#+TITLE: sdp header extract
#+AUTHOR: Ralph Amissah
#+EMAIL: [[mailto:ralph.amissah@gmail.com][ralph.amissah@gmail.com]]
#+DESCRIPTION for documents - structuring, publishing in multiple formats and search
#+KEYWORDS
#+LANGUAGE: en
#+STARTUP: indent content
#+OPTIONS: H:3 num:nil toc:t \n:nil @:t ::t |:t ^:nil _:nil -:t f:t *:t <:t
#+OPTIONS: TeX:t LaTeX:t skip:nil d:nil todo:t pri:nil tags:not-in-toc
#+OPTIONS: author:nil email:nil creator:nil timestamp:nil
#+PROPERTY: header-args :padline no :exports code :noweb yes
#+EXPORT_SELECT_TAGS: export
#+EXPORT_EXCLUDE_TAGS: noexport
#+FILETAGS: :sdp:rel:ao:
#+TAGS: assert(a) class(c) debug(d) mixin(m) sdp(s) tangle(T) template(t) WEB(W) noexport(n)

[[./sdp.org][sdp]]  [[./][org/]]
* 0. Header Hub                                :module:sdp:ao_conf_make_meta:
** module template

#+BEGIN_SRC d :tangle ../src/sdp/ao/conf_make_meta.d
/++
  extract native/orig header return associative array<BR>

  the header is passed as text (lopped off top of a sisu markup file until the
  required first heading ^A~), determine whether is a native header or sdlang one
  with a regex check if whether it contains the "native header" required tag/field
  @title: then process accordingly as a "native header" or "sdlang header"
  converting the metadata and make instructions to a common json format used by
  program internally. Moved to associative array.
+/
module sdp.ao.conf_make_meta;
template SiSUheaderExtractHub() {
  import
    std.exception,
    std.regex,
    std.stdio,
    std.traits,
    std.typecons,
    std.utf,
    std.conv : to;
  import sdlang;
  import
    sdp.ao.conf_make_meta_native,
    sdp.ao.conf_make_meta_sdlang,
    sdp.ao.rgx;
  mixin SiSUrgxInit;
  mixin SiSUheaderExtractNative;
  mixin SiSUheaderExtractSDLang;
  auto rgx = Rgx();
  auto SiSUheaderExtractHub(Src, DocMake)(
    Src     header_src,
    DocMake conf_doc_make_aa
  ) {
    debug(asserts){
      static assert(is(typeof(header_src)       == char[]));
      static assert(is(typeof(conf_doc_make_aa) == string[string][string]));
    }
    auto head_native = HeaderDocMetadataAndMakeNativeToAA();
    auto head_sdlang = HeaderExtractSDL();
    auto header_make_and_meta_tuple = (header_src.match(rgx.native_header_meta_title))
    ? (head_native.headerNativeToAA(header_src))
    : (head_sdlang.headerSDLangToAA(header_src, conf_doc_make_aa));
    static assert(!isTypeTuple!(header_make_and_meta_tuple));
    static assert(header_make_and_meta_tuple.length==2);
    return header_make_and_meta_tuple;
  }
}
#+END_SRC

* A. header sdlang (extract)            :module:sdp:ao_conf_make_meta_sdlang:
** module template

#+BEGIN_SRC d :tangle ../src/sdp/ao/conf_make_meta_sdlang.d
/++
  sdlang headers<BR>
  extract sdlang header return sdlang
+/
module sdp.ao.conf_make_meta_sdlang;
template SiSUheaderExtractSDLang() {
  import
    std.exception,
    std.regex,
    std.stdio,
    std.string,
    std.traits,
    std.typecons,
    std.utf,
    std.conv : to;
  import
    sdp.ao.defaults,
    sdp.ao.rgx;
  struct HeaderExtractSDL {
    mixin SiSUregisters;
    mixin SiSUrgxInit;
    auto rgx = Rgx();
    <<ao_conf_make_meta_sdl>>
  }
}
#+END_SRC

** entry points
*** _sdlang to associative array_

#+name: ao_conf_make_meta_sdl
#+BEGIN_SRC d
private auto sdlangToAAheaderMakeMeta(C,Tag)(C conf, Tag conf_sdlang) {
  debug(asserts){
    static assert(is(typeof(conf) == string[string][string]));
  }
  foreach (maintag, subtags; conf) {
    foreach (subtag, content; subtags) {
      if (maintag in conf_sdlang.maybe.tags) {
        Tag _maintag = conf_sdlang.getTag(maintag);
        if ((subtag in _maintag.maybe.tags)
        && (_maintag.getTagValues(subtag).length > 0)) {
          debug(headersdlang) {
            writeln(__LINE__, ": ", maintag, ":", subtag, ": ", _maintag.getTagValues(subtag)[0]);
          }
          if (_maintag.getTagValues(subtag).length == 1) {
            conf[maintag][subtag] =
              (_maintag.getTagValues(subtag)[0]).to!string;
          } else if (_maintag.getTagValues(subtag).length > 1) {
            foreach (st; _maintag.getTagValues(subtag)) {
              conf[maintag][subtag] ~=
                st.to!string ~ ";";
            }
          }
        } else if ((subtag in _maintag.maybe.attributes)
        && (_maintag.maybe.attributes[subtag][0].value.length > 0)) {
          debug(headersdlang) {
            writeln(__LINE__, ": ", maintag, ":", subtag, ": ", conf_sdlang.tags[maintag][0].attributes[subtag][0].value);
          }
          conf[maintag][subtag] =
            (conf_sdlang.tags[maintag][0].attributes[subtag][0].value).to!string;
        }
      }
    }
  }
  return conf;
}
#+END_SRC

*** _conf settings_ sdlang

#+name: ao_conf_make_meta_sdl
#+BEGIN_SRC d
private auto configSettingsSDLangToAAmake(Tag)(Tag conf_sdlang) {
  auto conf = sdlangToAAheaderMakeMeta(conf_aa_empty, conf_sdlang);
  return conf;
}
#+END_SRC

*** _conf make_ sdlang

#+name: ao_conf_make_meta_sdl
#+BEGIN_SRC d
private auto documentMakeSDLangToAAmake(Tag)(Tag document_make_sdlang) {
  auto dochead_make = sdlangToAAheaderMakeMeta(conf_aa_empty, document_make_sdlang);
  return dochead_make;
}
#+END_SRC

** header                                                       :header:sdl:
*** sdlang header parse and _extract root Tag_            :sdlang:root:tag:

#+name: ao_conf_make_meta_sdl
#+BEGIN_SRC d
final private auto headerMakeSDLang(Hs)(Hs src_header) {
  debug(asserts){
    static assert(is(typeof(src_header) == string));
  }
  scope(failure) {
    stderr.writefln(
      "%s\n%s\n%s:%s failed here:\n  src_header: %s",
      __MODULE__, __FUNCTION__,
      __FILE__, __LINE__,
      src_header,
    );
  }
  Tag sdl_root_header;
  try {
    sdl_root_header = parseSource(src_header);
  }
  catch(ParseException e) {
    stderr.writeln("SDLang problem with this document header:");
    stderr.writeln(src_header);
    // Error messages of the form:
    // myFile.sdl(5:28): Error: Invalid integer suffix.
    stderr.writeln(e.msg);
  }
  debug(sdlang) {
    writeln("header SDL:");
    writeln(__LINE__, ": ",  sdl_root_header.toSDLDocument());
    writeln(__LINE__, ": ",  sdl_root_header.maybe.namespaces);
    writeln("header make sdlang: ", sdl_root_header.toSDLDocument());
    writeln(__LINE__, ": ", sdl_root_header.getTagValues("title"));
    writeln(__LINE__, ": ", sdl_root_header.getTagValues("creator"));
    Tag creator = sdl_root_header.getTag("creator");
    if (creator !is null) {
      if ("author" in creator.maybe.tags) {
        writeln(__LINE__, ": ", creator.getTagValues("author"));
      } else if ("author" in creator.maybe.attributes) {
        writeln(__LINE__, ": ", creator.maybe.attributes["author"][0].value);
      }
    }
  }
  return sdl_root_header;
}
#+END_SRC

*** sdlang header _src text get_                           :sdlang:get:src:

#+name: ao_conf_make_meta_sdl
#+BEGIN_SRC d
private auto headerSDLangGet(Hs)(Hs src_header) {
  debug(asserts){
    static assert(is(typeof(src_header) == char[]));
  }
  char[][] source_header_arr =
    (cast(char[]) src_header).split(rgx.newline_eol_delimiter);
  char[] header_clean;
  foreach(header_line; source_header_arr) {
    if (!match(header_line, rgx.comments)) {
      header_clean ~= header_line ~ "\n";
    }
  }
  /+ get sdlang tags +/
  auto header_sdlang=headerMakeSDLang(to!string(header_clean));
  return header_sdlang; // sdlang.ast.Tag
}
#+END_SRC

*** sdlang header to _associative array_ make sdlTag in         :sdlang:aa:

#+name: ao_conf_make_meta_sdl
#+BEGIN_SRC d
private auto headerSDLangToAAmake(Tag,Ma)(Tag header_sdlang, Ma dochead_make) {
  debug(asserts){
    static assert(is(typeof(dochead_make) == string[string][string]));
  }
  dochead_make = sdlangToAAheaderMakeMeta(dochead_make, header_sdlang);
  auto dochead_meta = sdlangToAAheaderMakeMeta(meta_aa_empty, header_sdlang);
  if (dochead_meta["title"]["main"].empty) {
    {
      Tag _maintag = header_sdlang.getTag("title");
      if (_maintag !is null) {
        if ("main" in _maintag.maybe.tags) {
          dochead_meta["title"]["main"] =
            to!string(_maintag.getTagValues("main"));
        } else if ("main" !in _maintag.maybe.attributes) {
          writeln(__LINE__, ": ", _maintag.values[0]);
          dochead_meta["title"]["main"] =
            (_maintag.values[0]).to!string; // test that this exists
        }
      }
    }
  }
  if (!(dochead_meta["title"]["subtitle"].empty)
  && (dochead_meta["title"]["sub"].empty)) {
    dochead_meta["title"]["sub"] ~= dochead_meta["title"]["subtitle"];
  }
  dochead_meta["title"].remove("subtitle");
  if (dochead_meta["title"]["sub"].empty) {
    dochead_meta["title"]["full"] ~= dochead_meta["title"]["main"];
  } else {
    dochead_meta["title"]["full"] ~= format(
      "%s - %s",
      dochead_meta["title"]["main"],
      dochead_meta["title"]["sub"],
    );
  }
  dochead_meta["creator"]["author_raw"] = dochead_meta["creator"]["author"];
  string[] authors_arr;
  auto authors_raw_arr = dochead_meta["creator"]["author"].split(rgx.arr_delimiter);
  foreach (author_raw; authors_raw_arr) {
    authors_arr ~= author_raw.replace(rgx.raw_author_munge, "$2 $1");
  }
  dochead_meta["creator"]["author"] = join(authors_arr, ", ").chomp.chomp;
  auto t = tuple(dochead_make, dochead_meta);
  static assert(t.length==2);
  return t;
}
#+END_SRC

*** hub: get sdlang header and convert to associative array :hub:sdlang:aa:

#+name: ao_conf_make_meta_sdl
#+BEGIN_SRC d
private auto headerSDLangToAA(Hs,Ma)(Hs header_sdlang_src, Ma conf_doc_make_aa) {
  debug(asserts){
    static assert(is(typeof(header_sdlang_src) == char[]));
    static assert(is(typeof(conf_doc_make_aa) == string[string][string]));
  }
  auto header_sdlang_tag = headerSDLangGet(header_sdlang_src);
  auto header_aa_tuple = headerSDLangToAAmake(header_sdlang_tag, conf_doc_make_aa);
  return header_aa_tuple;
}
#+END_SRC

* B. header native                      :module:sdp:ao_conf_make_meta_native:
** Header Native                :header:native:markup_header_extract_native:

#+BEGIN_SRC d :tangle ../src/sdp/ao/conf_make_meta_native.d
/++
  native headers using<br>@title:<BR>:subtitle:<BR>type tags<BR>
  extract native/orig header return associative array
+/
module sdp.ao.conf_make_meta_native;
template SiSUheaderExtractNative() {
  import
    std.exception,
    std.regex,
    std.stdio,
    std.string,
    std.traits,
    std.typecons,
    std.utf,
    std.conv : to;
  import
    sdp.ao.defaults,
    sdp.ao.rgx;
  struct HeaderDocMetadataAndMakeNativeToAA {
    mixin SiSUregisters;
    mixin SiSUrgxInitFlags;
    mixin SiSUrgxInit;
    auto rgx = Rgx();
    enum State { off, on }
    string hm, hs;
    <<ao_markup_header_extract_native>>
  }
}
#+END_SRC

** native header document metadata in associative array                 :aa:

#+name: ao_markup_header_extract_native
#+BEGIN_SRC d
auto header_metadata_and_make_aa(H,Me,Ma)(
  H  header,
  Me dochead_meta,
  Ma dochead_make
)
in {
  debug(asserts){
    static assert(is(typeof(header)       == string));
    static assert(is(typeof(dochead_meta) == string[string][string]));
    static assert(is(typeof(dochead_make) == string[string][string]));
  }
}
body {
  scope(exit) {
    destroy(header);
    destroy(dochead_meta);
    destroy(dochead_make);
  }
  if (auto t = header.match(rgx.native_header_main)) {
    char[][] header_obj_spl =
      (cast(char[]) header).split(rgx.line_delimiter_ws_strip);
    auto hm = to!string(t.captures[1]);
    if (hm.match(rgx.main_headers)) {
      foreach (line; header_obj_spl) {
        if (auto m = line.match(rgx.native_header_main)) {
          if (!empty(m.captures[2])) {
            if (hm == "creator") {
              dochead_meta[hm]["author"] =
                to!string(m.captures[2]);
            } else if (hm == "title") {
              dochead_meta[hm]["main"] =
                to!string(m.captures[2]);
            } else if (hm == "publisher") {
              dochead_meta[hm]["name"] =
                to!string(m.captures[2]);
            }
          }
        } else if (auto s = match(line, rgx.native_header_sub)) {
          if (!empty(s.captures[2])) {
            auto hs = to!string(s.captures[1]);
            if ((hm == "make" )
            && (dochead_make[hm])) {
              switch (hm) {
              case "make":
                if (hs.match(rgx.native_subhead_make)) {
                  if (dochead_make[hm][hs]) {
                    dochead_make[hm][hs] = to!string(s.captures[2]);
                  }
                } else {
                  writeln("not a valid header type:", hm, ":", hs);
                  destroy(hm);
                  destroy(hs);
                }
                break;
              default:
                break;
              }
            } else if (dochead_meta[hm]) {
              switch (hm) {
              case "creator":
                if (hs.match(rgx.native_subhead_creator)) {
                  if (dochead_meta[hm][hs]) {
                    dochead_meta[hm][hs] =
                      to!string(s.captures[2]);
                  }
                } else {
                  writeln("not a valid header type:", hm, ":", hs);
                  destroy(hm);
                  destroy(hs);
                }
                break;
              case "title":
                if (hs.match(rgx.native_subhead_title)) {
                  if ((hs == "subtitle")
                  && (dochead_meta[hm]["sub"])) {
                    dochead_meta[hm]["sub"] =
                      to!string(s.captures[2]);
                  } else if (dochead_meta[hm][hs]) {
                    dochead_meta[hm][hs] =
                      to!string(s.captures[2]);
                  }
                } else {
                  writeln("not a valid header type:", hm, ":", hs);
                  destroy(hm);
                  destroy(hs);
                }
                break;
              case "rights":
                if (hs.match(rgx.native_subhead_rights)) {
                  if (dochead_meta[hm][hs]) {
                    dochead_meta[hm][hs] =
                      to!string(s.captures[2]);
                  }
                } else {
                  writeln("not a valid header type:", hm, ":", hs);
                  destroy(hm);
                  destroy(hs);
                }
                break;
              case "date":
                if (hs.match(rgx.native_subhead_date)) {
                  if (dochead_meta[hm][hs]) {
                    dochead_meta[hm][hs] =
                      to!string(s.captures[2]);
                  }
                } else {
                  writeln("not a valid header type:", hm, ":", hs);
                  destroy(hm);
                  destroy(hs);
                }
                break;
              case "original":
                if (hs.match(rgx.native_subhead_original)) {
                  if (dochead_meta[hm][hs]) {
                    dochead_meta[hm][hs] =
                      to!string(s.captures[2]);
                  }
                } else {
                  writeln("not a valid header type:", hm, ":", hs);
                  destroy(hm);
                  destroy(hs);
                }
                break;
              case "classify":
                if (hs.match(rgx.native_subhead_classify)) {
                  if (dochead_meta[hm][hs]) {
                    dochead_meta[hm][hs] =
                      to!string(s.captures[2]);
                  }
                } else {
                  writeln("not a valid header type:", hm, ":", hs);
                  destroy(hm);
                  destroy(hs);
                }
                break;
              case "identifier":
                if (hs.match(rgx.native_subhead_identifier)) {
                  if (dochead_meta[hm][hs]) {
                    dochead_meta[hm][hs] =
                      to!string(s.captures[2]);
                  }
                } else {
                  writeln("not a valid header type:", hm, ":", hs);
                  destroy(hm);
                  destroy(hs);
                }
                break;
              case "notes":
                if (hs.match(rgx.native_subhead_notes)) {
                  if (dochead_meta[hm][hs]) {
                    dochead_meta[hm][hs] =
                      to!string(s.captures[2]);
                  }
                } else {
                  writeln("not a valid header type:", hm, ":", hs);
                  destroy(hm);
                  destroy(hs);
                }
                break;
              case "publisher":
                if (hs.match(rgx.native_subhead_publisher)) {
                  if (dochead_meta[hm][hs]) {
                    dochead_meta[hm][hs] =
                      to!string(s.captures[2]);
                  }
                } else {
                  writeln("not a valid header type:", hm, ":", hs);
                  destroy(hm);
                  destroy(hs);
                }
                break;
              case "links":
                destroy(hm);
                destroy(hs);
                break;
              default:
                break;
              }
            }
          }
        }
      }
    } else {
      writeln("not a valid header type:", hm);
    }
  }
  auto t = tuple(dochead_meta, dochead_make);
  static assert(t.length==2);
  return t;
}
#+END_SRC

** native header extract to string object                           :string:

#+name: ao_markup_header_extract_native
#+BEGIN_SRC d
private auto native_header_extract(L,Lo,O,T)(
  L                line,
  return ref Lo    line_occur,
  return ref O     an_object,
  return ref T     type
) {
  debug(asserts){
    static assert(is(typeof(line)       == char[]));
    static assert(is(typeof(line_occur) == int[string]));
    static assert(is(typeof(an_object)  == string[string]));
    static assert(is(typeof(type)       == int[string]));
  }
  if (line.matchFirst(rgx.native_header_make)) {   /+ matched header_make +/
    debug(header1) { /+ writeln(line); +/ }
    type["header"]      = State.on;
    type["header_make"] = State.on;
    type["header_meta"] = State.off;
    ++line_occur["header_make"];
    an_object["body_nugget"] ~= line ~= "\n";
  } else if (line.matchFirst(rgx.native_header)) { /+ matched header_metadata +/
    /+ (generic header match and not previously caught by header_make) +/
    debug(header1) { /+ writeln(line); +/ }
    type["header"]      = State.on;
    type["header_make"] = State.off;
    type["header_meta"] = State.on;
    ++line_occur["header_meta"];
    an_object["body_nugget"] ~= line ~= "\n";
  } else if (type["header_make"] == State.on
  && (line_occur["header_make"] > State.off)) {     /+ header_make flag set +/
    if (line.matchFirst(rgx.native_header_sub)) {  /+ sub-header +/
      debug(header1) { /+ writeln(line); +/ }
      ++line_occur["header_make"];
      an_object["body_nugget"] ~= line ~= "\n";
    }
  } else if (type["header_meta"] == State.on
  && (line_occur["header_meta"] > State.off)) {     /+ header_metadata flag set +/
    if (line.matchFirst(rgx.native_header_sub)) {  /+ sub-header +/
      debug(header1) { /+ writeln(line); +/ }
      ++line_occur["header_meta"];
      an_object["body_nugget"] ~= line ~= "\n";
    }
  }
  return an_object;
}
#+END_SRC

** native header reset states                                        :reset:

#+name: ao_markup_header_extract_native
#+BEGIN_SRC d
auto header_reset_states_common(Lo,O,T)(
  return ref Lo    line_occur,
  return ref O     an_object,
  return ref T     type
) {
  debug(asserts){
    static assert(is(typeof(line_occur) == int[string]));
    static assert(is(typeof(an_object)  == string[string]));
    static assert(is(typeof(type)       == int[string]));
  }
  line_occur["header_make"] = State.off;
  line_occur["header_meta"] = State.off;
  type["header"] = State.off;
  an_object.remove("body_nugget");
  an_object.remove("is");
  an_object.remove("attrib");
}
#+END_SRC

** hub: native header start                                            :hub:

#+name: ao_markup_header_extract_native
#+BEGIN_SRC d
private auto headerNativeToAA(Hn)(Hn src_header) {
  debug(asserts){
    static assert(is(typeof(src_header) == char[]));
  }
  auto type = flags_type_init;
  type = [
   "header"          : State.off,
   "header_make"     : State.off,
   "header_meta"     : State.off,
  ];
  string[string] an_object;
  int[string] line_occur;
  auto dochead_make = conf_aa_empty;
  auto dochead_meta = meta_aa_empty;
  auto set_header = HeaderDocMetadataAndMakeNativeToAA();
  char[][] source_header_arr =
    (cast(char[]) src_header).split(rgx.newline_eol_delimiter);
  foreach(header_line; source_header_arr) {
    if (auto m = header_line.matchFirst(rgx.comment)) {
      /+ matched comment +/
      debug(comment) {
      }
      header_reset_states_common(line_occur, an_object, type);
    } else if ((header_line.matchFirst(rgx.native_header))
    || (type["header_make"] == State.on
    && (line_occur["header_make"] > State.off))
    || (type["header_meta"] == State.on
    && (line_occur["header_meta"] > State.off))) {
      if (header_line.length == 0) {
        /+ header_make instructions (current line empty) +/
        auto dochead_metadata_and_make =
          set_header.header_metadata_and_make_aa(strip(an_object["body_nugget"]), dochead_meta, dochead_make);
        static assert(!isTypeTuple!(dochead_metadata_and_make));
        dochead_meta = dochead_metadata_and_make[0];
        dochead_make = dochead_metadata_and_make[1];
        header_reset_states_common(line_occur, an_object, type);
        type["header_make"] = State.off;
        type["header_meta"] = State.off;
        debug(headersdlang) {
          writeln(dochead_metadata_and_make);
        }
      } else {
        an_object = native_header_extract(header_line, line_occur, an_object, type);
      }
    }
  }
  auto t = tuple(
    dochead_make,
    dochead_meta,
  );
  return t;
}
#+END_SRC