Method ValidateAndFormat_wikipedia

generator/osm2meta.cpp:287–330

}

std::string MetadataTagProcessorImpl::ValidateAndFormat_wikipedia(std::string v)
{
  using std::string;
  strings::Trim(v);
  // Normalize by converting full URL to "lang:title" if necessary
  // (some OSMers do not follow standard for wikipedia tag).
  static string const base = ".wikipedia.org/wiki/";
  auto const baseIndex = v.find(base);
  if (baseIndex != string::npos)
  {
    auto const baseSize = base.size();
    // Do not allow urls without article name after /wiki/.
    if (v.size() > baseIndex + baseSize)
    {
      auto const slashIndex = v.rfind('/', baseIndex);
      if (slashIndex != string::npos && slashIndex + 1 != baseIndex)
      {
        // Normalize article title according to OSM standards.
        string title = url::UrlDecode(v.substr(baseIndex + baseSize));
        replace(title.begin(), title.end(), '_', ' ');
        return v.substr(slashIndex + 1, baseIndex - slashIndex - 1) + ":" + title;
      }
    }
    LOG(LDEBUG, ("Invalid Wikipedia tag value:", v));
    return {};
  }
  // Standard case: "lang:Article Name With Spaces".
  // Language and article are at least 2 chars each.
  auto const colonIndex = v.find(':');
  if (colonIndex == string::npos || colonIndex < 2 || colonIndex + 2 > v.size())
  {
    LOG(LDEBUG, ("Invalid Wikipedia tag value:", v));
    return {};
  }
  // Check if it's not a random/invalid link.
  if (v.find("//") != string::npos || v.find(".org") != string::npos)
  {
    LOG(LDEBUG, ("Invalid Wikipedia tag value:", v));
    return {};
  }
  // Normalize to OSM standards.
  replace(v.begin() + colonIndex, v.end(), '_', ' ');
  return v;
}

std::string MetadataTagProcessorImpl::ValidateAndFormat_wikimedia_commons(std::string v)
{