[locale] Approach matching from a different angle

- add struct that splits a locale name into parts
- add tests that the splitting and joining works

src/modules/locale/CMakeLists.txt

@@ -22,6 +22,7 @@ calamares_add_plugin(locale
         Config.cpp
         LCLocaleDialog.cpp
         LocaleConfiguration.cpp
+        LocaleNames.cpp
         LocalePage.cpp
         LocaleViewStep.cpp
         SetTimezoneJob.cpp
@@ -39,7 +40,7 @@ calamares_add_plugin(locale
 
 calamares_add_test(
     localetest
-    SOURCES Tests.cpp Config.cpp LocaleConfiguration.cpp SetTimezoneJob.cpp timezonewidget/TimeZoneImage.cpp
+    SOURCES Tests.cpp Config.cpp LocaleConfiguration.cpp LocaleNames.cpp SetTimezoneJob.cpp timezonewidget/TimeZoneImage.cpp
     DEFINITIONS SOURCE_DIR="${CMAKE_CURRENT_LIST_DIR}/images" DEBUG_TIMEZONES=1
     LIBRARIES Qt5::Gui
 )

src/modules/locale/LocaleNames.cpp

@@ -0,0 +1,72 @@
+/* === This file is part of Calamares - <https://calamares.io> ===
+ *
+ *   SPDX-FileCopyrightText: 2022 Adriaan de Groot <groot@kde.org>
+ *   SPDX-License-Identifier: GPL-3.0-or-later
+ *
+ *   Calamares is Free Software: see the License-Identifier above.
+ *
+ */
+
+#include "LocaleNames.h"
+
+#include "utils/Logger.h"
+
+#include <QRegularExpression>
+
+LocaleNameParts
+LocaleNameParts::fromName( const QString& name )
+{
+    auto requireAndRemoveLeadingChar = []( QChar c, QString s )
+    {
+        if ( s.startsWith( c ) )
+        {
+            return s.remove( 0, 1 );
+        }
+        else
+        {
+            return QString();
+        }
+    };
+
+    auto parts = QRegularExpression( "^([a-zA-Z]+)(_[a-zA-Z]+)?(\\.[-a-zA-Z0-9]+)?(@[a-zA-Z]+)?" ).match( name );
+    const QString calamaresLanguage = parts.captured( 1 );
+    const QString calamaresCountry = requireAndRemoveLeadingChar( '_', parts.captured( 2 ) );
+    const QString calamaresEncoding = requireAndRemoveLeadingChar( '.', parts.captured( 3 ) );
+    const QString calamaresRegion = requireAndRemoveLeadingChar( '@', parts.captured( 4 ) );
+
+    if ( calamaresLanguage.isEmpty() )
+    {
+        return LocaleNameParts {};
+    }
+    else
+    {
+        return LocaleNameParts { calamaresLanguage, calamaresCountry, calamaresRegion, calamaresEncoding };
+    }
+}
+
+QString
+LocaleNameParts::name() const
+{
+    // We don't want QStringView to a temporary; force conversion
+    auto insertLeadingChar = []( QChar c, QString s ) -> QString
+    {
+        if ( s.isEmpty() )
+        {
+            return QString();
+        }
+        else
+        {
+            return c + s;
+        }
+    };
+
+    if ( !isValid() )
+    {
+        return QString();
+    }
+    else
+    {
+        return language + insertLeadingChar( '_', country ) + insertLeadingChar( '.', encoding )
+            + insertLeadingChar( '@', region );
+    }
+}

src/modules/locale/LocaleNames.h

@@ -0,0 +1,35 @@
+/* === This file is part of Calamares - <https://calamares.io> ===
+ *
+ *   SPDX-FileCopyrightText: 2022 Adriaan de Groot <groot@kde.org>
+ *   SPDX-License-Identifier: GPL-3.0-or-later
+ *
+ *   Calamares is Free Software: see the License-Identifier above.
+ *
+ */
+
+#ifndef LOCALENAMES_H
+#define LOCALENAMES_H
+
+#include <QString>
+
+/** @brief parts of a locale-name (e.g. "ar_LY.UTF-8", split apart)
+ *
+ * These are created from lines in `/usr/share/i18n/SUPPORTED`,
+ * which lists all the locales supported by the system (there
+ * are also other sources of the same).
+ *
+ */
+struct LocaleNameParts
+{
+    QString language;  // e.g. "ar"
+    QString country;  // e.g. "LY" (may be empty)
+    QString region;  // e.g. "@valencia" (may be empty)
+    QString encoding;  // e.g. "UTF-8" (may be empty)
+
+    bool isValid() const { return !language.isEmpty(); }
+    QString name() const;
+
+    static LocaleNameParts fromName( const QString& name );
+};
+
+#endif

src/modules/locale/Tests.cpp

@@ -9,6 +9,7 @@
 
 #include "Config.h"
 #include "LocaleConfiguration.h"
+#include "LocaleNames.h"
 #include "timezonewidget/TimeZoneImage.h"
 
 #include "Settings.h"
@@ -49,8 +50,11 @@ private Q_SLOTS:
     void testLanguageDetection();
     void testLanguageDetectionValencia();
 
-    // Check realistic language mapping for issue 2008
+    // Check that the test-data is available and ok
     void testKDENeonLanguageData();
+    void testLocaleNameParts();
+
+    // Check realistic language mapping for issue 2008
     void testLanguageMappingNeon_data();
     void testLanguageMappingNeon();
     void testLanguageMappingFreeBSD_data();
@@ -392,6 +396,10 @@ splitTestFileIntoLines( const QString& filename )
 void
 LocaleTests::testKDENeonLanguageData()
 {
+    if ( !m_KDEneonLocales.isEmpty() )
+    {
+        return;
+    }
     const QStringList neonLocales = splitTestFileIntoLines( QStringLiteral( "locale-data-neon" ) );
     cDebug() << "Loaded KDE neon locales test data" << neonLocales.front() << "to" << neonLocales.back();
     QCOMPARE( neonLocales.length(), 318 );  // wc -l tells me 318 lines
@@ -442,12 +450,14 @@ LocaleTests::MappingData()
 }
 
 
-void LocaleTests::testLanguageMappingNeon_data()
+void
+LocaleTests::testLanguageMappingNeon_data()
 {
     MappingData();
 }
 
-void LocaleTests::testLanguageMappingFreeBSD_data()
+void
+LocaleTests::testLanguageMappingFreeBSD_data()
 {
     MappingData();
 }
@@ -455,6 +465,7 @@ void LocaleTests::testLanguageMappingFreeBSD_data()
 void
 LocaleTests::testLanguageMappingNeon()
 {
+    testKDENeonLanguageData();
     QVERIFY( !m_KDEneonLocales.isEmpty() );
 
     QFETCH( QString, selectedLanguage );
@@ -471,6 +482,7 @@ LocaleTests::testLanguageMappingNeon()
 void
 LocaleTests::testLanguageMappingFreeBSD()
 {
+    testKDENeonLanguageData();
     QVERIFY( !m_FreeBSDLocales.isEmpty() );
 
     QFETCH( QString, selectedLanguage );
@@ -485,6 +497,45 @@ LocaleTests::testLanguageMappingFreeBSD()
     QCOMPARE( bsd.language(), expected );
 }
 
+void
+LocaleTests::testLocaleNameParts()
+{
+    testKDENeonLanguageData();
+    QVERIFY( !m_FreeBSDLocales.isEmpty() );
+    QVERIFY( !m_KDEneonLocales.isEmpty() );
+
+    // Example constant locales
+    {
+        auto c_parts = LocaleNameParts::fromName( QStringLiteral( "nl_NL.UTF-8" ) );
+        QCOMPARE( c_parts.language, QStringLiteral( "nl" ) );
+        QCOMPARE( c_parts.country, QStringLiteral( "NL" ) );
+        QCOMPARE( c_parts.encoding, QStringLiteral( "UTF-8" ) );
+        QVERIFY( c_parts.region.isEmpty() );
+    }
+    {
+        auto c_parts = LocaleNameParts::fromName( QStringLiteral( "C.UTF-8" ) );
+        QCOMPARE( c_parts.language, QStringLiteral( "C" ) );
+        QVERIFY( c_parts.country.isEmpty() );
+        QCOMPARE( c_parts.encoding, QStringLiteral( "UTF-8" ) );
+        QVERIFY( c_parts.region.isEmpty() );
+    }
+
+    // Check all the loaded test locales
+    for ( const auto& s : m_FreeBSDLocales )
+    {
+        auto parts = LocaleNameParts::fromName( s );
+        QVERIFY( parts.isValid() );
+        QCOMPARE( parts.name(), s );
+    }
+
+    for ( const auto& s : m_KDEneonLocales )
+    {
+        auto parts = LocaleNameParts::fromName( s );
+        QVERIFY( parts.isValid() );
+        QCOMPARE( parts.name(), s );
+    }
+}
+
 
 #include "utils/moc-warnings.h"