More hacking on --with-locales

Propagate the restriction of locales into <config_locales.h>. Note that in the
normal case, with no locale restrictions, all the WITH_LOCALE_xx macros are
zero anyway, but WITH_LOCALE_ALL is one.

Restrict which character encodings are handled in
sal/textencsal/textenc/tables.cxx based on the WITH_LOCALE_ macros. (Don't
simply always do it for iOS.)

Massage the affected unit tests to not crash when only partial character
encoding information is present.

Change-Id: Ie2c882c262ebd0d2b37dde66b8fe3c3e2570da14

config_host/config_locales.h.in

+/* Configuration of restriction on supported locales, by the
+ * --with-locales option.
+ *
+ * (Note that this is totally unrelated to what UI language(s) were
+ * configured with the --with-lang option.)
+ *
+ * Normally, especially for desktop OSes, LibreOffice is built to
+ * support all the locales and character encodings that it has code
+ * (data) for in the source code. By using the --with-locales
+ * configure switch, this can be restricted.
+ *
+ * This file defines feature test macros for those languages for which
+ * at least one locale is to be supported. In some cases, a feature
+ * test macro for a script is provided in addition or instead.
+ *
+ * Only languages for which there exists separate code and/or data in
+ * the source code need to be mentioned here.
+ */
+
+#ifndef CONFIG_LOCALES_H
+#define CONFIG_LOCALES_H
+
+/* In the normal case WITH_LOCALE_ALL is one and this
+ * overrides the others which are zero.
+ */
+#define WITH_LOCALE_ALL 0
+
+/* When --with-locales was used, WITH_LOCALE_ALL is zero and some of
+ * these are one.
+ */
+#define WITH_LOCALE_ca 0
+#define WITH_LOCALE_dz 0
+#define WITH_LOCALE_el 0
+#define WITH_LOCALE_he 0
+#define WITH_LOCALE_hi 0
+#define WITH_LOCALE_hr 0
+#define WITH_LOCALE_hu 0
+#define WITH_LOCALE_is 0
+#define WITH_LOCALE_ja 0
+#define WITH_LOCALE_ko 0
+#define WITH_LOCALE_ku 0
+#define WITH_LOCALE_ln 0
+#define WITH_LOCALE_my 0
+#define WITH_LOCALE_ne 0
+#define WITH_LOCALE_ro 0
+#define WITH_LOCALE_ru 0
+#define WITH_LOCALE_sid 0
+#define WITH_LOCALE_th 0
+#define WITH_LOCALE_tr 0
+#define WITH_LOCALE_uk 0
+#define WITH_LOCALE_zh 0
+
+#define WITH_LOCALE_FOR_SCRIPT_Cyrl 0
+#define WITH_LOCALE_FOR_SCRIPT_Deva 0
+
+#endif

configure.ac

@@ -12201,6 +12201,26 @@ AC_SUBST(WITH_POOR_HELP_LOCALIZATIONS)
 
 if test -n "$with_locales"; then
     WITH_LOCALES="$with_locales"
+
+    just_langs="`echo $WITH_LOCALES | sed -e 's/_[A-Z]*//g'`"
+    # Only languages and scripts for which we actually have ifdefs need to be handled. Also see
+    # config_host/config_locales.h.in
+    for locale in $WITH_LOCALES; do
+        lang=${locale%_*}
+
+        AC_DEFINE_UNQUOTED(WITH_LOCALE_$lang, 1)
+
+        case $lang in
+        hi|mr*ne)
+            AC_DEFINE(WITH_LOCALE_FOR_SCRIPT_Deva)
+            ;;
+        bg|ru)
+            AC_DEFINE(WITH_LOCALE_FOR_SCRIPT_Cyrl)
+            ;;
+        esac
+    done
+else
+    AC_DEFINE(WITH_LOCALE_ALL)
 fi
 AC_SUBST(WITH_LOCALES)
 
@@ -12759,6 +12779,7 @@ AC_CONFIG_HEADERS([config_host/config_gcc.h])
 AC_CONFIG_HEADERS([config_host/config_global.h])
 AC_CONFIG_HEADERS([config_host/config_graphite.h])
 AC_CONFIG_HEADERS([config_host/config_lgpl.h])
+AC_CONFIG_HEADERS([config_host/config_locales.h])
 AC_CONFIG_HEADERS([config_host/config_mpl.h])
 AC_CONFIG_HEADERS([config_host/config_orcus.h])
 AC_CONFIG_HEADERS([config_host/config_kde4.h])

i18npool/CustomTarget_collator.mk

@@ -55,20 +55,10 @@ $(i18npool_CODIR)/collator_%.cxx : \
 	$(call gb_Helper_abbreviate_dirs, \
 		$(call gb_Helper_execute,gencoll_rule) $< $@ $*)
 
-define i18npool_echo_one_lang
-	echo '#define LOCAL_RULE_$(1)' >>$@
-
-endef
-
 $(i18npool_CODIR)/lrl_include.hxx : \
 		$(SRCDIR)/i18npool/source/collator/data | $(i18npool_CODIR)/.dir
 	$(call gb_Output_announce,$(subst $(WORKDIR)/,,$@),$(true),ECH,1)
 	echo '#define LOCAL_RULE_LANGS "$(sort $(foreach txt,$(i18npool_COTXTS), \
 		$(firstword $(subst _, ,$(txt)))))"' > $@
-ifeq ($(WITH_LOCALES),)
-	echo '#define LOCAL_RULE_ALL' >>$@
-else
-	$(foreach txt,$(i18npool_COTXTS),$(call i18npool_echo_one_lang,$(subst .txt,,$(txt))))
-endif
 
 # vim: set noet sw=4 ts=4:

i18npool/source/collator/collator_unicode.cxx

@@ -17,8 +17,8 @@
  *   the License at http://www.apache.org/licenses/LICENSE-2.0 .
  */
 
+#include <config_locales.h>
 
-// generated list of languages
 #include "lrl_include.hxx"
 
 #include <rtl/ustrbuf.hxx>
@@ -156,23 +156,23 @@ Collator_Unicode::loadCollatorAlgorithm(const OUString& rAlgorithm, const lang::
 #else
             if (false) {
                 ;
-#if defined LOCAL_RULE_ALL || defined LOCAL_RULE_ca_charset
+#if WITH_LOCALE_ALL || WITH_LOCALE_CA
             } else if ( rLocale.Language == "ca" ) {
                 if ( rAlgorithm == "charset" )
                     func = get_collator_data_ca_charset;
 #endif
-#if defined LOCAL_RULE_ALL || defined LOCAL_RULE_dz_charset
+#if WITH_LOCALE_ALL || WITH_LOCALE_DZ
             } else if ( rLocale.Language == "dz" || rLocale.Language == "bo" ) {
                 // 'bo' Tibetan uses the same collation rules as 'dz' Dzongkha
                 if ( rAlgorithm == "charset" )
                     func = get_collator_data_dz_charset;
 #endif
-#if defined LOCAL_RULE_ALL || defined LOCAL_RULE_hu
+#if WITH_LOCALE_ALL || WITH_LOCALE_HU
             } else if ( rLocale.Language == "hu" ) {
                 if ( rAlgorithm == "charset" )
                     func = get_collator_data_hu_charset;
 #endif
-#if defined LOCAL_RULE_ALL || defined LOCAL_RULE_ja_charset
+#if WITH_LOCALE_ALL || WITH_LOCALE_JA
             } else if ( rLocale.Language == "ja" ) {
                 if ( rAlgorithm == "charset" )
                     func = get_collator_data_ja_charset;
@@ -181,37 +181,37 @@ Collator_Unicode::loadCollatorAlgorithm(const OUString& rAlgorithm, const lang::
                 else if ( rAlgorithm == "phonetic (alphanumeric last)" )
                     func = get_collator_data_ja_phonetic_alphanumeric_last;
 #endif
-#if defined LOCAL_RULE_ALL || defined LOCAL_RULE_ko_charset
+#if WITH_LOCALE_ALL || WITH_LOCALE_KO
             } else if ( rLocale.Language == "ko" ) {
                 if ( rAlgorithm == "charset" )
                     func = get_collator_data_ko_charset;
 #endif
-#if defined LOCAL_RULE_ALL || defined LOCAL_RULE_ku_charset
+#if WITH_LOCALE_ALL || WITH_LOCALE_KU
             } else if ( rLocale.Language == "ku" ) {
                 if ( rAlgorithm == "alphanumeric" )
                     func = get_collator_data_ku_alphanumeric;
 #endif
-#if defined LOCAL_RULE_ALL || defined LOCAL_RULE_ln_charset
+#if WITH_LOCALE_ALL || WITH_LOCALE_LN
             } else if ( rLocale.Language == "ln" ) {
                 if ( rAlgorithm == "charset" )
                     func = get_collator_data_ln_charset;
 #endif
-#if defined LOCAL_RULE_ALL || defined LOCAL_RULE_my_charset
+#if WITH_LOCALE_ALL || WITH_LOCALE_MY
             } else if ( rLocale.Language == "my" ) {
                 if ( rAlgorithm == "dictionary" )
                     func = get_collator_data_my_dictionary;
 #endif
-#if defined LOCAL_RULE_ALL || defined LOCAL_RULE_ne_charset
+#if WITH_LOCALE_ALL || WITH_LOCALE_NE
             } else if ( rLocale.Language == "ne" ) {
                 if ( rAlgorithm == "charset" )
                     func = get_collator_data_ne_charset;
 #endif
-#if defined LOCAL_RULE_ALL || defined LOCAL_RULE_sid_charset
+#if WITH_LOCALE_ALL || WITH_LOCALE_SID
             } else if ( rLocale.Language == "sid" ) {
                 if ( rAlgorithm == "charset" )
                     func = get_collator_data_sid_charset;
 #endif
-#if defined LOCAL_RULE_ALL || defined LOCAL_RULE_zh_charset
+#if WITH_LOCALE_ALL || WITH_LOCALE_ZH
             } else if ( rLocale.Language == "zh" && (rLocale.Country == "TW" || rLocale.Country == "HK" || rLocale.Country == "MO") ) {
                 if ( rAlgorithm == "charset" )
                     func = get_collator_data_zh_TW_charset;

sal/Module_sal.mk

@@ -54,8 +54,8 @@ $(eval $(call gb_Module_add_check_targets,sal,\
 	CppunitTest_sal_rtl_oustringbuffer \
 	CppunitTest_sal_rtl_oustring \
 	CppunitTest_sal_rtl_textenc \
-	CppunitTest_sal_rtl_strings \
 	CppunitTest_sal_rtl_uri \
+	CppunitTest_sal_rtl_strings \
 	CppunitTest_sal_rtl_uuid \
 	CppunitTest_sal_tcwf \
 	CppunitTest_sal_types \

sal/qa/rtl/uri/rtl_testuri.cxx

@@ -17,6 +17,8 @@
  *   the License at http://www.apache.org/licenses/LICENSE-2.0 .
  */
 
+#include <config_locales.h>
+
 #include "rtl/strbuf.hxx"
 #include "rtl/uri.hxx"
 #include "rtl/ustrbuf.hxx"
@@ -395,6 +397,7 @@ void Test::test_Uri() {
                 aText2, rtl_UriDecodeWithCharset, RTL_TEXTENCODING_ISO_8859_5)
              == aText1));
     }
+#if WITH_LOCALE_ALL || WITH_LOCALE_zh
     {
         sal_Unicode const aText1U[] = { ' ', '!', 0x028A, 0xD849, 0xDD13, 0 };
         aText1 = rtl::OUString(aText1U);
@@ -411,7 +414,7 @@ void Test::test_Uri() {
                 aText2, rtl_UriDecodeWithCharset, RTL_TEXTENCODING_GB_18030)
              == aText1));
     }
-
+#endif
     // Check strict mode:
 
     {
@@ -434,6 +437,7 @@ void Test::test_Uri() {
                 aText1, rtl_UriDecodeStrict, RTL_TEXTENCODING_UTF8)
              == aText2));
     }
+#if WITH_LOCALE_ALL || WITH_LOCALE_zh
     {
         aText1 = rtl::OUString("%81 ");
         aText2 = rtl::OUString();
@@ -472,7 +476,7 @@ void Test::test_Uri() {
                 aText1, rtl_UriDecodeStrict, RTL_TEXTENCODING_GB_18030)
              == aText2));
     }
-
+#endif
     // Check rtl_UriEncodeStrictKeepEscapes mode:
 
     {

sal/textenc/tables.cxx

@@ -17,6 +17,8 @@
  *   the License at http://www.apache.org/licenses/LICENSE-2.0 .
  */
 
+#include <config_locales.h>
+
 #include "sal/config.h"
 
 #include <cassert>
@@ -158,48 +160,101 @@ extern "C" SAL_DLLPUBLIC_EXPORT ImplTextEncodingData const *
             NULL, /* TODO! APPLE_THAI */
             &aImplAPPLETURKISHTextEncodingData, /* APPLE_TURKISH */
             &aImplAPPLEUKRAINIANTextEncodingData, /* APPLE_UKRAINIAN */
+#if WITH_LOCALE_ALL || WITH_LOCALE_zh
             &aImplAPPLECHINSIMPTextEncodingData, /* APPLE_CHINSIMP */
             &aImplAPPLECHINTRADTextEncodingData, /* APPLE_CHINTRAD */
+#else
+            NULL,
+            NULL,
+#endif
+#if WITH_LOCALE_ALL || WITH_LOCALE_ja
             &aImplAPPLEJAPANESETextEncodingData, /* APPLE_JAPANESE */
+#else
+            NULL,
+#endif
+#if WITH_LOCALE_ALL || WITH_LOCALE_ko
             &aImplAPPLEKOREANTextEncodingData, /* APPLE_KOREAN */
+#else
+            NULL,
+#endif
             &aImplMS932TextEncodingData, /* MS_932 */
             &aImplMS936TextEncodingData, /* MS_936 */
             &aImplMS949TextEncodingData, /* MS_949 */
             &aImplMS950TextEncodingData, /* MS_950 */
+#if WITH_LOCALE_ALL || WITH_LOCALE_ja
             &aImplSJISTextEncodingData, /* SHIFT_JIS */
+#else
+            NULL,
+#endif
+#if WITH_LOCALE_ALL || WITH_LOCALE_zh
             &aImplGB2312TextEncodingData, /* GB_2312 */
             &aImplGBT12345TextEncodingData, /* GBT_12345 */
             &aImplGBKTextEncodingData, /* GBK */
             &aImplBIG5TextEncodingData, /* BIG5 */
-#ifndef IOS
+#else
+            NULL,
+            NULL,
+            NULL,
+            NULL,
+#endif
+#if WITH_LOCALE_ALL || WITH_LOCALE_ja
             &aImplEUCJPTextEncodingData, /* EUC_JP */
+#else
+            NULL,
+#endif
+#if WITH_LOCALE_ALL || WITH_LOCALE_zh
             &aImplEUCCNTextEncodingData, /* EUC_CN */
             &aImplEucTwTextEncodingData, /* EUC_TW */
+#else
+            NULL,
+            NULL,
+#endif
+#if WITH_LOCALE_ALL || WITH_LOCALE_ja
             &aImplIso2022JpTextEncodingData, /* ISO_2022_JP */
+#else
+            NULL,
+#endif
+#if WITH_LOCALE_ALL || WITH_LOCALE_zh
             &aImplIso2022CnTextEncodingData, /* ISO_2022_CN */
 #else
-            NULL, /* EUC_JP */
-            NULL, /* EUC_CN */
-            NULL, /* EUC_TW */
-            NULL, /* ISO_2022_JP */
-            NULL, /* ISO_2022_CN */
+            NULL,
 #endif
             &aImplKOI8RTextEncodingData, /* KOI8_R */
             &aImplUTF7TextEncodingData, /* UTF7 */
             NULL, /* UTF8, see above */
             &aImplISO885910TextEncodingData, /* ISO_8859_10 */
             &aImplISO885913TextEncodingData, /* ISO_8859_13 */
+#if WITH_LOCALE_ALL || WITH_LOCALE_ko
             &aImplEUCKRTextEncodingData, /* EUC_KR */
             &aImplIso2022KrTextEncodingData, /* ISO_2022_KR */
+#else
+            NULL,
+            NULL,
+#endif
+#if WITH_LOCALE_ALL || WITH_LOCALE_ja
             &aImplJISX0201TextEncodingData, /* JIS_X_0201 */
             &aImplJISX0208TextEncodingData, /* JIS_X_0208 */
             &aImplJISX0212TextEncodingData, /* JIS_X_0212 */
+#else
+            NULL,
+            NULL,
+            NULL,
+#endif
             &aImplMS1361TextEncodingData, /* MS_1361 */
+#if WITH_LOCALE_ALL || WITH_LOCALE_zh
             &aImplGb18030TextEncodingData, /* GB_18030 */
             &aImplBig5HkscsTextEncodingData, /* BIG5_HKSCS */
+#else
+            NULL,
+            NULL,
+#endif
             &aImplTis620TextEncodingData, /* TIS_620 */
             &aImplKoi8UTextEncodingData, /* KOI8_U */
+#if WITH_LOCALE_ALL || WITH_LOCALE_FOR_SCRIPT_Deva
             &aImplIsciiDevanagariTextEncodingData, /* ISCII_DEVANAGARI */
+#else
+            NULL,
+#endif
             NULL, /* JAVA_UTF8, see above */
             &adobeStandardEncodingData, /* ADOBE_STANDARD */
             &adobeSymbolEncodingData, /* ADOBE_SYMBOL */