Rework and improve locale validation

Author: ids1024

src/desktopfilereader.cpp

@@ -136,16 +136,50 @@ namespace linuxdeploy {
                                         throw ParseError(errorPrefix + "invalid ] position");
                                     }
 
-                                    // locale may only contain A-Za-z- characters according to specification
-                                    for (const char c : entryName) {
+                                    // splitting the locale into the three parts (of which only one is mandatory) is easier
+                                    // if we strip the surrounding brackets first
+                                    const auto innerPart = entryLocale.substr(1, entryLocale.size() - 2);
+
+                                    // will be std::string::npos if necessary
+                                    const auto stopPosition = innerPart.find('.');
+                                    const auto atPosition = innerPart.find('@');
+
+                                    auto langCountry = innerPart.substr(0, std::min(stopPosition, atPosition));
+                                    std::string encoding, modifier;
+
+                                    if (stopPosition != std::string::npos) {
+                                        encoding = innerPart.substr(stopPosition + 1, atPosition);
+                                    }
+                                    if (atPosition != std::string::npos) {
+                                        modifier = innerPart.substr(atPosition + 1);
+                                    }
+
+                                    // lang_COUNTRY may only contain A-Za-z- characters according to specification
+                                    for (const char c : langCountry) {
+                                        if (!(
+                                            (c >= 'A' && c <= 'Z') ||
+                                            (c >= 'a' && c <= 'z') ||
+                                            (c == '_')
+                                        )) {
+                                            throw ParseError("Locale " + key + " lang_COUNTRY contains invalid character " + std::string{c});
+                                        }
+                                    }
+
+                                    // encoding may only contain A-Za-z- characters according to specification
+                                    // if the encoding is empty, this loop is a no-op
+                                    for (const char c : encoding) {
                                         if (!(
                                             (c >= 'A' && c <= 'Z') ||
                                             (c >= 'a' && c <= 'z') ||
-                                            (c == '_') || (c == '@')
+                                            (c >= '0' && c <= '9') ||
+                                            (c == '-')
                                         )) {
-                                            throw ParseError("Key " + key + " contains invalid character " + std::string{c});
+                                            throw ParseError("Locale " + key + " encoding contains invalid character " + std::string{c});
                                         }
                                     }
+
+                                    // TODO: test modifier properly
+                                    (void) modifier;
                                 }
 
                                 auto& section = sections[currentSectionName];