot/nmakefile index d713db0..dbf57fb 100644 --- a/dist/examples/cxx/hybrid/polyroot/nmakefile +++ b/dist/examples/cxx/hybrid/polyroot/nmakefile @@ -6,6 +6,10 @@ root = ..\..\..\.. # EXTRA_CPPFLAGS = /I$(root)\libxsde +!if "$(XSDE_ENCODING)" == "iso8859-1" +EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding iso8859-1 +!endif + !if "$(XSDE_STL)" == "n" EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-stl !endif diff --git a/dist/examples/cxx/hybrid/streaming/makefile b/dist/examples/cxx/hybrid/streaming/makefile index 6c97f8b..bb67f6a 100644 --- a/dist/examples/cxx/hybrid/streaming/makefile +++ b/dist/examples/cxx/hybrid/streaming/makefile @@ -6,6 +6,10 @@ include $(root)/build/cxx/rules.make # EXTRA_CPPFLAGS := -I$(root)/libxsde +ifeq ($(XSDE_ENCODING),iso8859-1) +EXTRA_XSDFLAGS += --char-encoding iso8859-1 +endif + ifeq ($(XSDE_STL),n) EXTRA_XSDFLAGS += --no-stl endif diff --git a/dist/examples/cxx/hybrid/streaming/nmakefile b/dist/examples/cxx/hybrid/streaming/nmakefile index 76a0c14..e9b5821 100644 --- a/dist/examples/cxx/hybrid/streaming/nmakefile +++ b/dist/examples/cxx/hybrid/streaming/nmakefile @@ -6,6 +6,10 @@ root = ..\..\..\.. # EXTRA_CPPFLAGS = /I$(root)\libxsde +!if "$(XSDE_ENCODING)" == "iso8859-1" +EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding iso8859-1 +!endif + !if "$(XSDE_STL)" == "n" EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-stl !endif diff --git a/dist/examples/cxx/hybrid/wildcard/makefile b/dist/examples/cxx/hybrid/wildcard/makefile index 0c8fd1f..3087253 100644 --- a/dist/examples/cxx/hybrid/wildcard/makefile +++ b/dist/examples/cxx/hybrid/wildcard/makefile @@ -6,6 +6,10 @@ include $(root)/build/cxx/rules.make # EXTRA_CPPFLAGS := -I$(root)/libxsde +ifeq ($(XSDE_ENCODING),iso8859-1) +EXTRA_XSDFLAGS += --char-encoding iso8859-1 +endif + ifeq ($(XSDE_LONGLONG),n) EXTRA_XSDFLAGS += --no-long-long endif diff --git a/dist/examples/cxx/hybrid/wildcard/nmakefile b/dist/examples/cxx/hybrid/wildcard/nmakefile index d4d8d06..0634fa8 100644 --- a/dist/examples/cxx/hybrid/wildcard/nmakefile +++ b/dist/examples/cxx/hybrid/wildcard/nmakefile @@ -6,6 +6,10 @@ root = ..\..\..\.. # EXTRA_CPPFLAGS = /I$(root)\libxsde +!if "$(XSDE_ENCODING)" == "iso8859-1" +EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding iso8859-1 +!endif + !if "$(XSDE_LONGLONG)" == "n" EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-long-long !endif diff --git a/dist/examples/cxx/parser/generated/makefile b/dist/examples/cxx/parser/generated/makefile index bf3b512..6f77af4 100644 --- a/dist/examples/cxx/parser/generated/makefile +++ b/dist/examples/cxx/parser/generated/makefile @@ -14,6 +14,10 @@ EXTRA_XSDFLAGS := --generate-print-impl --generate-test-driver # EXTRA_XSDFLAGS += --force-overwrite +ifeq ($(XSDE_ENCODING),iso8859-1) +EXTRA_XSDFLAGS += --char-encoding iso8859-1 +endif + ifeq ($(XSDE_STL),n) EXTRA_XSDFLAGS += --no-stl endif diff --git a/dist/examples/cxx/parser/generated/nmakefile b/dist/examples/cxx/parser/generated/nmakefile index ad0900c..c801757 100644 --- a/dist/examples/cxx/parser/generated/nmakefile +++ b/dist/examples/cxx/parser/generated/nmakefile @@ -14,6 +14,10 @@ EXTRA_XSDFLAGS = --generate-print-impl --generate-test-driver # EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --force-overwrite +!if "$(XSDE_ENCODING)" == "iso8859-1" +EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding iso8859-1 +!endif + !if "$(XSDE_STL)" == "n" EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-stl !endif diff --git a/dist/examples/cxx/parser/hello/makefile b/dist/examples/cxx/parser/hello/makefile index 421d69e..a136306 100644 --- a/dist/examples/cxx/parser/hello/makefile +++ b/dist/examples/cxx/parser/hello/makefile @@ -6,6 +6,10 @@ include $(root)/build/cxx/rules.make # EXTRA_CPPFLAGS := -I$(root)/libxsde +ifeq ($(XSDE_ENCODING),iso8859-1) +EXTRA_XSDFLAGS += --char-encoding iso8859-1 +endif + ifeq ($(XSDE_LONGLONG),n) EXTRA_XSDFLAGS += --no-long-long endif diff --git a/dist/examples/cxx/parser/hello/nmakefile b/dist/examples/cxx/parser/hello/nmakefile index 40d1307..60f226a 100644 --- a/dist/examples/cxx/parser/hello/nmakefile +++ b/dist/examples/cxx/parser/hello/nmakefile @@ -6,6 +6,10 @@ root = ..\..\..\.. # EXTRA_CPPFLAGS = /I$(root)\libxsde +!if "$(XSDE_ENCODING)" == "iso8859-1" +EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding iso8859-1 +!endif + !if "$(XSDE_LONGLONG)" == "n" EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-long-long !endif diff --git a/dist/examples/cxx/parser/library/makefile b/dist/examples/cxx/parser/library/makefile index 0221150..a142b56 100644 --- a/dist/examples/cxx/parser/library/makefile +++ b/dist/examples/cxx/parser/library/makefile @@ -6,6 +6,10 @@ include $(root)/build/cxx/rules.make # EXTRA_CPPFLAGS += -I$(root)/libxsde +ifeq ($(XSDE_ENCODING),iso8859-1) +EXTRA_XSDFLAGS += --char-encoding iso8859-1 +endif + ifeq ($(XSDE_LONGLONG),n) EXTRA_XSDFLAGS += --no-long-long endif diff --git a/dist/examples/cxx/parser/library/nmakefile b/dist/examples/cxx/parser/library/nmakefile index 21ffebd..fe82086 100644 --- a/dist/examples/cxx/parser/library/nmakefile +++ b/dist/examples/cxx/parser/library/nmakefile @@ -6,6 +6,10 @@ root = ..\..\..\.. # EXTRA_CPPFLAGS = /I$(root)\libxsde +!if "$(XSDE_ENCODING)" == "iso8859-1" +EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding iso8859-1 +!endif + !if "$(XSDE_LONGLONG)" == "n" EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-long-long !endif diff --git a/dist/examples/cxx/parser/minimal/makefile b/dist/examples/cxx/parser/minimal/makefile index fb0d483..aba65d9 100644 --- a/dist/examples/cxx/parser/minimal/makefile +++ b/dist/examples/cxx/parser/minimal/makefile @@ -8,6 +8,10 @@ EXTRA_CPPFLAGS += -I$(root)/libxsde EXTRA_XSDFLAGS := --no-stl --no-exceptions +ifeq ($(XSDE_ENCODING),iso8859-1) +EXTRA_XSDFLAGS += --char-encoding iso8859-1 +endif + ifeq ($(XSDE_IOSTREAM),n) EXTRA_XSDFLAGS += --no-iostream endif diff --git a/dist/examples/cxx/parser/minimal/nmakefile b/dist/examples/cxx/parser/minimal/nmakefile index addf4c8..4e81aff 100644 --- a/dist/examples/cxx/parser/minimal/nmakefile +++ b/dist/examples/cxx/parser/minimal/nmakefile @@ -8,6 +8,10 @@ EXTRA_CPPFLAGS = /I$(root)\libxsde EXTRA_XSDFLAGS = --no-stl --no-exceptions +!if "$(XSDE_ENCODING)" == "iso8859-1" +EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding iso8859-1 +!endif + !if "$(XSDE_IOSTREAM)" == "n" EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-iostream !endif diff --git a/dist/examples/cxx/parser/mixed/makefile b/dist/examples/cxx/parser/mixed/makefile index 44d988e..b4e7426 100644 --- a/dist/examples/cxx/parser/mixed/makefile +++ b/dist/examples/cxx/parser/mixed/makefile @@ -6,6 +6,10 @@ include $(root)/build/cxx/rules.make # EXTRA_CPPFLAGS += -I$(root)/libxsde +ifeq ($(XSDE_ENCODING),iso8859-1) +EXTRA_XSDFLAGS += --char-encoding iso8859-1 +endif + ifeq ($(XSDE_LONGLONG),n) EXTRA_XSDFLAGS += --no-long-long endif diff --git a/dist/examples/cxx/parser/mixed/nmakefile b/dist/examples/cxx/parser/mixed/nmakefile index d25c2ae..bba2da4 100644 --- a/dist/examples/cxx/parser/mixed/nmakefile +++ b/dist/examples/cxx/parser/mixed/nmakefile @@ -6,6 +6,10 @@ root = ..\..\..\.. # EXTRA_CPPFLAGS = /I$(root)\libxsde +!if "$(XSDE_ENCODING)" == "iso8859-1" +EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding iso8859-1 +!endif + !if "$(XSDE_LONGLONG)" == "n" EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-long-long !endif diff --git a/dist/examples/cxx/parser/multiroot/makefile b/dist/examples/cxx/parser/multiroot/makefile index 8fdaeec..b3fb1f6 100644 --- a/dist/examples/cxx/parser/multiroot/makefile +++ b/dist/examples/cxx/parser/multiroot/makefile @@ -6,6 +6,10 @@ include $(root)/build/cxx/rules.make # EXTRA_CPPFLAGS += -I$(root)/libxsde +ifeq ($(XSDE_ENCODING),iso8859-1) +EXTRA_XSDFLAGS += --char-encoding iso8859-1 +endif + ifeq ($(XSDE_LONGLONG),n) EXTRA_XSDFLAGS += --no-long-long endif diff --git a/dist/examples/cxx/parser/multiroot/nmakefile b/dist/examples/cxx/parser/multiroot/nmakefile index 632b38b..779daa7 100644 --- a/dist/examples/cxx/parser/multiroot/nmakefile +++ b/dist/examples/cxx/parser/multiroot/nmakefile @@ -6,6 +6,10 @@ root = ..\..\..\.. # EXTRA_CPPFLAGS = /I$(root)\libxsde +!if "$(XSDE_ENCODING)" == "iso8859-1" +EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding iso8859-1 +!endif + !if "$(XSDE_LONGLONG)" == "n" EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-long-long !endif diff --git a/dist/examples/cxx/parser/polymorphism/makefile b/dist/examples/cxx/parser/polymorphism/makefile index 368401a..c970f3a 100644 --- a/dist/examples/cxx/parser/polymorphism/makefile +++ b/dist/examples/cxx/parser/polymorphism/makefile @@ -6,6 +6,10 @@ include $(root)/build/cxx/rules.make # EXTRA_CPPFLAGS += -I$(root)/libxsde +ifeq ($(XSDE_ENCODING),iso8859-1) +EXTRA_XSDFLAGS += --char-encoding iso8859-1 +endif + ifeq ($(XSDE_LONGLONG),n) EXTRA_XSDFLAGS += --no-long-long endif diff --git a/dist/examples/cxx/parser/polymorphism/nmakefile b/dist/examples/cxx/parser/polymorphism/nmakefile index a35b625..30c50cd 100644 --- a/dist/examples/cxx/parser/polymorphism/nmakefile +++ b/dist/examples/cxx/parser/polymorphism/nmakefile @@ -6,6 +6,10 @@ root = ..\..\..\.. # EXTRA_CPPFLAGS = /I$(root)\libxsde +!if "$(XSDE_ENCODING)" == "iso8859-1" +EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding iso8859-1 +!endif + !if "$(XSDE_LONGLONG)" == "n" EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-long-long !endif diff --git a/dist/examples/cxx/parser/polyroot/makefile b/dist/examples/cxx/parser/polyroot/makefile index 69e7c10..898472a 100644 --- a/dist/examples/cxx/parser/polyroot/makefile +++ b/dist/examples/cxx/parser/polyroot/makefile @@ -6,6 +6,10 @@ include $(root)/build/cxx/rules.make # EXTRA_CPPFLAGS += -I$(root)/libxsde +ifeq ($(XSDE_ENCODING),iso8859-1) +EXTRA_XSDFLAGS += --char-encoding iso8859-1 +endif + ifeq ($(XSDE_LONGLONG),n) EXTRA_XSDFLAGS += --no-long-long endif diff --git a/dist/examples/cxx/parser/polyroot/nmakefile b/dist/examples/cxx/parser/polyroot/nmakefile index f3c4ff0..220d824 100644 --- a/dist/examples/cxx/parser/polyroot/nmakefile +++ b/dist/examples/cxx/parser/polyroot/nmakefile @@ -6,6 +6,10 @@ root = ..\..\..\.. # EXTRA_CPPFLAGS = /I$(root)\libxsde +!if "$(XSDE_ENCODING)" == "iso8859-1" +EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding iso8859-1 +!endif + !if "$(XSDE_LONGLONG)" == "n" EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-long-long !endif diff --git a/dist/examples/cxx/parser/wildcard/makefile b/dist/examples/cxx/parser/wildcard/makefile index 7d117bc..40591b3 100644 --- a/dist/examples/cxx/parser/wildcard/makefile +++ b/dist/examples/cxx/parser/wildcard/makefile @@ -6,6 +6,10 @@ include $(root)/build/cxx/rules.make # EXTRA_CPPFLAGS := -I$(root)/libxsde +ifeq ($(XSDE_ENCODING),iso8859-1) +EXTRA_XSDFLAGS += --char-encoding iso8859-1 +endif + ifeq ($(XSDE_LONGLONG),n) EXTRA_XSDFLAGS += --no-long-long endif diff --git a/dist/examples/cxx/parser/wildcard/nmakefile b/dist/examples/cxx/parser/wildcard/nmakefile index c912a1c..99ab82f 100644 --- a/dist/examples/cxx/parser/wildcard/nmakefile +++ b/dist/examples/cxx/parser/wildcard/nmakefile @@ -6,6 +6,10 @@ root = ..\..\..\.. # EXTRA_CPPFLAGS = /I$(root)\libxsde +!if "$(XSDE_ENCODING)" == "iso8859-1" +EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding iso8859-1 +!endif + !if "$(XSDE_LONGLONG)" == "n" EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-long-long !endif diff --git a/dist/examples/cxx/serializer/hello/makefile b/dist/examples/cxx/serializer/hello/makefile index a498aa4..90a752d 100644 --- a/dist/examples/cxx/serializer/hello/makefile +++ b/dist/examples/cxx/serializer/hello/makefile @@ -6,6 +6,10 @@ include $(root)/build/cxx/rules.make # EXTRA_CPPFLAGS := -I$(root)/libxsde +ifeq ($(XSDE_ENCODING),iso8859-1) +EXTRA_XSDFLAGS += --char-encoding iso8859-1 +endif + ifeq ($(XSDE_LONGLONG),n) EXTRA_XSDFLAGS += --no-long-long endif diff --git a/dist/examples/cxx/serializer/hello/nmakefile b/dist/examples/cxx/serializer/hello/nmakefile index f3b3d2c..f1043a3 100644 --- a/dist/examples/cxx/serializer/hello/nmakefile +++ b/dist/examples/cxx/serializer/hello/nmakefile @@ -6,6 +6,10 @@ root = ..\..\..\.. # EXTRA_CPPFLAGS = /I$(root)\libxsde +!if "$(XSDE_ENCODING)" == "iso8859-1" +EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding iso8859-1 +!endif + !if "$(XSDE_LONGLONG)" == "n" EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-long-long !endif diff --git a/dist/examples/cxx/serializer/library/makefile b/dist/examples/cxx/serializer/library/makefile index cf275b9..a6166a3 100644 --- a/dist/examples/cxx/serializer/library/makefile +++ b/dist/examples/cxx/serializer/library/makefile @@ -6,6 +6,10 @@ include $(root)/build/cxx/rules.make # EXTRA_CPPFLAGS += -I$(root)/libxsde +ifeq ($(XSDE_ENCODING),iso8859-1) +EXTRA_XSDFLAGS += --char-encoding iso8859-1 +endif + ifeq ($(XSDE_LONGLONG),n) EXTRA_XSDFLAGS += --no-long-long endif diff --git a/dist/examples/cxx/serializer/library/nmakefile b/dist/examples/cxx/serializer/library/nmakefile index 9399f5d..71a21f0 100644 --- a/dist/examples/cxx/serializer/library/nmakefile +++ b/dist/examples/cxx/serializer/library/nmakefile @@ -6,6 +6,10 @@ root = ..\..\..\.. # EXTRA_CPPFLAGS = /I$(root)\libxsde +!if "$(XSDE_ENCODING)" == "iso8859-1" +EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding iso8859-1 +!endif + !if "$(XSDE_LONGLONG)" == "n" EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-long-long !endif diff --git a/dist/examples/cxx/serializer/minimal/makefile b/dist/examples/cxx/serializer/minimal/makefile index 4c8c23c..5c17e24 100644 --- a/dist/examples/cxx/serializer/minimal/makefile +++ b/dist/examples/cxx/serializer/minimal/makefile @@ -8,6 +8,10 @@ EXTRA_CPPFLAGS += -I$(root)/libxsde EXTRA_XSDFLAGS := --no-stl --no-exceptions +ifeq ($(XSDE_ENCODING),iso8859-1) +EXTRA_XSDFLAGS += --char-encoding iso8859-1 +endif + ifeq ($(XSDE_IOSTREAM),n) EXTRA_XSDFLAGS += --no-iostream endif diff --git a/dist/examples/cxx/serializer/minimal/nmakefile b/dist/examples/cxx/serializer/minimal/nmakefile index 7b1def3..0184770 100644 --- a/dist/examples/cxx/serializer/minimal/nmakefile +++ b/dist/examples/cxx/serializer/minimal/nmakefile @@ -8,6 +8,10 @@ EXTRA_CPPFLAGS = /I$(root)\libxsde EXTRA_XSDFLAGS = --no-stl --no-exceptions +!if "$(XSDE_ENCODING)" == "iso8859-1" +EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding iso8859-1 +!endif + !if "$(XSDE_IOSTREAM)" == "n" EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-iostream !endif diff --git a/dist/examples/cxx/serializer/polymorphism/makefile b/dist/examples/cxx/serializer/polymorphism/makefile index a602422..6f5b4c8 100644 --- a/dist/examples/cxx/serializer/polymorphism/makefile +++ b/dist/examples/cxx/serializer/polymorphism/makefile @@ -6,6 +6,10 @@ include $(root)/build/cxx/rules.make # EXTRA_CPPFLAGS += -I$(root)/libxsde +ifeq ($(XSDE_ENCODING),iso8859-1) +EXTRA_XSDFLAGS += --char-encoding iso8859-1 +endif + ifeq ($(XSDE_LONGLONG),n) EXTRA_XSDFLAGS += --no-long-long endif diff --git a/dist/examples/cxx/serializer/polymorphism/nmakefile b/dist/examples/cxx/serializer/polymorphism/nmakefile index ad10eea..75e9d7c 100644 --- a/dist/examples/cxx/serializer/polymorphism/nmakefile +++ b/dist/examples/cxx/serializer/polymorphism/nmakefile @@ -6,6 +6,10 @@ root = ..\..\..\.. # EXTRA_CPPFLAGS = /I$(root)\libxsde +!if "$(XSDE_ENCODING)" == "iso8859-1" +EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding iso8859-1 +!endif + !if "$(XSDE_LONGLONG)" == "n" EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-long-long !endif diff --git a/dist/examples/cxx/serializer/polyroot/makefile b/dist/examples/cxx/serializer/polyroot/makefile index a602422..6f5b4c8 100644 --- a/dist/examples/cxx/serializer/polyroot/makefile +++ b/dist/examples/cxx/serializer/polyroot/makefile @@ -6,6 +6,10 @@ include $(root)/build/cxx/rules.make # EXTRA_CPPFLAGS += -I$(root)/libxsde +ifeq ($(XSDE_ENCODING),iso8859-1) +EXTRA_XSDFLAGS += --char-encoding iso8859-1 +endif + ifeq ($(XSDE_LONGLONG),n) EXTRA_XSDFLAGS += --no-long-long endif diff --git a/dist/examples/cxx/serializer/polyroot/nmakefile b/dist/examples/cxx/serializer/polyroot/nmakefile index ad10eea..75e9d7c 100644 --- a/dist/examples/cxx/serializer/polyroot/nmakefile +++ b/dist/examples/cxx/serializer/polyroot/nmakefile @@ -6,6 +6,10 @@ root = ..\..\..\.. # EXTRA_CPPFLAGS = /I$(root)\libxsde +!if "$(XSDE_ENCODING)" == "iso8859-1" +EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding iso8859-1 +!endif + !if "$(XSDE_LONGLONG)" == "n" EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-long-long !endif diff --git a/dist/examples/cxx/serializer/wildcard/makefile b/dist/examples/cxx/serializer/wildcard/makefile index 9c99a98..742d8ff 100644 --- a/dist/examples/cxx/serializer/wildcard/makefile +++ b/dist/examples/cxx/serializer/wildcard/makefile @@ -6,6 +6,10 @@ include $(root)/build/cxx/rules.make # EXTRA_CPPFLAGS += -I$(root)/libxsde +ifeq ($(XSDE_ENCODING),iso8859-1) +EXTRA_XSDFLAGS += --char-encoding iso8859-1 +endif + ifeq ($(XSDE_LONGLONG),n) EXTRA_XSDFLAGS += --no-long-long endif diff --git a/dist/examples/cxx/serializer/wildcard/nmakefile b/dist/examples/cxx/serializer/wildcard/nmakefile index a6c0feb..9ca6d8c 100644 --- a/dist/examples/cxx/serializer/wildcard/nmakefile +++ b/dist/examples/cxx/serializer/wildcard/nmakefile @@ -6,6 +6,10 @@ root = ..\..\..\.. # EXTRA_CPPFLAGS = /I$(root)\libxsde +!if "$(XSDE_ENCODING)" == "iso8859-1" +EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --char-encoding iso8859-1 +!endif + !if "$(XSDE_LONGLONG)" == "n" EXTRA_XSDFLAGS = $(EXTRA_XSDFLAGS) --no-long-long !endif diff --git a/dist/libxsde/xsde/makefile b/dist/libxsde/xsde/makefile index 9abff7a..60c5ba3 100644 --- a/dist/libxsde/xsde/makefile +++ b/dist/libxsde/xsde/makefile @@ -11,6 +11,10 @@ src += c/genx/genx.c c/genx/char-props.c src += cxx/string.cxx cxx/ro-string.cxx cxx/stack.cxx +ifeq ($(XSDE_ENCODING),iso8859-1) +src += cxx/iso8859-1.cxx +endif + ifeq ($(XSDE_STL),n) src += cxx/strdupx.cxx endif @@ -506,6 +510,20 @@ endif ifeq ($(XSDE_BYTEORDER),b) @echo $(h)define XSDE_BYTEORDER 4321 >>$@ endif +ifeq ($(XSDE_ENCODING),utf8) + @echo $(h)define XSDE_ENCODING_UTF8 >>$@ +else +ifeq ($(XSDE_ENCODING),) + @echo $(h)define XSDE_ENCODING_UTF8 >>$@ +else + @echo $(h)undef XSDE_ENCODING_UTF8 >>$@ +endif +endif +ifeq ($(XSDE_ENCODING),iso8859-1) + @echo $(h)define XSDE_ENCODING_ISO8859_1 >>$@ +else + @echo $(h)undef XSDE_ENCODING_ISO8859_1 >>$@ +endif ifeq ($(XSDE_STL),y) @echo $(h)define XSDE_STL >>$@ endif diff --git a/dist/libxsde/xsde/nmakefile b/dist/libxsde/xsde/nmakefile index dd806f7..90fc489 100644 --- a/dist/libxsde/xsde/nmakefile +++ b/dist/libxsde/xsde/nmakefile @@ -13,6 +13,10 @@ src = $(src) c\genx\genx.c c\genx\char-props.c src = $(src) cxx\string.cxx cxx\ro-string.cxx cxx\stack.cxx +!if "$(XSDE_ENCODING)" == "iso8859-1" +src = $(src) cxx\iso8859-1.cxx +!endif + !if "$(XSDE_STL)" == "n" src = $(src) cxx\strdupx.cxx !endif @@ -501,6 +505,20 @@ config.h: !if "$(XSDE_BYTEORDER)" == "b" @echo #define XSDE_BYTEORDER 4321 >>$@ !endif +!if "$(XSDE_ENCODING)" == "utf8" + @echo #define XSDE_ENCODING_UTF8 >>$@ +!else +!if "$(XSDE_ENCODING)" == "" + @echo #define XSDE_ENCODING_UTF8 >>$@ +!else + @echo #undef XSDE_ENCODING_UTF8 >>$@ +!endif +!endif +!if "$(XSDE_ENCODING)" == "iso8859-1" + @echo #define XSDE_ENCODING_ISO8859_1 >>$@ +!else + @echo #undef XSDE_ENCODING_ISO8859_1 >>$@ +!endif !if "$(XSDE_STL)" == "y" @echo #define XSDE_STL >>$@ !endif diff --git a/dist/tests/cxx/hybrid/makefile b/dist/tests/cxx/hybrid/makefile index 3cd2d7d..ddcf464 100644 --- a/dist/tests/cxx/hybrid/makefile +++ b/dist/tests/cxx/hybrid/makefile @@ -17,6 +17,10 @@ endif ifeq ($(XSDE_IOSTREAM),y) dirs += built-in default enumeration list test-template union +ifeq ($(XSDE_ENCODING),iso8859-1) +dirs += iso8859-1 +endif + ifeq ($(XSDE_CDR),y) dirs += binary/cdr endif diff --git a/dist/tests/cxx/hybrid/nmakefile b/dist/tests/cxx/hybrid/nmakefile index 9052637..1127649 100644 --- a/dist/tests/cxx/hybrid/nmakefile +++ b/dist/tests/cxx/hybrid/nmakefile @@ -17,6 +17,10 @@ dirs = $(dirs) iterator !if "$(XSDE_IOSTREAM)" == "y" dirs = $(dirs) built-in default enumeration list test-template union +!if "$(XSDE_ENCODING)" == "iso8859-1" +dirs = $(dirs) iso8859-1 +!endif + !if "$(XSDE_CDR)" == "y" dirs = $(dirs) binary/cdr !endif diff --git a/documentation/cxx/hybrid/guide/index.xhtml b/documentation/cxx/hybrid/guide/index.xhtml index 8c46932..dead3dc 100644 --- a/documentation/cxx/hybrid/guide/index.xhtml +++ b/documentation/cxx/hybrid/guide/index.xhtml @@ -1265,15 +1265,34 @@ main (int argc, char* argv[]) Compiler Command Line Manual</a>. </p> - <p>While the XML documents can use various encodings, the Embedded - C++/Hybrid mapping always delivers character data to the application - in the UTF-8 encoding. The underlying XML parser used by the mapping - includes built-in support for XML documents encoded in UTF-8, UTF-16, - ISO-8859-1, and US-ASCII. Other encodings can be supported by providing - application-specific decoder functions. C++/Hybrid also expects character - data supplied by the application to be in the UTF-8 encoding. The - underlying XML serializer used by the mapping produces the resulting - XML in the UTF-8 encoding as well.</p> + <p>While the XML documents can use various encodings, the C++/Hybrid + object model always stores character data in the same encoding, + called application encoding. The application encoding can either be + UTF-8 (default) or ISO-8859-1. To select a particular encoding, configure + the XSD/e runtime library accordingly and pass the <code>--char-encoding</code> + option to the XSD/e compiler when translating your schemas.</p> + + <p>When using ISO-8859-1 as the application encoding, XML documents + being parsed may contain characters with Unicode values greater + than 0xFF which are unrepresentable in the ISO-8859-1 encoding. + By default, in such situations parsing will terminate with + an error. However, you can suppress the error by providing a + replacement character that should be used instead of + unrepresentable characters, for example:</p> + + <pre class="c++"> +xml_schema::iso8859_1::unrep_char ('?'); + </pre> + + <p>To revert to the default behavior, set the replacement character + to <code>'\0'</code>.</p> + + <p>The underlying XML parser used by the mapping includes built-in + support for XML documents encoded in UTF-8, UTF-16, ISO-8859-1, + and US-ASCII. Other encodings can be supported by providing + application-specific decoder functions. The underlying XML + serializer used by C++/Hybrid produces the resulting + XML documents in the UTF-8 encoding.</p> <h2><a name="3.1">3.1 Standard Template Library</a></h2> diff --git a/documentation/cxx/parser/guide/index.xhtml b/documentation/cxx/parser/guide/index.xhtml index 305c420..6a019c5 100644 --- a/documentation/cxx/parser/guide/index.xhtml +++ b/documentation/cxx/parser/guide/index.xhtml @@ -1991,8 +1991,27 @@ age: 28 <p>While the XML documents can use various encodings, the Embedded C++/Parser mapping always delivers character data to the application - in the UTF-8 encoding. The underlying XML parser used by the - Embedded C++/Parser mapping includes built-in support for XML + in the same encoding. The application encoding can either be UTF-8 + (default) or ISO-8859-1. To select a particular encoding, configure + the XSD/e runtime library accordingly and pass the <code>--char-encoding</code> + option to the XSD/e compiler when translating your schemas.</p> + + <p>When using ISO-8859-1 as the application encoding, XML documents + being parsed may contain characters with Unicode values greater + than 0xFF which are unrepresentable in the ISO-8859-1 encoding. + By default, in such situations parsing will terminate with + an error. However, you can suppress the error by providing a + replacement character that should be used instead of + unrepresentable characters, for example:</p> + + <pre class="c++"> +xml_schema::iso8859_1::unrep_char ('?'); + </pre> + + <p>To revert to the default behavior, set the replacement character + to <code>'\0'</code>.</p> + + <p>The Embedded C++/Parser mapping includes built-in support for XML documents encoded in UTF-8, UTF-16, ISO-8859-1, and US-ASCII. Other encodings can be supported by providing application-specific decoder functions.</p> diff --git a/documentation/cxx/serializer/guide/index.xhtml b/documentation/cxx/serializer/guide/index.xhtml index 5abb31f..34ef798 100644 --- a/documentation/cxx/serializer/guide/index.xhtml +++ b/documentation/cxx/serializer/guide/index.xhtml @@ -2609,10 +2609,13 @@ private: </p> <p>The Embedded C++/Serializer mapping always expects character data - supplied by the application to be in the UTF-8 encoding. The - underlying XML serializer used by the Embedded C++/Serializer - mapping produces the resulting XML in the UTF-8 encoding as - well.</p> + supplied by the application to be in the same encoding. The + application encoding can either be UTF-8 (default) or ISO-8859-1. + To select a particular encoding, configure the XSD/e runtime library + accordingly and pass the <code>--char-encoding</code> option to the + XSD/e compiler when translating your schemas. The underlying XML + serializer used by the Embedded C++/Serializer mapping produces + the resulting XML documents in the UTF-8 encoding.</p> <h2><a name="6.1">6.1 Standard Template Library</a></h2> diff --git a/documentation/xsde.1 b/documentation/xsde.1 index 0c0dc72..afafa8c 100644 --- a/documentation/xsde.1 +++ b/documentation/xsde.1 @@ -175,6 +175,19 @@ Write generated files to .I dir instead of the current directory. +.IP "\fB\--char-encoding \fIenc\fR" +Specify the application character encoding. Valid values are +.B utf8 +(default) and +.BR iso8859-1 . +Note that this encoding is not the same as the XML document encoding +that is being parsed or serialized. Rather, it is the encoding that +is used inside the application. When an XML document is parsed, the +character data is automatically converted to the application encoding. +Similarly, when an XML document is serialized, the data in the +application encoding is automatically converted to the resulting +document encoding. + .IP "\fB\--no-stl\fR" Generate code that does not use the Standard Template Library (STL). diff --git a/documentation/xsde.xhtml b/documentation/xsde.xhtml index 9328e99..31c2f03 100644 --- a/documentation/xsde.xhtml +++ b/documentation/xsde.xhtml @@ -166,6 +166,17 @@ <dd>Write generated files to <code><i>dir</i></code> instead of the current directory.</dd> + <dt><code><b>--char-encoding</b> <i>enc</i></code></dt> + <dd>Specify the application character encoding. Valid values are + <code><b>utf8</b></code> (default) and <code><b>iso8859-1</b></code>. + Note that this encoding is not the same as the XML document encoding + that is being parsed or serialized. Rather, it is the encoding that + is used inside the application. When an XML document is parsed, the + character data is automatically converted to the application encoding. + Similarly, when an XML document is serialized, the data in the + application encoding is automatically converted to the resulting + document encoding.</dd> + <dt><code><b>--no-stl</b></code></dt> <dd>Generate code that does not use the Standard Template Library (STL).</dd> diff --git a/libxsde/xsde/cxx/iso8859-1.cxx b/libxsde/xsde/cxx/iso8859-1.cxx new file mode 100644 index 0000000..b52a284 --- /dev/null +++ b/libxsde/xsde/cxx/iso8859-1.cxx @@ -0,0 +1,170 @@ +// file : xsde/cxx/iso8859-1.cxx +// author : Boris Kolpackov <boris@codesynthesis.com> +// copyright : Copyright (c) 2005-2010 Code Synthesis Tools CC +// license : GNU GPL v2 + exceptions; see accompanying LICENSE file + +#include <xsde/cxx/iso8859-1.hxx> + +namespace xsde +{ + namespace cxx + { + char iso8859_1::unrep_char_ = '\0'; + + bool iso8859_1:: + to (const char* s, char* r) + { + for (; *s != 0; ++s) + { + unsigned char c (*s); + + if (c < 0x80) + { + // Fast path. + // + *r++ = c; + continue; + } + else if ((c >> 5) == 0x06) + { + // UTF-8: 110yyyyy 10zzzzzz + // Unicode: 00000yyy yyzzzzzz + // + unsigned int u = (c & 0x1F) << 6; + c = *++s; + u |= c & 0x3F; + + if (u < 0x100) + { + *r++ = static_cast<char> (u); + continue; + } + } + else if ((c >> 4) == 0x0E) + { + // UTF-8: 1110xxxx 10yyyyyy 10zzzzzz + // Unicode: xxxxyyyy yyzzzzzz + // + s += 2; + } + else if ((c >> 3) == 0x1E) + { + // UTF-8: 000wwwxx xxxxyyyy yyzzzzzz + // Unicode: 11110www 10xxxxxx 10yyyyyy 10zzzzzz + // + s += 3; + } + + if (unrep_char_ != '\0') + *r++ = unrep_char_; + else + return false; + } + + *r = '\0'; + + return true; + } + + bool iso8859_1:: + to (const char* s, size_t n, char* r) + { + const char* end = s + n; + + for (; s != end; ++s) + { + unsigned char c (*s); + + if (c < 0x80) + { + // Fast path. + // + *r++ = c; + continue; + } + else if ((c >> 5) == 0x06) + { + // UTF-8: 110yyyyy 10zzzzzz + // Unicode: 00000yyy yyzzzzzz + // + unsigned int u = (c & 0x1F) << 6; + c = *++s; + u |= c & 0x3F; + + if (u < 0x100) + { + *r++ = static_cast<char> (u); + continue; + } + } + else if ((c >> 4) == 0x0E) + { + // UTF-8: 1110xxxx 10yyyyyy 10zzzzzz + // Unicode: xxxxyyyy yyzzzzzz + // + s += 2; + } + else if ((c >> 3) == 0x1E) + { + // UTF-8: 000wwwxx xxxxyyyy yyzzzzzz + // Unicode: 11110www 10xxxxxx 10yyyyyy 10zzzzzz + // + s += 3; + } + + if (unrep_char_ != '\0') + *r++ = unrep_char_; + else + return false; + } + + *r = '\0'; + + return true; + } + + void iso8859_1:: + from (const char* s, char* r) + { + for (; *s != 0; ++s) + { + unsigned char c (*s); + + if (c < 0x80) + *r++ = c; + else + { + // 2-byte sequence. + // + *r++ = (c >> 6) | 0xC0; + *r++ = (c | 0x80) & 0xBF; + } + } + + *r = '\0'; + } + + void iso8859_1:: + from (const char* s, size_t n, char* r) + { + const char* end = s + n; + + for (; s != end; ++s) + { + unsigned char c (*s); + + if (c < 0x80) + *r++ = c; + else + { + // 2-byte sequence. + // + *r++ = (c >> 6) | 0xC0; + *r++ = (c | 0x80) & 0xBF; + } + } + + *r = '\0'; + } + } +} diff --git a/libxsde/xsde/cxx/iso8859-1.hxx b/libxsde/xsde/cxx/iso8859-1.hxx new file mode 100644 index 0000000..5809886 --- /dev/null +++ b/libxsde/xsde/cxx/iso8859-1.hxx @@ -0,0 +1,85 @@ +// file : xsde/cxx/iso8859-1.hxx +// author : Boris Kolpackov <boris@codesynthesis.com> +// copyright : Copyright (c) 2005-2010 Code Synthesis Tools CC +// license : GNU GPL v2 + exceptions; see accompanying LICENSE file + +#ifndef XSDE_CXX_ISO8859_1_HXX +#define XSDE_CXX_ISO8859_1_HXX + +#include <stddef.h> // size_t + +#include <xsde/cxx/config.hxx> + +namespace xsde +{ + namespace cxx + { + class iso8859_1 + { + public: + // Get/set a replacement for unrepresentable characters. If set to + // 0 (the default value), the conversion fails if such a character + // is encountered. + // + static char + unrep_char () + { + return unrep_char_; + } + + static void + unrep_char (char c) + { + unrep_char_ = c; + } + + public: + // Test whether a UTF-8 string is an ASCII string. If these + // functions return false, the iso_size argument contains + // the size of the buffer needed to convert this string to + // ISO-8859-1. + // + static bool + ascii_utf (const char* utf_str, size_t& iso_size); + + static bool + ascii_utf (const char* utf_str, size_t utf_size, size_t& iso_size); + + // Test whether an ISO-8859-1 string is an ASCII string. If + // these functions return false, the utf_size argument contains + // the size of the buffer needed to convert this string to UTF-8. + // + static bool + ascii_iso (const char* iso_str, size_t& utf_size); + + static bool + ascii_iso (const char* iso_str, size_t iso_size, size_t& utf_size); + + // Convert from UTF-8 to ISO-8859-1. Return false if an + // unrepresentable character was encountered and no replacement + // was specified. + // + // + static bool + to (const char* utf_str, char* iso_str); + + static bool + to (const char* utf_str, size_t utf_size, char* iso_str); + + // Convert from ISO-8859-1 to UTF-8. + // + static void + from (const char* iso_str, char* utf_str); + + static void + from (const char* iso_str, size_t iso_size, char* utf_str); + + private: + static char unrep_char_; + }; + } +} + +#include <xsde/cxx/iso8859-1.ixx> + +#endif // XSDE_CXX_ISO8859_1_HXX diff --git a/libxsde/xsde/cxx/iso8859-1.ixx b/libxsde/xsde/cxx/iso8859-1.ixx new file mode 100644 index 0000000..716be53 --- /dev/null +++ b/libxsde/xsde/cxx/iso8859-1.ixx @@ -0,0 +1,128 @@ +// file : xsde/cxx/iso8859-1.ixx +// author : Boris Kolpackov <boris@codesynthesis.com> +// copyright : Copyright (c) 2005-2010 Code Synthesis Tools CC +// license : GNU GPL v2 + exceptions; see accompanying LICENSE file + +namespace xsde +{ + namespace cxx + { + inline bool iso8859_1:: + ascii_utf (const char* s, size_t& rn) + { + // NULL string is an ASCII string. + // + if (s == 0) + return true; + + bool r = true; + size_t l = 0; + + for (; *s != 0; ++s, ++l) + { + unsigned char c (*s); + + if (c > 0x7F) + { + if ((c >> 5) == 0x06) + ++s; + else if ((c >> 4) == 0x0E) + s += 2; + else if ((c >> 3) == 0x1E) + s += 3; + + r = false; + } + } + + rn = l + 1; + return r; + } + + inline bool iso8859_1:: + ascii_utf (const char* s, size_t n, size_t& rn) + { + // NULL string is an ASCII string. + // + if (s == 0) + return true; + + bool r = true; + size_t l = 0; + const char* end = s + n; + + for (; s != end; ++s, ++l) + { + unsigned char c (*s); + + if (c > 0x7F) + { + if ((c >> 5) == 0x06) + ++s; + else if ((c >> 4) == 0x0E) + s += 2; + else if ((c >> 3) == 0x1E) + s += 3; + + r = false; + } + } + + rn = l + 1; + return r; + } + + inline bool iso8859_1:: + ascii_iso (const char* s, size_t& rn) + { + // NULL string is an ASCII string. + // + if (s == 0) + return true; + + bool r = true; + size_t l = 0; + + for (; *s != 0; ++s, ++l) + { + unsigned char c (*s); + + if (c > 0x7F) + { + ++l; // 2-byte surrogate. + r = false; + } + } + + rn = l + 1; + return r; + } + + inline bool iso8859_1:: + ascii_iso (const char* s, size_t n, size_t& rn) + { + // NULL string is an ASCII string. + // + if (s == 0) + return true; + + bool r = true; + size_t l = 0; + const char* end = s + n; + + for (; s != end; ++s, ++l) + { + unsigned char c (*s); + + if (c > 0x7F) + { + ++l; // 2-byte surrogate. + r = false; + } + } + + rn = l + 1; + return r; + } + } +} diff --git a/libxsde/xsde/cxx/parser/expat/document.cxx b/libxsde/xsde/cxx/parser/expat/document.cxx index 4bb3ec0..61cc1d0 100644 --- a/libxsde/xsde/cxx/parser/expat/document.cxx +++ b/libxsde/xsde/cxx/parser/expat/document.cxx @@ -17,6 +17,10 @@ # include <fstream> #endif +#ifdef XSDE_ENCODING_ISO8859_1 +# include <xsde/cxx/iso8859-1.hxx> +#endif + #ifdef XSDE_EXCEPTIONS # include <xsde/cxx/parser/exceptions.hxx> #endif @@ -58,6 +62,9 @@ namespace xsde #endif : first_ (true), xml_parser_ (0), context_ (0), parser_ (&p) { +#ifdef XSDE_ENCODING_ISO8859_1 + xml_error_ = 0; +#endif #ifdef XSDE_POLYMORPHIC polymorphic_ = poly; #endif @@ -77,6 +84,9 @@ namespace xsde #endif : first_ (true), xml_parser_ (0), context_ (0), parser_ (&p) { +#ifdef XSDE_ENCODING_ISO8859_1 + xml_error_ = 0; +#endif #ifdef XSDE_POLYMORPHIC polymorphic_ = poly; #endif @@ -93,6 +103,9 @@ namespace xsde #endif : first_ (true), xml_parser_ (0), context_ (0), parser_ (&p) { +#ifdef XSDE_ENCODING_ISO8859_1 + xml_error_ = 0; +#endif #ifdef XSDE_POLYMORPHIC polymorphic_ = poly; #endif @@ -110,6 +123,9 @@ namespace xsde #endif : first_ (true), xml_parser_ (0), context_ (0), parser_ (&p) { +#ifdef XSDE_ENCODING_ISO8859_1 + xml_error_ = 0; +#endif #ifdef XSDE_POLYMORPHIC polymorphic_ = poly; #endif @@ -120,6 +136,9 @@ namespace xsde document_pimpl () : first_ (true), xml_parser_ (0), context_ (0), parser_ (0) { +#ifdef XSDE_ENCODING_ISO8859_1 + xml_error_ = 0; +#endif #ifdef XSDE_POLYMORPHIC polymorphic_ = false; #endif @@ -131,6 +150,9 @@ namespace xsde : first_ (true), xml_parser_ (0), context_ (0), parser_ (0), polymorphic_ (true) { +#ifdef XSDE_ENCODING_ISO8859_1 + xml_error_ = 0; +#endif init_root_name (0, n); } @@ -139,6 +161,9 @@ namespace xsde : first_ (true), xml_parser_ (0), context_ (0), parser_ (0), polymorphic_ (true) { +#ifdef XSDE_ENCODING_ISO8859_1 + xml_error_ = 0; +#endif init_root_name (ns, n); } @@ -148,6 +173,9 @@ namespace xsde : first_ (true), xml_parser_ (0), context_ (0), parser_ (0), polymorphic_ (true) { +#ifdef XSDE_ENCODING_ISO8859_1 + xml_error_ = 0; +#endif init_root_name (0, n.c_str ()); } @@ -156,6 +184,9 @@ namespace xsde : first_ (true), xml_parser_ (0), context_ (0), parser_ (0), polymorphic_ (true) { +#ifdef XSDE_ENCODING_ISO8859_1 + xml_error_ = 0; +#endif init_root_name (ns.c_str (), n.c_str ()); } #endif // XSDE_STL @@ -208,6 +239,10 @@ namespace xsde #ifndef XSDE_EXCEPTIONS error_ = error (); #endif + +#ifdef XSDE_ENCODING_ISO8859_1 + xml_error_ = 0; +#endif first_ = true; if (parser_) @@ -405,6 +440,18 @@ namespace xsde // if (e == XML_ERROR_ABORTED) { +#ifdef XSDE_ENCODING_ISO8859_1 + if (xml_error_ != 0) + { +#ifdef XSDE_EXCEPTIONS + throw xml (xml_error_, l, c); +#else + error_ = error (xml_error_, l, c); + return; +#endif + } +#endif + // Got to be either a system, schema, or application // level error. // @@ -584,11 +631,35 @@ namespace xsde return; } + // Convert to application encoding. + // +#ifdef XSDE_ENCODING_UTF8 + const char* ans_name = ns_name; +#elif defined(XSDE_ENCODING_ISO8859_1) + const char* ans_name; + size_t an; + string buf; + + if (iso8859_1::ascii_utf (ns_name, an)) + ans_name = ns_name; + else + { + ans_name = conv_name (ns_name, an, buf); + + if (ans_name == 0) + { + XML_StopParser (xml_parser_, false); + return; + } + } + +#endif // XSDE_ENCODING_* + const char* ns_p; const char* name_p; size_t ns_s, name_s; - bits::split_name (ns_name, ns_p, ns_s, name_p, name_s); + bits::split_name (ans_name, ns_p, ns_s, name_p, name_s); parser_state& cur = context_.current_; @@ -602,7 +673,10 @@ namespace xsde if (polymorphic_) { - // Search for the xsi:type attribute. + // Search for the xsi:type attribute. The name and + // namespace of the attribute that we are looking for + // are ASCII strings so we don't need to convert it to + // the application encoding. // const XML_Char** p = atts; // VC8 can't handle p (atts) for (; *p != 0; p += 2) @@ -622,9 +696,34 @@ namespace xsde // @@ Need proper QName validation. // + + // Convert the type name to the application encoding. + // +#ifdef XSDE_ENCODING_UTF8 + const char* av = *(p + 1); +#elif defined(XSDE_ENCODING_ISO8859_1) + const char* av; + size_t an; + string buf; + + if (iso8859_1::ascii_utf (*(p + 1), an)) + av = *(p + 1); + else + { + av = conv_data (*(p + 1), an, buf); + + if (av == 0) + { + XML_StopParser (xml_parser_, false); + return; + } + } + +#endif // XSDE_ENCODING_* + // Get the qualified type name and try to resolve it. // - ro_string qn (*(p + 1)); + ro_string qn (av); ro_string tp, tn; size_t pos = qn.find (':'); @@ -834,10 +933,53 @@ namespace xsde { for (; *atts != 0; atts += 2) { - bits::split_name (*atts, ns_p, ns_s, name_p, name_s); + // Convert the type name to the application encoding. + // +#ifdef XSDE_ENCODING_UTF8 + const char* aname = *atts; + const char* avalue = *(atts + 1); +#elif defined(XSDE_ENCODING_ISO8859_1) + const char* aname; + const char* avalue; + size_t an; + string nbuf, vbuf; + + // Name. + // + if (iso8859_1::ascii_utf (*atts, an)) + aname = *atts; + else + { + aname = conv_name (*atts, an, nbuf); + + if (aname == 0) + { + XML_StopParser (xml_parser_, false); + return; + } + } + + // Value. + // + if (iso8859_1::ascii_utf (*(atts + 1), an)) + avalue = *(atts + 1); + else + { + avalue = conv_data (*(atts + 1), an, vbuf); + + if (avalue == 0) + { + XML_StopParser (xml_parser_, false); + return; + } + } + +#endif // XSDE_ENCODING_* + + bits::split_name (aname, ns_p, ns_s, name_p, name_s); const ro_string ns (ns_p, ns_s), name (name_p, name_s); - const ro_string value (*(atts + 1)); + const ro_string value (avalue); if (!cur.any_) cur.parser_->_attribute (ns, name, value); @@ -867,19 +1009,41 @@ namespace xsde return; } + // Convert to application encoding. + // +#ifdef XSDE_ENCODING_UTF8 + const char* ans_name = ns_name; +#elif defined(XSDE_ENCODING_ISO8859_1) + const char* ans_name; + size_t an; + string buf; + + if (iso8859_1::ascii_utf (ns_name, an)) + ans_name = ns_name; + else + { + ans_name = conv_name (ns_name, an, buf); + + if (ans_name == 0) + { + XML_StopParser (xml_parser_, false); + return; + } + } + +#endif // XSDE_ENCODING_* + const char* ns_p; const char* name_p; size_t ns_s, name_s; - bits::split_name (ns_name, ns_p, ns_s, name_p, name_s); + bits::split_name (ans_name, ns_p, ns_s, name_p, name_s); const ro_string ns (ns_p, ns_s); const ro_string name (name_p, name_s); parser_state& cur = context_.current_; - // @@ Error propagation. - // if (cur.depth_ == 0) { // The "normal" case: call _post to pop the parser and then @@ -980,7 +1144,37 @@ namespace xsde if (n != 0 && (cur.depth_ == 0 || cur.any_)) { - const ro_string str (s, n); + // Convert to application encoding. + // +#ifdef XSDE_ENCODING_UTF8 + const char* as = s; + size_t an = n; +#elif defined(XSDE_ENCODING_ISO8859_1) + const char* as; + size_t an; + string buf; + + if (iso8859_1::ascii_utf (s, n, an)) + { + as = s; + an = n; + } + else + { + as = conv_data (s, n, an, buf); + + if (as == 0) + { + XML_StopParser (xml_parser_, false); + return; + } + + --an; // Discount trailing zero. + } + +#endif // XSDE_ENCODING_* + + const ro_string str (as, an); if (!cur.any_) cur.parser_->_characters (str); @@ -1003,22 +1197,66 @@ namespace xsde // if (polymorphic_) { -#if defined (XSDE_STL) - prefixes_.push_back (p ? p : ""); - prefix_namespaces_.push_back (ns ? ns : ""); -#else -#if defined (XSDE_EXCEPTIONS) - prefixes_.push_back_copy (p ? p : ""); - prefix_namespaces_.push_back_copy (ns ? ns : ""); + // Convert the type name to the application encoding. + // +#ifdef XSDE_ENCODING_UTF8 + const char* ap = p; + const char* ans = ns; +#elif defined(XSDE_ENCODING_ISO8859_1) + const char* ap; + const char* ans; + size_t an; + string pbuf, nsbuf; + + // Prefix. + // + if (p == 0) + ap = 0; + else if (iso8859_1::ascii_utf (p, an)) + ap = p; + else + { + ap = conv_name (p, an, pbuf); + + if (ap == 0) + { + XML_StopParser (xml_parser_, false); + return; + } + } + + // Namespace. + // + if (ns == 0) + ans = 0; + else if (iso8859_1::ascii_utf (ns, an)) + ans = ns; + else + { + ans = conv_data (ns, an, nsbuf); + + if (ans == 0) + { + XML_StopParser (xml_parser_, false); + return; + } + } +#endif // XSDE_ENCODING_* + +#ifdef XSDE_STL + prefixes_.push_back (ap ? ap : ""); + prefix_namespaces_.push_back (ans ? ans : ""); +#elif defined (XSDE_EXCEPTIONS) + prefixes_.push_back_copy (ap ? ap : ""); + prefix_namespaces_.push_back_copy (ans ? ans : ""); #else - if (prefixes_.push_back_copy (p ? p : "" ) || - prefix_namespaces_.push_back_copy (ns ? ns : "")) + if (prefixes_.push_back_copy (ap ? ap : "" ) || + prefix_namespaces_.push_back_copy (ans ? ans : "")) { context_.sys_error (sys_error::no_memory); XML_StopParser (xml_parser_, false); } #endif -#endif } } @@ -1038,6 +1276,105 @@ namespace xsde } } #endif // XSDE_POLYMORPHIC + + // + // Support for ISO-8859-1 conversion. + // + +#ifdef XSDE_ENCODING_ISO8859_1 + const char* document_pimpl:: + conv_data (const XML_Char* utf_s, size_t iso_n, string& var) + { + char* buf; + + if (iso_n <= sizeof (data_buf_)) + buf = data_buf_; + else + { + buf = new char[iso_n]; + +#ifndef XSDE_EXCEPTIONS + if (buf == 0) + { + context_.sys_error (sys_error::no_memory); + return 0; + } +#endif + var.attach (buf, iso_n - 1); + } + + if (!iso8859_1::to (utf_s, buf)) + { + xml_error_ = XML_ERROR_UNREPRESENTABLE; + return 0; + } + + return buf; + } + + const char* document_pimpl:: + conv_data (const XML_Char* utf_s, + size_t utf_n, + size_t iso_n, + string& var) + { + char* buf; + + if (iso_n <= sizeof (data_buf_)) + buf = data_buf_; + else + { + buf = new char[iso_n]; + +#ifndef XSDE_EXCEPTIONS + if (buf == 0) + { + context_.sys_error (sys_error::no_memory); + return 0; + } +#endif + var.attach (buf, iso_n - 1); + } + + if (!iso8859_1::to (utf_s, utf_n, buf)) + { + xml_error_ = XML_ERROR_UNREPRESENTABLE; + return 0; + } + + return buf; + } + + const char* document_pimpl:: + conv_name (const XML_Char* utf_s, size_t iso_n, string& var) + { + char* buf; + + if (iso_n <= sizeof (name_buf_)) + buf = name_buf_; + else + { + buf = new char[iso_n]; + +#ifndef XSDE_EXCEPTIONS + if (buf == 0) + { + context_.sys_error (sys_error::no_memory); + return 0; + } +#endif + var.attach (buf, iso_n - 1); + } + + if (!iso8859_1::to (utf_s, buf)) + { + xml_error_ = XML_ERROR_UNREPRESENTABLE; + return 0; + } + + return buf; + } +#endif // XSDE_ENCODING_ISO8859_1 } } } diff --git a/libxsde/xsde/cxx/parser/expat/document.hxx b/libxsde/xsde/cxx/parser/expat/document.hxx index afa7625..df3bf21 100644 --- a/libxsde/xsde/cxx/parser/expat/document.hxx +++ b/libxsde/xsde/cxx/parser/expat/document.hxx @@ -323,6 +323,27 @@ namespace xsde error error_; #endif + // Support for ISO-8859-1 conversion. + // +#ifdef XSDE_ENCODING_ISO8859_1 + protected: + const char* + conv_data (const XML_Char* utf_s, size_t iso_n, string& var); + + const char* + conv_data (const XML_Char* utf_s, + size_t utf_n, + size_t iso_n, + string& var); + + const char* + conv_name (const XML_Char* utf_s, size_t iso_n, string& var); + + xml_error xml_error_; + + char data_buf_[256]; + char name_buf_[128]; +#endif private: void init_root_name (const char* ns, const char* name); diff --git a/libxsde/xsde/cxx/parser/expat/document.ixx b/libxsde/xsde/cxx/parser/expat/document.ixx index 72451ee..16d6ef7 100644 --- a/libxsde/xsde/cxx/parser/expat/document.ixx +++ b/libxsde/xsde/cxx/parser/expat/document.ixx @@ -50,4 +50,3 @@ namespace xsde } } } - diff --git a/libxsde/xsde/cxx/parser/expat/xml-error.cxx b/libxsde/xsde/cxx/parser/expat/xml-error.cxx index cb7b904..b5f15a4 100644 --- a/libxsde/xsde/cxx/parser/expat/xml-error.cxx +++ b/libxsde/xsde/cxx/parser/expat/xml-error.cxx @@ -13,13 +13,19 @@ namespace xsde { namespace expat { + static const char* const text_[] = + { + "character unrepresentable in application encoding" + }; + const char* xml_error_text (xml_error e) { - return XML_ErrorString (e); + return e < 0x10000 + ? XML_ErrorString (static_cast<XML_Error> (e)) + : text_[e & 0xFFFF]; } } } } } - diff --git a/libxsde/xsde/cxx/parser/expat/xml-error.hxx b/libxsde/xsde/cxx/parser/expat/xml-error.hxx index 6e28687..5871b48 100644 --- a/libxsde/xsde/cxx/parser/expat/xml-error.hxx +++ b/libxsde/xsde/cxx/parser/expat/xml-error.hxx @@ -8,6 +8,11 @@ #include <xsde/c/expat/expat.h> +enum XML_Error_Ex +{ + XML_ERROR_UNREPRESENTABLE = 0x10000 +}; + namespace xsde { namespace cxx @@ -16,7 +21,10 @@ namespace xsde { namespace expat { - typedef XML_Error xml_error; + // The xml_error type can hold values from XML_Error and + // XML_Error_Ex enums. + // + typedef unsigned int xml_error; const char* xml_error_text (xml_error); diff --git a/libxsde/xsde/cxx/serializer/context.cxx b/libxsde/xsde/cxx/serializer/context.cxx index 650bed2..faa1ed4 100644 --- a/libxsde/xsde/cxx/serializer/context.cxx +++ b/libxsde/xsde/cxx/serializer/context.cxx @@ -3,6 +3,8 @@ // copyright : Copyright (c) 2005-2010 Code Synthesis Tools CC // license : GNU GPL v2 + exceptions; see accompanying LICENSE file +#include <assert.h> + #include <xsde/cxx/config.hxx> #include <xsde/cxx/ro-string.hxx> @@ -11,6 +13,10 @@ # include <xsde/cxx/serializer/exceptions.hxx> #endif +#ifdef XSDE_ENCODING_ISO8859_1 +# include <xsde/cxx/iso8859-1.hxx> +#endif + #include <xsde/cxx/serializer/context.hxx> namespace xsde @@ -19,6 +25,625 @@ namespace xsde { namespace serializer { + // + // +#ifdef XSDE_EXCEPTIONS + void context:: +#else + bool context:: +#endif + start_element (const char* name) + { + // Convert from application encoding. + // +#ifdef XSDE_ENCODING_UTF8 + const char* uname = name; +#elif defined(XSDE_ENCODING_ISO8859_1) + const char* uname; + size_t un; + string buf; + + if (iso8859_1::ascii_iso (name, un)) + uname = name; + else + { + uname = conv_name (name, un, name_buf1_, buf); + +#ifndef XSDE_EXCEPTIONS + if (uname == 0) + return false; +#endif + } +#endif // XSDE_ENCODING_* + + genxStatus e = genxStartElementLiteral ( + xml_serializer_, + 0, + reinterpret_cast<constUtf8> (uname)); + +#ifdef XSDE_EXCEPTIONS + if (e != GENX_SUCCESS) + throw_xml_error (e); +#else + if (e != GENX_SUCCESS) + xml_error (e); + + return e == GENX_SUCCESS; +#endif + } + + // + // +#ifdef XSDE_EXCEPTIONS + void context:: +#else + bool context:: +#endif + start_element (const char* ns, const char* name) + { + // Convert from application encoding. + // +#ifdef XSDE_ENCODING_UTF8 + const char* uns = ns; + const char* uname = name; +#elif defined(XSDE_ENCODING_ISO8859_1) + const char* uns; + const char* uname; + size_t un; + string buf1, buf2; + + if (iso8859_1::ascii_iso (ns, un)) + uns = ns; + else + { + uns = conv_name (ns, un, name_buf1_, buf1); + +#ifndef XSDE_EXCEPTIONS + if (uns == 0) + return false; +#endif + } + + if (iso8859_1::ascii_iso (name, un)) + uname = name; + else + { + uname = conv_name (name, un, name_buf2_, buf2); + +#ifndef XSDE_EXCEPTIONS + if (uname == 0) + return false; +#endif + } +#endif // XSDE_ENCODING_* + + genxStatus e = genxStartElementLiteral ( + xml_serializer_, + reinterpret_cast<constUtf8> (uns), + reinterpret_cast<constUtf8> (uname)); + +#ifdef XSDE_EXCEPTIONS + if (e != GENX_SUCCESS) + throw_xml_error (e); +#else + if (e != GENX_SUCCESS) + xml_error (e); + + return e == GENX_SUCCESS; +#endif + } + + // + // +#ifdef XSDE_EXCEPTIONS + void context:: +#else + bool context:: +#endif + start_attribute (const char* name) + { + // Convert from application encoding. + // +#ifdef XSDE_ENCODING_UTF8 + const char* uname = name; +#elif defined(XSDE_ENCODING_ISO8859_1) + const char* uname; + size_t un; + string buf; + + if (iso8859_1::ascii_iso (name, un)) + uname = name; + else + { + uname = conv_name (name, un, name_buf1_, buf); + +#ifndef XSDE_EXCEPTIONS + if (uname == 0) + return false; +#endif + } +#endif // XSDE_ENCODING_* + + genxStatus e = genxStartAttributeLiteral ( + xml_serializer_, + 0, + reinterpret_cast<constUtf8> (uname)); + +#ifdef XSDE_EXCEPTIONS + if (e != GENX_SUCCESS) + throw_xml_error (e); +#else + if (e != GENX_SUCCESS) + xml_error (e); + + return e == GENX_SUCCESS; +#endif + } + + // + // +#ifdef XSDE_EXCEPTIONS + void context:: +#else + bool context:: +#endif + start_attribute (const char* ns, const char* name) + { + // Convert from application encoding. + // +#ifdef XSDE_ENCODING_UTF8 + const char* uns = ns; + const char* uname = name; +#elif defined(XSDE_ENCODING_ISO8859_1) + const char* uns; + const char* uname; + size_t un; + string buf1, buf2; + + if (iso8859_1::ascii_iso (ns, un)) + uns = ns; + else + { + uns = conv_name (ns, un, name_buf1_, buf1); + +#ifndef XSDE_EXCEPTIONS + if (uns == 0) + return false; +#endif + } + + if (iso8859_1::ascii_iso (name, un)) + uname = name; + else + { + uname = conv_name (name, un, name_buf2_, buf2); + +#ifndef XSDE_EXCEPTIONS + if (uname == 0) + return false; +#endif + } +#endif // XSDE_ENCODING_* + + genxStatus e = genxStartAttributeLiteral ( + xml_serializer_, + reinterpret_cast<constUtf8> (uns), + reinterpret_cast<constUtf8> (uname)); + +#ifdef XSDE_EXCEPTIONS + if (e != GENX_SUCCESS) + throw_xml_error (e); +#else + if (e != GENX_SUCCESS) + xml_error (e); + + return e == GENX_SUCCESS; +#endif + } + + // + // +#ifdef XSDE_EXCEPTIONS + void context:: +#else + bool context:: +#endif + attribute (const char* name, const char* value) + { + // Convert from application encoding. + // +#ifdef XSDE_ENCODING_UTF8 + const char* uname = name; + const char* uvalue = value; +#elif defined(XSDE_ENCODING_ISO8859_1) + const char* uname; + const char* uvalue; + size_t un; + string buf1, buf2; + + if (iso8859_1::ascii_iso (name, un)) + uname = name; + else + { + uname = conv_name (name, un, name_buf1_, buf1); + +#ifndef XSDE_EXCEPTIONS + if (uname == 0) + return false; +#endif + } + + if (iso8859_1::ascii_iso (value, un)) + uvalue = value; + else + { + uvalue = conv_data (value, un, buf2); + +#ifndef XSDE_EXCEPTIONS + if (uvalue == 0) + return false; +#endif + } +#endif // XSDE_ENCODING_* + + genxStatus e = genxAddAttributeLiteral ( + xml_serializer_, + 0, + reinterpret_cast<constUtf8> (uname), + reinterpret_cast<constUtf8> (uvalue)); + +#ifdef XSDE_EXCEPTIONS + if (e != GENX_SUCCESS) + throw_xml_error (e); +#else + if (e != GENX_SUCCESS) + xml_error (e); + + return e == GENX_SUCCESS; +#endif + } + + // + // +#ifdef XSDE_EXCEPTIONS + void context:: +#else + bool context:: +#endif + attribute (const char* ns, const char* name, const char* value) + { + // Convert from application encoding. + // +#ifdef XSDE_ENCODING_UTF8 + const char* uns = ns; + const char* uname = name; + const char* uvalue = value; +#elif defined(XSDE_ENCODING_ISO8859_1) + const char* uns; + const char* uname; + const char* uvalue; + size_t un; + string buf1, buf2, buf3; + + if (iso8859_1::ascii_iso (ns, un)) + uns = ns; + else + { + uns = conv_name (ns, un, name_buf1_, buf1); + +#ifndef XSDE_EXCEPTIONS + if (uns == 0) + return false; +#endif + } + + if (iso8859_1::ascii_iso (name, un)) + uname = name; + else + { + uname = conv_name (name, un, name_buf2_, buf2); + +#ifndef XSDE_EXCEPTIONS + if (uname == 0) + return false; +#endif + } + + if (iso8859_1::ascii_iso (value, un)) + uvalue = value; + else + { + uvalue = conv_data (value, un, buf3); + +#ifndef XSDE_EXCEPTIONS + if (uvalue == 0) + return false; +#endif + } +#endif // XSDE_ENCODING_* + + genxStatus e = genxAddAttributeLiteral ( + xml_serializer_, + reinterpret_cast<constUtf8> (uns), + reinterpret_cast<constUtf8> (uname), + reinterpret_cast<constUtf8> (uvalue)); + +#ifdef XSDE_EXCEPTIONS + if (e != GENX_SUCCESS) + throw_xml_error (e); +#else + if (e != GENX_SUCCESS) + xml_error (e); + + return e == GENX_SUCCESS; +#endif + } + + // + // +#ifdef XSDE_EXCEPTIONS + void context:: +#else + bool context:: +#endif + characters (const char* s) + { + // Convert from application encoding. + // +#ifdef XSDE_ENCODING_UTF8 + const char* us = s; +#elif defined(XSDE_ENCODING_ISO8859_1) + const char* us; + size_t un; + string buf; + + if (iso8859_1::ascii_iso (s, un)) + us = s; + else + { + us = conv_data (s, un, buf); + +#ifndef XSDE_EXCEPTIONS + if (us == 0) + return false; +#endif + } +#endif // XSDE_ENCODING_* + + genxStatus e = genxAddText ( + xml_serializer_, + reinterpret_cast<constUtf8> (us)); + +#ifdef XSDE_EXCEPTIONS + if (e != GENX_SUCCESS) + throw_xml_error (e); +#else + if (e != GENX_SUCCESS) + xml_error (e); + + return e == GENX_SUCCESS; +#endif + } + + // + // +#ifdef XSDE_EXCEPTIONS + void context:: +#else + bool context:: +#endif + characters (const char* s, size_t n) + { + // Convert from application encoding. + // +#ifdef XSDE_ENCODING_UTF8 + const char* us = s; + size_t un = n; +#elif defined(XSDE_ENCODING_ISO8859_1) + const char* us; + size_t un; + string buf; + + if (iso8859_1::ascii_iso (s, n, un)) + { + us = s; + un = n; + } + else + { + us = conv_data (s, n, un, buf); + +#ifndef XSDE_EXCEPTIONS + if (us == 0) + return false; +#endif + --un; // Discount trailing zero. + } +#endif // XSDE_ENCODING_* + + genxStatus e = genxAddCountedText ( + xml_serializer_, + reinterpret_cast<constUtf8> (us), + static_cast<int> (un)); + +#ifdef XSDE_EXCEPTIONS + if (e != GENX_SUCCESS) + throw_xml_error (e); +#else + if (e != GENX_SUCCESS) + xml_error (e); + + return e == GENX_SUCCESS; +#endif + } + + // + // +#ifdef XSDE_EXCEPTIONS + void context:: +#else + bool context:: +#endif + declare_namespace (const char* ns, const char* prefix) + { + // Convert from application encoding. + // +#ifdef XSDE_ENCODING_UTF8 + const char* uns = ns; + const char* uprefix = prefix; +#elif defined(XSDE_ENCODING_ISO8859_1) + const char* uns; + const char* uprefix; + size_t un; + string buf1, buf2; + + if (iso8859_1::ascii_iso (ns, un)) + uns = ns; + else + { + uns = conv_name (ns, un, name_buf1_, buf1); + +#ifndef XSDE_EXCEPTIONS + if (uns == 0) + return false; +#endif + } + + if (iso8859_1::ascii_iso (prefix, un)) + uprefix = prefix; + else + { + uprefix = conv_name (prefix, un, name_buf2_, buf2); + +#ifndef XSDE_EXCEPTIONS + if (uprefix == 0) + return false; +#endif + } +#endif // XSDE_ENCODING_* + + genxStatus e = genxAddNamespaceLiteral ( + xml_serializer_, + reinterpret_cast<constUtf8> (uns), + reinterpret_cast<constUtf8> (uprefix)); + +#ifdef XSDE_EXCEPTIONS + if (e != GENX_SUCCESS) + throw_xml_error (e); +#else + if (e != GENX_SUCCESS) + xml_error (e); + + return e == GENX_SUCCESS; +#endif + } + + // + // +#ifdef XSDE_EXCEPTIONS + void context:: +#else + bool context:: +#endif + declare_default_namespace (const char* ns) + { + // Convert from application encoding. + // +#ifdef XSDE_ENCODING_UTF8 + const char* uns = ns; +#elif defined(XSDE_ENCODING_ISO8859_1) + const char* uns; + size_t un; + string buf; + + if (iso8859_1::ascii_iso (ns, un)) + uns = ns; + else + { + uns = conv_name (ns, un, name_buf1_, buf); + +#ifndef XSDE_EXCEPTIONS + if (uns == 0) + return false; +#endif + } +#endif // XSDE_ENCODING_* + + genxStatus e = genxAddNamespaceLiteral ( + xml_serializer_, + reinterpret_cast<constUtf8> (uns), + reinterpret_cast<constUtf8> ("")); + +#ifdef XSDE_EXCEPTIONS + if (e != GENX_SUCCESS) + throw_xml_error (e); +#else + if (e != GENX_SUCCESS) + xml_error (e); + + return e == GENX_SUCCESS; +#endif + } + + // + // + const char* context:: + lookup_namespace_prefix (const char* ns) + { + // Convert from application encoding. + // +#ifdef XSDE_ENCODING_UTF8 + const char* uns = ns; +#elif defined(XSDE_ENCODING_ISO8859_1) + const char* uns; + size_t un; + string buf; + + if (iso8859_1::ascii_iso (ns, un)) + uns = ns; + else + { + uns = conv_name (ns, un, name_buf1_, buf); + +#ifndef XSDE_EXCEPTIONS + if (uns == 0) + return false; +#endif + } +#endif // XSDE_ENCODING_* + + genxStatus e; + genxNamespace gns = genxDeclareNamespace ( + xml_serializer_, reinterpret_cast<constUtf8> (uns), 0, &e); + + if (e != GENX_SUCCESS) + { +#ifdef XSDE_EXCEPTIONS + throw_xml_error (e); +#else + xml_error (e); +#endif + return 0; + } + + const char* p = reinterpret_cast<const char*> ( + genxGetNamespacePrefix (gns)); + + // Convert to application encoding. + // +#ifdef XSDE_ENCODING_UTF8 + return p; +#elif defined(XSDE_ENCODING_ISO8859_1) + // At the moment we can only support ASCII prefixes since there is + // no good way to store the converted string. + // + assert (iso8859_1::ascii_utf (p, un)); + return p; +#endif // XSDE_ENCODING_* + } + + // + // #ifdef XSDE_POLYMORPHIC #ifdef XSDE_EXCEPTIONS void context:: @@ -102,6 +727,83 @@ namespace xsde } } #endif + +#ifdef XSDE_ENCODING_ISO8859_1 + const char* context:: + conv_data (const char* iso_s, size_t utf_n, string& var) + { + char* buf; + + if (utf_n <= sizeof (data_buf_)) + buf = data_buf_; + else + { + buf = new char[utf_n]; + +#ifndef XSDE_EXCEPTIONS + if (buf == 0) + { + sys_error (sys_error::no_memory); + return 0; + } +#endif + var.attach (buf, utf_n - 1); + } + + iso8859_1::from (iso_s, buf); + return buf; + } + + const char* context:: + conv_data (const char* iso_s, size_t iso_n, size_t utf_n, string& var) + { + char* buf; + + if (utf_n <= sizeof (data_buf_)) + buf = data_buf_; + else + { + buf = new char[utf_n]; + +#ifndef XSDE_EXCEPTIONS + if (buf == 0) + { + sys_error (sys_error::no_memory); + return 0; + } +#endif + var.attach (buf, utf_n - 1); + } + + iso8859_1::from (iso_s, iso_n, buf); + return buf; + } + + const char* context:: + conv_name (const char* iso_s, size_t utf_n, char* fix, string& var) + { + char* buf; + + if (utf_n <= sizeof (name_buf1_)) + buf = fix; + else + { + buf = new char[utf_n]; + +#ifndef XSDE_EXCEPTIONS + if (buf == 0) + { + sys_error (sys_error::no_memory); + return 0; + } +#endif + var.attach (buf, utf_n - 1); + } + + iso8859_1::from (iso_s, buf); + return buf; + } +#endif // XSDE_ENCODING_ISO8859_1 } } } diff --git a/libxsde/xsde/cxx/serializer/context.hxx b/libxsde/xsde/cxx/serializer/context.hxx index d09bb4d..d2cec00 100644 --- a/libxsde/xsde/cxx/serializer/context.hxx +++ b/libxsde/xsde/cxx/serializer/context.hxx @@ -11,6 +11,7 @@ #include <stddef.h> // size_t #include <xsde/c/genx/genx.h> +#include <xsde/cxx/string.hxx> #ifndef XSDE_EXCEPTIONS # include <xsde/cxx/sys-error.hxx> @@ -251,6 +252,23 @@ namespace xsde const void* type_id_; #endif + // Support for ISO-8859-1 conversion. + // +#ifdef XSDE_ENCODING_ISO8859_1 + protected: + const char* + conv_data (const char* iso_s, size_t utf_n, string& var); + + const char* + conv_data (const char* iso_s, size_t iso_n, size_t utf_n, string& var); + + const char* + conv_name (const char* iso_s, size_t utf_n, char* fix, string& var); + + char data_buf_[256]; + char name_buf1_[128]; + char name_buf2_[128]; // Keep buf1 and buf2 sizes the same. +#endif }; } } diff --git a/libxsde/xsde/cxx/serializer/context.ixx b/libxsde/xsde/cxx/serializer/context.ixx index 04cfe87..3612f0a 100644 --- a/libxsde/xsde/cxx/serializer/context.ixx +++ b/libxsde/xsde/cxx/serializer/context.ixx @@ -45,56 +45,6 @@ namespace xsde #else inline bool context:: #endif - start_element (const char* name) - { - genxStatus e = genxStartElementLiteral ( - xml_serializer_, - 0, - reinterpret_cast<constUtf8> (name)); - -#ifdef XSDE_EXCEPTIONS - if (e != GENX_SUCCESS) - throw_xml_error (e); -#else - if (e != GENX_SUCCESS) - xml_error (e); - - return e == GENX_SUCCESS; -#endif - } - - // - // -#ifdef XSDE_EXCEPTIONS - inline void context:: -#else - inline bool context:: -#endif - start_element (const char* ns, const char* name) - { - genxStatus e = genxStartElementLiteral ( - xml_serializer_, - reinterpret_cast<constUtf8> (ns), - reinterpret_cast<constUtf8> (name)); - -#ifdef XSDE_EXCEPTIONS - if (e != GENX_SUCCESS) - throw_xml_error (e); -#else - if (e != GENX_SUCCESS) - xml_error (e); - - return e == GENX_SUCCESS; -#endif - } - - // - // -#ifdef XSDE_EXCEPTIONS - inline void context:: -#else - inline bool context:: -#endif end_element () { genxStatus e = genxEndElement (xml_serializer_); @@ -117,56 +67,6 @@ namespace xsde #else inline bool context:: #endif - start_attribute (const char* name) - { - genxStatus e = genxStartAttributeLiteral ( - xml_serializer_, - 0, - reinterpret_cast<constUtf8> (name)); - -#ifdef XSDE_EXCEPTIONS - if (e != GENX_SUCCESS) - throw_xml_error (e); -#else - if (e != GENX_SUCCESS) - xml_error (e); - - return e == GENX_SUCCESS; -#endif - } - - // - // -#ifdef XSDE_EXCEPTIONS - inline void context:: -#else - inline bool context:: -#endif - start_attribute (const char* ns, const char* name) - { - genxStatus e = genxStartAttributeLiteral ( - xml_serializer_, - reinterpret_cast<constUtf8> (ns), - reinterpret_cast<constUtf8> (name)); - -#ifdef XSDE_EXCEPTIONS - if (e != GENX_SUCCESS) - throw_xml_error (e); -#else - if (e != GENX_SUCCESS) - xml_error (e); - - return e == GENX_SUCCESS; -#endif - } - - // - // -#ifdef XSDE_EXCEPTIONS - inline void context:: -#else - inline bool context:: -#endif end_attribute () { genxStatus e = genxEndAttribute (xml_serializer_); @@ -189,157 +89,6 @@ namespace xsde #else inline bool context:: #endif - attribute (const char* name, const char* value) - { - genxStatus e = genxAddAttributeLiteral ( - xml_serializer_, - 0, - reinterpret_cast<constUtf8> (name), - reinterpret_cast<constUtf8> (value)); - -#ifdef XSDE_EXCEPTIONS - if (e != GENX_SUCCESS) - throw_xml_error (e); -#else - if (e != GENX_SUCCESS) - xml_error (e); - - return e == GENX_SUCCESS; -#endif - } - - // - // -#ifdef XSDE_EXCEPTIONS - inline void context:: -#else - inline bool context:: -#endif - attribute (const char* ns, const char* name, const char* value) - { - genxStatus e = genxAddAttributeLiteral ( - xml_serializer_, - reinterpret_cast<constUtf8> (ns), - reinterpret_cast<constUtf8> (name), - reinterpret_cast<constUtf8> (value)); - -#ifdef XSDE_EXCEPTIONS - if (e != GENX_SUCCESS) - throw_xml_error (e); -#else - if (e != GENX_SUCCESS) - xml_error (e); - - return e == GENX_SUCCESS; -#endif - } - - // - // -#ifdef XSDE_EXCEPTIONS - inline void context:: -#else - inline bool context:: -#endif - characters (const char* s) - { - genxStatus e = genxAddText ( - xml_serializer_, - reinterpret_cast<constUtf8> (s)); - -#ifdef XSDE_EXCEPTIONS - if (e != GENX_SUCCESS) - throw_xml_error (e); -#else - if (e != GENX_SUCCESS) - xml_error (e); - - return e == GENX_SUCCESS; -#endif - } - - // - // -#ifdef XSDE_EXCEPTIONS - inline void context:: -#else - inline bool context:: -#endif - characters (const char* s, size_t n) - { - genxStatus e = genxAddCountedText ( - xml_serializer_, - reinterpret_cast<constUtf8> (s), - static_cast<int> (n)); - -#ifdef XSDE_EXCEPTIONS - if (e != GENX_SUCCESS) - throw_xml_error (e); -#else - if (e != GENX_SUCCESS) - xml_error (e); - - return e == GENX_SUCCESS; -#endif - } - - // - // -#ifdef XSDE_EXCEPTIONS - inline void context:: -#else - inline bool context:: -#endif - declare_namespace (const char* ns, const char* prefix) - { - genxStatus e = genxAddNamespaceLiteral ( - xml_serializer_, - reinterpret_cast<constUtf8> (ns), - reinterpret_cast<constUtf8> (prefix)); - -#ifdef XSDE_EXCEPTIONS - if (e != GENX_SUCCESS) - throw_xml_error (e); -#else - if (e != GENX_SUCCESS) - xml_error (e); - - return e == GENX_SUCCESS; -#endif - } - - // - // -#ifdef XSDE_EXCEPTIONS - inline void context:: -#else - inline bool context:: -#endif - declare_default_namespace (const char* ns) - { - genxStatus e = genxAddNamespaceLiteral ( - xml_serializer_, - reinterpret_cast<constUtf8> (ns), - reinterpret_cast<constUtf8> ("")); - -#ifdef XSDE_EXCEPTIONS - if (e != GENX_SUCCESS) - throw_xml_error (e); -#else - if (e != GENX_SUCCESS) - xml_error (e); - - return e == GENX_SUCCESS; -#endif - } - - // - // -#ifdef XSDE_EXCEPTIONS - inline void context:: -#else - inline bool context:: -#endif clear_default_namespace () { genxStatus e = genxUnsetDefaultNamespace (xml_serializer_); @@ -357,29 +106,6 @@ namespace xsde // // - inline const char* context:: - lookup_namespace_prefix (const char* ns) - { - genxStatus e; - genxNamespace gns = genxDeclareNamespace ( - xml_serializer_, reinterpret_cast<constUtf8> (ns), 0, &e); - - if (e != GENX_SUCCESS) - { -#ifdef XSDE_EXCEPTIONS - throw_xml_error (e); -#else - xml_error (e); -#endif - return 0; - } - - return reinterpret_cast<const char*> ( - genxGetNamespacePrefix (gns)); - } - - // - // #ifndef XSDE_EXCEPTIONS inline int context:: app_error () const diff --git a/libxsde/xsde/makefile b/libxsde/xsde/makefile index 468d19e..3a0465c 100644 --- a/libxsde/xsde/makefile +++ b/libxsde/xsde/makefile @@ -10,6 +10,10 @@ c_tun += c/genx/genx.c c/genx/char-props.c cxx_tun := cxx/string.cxx cxx/ro-string.cxx cxx/stack.cxx +ifeq ($(xsde_encoding),iso8859-1) +cxx_tun += cxx/iso8859-1.cxx +endif + ifeq ($(xsde_stl),n) cxx_tun += cxx/strdupx.cxx endif @@ -517,6 +521,16 @@ ifeq ($(xsde_byteorder),l) else @echo '#define XSDE_BYTEORDER 4321' >>$@ endif +ifeq ($(xsde_encoding),utf8) + @echo '#define XSDE_ENCODING_UTF8' >>$@ +else + @echo '#undef XSDE_ENCODING_UTF8' >>$@ +endif +ifeq ($(xsde_encoding),iso8859-1) + @echo '#define XSDE_ENCODING_ISO8859_1' >>$@ +else + @echo '#undef XSDE_ENCODING_ISO8859_1' >>$@ +endif ifeq ($(xsde_stl),y) @echo '#define XSDE_STL' >>$@ else diff --git a/tests/cxx/hybrid/iso8859-1/driver.cxx b/tests/cxx/hybrid/iso8859-1/driver.cxx new file mode 100644 index 0000000..ec68dbf --- /dev/null +++ b/tests/cxx/hybrid/iso8859-1/driver.cxx @@ -0,0 +1,95 @@ +// file : tests/cxx/hybrid/iso8859-1/driver.cxx +// author : Boris Kolpackov <boris@codesynthesis.com> +// copyright : Copyright (c) 2006-2010 Code Synthesis Tools CC +// license : GNU GPL v2 + exceptions; see accompanying LICENSE file + +// Test support for the ISO-8859-1 encoding. +// + +#include <assert.h> + +#include <string> +#include <iostream> + +#include "test.hxx" +#include "test-pimpl.hxx" +#include "test-simpl.hxx" + +#include <xsde/cxx/iso8859-1.hxx> + +using namespace std; +using namespace test; + +int +main (int argc, char* argv[]) +{ + if (argc != 2) + { + cerr << "usage: " << argv[0] << " test.xml" << endl; + return 1; + } + + // Parse. + // + root_paggr root_p; + + xml_schema::document_pimpl doc_p ( + root_p.root_parser (), + root_p.root_namespace (), + root_p.root_name ()); + +#ifdef XSDE_EXCEPTIONS + try + { + root_p.pre (); + doc_p.parse (argv[1]); + type* r = root_p.post (); + assert (false); + } + catch (const xml_schema::parser_exception& e) + { + // cerr << e.line () << ":" << e.column () << ": " << e.text () << endl; + } +#else + root_p.pre (); + doc_p.parse (argv[1]); + assert (doc_p._error ()); +#endif + + doc_p.reset (); + xml_schema::iso8859_1::unrep_char ('?'); + + root_p.pre (); + doc_p.parse (argv[1]); + type* r = root_p.post (); + + assert (r->a ()[0] == string ("abc")); + assert (r->a ()[1] == string ("\xE6")); + assert (r->a ()[2] == string ("\xA2\xA3\xA4\xA5")); + assert (r->a ()[3] == string ("???")); + assert (r->a ()[4] == string ("longlonglonglonglonglonglonglonglonglonglonglong\xE6longlonglonglong")); + + assert (r->b ()[0] == strenum::abc); + assert (r->b ()[1] == strenum::a_c); + assert (r->b ()[1].string () == string ("a\xE2""c")); + assert (r->b ()[2] == strenum::cxx__bc); + + assert (r->element_name_with_special_char__ () == string ("longlonglong\xA2\xA3\xA4\xA5longlong")); + + // Serialize. + // + root_saggr root_s; + + xml_schema::document_simpl doc_s ( + root_s.root_serializer (), + root_s.root_namespace (), + root_s.root_name ()); + + doc_s.add_prefix ("t", "test"); + + root_s.pre (*r); + doc_s.serialize (cout); + root_s.post (); + + delete r; +} diff --git a/tests/cxx/hybrid/iso8859-1/makefile b/tests/cxx/hybrid/iso8859-1/makefile new file mode 100644 index 0000000..2cc1f57 --- /dev/null +++ b/tests/cxx/hybrid/iso8859-1/makefile @@ -0,0 +1,108 @@ +# file : tests/cxx/hybrid/iso8859-1/makefile +# author : Boris Kolpackov <boris@codesynthesis.com> +# copyright : Copyright (c) 2006-2010 Code Synthesis Tools CC +# license : GNU GPL v2 + exceptions; see accompanying LICENSE file + +include $(dir $(lastword $(MAKEFILE_LIST)))../../../../build/bootstrap.make + +xsd := test.xsd +cxx := driver.cxx + +obj := $(addprefix $(out_base)/,\ +$(cxx:.cxx=.o) \ +$(xsd:.xsd=.o) \ +$(xsd:.xsd=-pskel.o) \ +$(xsd:.xsd=-pimpl.o) \ +$(xsd:.xsd=-sskel.o) \ +$(xsd:.xsd=-simpl.o)) + +dep := $(obj:.o=.o.d) + +xsde.l := $(out_root)/libxsde/xsde/xsde.l +xsde.l.cpp-options := $(out_root)/libxsde/xsde/xsde.l.cpp-options + +driver := $(out_base)/driver +test := $(out_base)/.test +dist := $(out_base)/.dist +dist-win := $(out_base)/.dist-win +clean := $(out_base)/.clean + + +# Build. +# +$(driver): $(obj) $(xsde.l) + +$(obj) $(dep): $(xsde.l.cpp-options) + +genf := $(xsd:.xsd=.hxx) $(xsd:.xsd=.cxx) \ + $(xsd:.xsd=-pskel.hxx) $(xsd:.xsd=-pskel.cxx) \ + $(xsd:.xsd=-pimpl.hxx) $(xsd:.xsd=-pimpl.cxx) \ + $(xsd:.xsd=-sskel.hxx) $(xsd:.xsd=-sskel.cxx) \ + $(xsd:.xsd=-simpl.hxx) $(xsd:.xsd=-simpl.cxx) + +gen := $(addprefix $(out_base)/,$(genf)) + +$(gen): $(out_root)/xsde/xsde +$(gen): xsde := $(out_root)/xsde/xsde +$(gen) $(dist) $(dist-win): xsde_options += --generate-parser \ +--generate-serializer --generate-aggregate + +$(call include-dep,$(dep)) + +# Convenience alias for default target. +# +$(out_base)/: $(driver) + + +# Test. +# +$(test): driver := $(driver) +$(test): $(driver) $(src_base)/test-000.xml $(src_base)/test-000.std + $(call message,test $$1,$$1 $(src_base)/test-000.xml | diff -u $(src_base)/test-000.std -,$(driver)) + + +# Dist. +# +$(dist) $(dist-win): opt := -src $(src_base) -cmd cxx-hybrid -xsd "$(xsd)" \ +-cxx "$(cxx)" -gen "$(genf)" -opt "$(xsde_options)" -out $(dist_prefix) + +$(dist): + $(call message,install $(src_base),$(scf_root)/dist $(opt)) + +$(dist-win): + $(call message,install $(src_base),$(scf_root)/dist -win $(opt)) + + +# Clean. +# +$(clean): $(driver).o.clean \ + $(addsuffix .cxx.clean,$(obj)) \ + $(addsuffix .cxx.clean,$(dep)) \ + $(addprefix $(out_base)/,$(xsd:.xsd=.cxx.xsd.clean)) + + +# Generated .gitignore. +# +ifeq ($(out_base),$(src_base)) +$(gen): | $(out_base)/.gitignore +$(driver): | $(out_base)/.gitignore + +$(out_base)/.gitignore: files := driver $(genf) +$(clean): $(out_base)/.gitignore.clean + +$(call include,$(bld_root)/git/gitignore.make) +endif + + +# How to. +# +$(call include,$(bld_root)/cxx/o-e.make) +$(call include,$(bld_root)/cxx/cxx-o.make) +$(call include,$(bld_root)/cxx/cxx-d.make) +$(call include,$(scf_root)/xsde/hybrid/xsd-cxx.make) + + +# Dependencies. +# +$(call import,$(src_root)/xsde/makefile) +$(call import,$(src_root)/libxsde/xsde/makefile) diff --git a/tests/cxx/hybrid/iso8859-1/test-000.std b/tests/cxx/hybrid/iso8859-1/test-000.std new file mode 100644 index 0000000..2631b99 --- /dev/null +++ b/tests/cxx/hybrid/iso8859-1/test-000.std @@ -0,0 +1 @@ +<t:root xmlns:t="test" x="longlonglong¢£¤¥longlong"><a>abc</a><a>æ</a><a>¢£¤¥</a><a>???</a><a>longlonglonglonglonglonglonglonglonglonglonglongælonglonglonglong</a><b>abc</b><b>aâc</b><b>âòbc</b><element_name_with_special_char_â>longlonglong¢£¤¥longlong</element_name_with_special_char_â></t:root>
\ No newline at end of file diff --git a/tests/cxx/hybrid/iso8859-1/test-000.xml b/tests/cxx/hybrid/iso8859-1/test-000.xml new file mode 100644 index 0000000..fd61446 --- /dev/null +++ b/tests/cxx/hybrid/iso8859-1/test-000.xml @@ -0,0 +1,15 @@ +<t:root xmlns:t="test" x="longlonglong¢£¤¥longlong"> + + <a>abc</a> + <a>æ</a> + <a>¢£¤¥</a> + <a>Āꪪ򪪪</a> + <a>longlonglonglonglonglonglonglonglonglonglonglongælonglonglonglong</a> + + <b>abc</b> + <b>aâc</b> + <b>âòbc</b> + + <element_name_with_special_char_â>longlonglong¢£¤¥longlong</element_name_with_special_char_â> + +</t:root> diff --git a/tests/cxx/hybrid/iso8859-1/test.xsd b/tests/cxx/hybrid/iso8859-1/test.xsd new file mode 100644 index 0000000..242b84a --- /dev/null +++ b/tests/cxx/hybrid/iso8859-1/test.xsd @@ -0,0 +1,23 @@ +<?xml version="1.0"?> +<schema xmlns="http://www.w3.org/2001/XMLSchema" xmlns:t="test" targetNamespace="test"> + + <simpleType name="strenum"> + <restriction base="string"> + <enumeration value="abc"/> + <enumeration value="aâc"/> + <enumeration value="âòbc"/> + </restriction> + </simpleType> + + <complexType name="type"> + <sequence> + <element name="a" type="string" maxOccurs="unbounded"/> + <element name="b" type="t:strenum" maxOccurs="unbounded"/> + <element name="element_name_with_special_char_â" type="string"/> + </sequence> + <attribute name="x" type="string"/> + </complexType> + + <element name="root" type="t:type"/> + +</schema> diff --git a/tests/cxx/hybrid/makefile b/tests/cxx/hybrid/makefile index 60960c3..8fb3f3f 100644 --- a/tests/cxx/hybrid/makefile +++ b/tests/cxx/hybrid/makefile @@ -10,7 +10,7 @@ include $(dir $(lastword $(MAKEFILE_LIST)))../../../build/bootstrap.make # all_tests := sequences polymorphism iterator built-in default enumeration \ -list recursive test-template union binary/cdr binary/xdr choice +iso8859-1 list recursive test-template union binary/cdr binary/xdr choice build_tests := sequences @@ -27,6 +27,10 @@ endif ifeq ($(xsde_iostream),y) build_tests += built-in default enumeration list test-template union +ifeq ($(xsde_encoding),iso8859-1) +build_tests += iso8859-1 +endif + ifeq ($(xsde_cdr),y) build_tests += binary/cdr endif diff --git a/xsde/cxx/elements.cxx b/xsde/cxx/elements.cxx index be65fc1..f8d7132 100644 --- a/xsde/cxx/elements.cxx +++ b/xsde/cxx/elements.cxx @@ -114,6 +114,7 @@ namespace CXX SemanticGraph::Schema& root, Char const* name_key, NarrowString const& char_type__, + NarrowString const& char_encoding__, Boolean include_with_brackets__, NarrowString const& include_prefix__, NarrowString const& esymbol, @@ -128,6 +129,7 @@ namespace CXX schema_root (root), ename_key (ename_key_), char_type (char_type_), + char_encoding (char_encoding_), L (L_), string_type (string_type_), include_with_brackets (include_with_brackets_), @@ -139,6 +141,7 @@ namespace CXX xs_ns_ (0), ename_key_ (name_key), char_type_ (char_type__), + char_encoding_ (char_encoding__), L_ (char_type == L"wchar_t" ? L"L" : L""), include_with_brackets_ (include_with_brackets__), include_prefix_ (include_prefix__), @@ -819,6 +822,126 @@ namespace CXX } String + strlit_iso8859_1 (String const& str) + { + String r; + Size n (str.size ()); + + // In most common cases we will have that many chars. + // + r.reserve (n + 2); + + r += '"'; + + Boolean escape (false); + + for (Size i (0); i < n; ++i) + { + UnsignedLong u (Context::unicode_char (str, i)); // May advance i. + + // [256 - ] - unrepresentable + // [127 - 255] - \xXX + // [32 - 126] - as is + // [0 - 31] - \X or \xXX + // + + if (u < 32) + { + switch (u) + { + case L'\n': + { + r += L"\\n"; + break; + } + case L'\t': + { + r += L"\\t"; + break; + } + case L'\v': + { + r += L"\\v"; + break; + } + case L'\b': + { + r += L"\\b"; + break; + } + case L'\r': + { + r += L"\\r"; + break; + } + case L'\f': + { + r += L"\\f"; + break; + } + case L'\a': + { + r += L"\\a"; + break; + } + default: + { + r += charlit (u); + escape = true; + break; + } + } + } + else if (u < 127) + { + if (escape) + { + // Close and open the string so there are no clashes. + // + r += '"'; + r += '"'; + + escape = false; + } + + switch (u) + { + case L'"': + { + r += L"\\\""; + break; + } + case L'\\': + { + r += L"\\\\"; + break; + } + default: + { + r += static_cast<WideChar> (u); + break; + } + } + } + else if (u < 256) + { + r += charlit (u); + escape = true; + } + else + { + // Unrepresentable character. + // + throw UnrepresentableCharacter (str, i + 1); + } + } + + r += '"'; + + return r; + } + + String strlit_utf32 (String const& str) { String r; @@ -936,7 +1059,12 @@ namespace CXX strlit (String const& str) { if (char_type == L"char") - return strlit_utf8 (str); + { + if (char_encoding == L"iso8859-1") + return strlit_iso8859_1 (str); + else + return strlit_utf8 (str); + } else return strlit_utf32 (str); } diff --git a/xsde/cxx/elements.hxx b/xsde/cxx/elements.hxx index d5c5a3c..67b51db 100644 --- a/xsde/cxx/elements.hxx +++ b/xsde/cxx/elements.hxx @@ -35,6 +35,30 @@ namespace CXX // Exceptions. // + struct UnrepresentableCharacter + { + UnrepresentableCharacter (String const& str, Size pos) + : str_ (str), pos_ (pos) + { + } + + String const& + string () const + { + return str_; + } + + Size + position () const + { + return pos_; + } + + private: + String str_; + Size pos_; + }; + struct NoNamespaceMapping { NoNamespaceMapping (SemanticGraph::Path const& file, @@ -124,6 +148,7 @@ namespace CXX SemanticGraph::Schema& root, Char const* name_key, NarrowString const& char_type__, + NarrowString const& char_encoding__, Boolean include_with_brackets__, NarrowString const& include_prefix__, NarrowString const& esymbol, @@ -141,6 +166,7 @@ namespace CXX schema_root (c.schema_root), ename_key (c.ename_key), char_type (c.char_type), + char_encoding (c.char_encoding), L (c.L), string_type (c.string_type), include_with_brackets (c.include_with_brackets), @@ -167,6 +193,7 @@ namespace CXX schema_root (c.schema_root), ename_key (c.ename_key), char_type (c.char_type), + char_encoding (c.char_encoding), L (c.L), string_type (c.string_type), include_with_brackets (c.include_with_brackets), @@ -296,6 +323,7 @@ namespace CXX NarrowString const& ename_key; String& char_type; + String& char_encoding; String& L; // string literal prefix String& string_type; @@ -315,6 +343,7 @@ namespace CXX NarrowString const ename_key_; String char_type_; + String char_encoding_; String L_; String string_type_; diff --git a/xsde/cxx/hybrid/cli.hxx b/xsde/cxx/hybrid/cli.hxx index ff89942..3834767 100644 --- a/xsde/cxx/hybrid/cli.hxx +++ b/xsde/cxx/hybrid/cli.hxx @@ -23,6 +23,7 @@ namespace CXX typedef Char const Key[]; + extern Key char_encoding; extern Key no_stl; extern Key no_iostream; extern Key no_exceptions; @@ -112,6 +113,7 @@ namespace CXX extern Key proprietary_license; typedef Cult::CLI::Options< + char_encoding, NarrowString, no_stl, Boolean, no_iostream, Boolean, no_exceptions, Boolean, diff --git a/xsde/cxx/hybrid/elements.cxx b/xsde/cxx/hybrid/elements.cxx index e08cf8b..27e983a 100644 --- a/xsde/cxx/hybrid/elements.cxx +++ b/xsde/cxx/hybrid/elements.cxx @@ -20,6 +20,7 @@ namespace CXX root, "name", "char", + ops.value<CLI::char_encoding> (), ops.value<CLI::include_with_brackets> (), ops.value<CLI::include_prefix> (), "", // export symbol diff --git a/xsde/cxx/hybrid/generator.cxx b/xsde/cxx/hybrid/generator.cxx index 311d5d7..14cba6a 100644 --- a/xsde/cxx/hybrid/generator.cxx +++ b/xsde/cxx/hybrid/generator.cxx @@ -111,6 +111,7 @@ namespace CXX { namespace CLI { + extern Key char_encoding = "char-encoding"; extern Key no_stl = "no-stl"; extern Key no_iostream = "no-iostream"; extern Key no_exceptions = "no-exceptions"; @@ -207,6 +208,12 @@ namespace CXX std::wostream& e (wcerr); ::CLI::Indent::Clip< ::CLI::OptionsUsage, WideChar> clip (e); + e << "--char-encoding <enc>" << endl + << " Specify the character encoding that should be\n" + << " used in the object model. Valid values are 'utf8'\n" + << " (default) and 'iso8859-1'." + << endl; + e << "--no-stl" << endl << " Generate code that does not use STL." << endl; @@ -704,6 +711,8 @@ namespace CXX { CLI::OptionsSpec spec; + spec.option<CLI::char_encoding> ().default_value ("utf8"); + spec.option<CLI::pskel_file_suffix> ().default_value ("-pskel"); spec.option<CLI::sskel_file_suffix> ().default_value ("-sskel"); spec.option<CLI::pskel_type_suffix> ().default_value ("_pskel"); @@ -814,6 +823,7 @@ namespace CXX Evptr<P::Options> r (new P::Options); + r->value<P::char_encoding> () = h.value<H::char_encoding> (); r->value<P::no_stl> () = h.value<H::no_stl> (); r->value<P::no_iostream> () = h.value<H::no_iostream> (); r->value<P::no_exceptions> () = h.value<H::no_exceptions> (); @@ -895,6 +905,7 @@ namespace CXX Evptr<S::Options> r (new S::Options); + r->value<S::char_encoding> () = h.value<H::char_encoding> (); r->value<S::no_stl> () = h.value<H::no_stl> (); r->value<S::no_iostream> () = h.value<H::no_iostream> (); r->value<S::no_exceptions> () = h.value<H::no_exceptions> (); @@ -1527,6 +1538,25 @@ namespace CXX hxx << "#include <xsde/cxx/config.hxx>" << endl << endl; + if (ops.value<CLI::char_encoding> () == "iso8859-1") + { + hxx << "#ifndef XSDE_ENCODING_ISO8859_1" << endl + << "#error the generated code uses the ISO-8859-1 encoding" << + "while the XSD/e runtime does not (reconfigure the runtime " << + "or change the --char-encoding value)" << endl + << "#endif" << endl + << endl; + } + else + { + hxx << "#ifndef XSDE_ENCODING_UTF8" << endl + << "#error the generated code uses the UTF-8 encoding" << + "while the XSD/e runtime does not (reconfigure the runtime " << + "or change the --char-encoding value)" << endl + << "#endif" << endl + << endl; + } + if (ops.value<CLI::no_stl> ()) { hxx << "#ifdef XSDE_STL" << endl @@ -2613,6 +2643,14 @@ namespace CXX return sloc; } + catch (UnrepresentableCharacter const& e) + { + wcerr << "error: character at position " << e.position () << " " + << "in string '" << e.string () << "' is unrepresentable in " + << "the target encoding" << endl; + + throw Failed (); + } catch (NoNamespaceMapping const& e) { wcerr << e.file () << ":" << e.line () << ":" << e.column () diff --git a/xsde/cxx/hybrid/parser-name-processor.cxx b/xsde/cxx/hybrid/parser-name-processor.cxx index a8d3639..afb0df4 100644 --- a/xsde/cxx/hybrid/parser-name-processor.cxx +++ b/xsde/cxx/hybrid/parser-name-processor.cxx @@ -36,6 +36,7 @@ namespace CXX root, "name", "char", + ops.value<CLI::char_encoding> (), ops.value<CLI::include_with_brackets> (), ops.value<CLI::include_prefix> (), "", // export symbol diff --git a/xsde/cxx/hybrid/serializer-name-processor.cxx b/xsde/cxx/hybrid/serializer-name-processor.cxx index 34f205a..5da88b9 100644 --- a/xsde/cxx/hybrid/serializer-name-processor.cxx +++ b/xsde/cxx/hybrid/serializer-name-processor.cxx @@ -35,6 +35,7 @@ namespace CXX root, "name", "char", + ops.value<CLI::char_encoding> (), ops.value<CLI::include_with_brackets> (), ops.value<CLI::include_prefix> (), "", // export symbol diff --git a/xsde/cxx/hybrid/tree-name-processor.cxx b/xsde/cxx/hybrid/tree-name-processor.cxx index 2ee30ef..b2a2c69 100644 --- a/xsde/cxx/hybrid/tree-name-processor.cxx +++ b/xsde/cxx/hybrid/tree-name-processor.cxx @@ -38,6 +38,7 @@ namespace CXX root, "name", "char", + ops.value<CLI::char_encoding> (), ops.value<CLI::include_with_brackets> (), ops.value<CLI::include_prefix> (), "", // export symbol diff --git a/xsde/cxx/hybrid/tree-type-map.cxx b/xsde/cxx/hybrid/tree-type-map.cxx index 5f1c36b..1b22ff9 100644 --- a/xsde/cxx/hybrid/tree-type-map.cxx +++ b/xsde/cxx/hybrid/tree-type-map.cxx @@ -27,6 +27,7 @@ namespace CXX root, "name", "char", + ops.value<CLI::char_encoding> (), ops.value<CLI::include_with_brackets> (), ops.value<CLI::include_prefix> (), "", // export symbol diff --git a/xsde/cxx/hybrid/validator.cxx b/xsde/cxx/hybrid/validator.cxx index f95bd96..716f072 100644 --- a/xsde/cxx/hybrid/validator.cxx +++ b/xsde/cxx/hybrid/validator.cxx @@ -489,6 +489,18 @@ namespace CXX Boolean valid (true); ValidationContext ctx (root, options, disabled_warnings, valid); + // + // + NarrowString enc (options.value<CLI::char_encoding> ()); + + if (enc != "utf8" && enc != "iso8859-1") + { + wcerr << "error: unknown encoding '" << enc.c_str () << "'" << endl; + return false; + } + + // + // Boolean par (options.value<CLI::generate_parser> ()); Boolean ser (options.value<CLI::generate_serializer> ()); Boolean agg (options.value<CLI::generate_aggregate> ()); diff --git a/xsde/cxx/parser/cli.hxx b/xsde/cxx/parser/cli.hxx index 62a1891..a50c3ff 100644 --- a/xsde/cxx/parser/cli.hxx +++ b/xsde/cxx/parser/cli.hxx @@ -24,6 +24,7 @@ namespace CXX typedef Char const Key[]; extern Key type_map; + extern Key char_encoding; extern Key no_stl; extern Key no_iostream; extern Key no_exceptions; @@ -86,6 +87,7 @@ namespace CXX typedef Cult::CLI::Options< type_map, Cult::Containers::Vector<NarrowString>, + char_encoding, NarrowString, no_stl, Boolean, no_iostream, Boolean, no_exceptions, Boolean, diff --git a/xsde/cxx/parser/elements.cxx b/xsde/cxx/parser/elements.cxx index 4b624ff..3c62b5b 100644 --- a/xsde/cxx/parser/elements.cxx +++ b/xsde/cxx/parser/elements.cxx @@ -20,6 +20,7 @@ namespace CXX root, "p:name", "char", + ops.value<CLI::char_encoding> (), ops.value<CLI::include_with_brackets> (), ops.value<CLI::include_prefix> (), "", // export symbol diff --git a/xsde/cxx/parser/generator.cxx b/xsde/cxx/parser/generator.cxx index e59180c..1936286 100644 --- a/xsde/cxx/parser/generator.cxx +++ b/xsde/cxx/parser/generator.cxx @@ -113,6 +113,7 @@ namespace CXX namespace CLI { extern Key type_map = "type-map"; + extern Key char_encoding = "char-encoding"; extern Key no_stl = "no-stl"; extern Key no_iostream = "no-iostream"; extern Key no_exceptions = "no-exceptions"; @@ -188,6 +189,12 @@ namespace CXX << " order of appearance and the first match is used." << endl; + e << "--char-encoding <enc>" << endl + << " Specify the character encoding that should be\n" + << " used for the extracted text data. Valid values\n" + << " are 'utf8' (default) and 'iso8859-1'." + << endl; + e << "--no-stl" << endl << " Generate code that does not use STL." << endl; @@ -514,6 +521,8 @@ namespace CXX { CLI::OptionsSpec spec; + spec.option<CLI::char_encoding> ().default_value ("utf8"); + spec.option<CLI::skel_file_suffix> ().default_value ("-pskel"); spec.option<CLI::skel_type_suffix> ().default_value ("_pskel"); spec.option<CLI::impl_file_suffix> ().default_value ("-pimpl"); @@ -1207,6 +1216,25 @@ namespace CXX hxx << "#include <xsde/cxx/config.hxx>" << endl << endl; + if (ops.value<CLI::char_encoding> () == "iso8859-1") + { + hxx << "#ifndef XSDE_ENCODING_ISO8859_1" << endl + << "#error the generated code uses the ISO-8859-1 encoding" << + "while the XSD/e runtime does not (reconfigure the runtime " << + "or change the --char-encoding value)" << endl + << "#endif" << endl + << endl; + } + else + { + hxx << "#ifndef XSDE_ENCODING_UTF8" << endl + << "#error the generated code uses the UTF-8 encoding" << + "while the XSD/e runtime does not (reconfigure the runtime " << + "or change the --char-encoding value)" << endl + << "#endif" << endl + << endl; + } + if (ops.value<CLI::no_stl> ()) { hxx << "#ifdef XSDE_STL" << endl @@ -1585,6 +1613,14 @@ namespace CXX return sloc; } + catch (UnrepresentableCharacter const& e) + { + wcerr << "error: character at position " << e.position () << " " + << "in string '" << e.string () << "' is unrepresentable in " + << "the target encoding" << endl; + + throw Failed (); + } catch (NoNamespaceMapping const& e) { wcerr << e.file () << ":" << e.line () << ":" << e.column () diff --git a/xsde/cxx/parser/name-processor.cxx b/xsde/cxx/parser/name-processor.cxx index 658b70a..a168c28 100644 --- a/xsde/cxx/parser/name-processor.cxx +++ b/xsde/cxx/parser/name-processor.cxx @@ -34,6 +34,7 @@ namespace CXX root, "p:name", "char", + ops.value<CLI::char_encoding> (), ops.value<CLI::include_with_brackets> (), ops.value<CLI::include_prefix> (), "", // export symbol diff --git a/xsde/cxx/parser/parser-header.cxx b/xsde/cxx/parser/parser-header.cxx index f3dd45c..6ef9578 100644 --- a/xsde/cxx/parser/parser-header.cxx +++ b/xsde/cxx/parser/parser-header.cxx @@ -1634,6 +1634,14 @@ namespace CXX << "typedef xsde::cxx::parser::context parser_context;" << endl; + if (char_encoding == L"iso8859-1") + { + os << "// ISO-8859-1 transcoder." << endl + << "//" << endl + << "using xsde::cxx::iso8859_1;" + << endl; + } + post (ns); } }; @@ -1675,8 +1683,12 @@ namespace CXX else { ctx.os << "#include <xsde/config.h>" << endl - << "#include <xsde/cxx/ro-string.hxx>" << endl - << endl; + << "#include <xsde/cxx/ro-string.hxx>" << endl; + + if (ctx.char_encoding == L"iso8859-1") + ctx.os << "#include <xsde/cxx/iso8859-1.hxx>" << endl; + + ctx.os << endl; // Data types. // diff --git a/xsde/cxx/parser/validator.cxx b/xsde/cxx/parser/validator.cxx index 96170de..33fe2e2 100644 --- a/xsde/cxx/parser/validator.cxx +++ b/xsde/cxx/parser/validator.cxx @@ -601,6 +601,16 @@ namespace CXX // // + NarrowString enc (options.value<CLI::char_encoding> ()); + + if (enc != "utf8" && enc != "iso8859-1") + { + wcerr << "error: unknown encoding '" << enc.c_str () << "'" << endl; + return false; + } + + // + // if (options.value<CLI::generate_noop_impl> () && options.value<CLI::generate_print_impl> ()) { diff --git a/xsde/cxx/serializer/cli.hxx b/xsde/cxx/serializer/cli.hxx index 41bdde2..d65ca1a 100644 --- a/xsde/cxx/serializer/cli.hxx +++ b/xsde/cxx/serializer/cli.hxx @@ -24,6 +24,7 @@ namespace CXX typedef Char const Key[]; extern Key type_map; + extern Key char_encoding; extern Key no_stl; extern Key no_iostream; extern Key no_exceptions; @@ -85,6 +86,7 @@ namespace CXX typedef Cult::CLI::Options< type_map, Cult::Containers::Vector<NarrowString>, + char_encoding, NarrowString, no_stl, Boolean, no_iostream, Boolean, no_exceptions, Boolean, diff --git a/xsde/cxx/serializer/elements.cxx b/xsde/cxx/serializer/elements.cxx index 495b867..f59a102 100644 --- a/xsde/cxx/serializer/elements.cxx +++ b/xsde/cxx/serializer/elements.cxx @@ -20,6 +20,7 @@ namespace CXX root, "s:name", "char", + ops.value<CLI::char_encoding> (), ops.value<CLI::include_with_brackets> (), ops.value<CLI::include_prefix> (), "", // export symbol diff --git a/xsde/cxx/serializer/generator.cxx b/xsde/cxx/serializer/generator.cxx index c15d603..cf6091d 100644 --- a/xsde/cxx/serializer/generator.cxx +++ b/xsde/cxx/serializer/generator.cxx @@ -111,6 +111,7 @@ namespace CXX namespace CLI { extern Key type_map = "type-map"; + extern Key char_encoding = "char-encoding"; extern Key no_stl = "no-stl"; extern Key no_iostream = "no-iostream"; extern Key no_exceptions = "no-exceptions"; @@ -185,6 +186,12 @@ namespace CXX << " order of appearance and the first match is used." << endl; + e << "--char-encoding <enc>" << endl + << " Specify the character encoding that is used in\n" + << " the text data being serialized. Valid values are\n" + << " 'utf8' (default) and 'iso8859-1'." + << endl; + e << "--no-stl" << endl << " Generate code that does not use STL." << endl; @@ -507,6 +514,8 @@ namespace CXX { CLI::OptionsSpec spec; + spec.option<CLI::char_encoding> ().default_value ("utf8"); + spec.option<CLI::skel_file_suffix> ().default_value ("-sskel"); spec.option<CLI::skel_type_suffix> ().default_value ("_sskel"); spec.option<CLI::impl_file_suffix> ().default_value ("-simpl"); @@ -1193,6 +1202,25 @@ namespace CXX hxx << "#include <xsde/cxx/config.hxx>" << endl << endl; + if (ops.value<CLI::char_encoding> () == "iso8859-1") + { + hxx << "#ifndef XSDE_ENCODING_ISO8859_1" << endl + << "#error the generated code uses the ISO-8859-1 encoding" << + "while the XSD/e runtime does not (reconfigure the runtime " << + "or change the --char-encoding value)" << endl + << "#endif" << endl + << endl; + } + else + { + hxx << "#ifndef XSDE_ENCODING_UTF8" << endl + << "#error the generated code uses the UTF-8 encoding" << + "while the XSD/e runtime does not (reconfigure the runtime " << + "or change the --char-encoding value)" << endl + << "#endif" << endl + << endl; + } + if (ops.value<CLI::no_stl> ()) { hxx << "#ifdef XSDE_STL" << endl @@ -1570,6 +1598,14 @@ namespace CXX return sloc; } + catch (UnrepresentableCharacter const& e) + { + wcerr << "error: character at position " << e.position () << " " + << "in string '" << e.string () << "' is unrepresentable in " + << "the target encoding" << endl; + + throw Failed (); + } catch (NoNamespaceMapping const& e) { wcerr << e.file () << ":" << e.line () << ":" << e.column () diff --git a/xsde/cxx/serializer/name-processor.cxx b/xsde/cxx/serializer/name-processor.cxx index e388169..ddc9c4e 100644 --- a/xsde/cxx/serializer/name-processor.cxx +++ b/xsde/cxx/serializer/name-processor.cxx @@ -34,6 +34,7 @@ namespace CXX root, "s:name", "char", + ops.value<CLI::char_encoding> (), ops.value<CLI::include_with_brackets> (), ops.value<CLI::include_prefix> (), "", // export symbol diff --git a/xsde/cxx/serializer/validator.cxx b/xsde/cxx/serializer/validator.cxx index 50948b9..c163459 100644 --- a/xsde/cxx/serializer/validator.cxx +++ b/xsde/cxx/serializer/validator.cxx @@ -410,6 +410,16 @@ namespace CXX // // + NarrowString enc (options.value<CLI::char_encoding> ()); + + if (enc != "utf8" && enc != "iso8859-1") + { + wcerr << "error: unknown encoding '" << enc.c_str () << "'" << endl; + return false; + } + + // + // { Boolean ref (options.value<CLI::root_element_first> ()); Boolean rel (options.value<CLI::root_element_last> ()); |