[editor] Check OSM max char length in value validation

Signed-off-by: map-per <map-per@gmx.de>

libs/editor/changeset_wrapper.cpp

@@ -243,14 +243,11 @@ void ChangesetWrapper::Modify(editor::XMLFeature node)
 
 void ChangesetWrapper::AddChangesetTag(std::string key, std::string value)
 {
-  value = strings::EscapeForXML(value);
+  // Truncate to 254 characters as OSM has a length limit of 255
+  if (strings::Truncate(value, kMaximumOsmChars))
+    value += "…";
 
-  //OSM has a length limit of 255 characters
-  if (value.length() > kMaximumOsmChars)
-  {
-    LOG(LWARNING, ("value is too long for OSM 255 char limit: ", value));
-    value = value.substr(0, kMaximumOsmChars - 3).append("...");
-  }
+  value = strings::EscapeForXML(value);
 
   m_changesetComments.insert_or_assign(std::move(key), std::move(value));
 }

libs/indexer/editable_map_object.cpp

@@ -479,6 +479,9 @@ bool EditableMapObject::CheckHouseNumberWhenIsAddress() const
 // static
 bool EditableMapObject::ValidateFlats(string const & flats)
 {
+  if (strings::CountChar(flats) > kMaximumOsmChars)
+    return false;
+
   for (auto it = strings::SimpleTokenizer(flats, ";"); it; ++it)
   {
     string_view token = *it;
@@ -516,6 +519,9 @@ bool EditableMapObject::ValidatePhoneList(string const & phone)
   if (phone.empty())
     return true;
 
+  if (strings::CountChar(phone) > kMaximumOsmChars)
+    return false;
+
   auto constexpr kMaxNumberLen = 15;
   auto constexpr kMinNumberLen = 5;
 
@@ -556,6 +562,9 @@ bool EditableMapObject::ValidateEmail(string const & email)
   if (email.empty())
     return true;
 
+  if (strings::CountChar(email) > kMaximumOsmChars)
+    return false;
+
   if (strings::IsASCIIString(email))
   {
     static auto const s_emailRegex = regex(R"([^@\s]+@[a-zA-Z0-9-]+(\.[a-zA-Z0-9-]+)+$)");
@@ -589,6 +598,9 @@ bool EditableMapObject::ValidateLevel(string const & level)
   if (level.empty())
     return true;
 
+  if (strings::CountChar(level) > kMaximumOsmChars)
+    return false;
+
   if (level.front() == ';' || level.back() == ';' || level.find(";;") != std::string::npos)
     return false;
 
@@ -633,6 +645,9 @@ bool EditableMapObject::ValidateName(string const & name)
   if (name.empty())
     return true;
 
+  if (strings::CountChar(name) > kMaximumOsmChars)
+    return false;
+
   static std::u32string_view constexpr excludedSymbols = U"^§><*=_±√•÷׶";
 
   using Iter = utf8::unchecked::iterator<string::const_iterator>;

libs/indexer/editable_map_object.hpp

@@ -70,6 +70,7 @@ class EditableMapObject : public MapObject
 {
 public:
   static uint8_t constexpr kMaximumLevelsEditableByUsers = 50;
+  static int constexpr kMaximumOsmChars = 255;
 
   bool IsNameEditable() const;
   bool IsAddressEditable() const;

libs/indexer/validate_and_format_contacts.cpp

@@ -408,6 +408,10 @@ bool ValidateWebsite(string const & site)
 
   auto const startPos = GetProtocolNameLength(site);
 
+  // check lengt and leave room for addition of 'http://'
+  if (strings::CountChar(site) > (IsProtocolSpecified(site) ? kMaximumOsmChars : kMaximumOsmChars - 7))
+    return false;
+
   if (startPos >= site.size())
     return false;
 
@@ -429,6 +433,9 @@ bool ValidateFacebookPage(string const & page)
   if (page.empty())
     return true;
 
+  if (strings::CountChar(page) > kMaximumOsmChars)
+    return false;
+
   // Check if 'page' contains valid Facebook username or page name.
   // * length >= 5
   // * no forbidden symbols in the string
@@ -452,6 +459,9 @@ bool ValidateInstagramPage(string const & page)
   if (page.empty())
     return true;
 
+  if (strings::CountChar(page) > kMaximumOsmChars)
+    return false;
+
   // Rules are defined here: https://blog.jstassen.com/2016/03/code-regex-for-instagram-username-and-hashtags/
   if (regex_match(page, s_instaRegex))
     return true;
@@ -468,6 +478,9 @@ bool ValidateTwitterPage(string const & page)
   if (page.empty())
     return true;
 
+  if (strings::CountChar(page) > kMaximumOsmChars)
+    return false;
+
   if (!ValidateWebsite(page))
     return regex_match(page, s_twitterRegex);  // Rules are defined here: https://stackoverflow.com/q/11361044
 
@@ -480,6 +493,9 @@ bool ValidateVkPage(string const & page)
   if (page.empty())
     return true;
 
+  if (strings::CountChar(page) > kMaximumOsmChars)
+    return false;
+
   {
     // Check that page contains valid username. Rules took here: https://vk.com/faq18038
     // The page name must be between 5 and 32 characters.
@@ -513,6 +529,9 @@ bool ValidateLinePage(string const & page)
   if (page.empty())
     return true;
 
+  if (strings::CountChar(page) > kMaximumOsmChars)
+    return false;
+
   {
     // Check that linePage contains valid page name.
     // Rules are defined here: https://help.line.me/line/?contentId=10009904
@@ -536,6 +555,9 @@ bool ValidateFediversePage(string const & page)
   if (page.empty())
     return true;
 
+  if (strings::CountChar(page) > kMaximumOsmChars)
+    return false;
+
   // Match @username@instance.name format
   if (regex_match(page, s_fediverseRegex))
     return true;
@@ -575,6 +597,9 @@ bool ValidateBlueskyPage(string const & page)
   if (page.empty())
     return true;
 
+  if (strings::CountChar(page) > kMaximumOsmChars)
+    return false;
+
   // Match {@?}{user/domain.name} format
   if (regex_match(page, s_blueskyRegex))
     return true;

libs/indexer/validate_and_format_contacts.hpp

@@ -6,6 +6,8 @@
 
 namespace osm
 {
+static int constexpr kMaximumOsmChars = 255;
+
 std::string ValidateAndFormat_website(std::string const & v);
 std::string ValidateAndFormat_facebook(std::string const & v);
 std::string ValidateAndFormat_instagram(std::string const & v);