Fix: Un-escape the value before setting LANG

src/linux/init/config.cpp

@@ -2597,9 +2597,8 @@ try
         // If the current line contains the "LANG=" string, update the
         // environment block with the remainder of the line.
         //
-        // N.B. No validation is done on the contents of the string. If the
-        //      file contains multiple lines containing "LANG=" the last will
-        //      be used.
+        // N.B. If the file contains multiple lines containing "LANG=" the last
+        //      will be used.
         //
 
         auto Content = strstr(Line, LANG_ENV "=");
@@ -2617,7 +2616,8 @@ try
                 *SpecialCharacter = '\0';
             }
 
-            Environment.AddVariable(LANG_ENV, Content);
+            const auto Value = wsl::shared::string::UnescapeShell(wsl::shared::string::Trim(std::string{Content}));
+            Environment.AddVariable(LANG_ENV, Value);
         }
     }
 

src/shared/inc/stringshared.h

@@ -849,6 +849,107 @@ inline std::wstring FormatBytes(uint64_t bytes)
     }
 }
 
+template <typename TChar>
+inline std::basic_string<TChar> Trim(const std::basic_string<TChar>& input)
+{
+    constexpr TChar whitespace[] = {TChar(' '), TChar('\t'), TChar('\n'), TChar('\r'), TChar('\f'), TChar('\v'), TChar('\0')};
+    const auto first = input.find_first_not_of(whitespace);
+    if (first == std::basic_string<TChar>::npos)
+    {
+        return {};
+    }
+
+    const auto last = input.find_last_not_of(whitespace);
+    return input.substr(first, last - first + 1);
+}
+
+template <typename TChar>
+inline std::basic_string<TChar> UnescapeShell(const std::basic_string<TChar>& input)
+{
+    enum class Quote
+    {
+        None,
+        Single,
+        Double
+    };
+
+    Quote quote = Quote::None;
+    std::basic_string<TChar> output;
+    output.reserve(input.size());
+
+    for (size_t index = 0; index < input.size(); index += 1)
+    {
+        const auto current = input[index];
+        if (quote == Quote::Single)
+        {
+            if (current == TChar('\''))
+            {
+                quote = Quote::None;
+            }
+            else
+            {
+                output.push_back(current);
+            }
+
+            continue;
+        }
+
+        if (current == TChar('\''))
+        {
+            if (quote == Quote::Double)
+            {
+                output.push_back(current);
+            }
+            else
+            {
+                quote = Quote::Single;
+            }
+
+            continue;
+        }
+
+        if (current == TChar('"'))
+        {
+            quote = (quote == Quote::Double) ? Quote::None : Quote::Double;
+            continue;
+        }
+
+        if (current != TChar('\\'))
+        {
+            output.push_back(current);
+            continue;
+        }
+
+        if (++index == input.size())
+        {
+            // return the original string if the escape is invalid.
+            return input;
+        }
+
+        const auto escaped = input[index];
+        if (quote == Quote::None)
+        {
+            // "\\\n" out of escape means continue the line.
+            if (escaped != TChar('\n'))
+            {
+                output.push_back(escaped);
+            }
+        }
+        else if (escaped == TChar('"') || escaped == TChar('\\') || escaped == TChar('$') || escaped == TChar('`'))
+        {
+            output.push_back(escaped);
+        }
+        else if (escaped != TChar('\n'))
+        {
+            output.push_back(current);
+            output.push_back(escaped);
+        }
+    }
+
+    // return the original string if the escape is invalid.
+    return (quote == Quote::None) ? output : input;
+}
+
 } // namespace wsl::shared::string
 
 template <>

test/windows/SimpleTests.cpp

@@ -259,6 +259,41 @@ class SimpleTests
         auto upperCaseGuidStringWide = wideGuidStringNoBraces;
         std::transform(upperCaseGuidStringWide.begin(), upperCaseGuidStringWide.end(), upperCaseGuidStringWide.begin(), toupper);
         VERIFY_ARE_EQUAL(upperCaseGuidStringWide, wsl::shared::string::GuidToString<wchar_t>(guid, wsl::shared::string::GuidToStringFlags::Uppercase));
+
+        VERIFY_ARE_EQUAL(wsl::shared::string::Trim(std::string{" \tvalue\r\n"}), std::string{"value"});
+        VERIFY_ARE_EQUAL(wsl::shared::string::Trim(std::string{" \t\r\n"}), std::string{});
+        VERIFY_ARE_EQUAL(wsl::shared::string::Trim(std::wstring{L" \tvalue\r\n"}), std::wstring{L"value"});
+
+        const std::vector<std::pair<std::string, std::string>> shellStrings{
+            {"", ""},
+            {"plain", "plain"},
+            {"\"\"", ""},
+            {"''", ""},
+            {"\"double quoted\"", "double quoted"},
+            {"'single quoted'", "single quoted"},
+            {"one' two'\" three\"", "one two three"},
+            {"escaped\\ value", "escaped value"},
+            {"escaped\\q", "escapedq"},
+            {"escaped\\'quote", "escaped'quote"},
+            {"abc\\\nedf", "abcedf"},
+            {"\"abc\\\nedf\"", "abcedf"},
+            {"'abc\\\nedf'", "abc\\\nedf"},
+            {"\"escaped \\\"quote\\\" and \\\\ slash\"", "escaped \"quote\" and \\ slash"},
+            {"\"escaped \\$dollar and \\`backtick\"", "escaped $dollar and `backtick"},
+            {"\"literal \\q\"", "literal \\q"},
+            {"'literal \\ value'", "literal \\ value"},
+            {"unterminated'", "unterminated'"},
+            {"\"unterminated", "\"unterminated"},
+            {"trailing\\", "trailing\\"},
+            {"\"trailing\\", "\"trailing\\"}};
+
+        for (const auto& [input, expected] : shellStrings)
+        {
+            VERIFY_ARE_EQUAL(wsl::shared::string::UnescapeShell(input), expected);
+            VERIFY_ARE_EQUAL(
+                wsl::shared::string::UnescapeShell(wsl::shared::string::MultiByteToWide(input)),
+                wsl::shared::string::MultiByteToWide(expected));
+        }
     }
 
     TEST_METHOD(WindowsPathWithSpaces)

test/windows/UnitTests.cpp

@@ -525,7 +525,7 @@ class UnitTests
         DistroFileChange defaultLocale(L"/etc/default/locale", LxsstuLaunchWsl(L"test -f /etc/default/locale") == 0);
         DistroFileChange localeConf(L"/etc/locale.conf", LxsstuLaunchWsl(L"test -f /etc/locale.conf") == 0);
 
-        const auto readLang = []() { return LxsstuLaunchWslAndCaptureOutput(L"echo $LANG").first; };
+        const auto readLang = []() { return LxsstuLaunchWslAndCaptureOutput(L"printenv LANG").first; };
 
         // Only /etc/default/locale is present (Debian/Ubuntu).
         {
@@ -540,7 +540,7 @@ class UnitTests
         {
             defaultLocale.Delete();
             localeConf.Delete();
-            localeConf.SetContent(L"LANG=fr_FR.UTF-8\n");
+            localeConf.SetContent(L"LANG=\"fr_FR.UTF-8\"\n");
             TerminateDistribution();
             VERIFY_ARE_EQUAL(readLang(), L"fr_FR.UTF-8\n");
         }