Be more cautious when guessing language from filename

Use more strict validation when attempting to interpret part of filename
as language code. Don't try to parse things that don't look like
language tags, because Language::TryParse() is too permissive.

A particular failure mode is that ICU attempts to parse a language tag
in a way where a string with-many-components-like-this may be
interpreted as a valid language with e.g. nonsense country or variants.

Author: vslavik

src/language.cpp

@@ -64,6 +64,9 @@ const std::wregex RE_LANG_CODE(L"([a-z]){2,3}(_([A-Z]{2}|[0-9]{3}))?(@[a-z]+)?")
 // a more permissive variant of the same that TryNormalize() would fix
 const std::wregex RE_LANG_CODE_PERMISSIVE(L"([a-zA-Z]){2,3}([_-]([a-zA-Z]{2}|[0-9]{3}))?(@[a-zA-Z]+)?");
 
+// approximate match for BCP 47 language tags
+const std::wregex RE_LANG_CODE_BCP47(LR"(^[a-zA-Z]{2,3}(-[A-Z][a-z]{3})?(-([A-Z]{2}|\d{3}))?$)");
+
 // try some normalizations: s/-/_/, case adjustments
 void TryNormalize(std::wstring& s)
 {
@@ -324,6 +327,11 @@ bool Language::IsValidCode(const std::wstring& s)
     return std::regex_match(s, RE_LANG_CODE);
 }
 
+bool Language::IsPlausibleCode(const std::wstring& s)
+{
+    return std::regex_match(s, RE_LANG_CODE_PERMISSIVE) || std::regex_match(s, RE_LANG_CODE_BCP47);
+}
+
 std::string Language::Lang() const
 {
     return m_code.substr(0, m_code.find_first_of("_@"));
@@ -636,7 +644,8 @@ Language Language::TryGuessFromFilename(const wxString& filename, wxString *wild
     while (pos != wxString::npos)
     {
         auto part = name.substr(pos+1);
-        lang = Language::TryParseWithValidation(part);
+        if (Language::IsPlausibleCode(part))
+            lang = Language::TryParseWithValidation(part);
         if (lang.IsValid())
         {
             if (wildcard)
@@ -663,12 +672,16 @@ Language Language::TryGuessFromFilename(const wxString& filename, wxString *wild
         wxString rest, wmatch;
         if (dirs[i].EndsWith(".lproj", &rest))
         {
-            lang = Language::TryParseWithValidation(rest.ToStdWstring());
+            auto l = rest.ToStdWstring();
+            if (Language::IsPlausibleCode(l))
+                lang = Language::TryParseWithValidation(l);
             wmatch = "*.lproj";
         }
         else
         {
-            lang = Language::TryParseWithValidation(dirs[i].ToStdWstring());
+            auto l = dirs[i].ToStdWstring();
+            if (Language::IsPlausibleCode(l))
+                lang = Language::TryParseWithValidation(l);
             wmatch = "*";
         }
         if (lang.IsValid())

src/language.h

@@ -184,6 +184,12 @@ class Language
         Checks if @a s has the form of language code.
      */
     static bool IsValidCode(const std::wstring& s);
+    /**
+        Checks if @a s has the form of language code, being more permissive and allowing
+        e.g. BCP 47 -- i..e something suitable for passing to TryParse().
+     */
+    static bool IsPlausibleCode(const std::wstring& s);
+
 
     bool operator==(const Language& other) const { return m_code == other.m_code; }
     bool operator!=(const Language& other) const { return m_code != other.m_code; }