#pragma once /* Document Markup Language (DML) v1.0 parser * revision 0.04 */ #include namespace nall { namespace { struct DML { auto& setAllowHTML(bool allowHTML) { settings.allowHTML = allowHTML; return *this; } auto& setHost(const string& hostname) { settings.host = {hostname, "/"}; return *this; } auto& setPath(const string& pathname) { settings.path = pathname; return *this; } auto& setReader(const function& reader) { settings.reader = reader; return *this; } auto& setSectioned(bool sectioned) { settings.sectioned = sectioned; return *this; } auto parse(const string& filedata, const string& pathname) -> string; auto parse(const string& filename) -> string; private: struct Settings { bool allowHTML = true; string host = "localhost/"; string path; function reader; bool sectioned = true; } settings; struct State { string output; uint sections = 0; } state; auto parseDocument(const string& filedata, const string& pathname, uint depth) -> bool; auto parseBlock(string& block, const string& pathname, uint depth) -> bool; auto count(const string& text, char value) -> uint; auto escape(const string& text) -> string; auto markup(const string& text) -> string; }; auto DML::parse(const string& filedata, const string& pathname) -> string { settings.path = pathname; parseDocument(filedata, settings.path, 0); return state.output; } auto DML::parse(const string& filename) -> string { if(!settings.path) settings.path = Location::path(filename); string document = settings.reader ? settings.reader(filename) : string::read(filename); parseDocument(document, settings.path, 0); return state.output; } auto DML::parseDocument(const string& filedata, const string& pathname, uint depth) -> bool { if(depth >= 100) return false; //attempt to prevent infinite recursion with reasonable limit auto blocks = filedata.split("\n\n"); for(auto& block : blocks) parseBlock(block, pathname, depth); if(settings.sectioned && state.sections && depth == 0) state.output.append("\n"); return true; } auto DML::parseBlock(string& block, const string& pathname, uint depth) -> bool { if(!block.stripRight()) return true; auto lines = block.split("\n"); //include if(block.beginsWith("")) { string filename{pathname, block.trim("", 1L).strip()}; string document = settings.reader ? settings.reader(filename) : string::read(filename); parseDocument(document, Location::path(filename), depth + 1); } //html else if(block.beginsWith("\n") && settings.allowHTML) { for(auto n : range(lines.size())) { if(n == 0 || !lines[n].beginsWith(" ")) continue; state.output.append(lines[n].trimLeft(" ", 1L), "\n"); } } //section else if(block.beginsWith("# ")) { if(settings.sectioned) { if(state.sections++) state.output.append(""); state.output.append("
"); } auto content = lines.takeLeft().trimLeft("# ", 1L).split("::", 1L).strip(); auto data = markup(content[0]); auto name = escape(content(1, data.hash())); state.output.append("
", data); for(auto& line : lines) { if(!line.beginsWith("# ")) continue; state.output.append("", line.trimLeft("# ", 1L), ""); } state.output.append("
\n"); } //header else if(auto depth = count(block, '=')) { auto content = slice(lines.takeLeft(), depth + 1).split("::", 1L).strip(); auto data = markup(content[0]); auto name = escape(content(1, data.hash())); if(depth <= 6) { state.output.append("", data); for(auto& line : lines) { if(count(line, '=') != depth) continue; state.output.append("", slice(line, depth + 1), ""); } state.output.append("\n"); } } //navigation else if(count(block, '-')) { state.output.append("\n"); } //list else if(count(block, '*')) { uint level = 0; for(auto& line : lines) { if(auto depth = count(line, '*')) { while(level < depth) level++, state.output.append("
    \n"); while(level > depth) level--, state.output.append("
\n"); auto data = markup(slice(line, depth + 1)); state.output.append("
  • ", data, "
    • \n"); } } while(level--) state.output.append("\n"); } //quote else if(count(block, '>')) { uint level = 0; for(auto& line : lines) { if(auto depth = count(line, '>')) { while(level < depth) level++, state.output.append("
    \n"); while(level > depth) level--, state.output.append("
    \n"); auto data = markup(slice(line, depth + 1)); state.output.append(data, "\n"); } } while(level--) state.output.append("\n"); } //code else if(block.beginsWith(" ")) { state.output.append("
    ");
    for(auto& line : lines) {
      if(!line.beginsWith("  ")) continue;
      state.output.append(escape(line.trimLeft("  ", 1L)), "\n");
    }
    state.output.trimRight("\n", 1L).append("
    \n"); } //divider else if(block.equals("---")) { state.output.append("
    \n"); } //paragraph else { state.output.append("

    ", markup(block), "

    \n"); } return true; } auto DML::count(const string& text, char value) -> uint { for(uint n = 0; n < text.size(); n++) { if(text[n] != value) { if(text[n] == ' ') return n; break; } } return 0; } auto DML::escape(const string& text) -> string { string output; for(auto c : text) { if(c == '&') { output.append("&"); continue; } if(c == '<') { output.append("<"); continue; } if(c == '>') { output.append(">"); continue; } if(c == '"') { output.append("""); continue; } output.append(c); } return output; } auto DML::markup(const string& s) -> string { string t; boolean strong; boolean emphasis; boolean insertion; boolean deletion; boolean code; natural link, linkBase; natural embed, embedBase; for(uint n = 0; n < s.size();) { char a = s[n]; char b = s[n + 1]; if(!link && !embed) { if(a == '*' && b == '*') { t.append(strong.flip() ? "" : ""); n += 2; continue; } if(a == '/' && b == '/') { t.append(emphasis.flip() ? "" : ""); n += 2; continue; } if(a == '_' && b == '_') { t.append(insertion.flip() ? "" : ""); n += 2; continue; } if(a == '~' && b == '~') { t.append(deletion.flip() ? "" : ""); n += 2; continue; } if(a == '|' && b == '|') { t.append(code.flip() ? "" : ""); n += 2; continue; } if(a =='\\' && b =='\\') { t.append("
    "); n += 2; continue; } } if(!embed) { if(link == 0 && a == '[' && b == '[') { t.append(""); link = 2; n += 2; continue; } if(link == 1 && a == ']' && b == ']') { t.append("\">", slice(s, linkBase, n - linkBase), ""); n += 2; link = 0; continue; } if(link == 2 && a == ']' && b == ']') { t.append(""); n += 2; link = 0; continue; } if(link == 1 && a == '@' && b == '/') { t.append(settings.host); n += 2; continue; } } if(!link) { if(embed == 0 && a == '{' && b == '{') { t.append("\"");"); embed = 0; n += 2; continue; } if(embed == 1 && a == '@' && b == '/') { t.append(settings.host); n += 2; continue; } } if(a =='\\') { t.append(b); n += 2; continue; } if(a == '&') { t.append("&"); n++; continue; } if(a == '<') { t.append("<"); n++; continue; } if(a == '>') { t.append(">"); n++; continue; } if(a == '"') { t.append("""); n++; continue; } t.append(a); n++; } if(strong) t.append(""); if(emphasis) t.append(""); if(insertion) t.append(""); if(deletion) t.append(""); if(code) t.append(""); if(link == 1) t.append("\">", slice(s, linkBase, s.size() - linkBase), ""); if(link == 2) t.append(""); if(embed != 0) t.append("\">"); return t; } }}