Fix langAltValue::read() parsing

include/exiv2/error.hpp

@@ -249,6 +249,7 @@ namespace Exiv2 {
         kerInvalidXMP,
         kerTiffDirectoryTooLarge,
         kerInvalidTypeValue,
+        kerInvalidLangAltValue,
         kerInvalidMalloc,
         kerCorruptedMetadata,
         kerArithmeticOverflow,

man/man1/exiv2.1

@@ -584,12 +584,16 @@ The format of XMP \fBLangAlt\fP values includes an optional language qualifier:
 .B lang="\fIlanguage-code\fP\fI" text\fP
 .fi
 .sp
-lang="x-default" is used if the value doesn't start with a language qualifier.
+The double quotes around the \fIlanguage-code\fP are optional. If no languge qualifier
+is supplied, then the value of "x-default" is used. More information 
+on the language format can be found at: https://www.ietf.org/rfc/rfc3066.txt
 .sp 1
 .nf
 $ exiv2 -M'set Xmp.dc.title lang="de-DE" Euros' X.jpg
 $ exiv2 -M'set Xmp.dc.title lang="en-GB" Pounds' X.jpg
 $ exiv2 -M'set Xmp.dc.title lang="en-US" In God We Trust' X.jpg
+$ exiv2 -M'set Xmp.dc.title lang=fr-FR Euros' X.jpg
+$ exiv2 -M'set Xmp.dc.title lang=jp Yen' X.jpg
 $ exiv2 -M'set Xmp.dc.title All others pay cash' X.jpg
 .fi
 .sp 1

src/error.cpp

@@ -158,6 +158,8 @@ namespace {
           N_("tiff directory length is too large") },
         { Exiv2::kerInvalidTypeValue,
           N_("invalid type in tiff structure") },
+        { Exiv2::kerInvalidLangAltValue,
+          N_("Invalid LangAlt value `%1'") }, // %1=value
         { Exiv2::kerInvalidMalloc,
           N_("invalid memory allocation request") },
         { Exiv2::kerCorruptedMetadata,

src/value.cpp

@@ -856,18 +856,38 @@ namespace Exiv2 {
     }
 
     int LangAltValue::read(const std::string& buf)
-    {
+    {        
         std::string b = buf;
         std::string lang = "x-default";
         if (buf.length() > 5 && buf.substr(0, 5) == "lang=") {
+            static const char* ALPHA = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz";
+            static const char* ALPHA_NUM = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789";
+
             std::string::size_type pos = buf.find_first_of(' ');
             lang = buf.substr(5, pos-5);
             // Strip quotes (so you can also specify the language without quotes)
-            if (lang[0] == '"') lang = lang.substr(1);
-            if (lang[lang.length()-1] == '"') lang = lang.substr(0, lang.length()-1);
+            if (lang[0] == '"') {
+                lang = lang.substr(1);
+
+                if (lang == "" || lang.find('"') != lang.length()-1)
+                    throw Error(kerInvalidLangAltValue, buf);
+
+                lang = lang.substr(0, lang.length()-1);
+            }
+
+            if (lang == "") throw Error(kerInvalidLangAltValue, buf);
+
+            // Check language is in the correct format (see https://www.ietf.org/rfc/rfc3066.txt)
+            std::string::size_type charPos = lang.find_first_not_of(ALPHA);
+            if (charPos != std::string::npos) {
+                if (lang[charPos] != '-' || lang.find_first_not_of(ALPHA_NUM, charPos+1) != std::string::npos)
+                    throw Error(kerInvalidLangAltValue, buf);
+            }
+
             b.clear();
             if (pos != std::string::npos) b = buf.substr(pos+1);
         }
+
         value_[lang] = b;
         return 0;
     }

unitTests/CMakeLists.txt

@@ -33,6 +33,7 @@ add_executable(unit_tests
     test_slice.cpp
     test_tiffheader.cpp
     test_types.cpp
+    test_LangAltValueRead.cpp
     gtestwrapper.h
     ${unit_tests_exiv2lib_SOURCES}
 )

unitTests/test_LangAltValueRead.cpp

@@ -0,0 +1,199 @@
+#include <exiv2/exiv2.hpp>
+
+#include "gtestwrapper.h"
+
+using namespace Exiv2;
+
+// The tests corrispond to those in issue https://github.com/Exiv2/exiv2/issues/1481
+
+
+// 1. No language value
+TEST(LangAltValueReadTest, noLangugeValBeforeSpace)
+{
+    XmpParser::initialize();
+    ::atexit(XmpParser::terminate);
+
+    Exiv2::XmpData xmpData;
+    try {
+        xmpData["Xmp.dc.title"] = "lang= test1-1";
+    }
+    catch (AnyError& e) {
+        ASSERT_EQ(e.code(),Exiv2::kerInvalidLangAltValue);
+    }
+    catch (...) {
+        ASSERT_TRUE(false);
+    }
+}
+
+TEST(LangAltValueReadTest, quoteThenNoLangugeValBeforeSpace)
+{
+    XmpParser::initialize();
+    ::atexit(XmpParser::terminate);
+
+    Exiv2::XmpData xmpData;
+    try {
+        xmpData["Xmp.dc.title"] = "lang=\" test1-2";
+    }
+    catch (AnyError& e) {
+        ASSERT_EQ(e.code(),Exiv2::kerInvalidLangAltValue);
+    }
+    catch (...) {
+        ASSERT_TRUE(false);
+    }
+}
+
+// 2. Empty language value
+TEST(LangAltValueReadTest, emptyDoubleQuotesLanguageValBeforeSpace)
+{
+    XmpParser::initialize();
+    ::atexit(XmpParser::terminate);
+
+    Exiv2::XmpData xmpData;
+    try {
+        xmpData["Xmp.dc.title"] = "lang=\"\" test2";
+    }
+    catch (AnyError& e) {
+        ASSERT_EQ(e.code(),Exiv2::kerInvalidLangAltValue);
+    }
+    catch (...) {
+        ASSERT_TRUE(false);
+    }
+}
+
+// 3. Mismatched and/or incorrect positioning of quotation marks
+TEST(LangAltValueReadTest, emptyDoubleQuotesLanguageValNoSpace)
+{
+    XmpParser::initialize();
+    ::atexit(XmpParser::terminate);
+
+    Exiv2::XmpData xmpData;
+    try {
+        xmpData["Xmp.dc.title"] = "lang=\"\"test3-1";
+    }
+    catch (AnyError& e) {
+        ASSERT_EQ(e.code(),Exiv2::kerInvalidLangAltValue);
+    }
+    catch (...) {
+        ASSERT_TRUE(false);
+    }
+}
+
+TEST(LangAltValueReadTest, oneDoubleQuotesLanguageValNoSpace)
+{
+    XmpParser::initialize();
+    ::atexit(XmpParser::terminate);
+
+    Exiv2::XmpData xmpData;
+    try {
+        xmpData["Xmp.dc.title"] = "lang=\"test3-2";
+    }
+    catch (AnyError& e) {
+        ASSERT_EQ(e.code(),Exiv2::kerInvalidLangAltValue);
+    }
+    catch (...) {
+        ASSERT_TRUE(false);
+    }
+}
+
+TEST(LangAltValueReadTest, oneDoubleQuotesLanguageValBeforeSpace)
+{
+    XmpParser::initialize();
+    ::atexit(XmpParser::terminate);
+
+    Exiv2::XmpData xmpData;
+    try {
+        xmpData["Xmp.dc.title"] = "lang=\"en-UK test3-3";
+    }
+    catch (AnyError& e) {
+        ASSERT_EQ(e.code(),Exiv2::kerInvalidLangAltValue);
+    }
+    catch (...) {
+        ASSERT_TRUE(false);
+    }
+}
+
+TEST(LangAltValueReadTest, languageValOneDoubleQuotesBeforeSpace)
+{
+    XmpParser::initialize();
+    ::atexit(XmpParser::terminate);
+
+    Exiv2::XmpData xmpData;
+    try {
+        xmpData["Xmp.dc.title"] = "lang=en-US\" test3-4";
+    }
+    catch (AnyError& e) {
+        ASSERT_EQ(e.code(),Exiv2::kerInvalidLangAltValue);
+    }
+    catch (...) {
+        ASSERT_TRUE(false);
+    }
+}
+
+TEST(LangAltValueReadTest, languageValOneDoubleQuotesNoSpace)
+{
+    XmpParser::initialize();
+    ::atexit(XmpParser::terminate);
+
+    Exiv2::XmpData xmpData;
+    try {
+        xmpData["Xmp.dc.title"] = "lang=test3-5\"";
+    }
+    catch (AnyError& e) {
+        ASSERT_EQ(e.code(),Exiv2::kerInvalidLangAltValue);
+    }
+    catch (...) {
+        ASSERT_TRUE(false);
+    }
+}
+
+TEST(LangAltValueReadTest, languageValTwoDoubleQuotesNoSpace)
+{
+    XmpParser::initialize();
+    ::atexit(XmpParser::terminate);
+
+    Exiv2::XmpData xmpData;
+    try {
+        xmpData["Xmp.dc.title"] = "lang=test3-6\"\"";
+    }
+    catch (AnyError& e) {
+        ASSERT_EQ(e.code(),Exiv2::kerInvalidLangAltValue);
+    }
+    catch (...) {
+        ASSERT_TRUE(false);
+    }
+}
+
+// 4. Invalid characters in language part
+TEST(LangAltValueReadTest, languageValExtraHyphenBeforeSpace)
+{
+    XmpParser::initialize();
+    ::atexit(XmpParser::terminate);
+
+    Exiv2::XmpData xmpData;
+    try {
+        xmpData["Xmp.dc.title"] = "lang=en-UK- test4-1";
+    }
+    catch (AnyError& e) {
+        ASSERT_EQ(e.code(),Exiv2::kerInvalidLangAltValue);
+    }
+    catch (...) {
+        ASSERT_TRUE(false);
+    }
+}
+
+TEST(LangAltValueReadTest, languageValWithInvalidCharBeforeSpace)
+{
+    XmpParser::initialize();
+    ::atexit(XmpParser::terminate);
+
+    Exiv2::XmpData xmpData;
+    try {
+        xmpData["Xmp.dc.title"] = "lang=en=UK test4-2";
+    }
+    catch (AnyError& e) {
+        ASSERT_EQ(e.code(),Exiv2::kerInvalidLangAltValue);
+    }
+    catch (...) {
+        ASSERT_TRUE(false);
+    }
+}