QString CleanSource::PreprocessSpecialCases(const QString &source)
{
QString newsource = source;
// For content files we want to retain the functionality offered by an
// auto-correcting html parser, but this can only handle one namespace
// in the <html> tag and simply strips tags with namespace prefixes it
// doesn't understand. The solution used here is to embed the namespace
// definition in the root tag itself.
QRegularExpression root_svg_tag_with_prefix("<\\s*svg\\s*:\\s*svg");
QString root_svg_embeddedNS = "<svg xmlns=\"http://www.w3.org/2000/svg\"";
newsource.replace(root_svg_tag_with_prefix, root_svg_embeddedNS);
// Once the root tag has an embedded namespace, strip the prefix from all other tags
QRegularExpression child_svg_tag_with_prefix("<\\s*svg\\s*:");
QString child_tag_no_prefix = "<";
newsource.replace(child_svg_tag_with_prefix, child_tag_no_prefix);
return newsource;
}
// neither svg nor math tags need a namespace prefix defined
// especially as epub3 now includes them into the html5 spec
// So we need to remove the svg prefix from the tags before
// processing them with gumbo
QString CleanSource::PreprocessSpecialCases(const QString &source)
{
QString newsource = source;
// remove prefix from root tag and add unprefixed svg namespace to it
QRegularExpression root_svg_tag_with_prefix("<\\s*svg\\s*:\\s*svg");
QString root_svg_embeddedNS = "<svg xmlns=\"http://www.w3.org/2000/svg\"";
newsource.replace(root_svg_tag_with_prefix, root_svg_embeddedNS);
// search for any prefixed svg namespace in that root tag and remove it
QRegularExpression svg_nsprefix(SVG_NAMESPACE_PREFIX);
QRegularExpressionMatch mo = svg_nsprefix.match(newsource);
if (mo.hasMatch()) {
newsource.replace(mo.capturedStart(1), mo.capturedLength(1), "");
}
// now strip the prefix from all child starting tags
QRegularExpression starting_child_svg_tag_with_prefix("<\\s*svg\\s*:");
QString starting_child_tag_no_prefix = "<";
newsource.replace(starting_child_svg_tag_with_prefix, starting_child_tag_no_prefix);
// do the same for any child ending tags
QRegularExpression ending_child_svg_tag_with_prefix("<\\s*/\\s*svg\\s*:");
QString ending_child_tag_no_prefix = "</";
newsource.replace(ending_child_svg_tag_with_prefix, ending_child_tag_no_prefix);
return newsource;
}
