i18nutil/source/utility/unicode.cxx | 32 ++++++++++++++++++++++++++++++++
include/i18nutil/unicode.hxx | 9 +++++++++
2 files changed, 41 insertions(+)
New commits:
commit ecc30ac20f559e47fc4a183942d73913d615ff3f
Author: Eike Rathke <er...@redhat.com>
AuthorDate: Wed Oct 5 01:29:02 2022 +0200
Commit: Eike Rathke <er...@redhat.com>
CommitDate: Wed Oct 5 12:05:01 2022 +0200
Introduce unicode::getScriptClassFromLanguageTag()
Change-Id: Ifb932ff2aabc5767571433627314f0b29c35c471
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/140953
Reviewed-by: Eike Rathke <er...@redhat.com>
Tested-by: Jenkins
diff --git a/i18nutil/source/utility/unicode.cxx
b/i18nutil/source/utility/unicode.cxx
index 5e479989eae9..33f1ca2f190e 100644
--- a/i18nutil/source/utility/unicode.cxx
+++ b/i18nutil/source/utility/unicode.cxx
@@ -24,6 +24,7 @@
#include <i18nutil/unicode.hxx>
#include <sal/log.hxx>
#include <unicode/numfmt.h>
+#include <unicode/uchar.h>
#include "unicode_data.h"
#include <rtl/character.hxx>
#include <o3tl/string_view.hxx>
@@ -190,6 +191,37 @@ sal_Int16
unicode::getScriptClassFromUScriptCode(UScriptCode eScript)
return nRet;
}
+sal_Int16 unicode::getScriptClassFromLanguageTag( const LanguageTag&
rLanguageTag )
+{
+ static UScriptCode nMaxScript =
static_cast<UScriptCode>(u_getIntPropertyMaxValue(UCHAR_SCRIPT));
+ constexpr int32_t nBuf = 42;
+ UScriptCode aBuf[nBuf];
+ if (rLanguageTag.hasScript())
+ {
+ aBuf[0] = static_cast<UScriptCode>(u_getPropertyValueEnum(
UCHAR_SCRIPT,
+ OUStringToOString( rLanguageTag.getScript(),
RTL_TEXTENCODING_ASCII_US).getStr()));
+ }
+ else
+ {
+ OUString aName;
+ if (rLanguageTag.getCountry().isEmpty())
+ aName = rLanguageTag.getLanguage();
+ else
+ aName = rLanguageTag.getLanguage() + "-" +
rLanguageTag.getCountry();
+ UErrorCode status = U_ZERO_ERROR;
+ const int32_t nScripts = uscript_getCode(
+ OUStringToOString( aName, RTL_TEXTENCODING_ASCII_US).getStr(),
+ aBuf, nBuf, &status);
+ // U_BUFFER_OVERFLOW_ERROR would be set with too many scripts for
buffer
+ // and required capacity returned, but really..
+ if (nScripts == 0 || !U_SUCCESS(status))
+ return css::i18n::ScriptType::LATIN;
+ }
+ if (aBuf[0] > nMaxScript)
+ return css::i18n::ScriptType::COMPLEX;
+ return getScriptClassFromUScriptCode( aBuf[0]);
+}
+
OString unicode::getExemplarLanguageForUScriptCode(UScriptCode eScript)
{
OString sRet;
diff --git a/include/i18nutil/unicode.hxx b/include/i18nutil/unicode.hxx
index 0ca14290981e..ebe50ce90384 100644
--- a/include/i18nutil/unicode.hxx
+++ b/include/i18nutil/unicode.hxx
@@ -82,6 +82,15 @@ public:
//Format a number as a percentage according to the rules of the given
//language, e.g. 100 -> "100%" for en-US vs "100 %" for de-DE
static OUString formatPercent(double dNumber, const LanguageTag& rLangTag);
+
+ /** Map a LanguageTag's language ISO 639 code or script ISO 15924 code or
+ language-script or locale to Latin/Asian/Complex/Weak. If more than one
+ script is used with a language(-country) tag then the first (default)
+ script is mapped for that language.
+
+ @return a css::i18n::ScriptType value.
+ */
+ static sal_Int16 getScriptClassFromLanguageTag(const LanguageTag&
rLanguageTag);
};
/*
