Metadata length is measured in characters

src/metadata.cpp

@@ -21,6 +21,7 @@
 
 #include <sstream>
 #include <regex>
+#include <unicode/unistr.h>
 
 namespace zim
 {
@@ -41,6 +42,10 @@ bool matchRegex(const std::string& regexStr, const std::string& text)
   return std::regex_match(text.begin(), text.end(), regex);
 }
 
+size_t getTextLength(const std::string& utf8EncodedString)
+{
+  return icu::UnicodeString::fromUTF8(utf8EncodedString).length();
+}
 
 #include "metadata_constraints.cpp"
 
@@ -99,14 +104,14 @@ Metadata::Errors Metadata::checkSimpleConstraints() const
     const auto& value = nv.second;
     try {
       const auto& rmr = getReservedMetadataRecord(name);
-      if ( value.size() < rmr.minLength ) {
+      if ( rmr.minLength != 0 && getTextLength(value) < rmr.minLength ) {
         std::ostringstream oss;
-        oss << name << " must be at least " << rmr.minLength << " bytes";
+        oss << name << " must contain at least " << rmr.minLength << " characters";
         errors.push_back(oss.str());
       }
-      if ( rmr.maxLength != 0 && value.size() > rmr.maxLength ) {
+      if ( rmr.maxLength != 0 && getTextLength(value) > rmr.maxLength ) {
         std::ostringstream oss;
-        oss << name << " must be at most " << rmr.maxLength << " bytes";
+        oss << name << " must contain at most " << rmr.maxLength << " characters";
         errors.push_back(oss.str());
       }
       if ( !rmr.regex.empty() && !matchRegex(rmr.regex, value) ) {

src/metadata_constraints.cpp

@@ -19,11 +19,8 @@ const Metadata::ReservedMetadataTable reservedMetadataInfoTable = {
   {
     "Illustration_48x48@1",
     MANDATORY,
-    67, // this is the lower limit on a syntactically valid PNG file
-        // (according to https://github.com/mathiasbynens/small)
-    10000, // this is roughly the size of the raw (i.e. without any compression)
-           // RGBA pixel data of a 48x48 image
-           // Question: how much PNG metadata shall we allow?
+    0, // There are no constraints on the illustration metadata size
+    0, // in order to avoid decoding it as UTF-8 encoded text
     PNG_REGEXP
   },
 };

test/metadata-test.cpp

@@ -82,7 +82,7 @@ TEST(Metadata, minSizeConstraints)
   ASSERT_FALSE(m.valid());
   ASSERT_EQ(m.check(),
       zim::Metadata::Errors({
-        "Title must be at least 1 bytes"
+        "Title must contain at least 1 characters"
       })
   );
   m.set("Title", "t");
@@ -96,7 +96,7 @@ TEST(Metadata, maxSizeConstraints)
   ASSERT_FALSE(m.valid());
   ASSERT_EQ(m.check(),
       zim::Metadata::Errors({
-        "Title must be at most 30 bytes"
+        "Title must contain at most 30 characters"
       })
   );
   m.set("Title", std::string(30, 'a'));