path: root/org/ao_header_extract.org

#+TITLE: sdp header extract
#+AUTHOR: Ralph Amissah
#+EMAIL: ralph.amissah@gmail.com
#+STARTUP: indent
#+LANGUAGE: en
#+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. sdlang to AA

#+name: ao_conf_make_meta_sdl
#+BEGIN_SRC d
private auto sdlangToAAmake(string[string][string] conf, Tag conf_sdlang) {
  foreach (maintag, subtags; conf) {
    foreach (subtag, content; subtags) {
      if (!(conf_sdlang.maybe.tags[maintag].empty)) {
        if (!(conf_sdlang.tags[maintag][0].maybe.attributes[subtag].empty)
        && (conf_sdlang.tags[maintag][0].attributes[subtag][0].value.length > 1)) {
          writeln(conf_sdlang.tags[maintag][0].attributes[subtag][0].value);
          conf[maintag][subtag] =
            to!string(conf_sdlang.tags[maintag][0].attributes[subtag][0].value);
        }
      }
    }
  }
  return conf;
}
#+END_SRC

* A. conf settings sdlang

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

* B. conf make sdlang

#+name: ao_conf_make_meta_sdl
#+BEGIN_SRC d
private auto documentMakeSDLangToAAmake(Tag document_make_sdlang) {
  auto dochead_make = sdlangToAAmake(make_aa, document_make_sdlang);
  /+
  hm = "links";
  if (!(document_make_sdlang.maybe.tags[hm].empty)) {
    /+ TODO
      stuff to fix
    +/
    // hs = "link";
    // if (!(document_make_sdlang.tags[hm][0].maybe.attributes[hs].empty)
    // && (document_make_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) {
    //   writeln(document_make_sdlang.tags[hm][0].attributes[hs][0].value);
    //   dochead_meta[hm][hs] =
    //     to!string(document_make_sdlang.tags[hm][0].attributes[hs][0].value);
    // }
  }
  +/
  // writeln(dochead_make);
  return dochead_make;
}
#+END_SRC

* C.a. header native                                          :header:native:

// mixin SiSUheader;
// auto set_header = HeaderDocMetadataAndMakeNativeToAA(); // reintroduce

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

#+name: ao_markup_header_extract_native
#+BEGIN_SRC d
auto header_metadata_and_make_aa(
  string header,
  string[string][string] dochead_meta,
  string[string][string] dochead_make
)
in { }
body {
  scope(exit) {
    destroy(header);
    destroy(dochead_meta);
    destroy(dochead_make);
  }
  if (auto t = match(header, rgx.native_header_main)) {
    char[][] header_obj_spl = split(
      cast(char[]) header,
      rgx.line_delimiter_ws_strip
    );
    auto hm = to!string(t.captures[1]);
    if (match(hm, rgx.main_headers)) {
      foreach (line; header_obj_spl) {
        if (auto m = match(line, 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 (match(hs, 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 (match(hs, 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 (match(hs, 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 (match(hs, 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 (match(hs, 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 (match(hs, 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 (match(hs, 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 (match(hs, 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 (match(hs, 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 (match(hs, 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);
                // if (match(hs, rgx.native_subhead_links)) {
                //   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;
              default:
                break;
              }
            }
          }
        }
      }
    } else {
      writeln("not a valid header type:", hm);
    }
  }
  auto t = tuple(dochead_meta, dochead_make);
  static assert(!isTypeTuple!(t));
  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(
  char[] line,
  ref int[string] line_occur,
  ref string[string] an_object,
  ref int[string] type
) {
  if (matchFirst(line, 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["obj"] ~= line ~= "\n";
  } else if (matchFirst(line, 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["obj"] ~= line ~= "\n";
  } else if (type["header_make"] == State.on
  && (line_occur["header_make"] > State.off)) {     /+ header_make flag set +/
    if (matchFirst(line, rgx.native_header_sub)) {  /+ sub-header +/
      debug(header1) { /+ writeln(line); +/ }
      ++line_occur["header_make"];
      an_object["obj"] ~= line ~= "\n";
    }
  } else if (type["header_meta"] == State.on
  && (line_occur["header_meta"] > State.off)) {     /+ header_metadata flag set +/
    if (matchFirst(line, rgx.native_header_sub)) {  /+ sub-header +/
      debug(header1) { /+ writeln(line); +/ }
      ++line_occur["header_meta"];
      an_object["obj"] ~= 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(
  ref int[string] line_occur,
  ref string[string] an_object,
  ref int[string] type
) {
  // line_occur["header"] = State.off;
  line_occur["header_make"] = State.off;
  line_occur["header_meta"] = State.off;
  type["header"] = State.off;
  // type["header_make"] = State.off;
  // type["header_meta"] = State.off;
  an_object.remove("obj");
  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(in char[] src_header) {
  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 = make_aa;
  auto dochead_meta = meta_aa;
  auto set_header = HeaderDocMetadataAndMakeNativeToAA();
  char[][] source_header_arr =
    split(cast(char[]) src_header, rgx.line_delimiter);
  foreach(header_line; source_header_arr) {
    if (auto m = matchFirst(header_line, rgx.comment)) {
      /+ matched comment +/
      debug(comment) {
        // tell_l("blue", header_line);
      }
      header_reset_states_common(line_occur, an_object, type);
      // type["header_make"] = State.off;
      // type["header_meta"] = State.off;
    } else if ((matchFirst(header_line, 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["obj"]), 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;
        writeln(dochead_metadata_and_make);
      } else {
        an_object = native_header_extract(header_line, line_occur, an_object, type);
      }
    } else {
      // writeln(__LINE__);
    }
  }
  auto t = tuple(
    dochead_make,
    dochead_meta,
  );
  return t;
}
#+END_SRC

* C.b. header sdlang                                             :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(in string src_header) {
  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(SDLangParseException 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(sdl_root_header.toSDLDocument());
  }
  return sdl_root_header;
}
#+END_SRC

** sdlang header get                                         :sdlang:get:src:

#+name: ao_conf_make_meta_sdl
#+BEGIN_SRC d
private auto headerSDLangGet(in char[] src_header) {
  char[][] source_header_arr =
    split(cast(char[]) src_header, rgx.line_delimiter);
  char[] header_clean;
  // TODO
  foreach(header_line; source_header_arr) {
    if (!match(header_line, rgx.comments)) {
      header_clean ~= header_line ~ "\n";
      // writeln(header_line);
    }
  }
  /+ get sdlang tags +/
  auto header_sdlang=headerMakeSDLang(to!string(header_clean));
  debug(sdlang) {
    writeln("--------------");
    stdout.rawWrite( header_sdlang.toSDLDocument() );
    writeln("--------------");
    Value test = header_sdlang.tags["title"][0].values[0];
    assert(test == typeid(string));
    writeln(header_sdlang.maybe.tags["title"]);
    writeln(header_sdlang.maybe.tags["title"][0].maybe.attributes["subtitle"]);
  }
  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 header_sdlang, string[string][string] dochead_make) {
  dochead_make = sdlangToAAmake(dochead_make, header_sdlang);
  auto dochead_meta = sdlangToAAmake(meta_aa, header_sdlang);
  /+
  /+ dochead +/
  string hm;
  string hs;
  /+ meta +/
  auto dochead_meta = meta_aa;
  hm = "title";
  if (!(header_sdlang.maybe.tags[hm].empty)) {
    /+ TODO Title REQUIRED +/
    hs = "main";
    if (!(header_sdlang.tags[hm].empty)
    && (header_sdlang.tags[hm][0].values[0].length > 1)) {
      writeln(header_sdlang.tags[hm][0].values[0]);
      dochead_meta[hm][hs] =
        to!string(header_sdlang.tags[hm][0].values[0]);
        // to!string(header_sdlang.tags[hm][0].attributes[hs][0].value);
    } else if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty)
    && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) {
      writeln(header_sdlang.tags[hm][0].attributes[hs][0].value);
      dochead_meta[hm][hs] =
        to!string(header_sdlang.tags[hm][0].attributes[hs][0].value);
    } else {
      writeln("Required header metadata Title, missing");
    }
    hs = "sub";
    if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty)
    && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) {
      writeln(header_sdlang.tags[hm][0].attributes[hs][0].value);
      dochead_meta[hm][hs] =
        to!string(header_sdlang.tags[hm][0].attributes[hs][0].value);
    } else if (!(header_sdlang.tags[hm][0].maybe.attributes["subtitle"].empty)
    && (header_sdlang.tags[hm][0].attributes["subtitle"][0].value.length > 1)) {
      writeln(header_sdlang.tags[hm][0].attributes["subtitle"][0].value);
      dochead_meta[hm][hs] =
        to!string(header_sdlang.tags[hm][0].attributes["subtitle"][0].value);
    }
    // full title composite (main + sub)
    // hs = "full";
    // dochead_meta[hm][hs] = dochead_meta[hm]["main"] ~ dochead_meta[hm]["sub"];
    hs = "language";
    if (!(header_sdlang.tags[hm][0].maybe.attributes["lang"].empty)
    && (header_sdlang.tags[hm][0].attributes["lang"][0].value.length > 1)) {
      writeln(header_sdlang.tags[hm][0].attributes["lang"][0].value);
      dochead_meta[hm][hs] =
        to!string(header_sdlang.tags[hm][0].attributes["lang"][0].value);
    } else if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty)
    && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) {
      writeln(header_sdlang.tags[hm][0].attributes[hs][0].value);
      dochead_meta[hm][hs] =
        to!string(header_sdlang.tags[hm][0].attributes[hs][0].value);
    }
  }
  hm = "creator";
  if (!(header_sdlang.maybe.tags[hm].empty)) {
    /+ Creator / Author REQUIRED +/
    /+ TODO
    - decide on representation for & deal with multiple authors;
    - author(s) full name;
    - author(s) surname & other parts
    +/
    hs = "author";
    if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty)
    && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) {
      writeln(header_sdlang.tags[hm][0].attributes[hs][0].value);
      dochead_meta[hm][hs] =
        to!string(header_sdlang.tags[hm][0].attributes[hs][0].value);
    }
  }
  hm = "links";
  if (!(header_sdlang.maybe.tags[hm].empty)) {
    /+ TODO
      stuff to fix
    +/
    // hs = "link";
    // if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty)
    // && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) {
    //   writeln(header_sdlang.tags[hm][0].attributes[hs][0].value);
    //   dochead_meta[hm][hs] =
    //     to!string(header_sdlang.tags[hm][0].attributes[hs][0].value);
    // }
  }
  hm = "publisher";
  if (!(header_sdlang.maybe.tags[hm].empty)) {
    hs = "name";
    if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty)
    && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) {
      writeln(header_sdlang.tags[hm][0].attributes[hs][0].value);
      dochead_meta[hm][hs] =
        to!string(header_sdlang.tags[hm][0].attributes[hs][0].value);
    }
  }
  }
  +/
  auto t = tuple(dochead_make, dochead_meta);
  static assert(!isTypeTuple!(t));
  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(char[] header_sdlang_src, string[string][string] conf_doc_make_aa) {
  auto header_sdlang_tag = headerSDLangGet(header_sdlang_src); // sdlang.ast.Tag
  auto header_aa_tuple = headerSDLangToAAmake(header_sdlang_tag, conf_doc_make_aa);
  return header_aa_tuple;
}
#+END_SRC

* tangles (code structure)                                           :tangle:
** 1. Header Hub                                 :ao_markup_header_extract:

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.

#+BEGIN_SRC d :tangle ../src/sdp/ao_conf_make_meta.d
/+
  extract native/orig header return associative array
+/
template SiSUheaderExtractHub() {
  private import
    std.regex;
  private import
    ao_rgx;
  struct HeaderDocMetadataAndMake {
    mixin SiSUheaderExtractNative;
    mixin SiSUheaderExtractSDLang;
    auto rgx = Rgx();
    private auto headerContentAA(char[] header_src, string[string][string] conf_doc_make_aa) {
      auto head_native = HeaderDocMetadataAndMakeNativeToAA();
      auto head_sdlang = HeaderExtractSDL();
      writeln(__LINE__);
      auto header_make_and_meta_tuple = (match(header_src, rgx.native_header_meta_title))
      ? (head_native.headerNativeToAA(header_src))
      : (head_sdlang.headerSDLangToAA(header_src, conf_doc_make_aa));
      writeln(__LINE__);
      static assert(!isTypeTuple!(header_make_and_meta_tuple));
      return header_make_and_meta_tuple;
    }
  }
}
#+END_SRC

** 2a. Header Native                      :ao_markup_header_extract_native:

#+BEGIN_SRC d :tangle ../src/sdp/ao_conf_make_meta_native.d
/+
  extract native/orig header return associative array
+/
template SiSUheaderExtractNative() {
  private import
    std.exception,
    std.regex,
    std.utf,
    std.conv : to;
  private import
    ao_rgx;
  struct HeaderDocMetadataAndMakeNativeToAA {
    mixin SiSUregisters;
    mixin SiSUrgxInitFlags;
    mixin RgxInit;
    auto rgx = Rgx();
    enum State { off, on }
    string hm, hs;
    <<ao_markup_header_extract_native>>
  }
}
#+END_SRC

** 2b. Header SDLang                      :ao_markup_header_extract_sdlang:

#+BEGIN_SRC d :tangle ../src/sdp/ao_conf_make_meta_sdlang.d
/+
  extract sdl header return sdl
+/
template SiSUheaderExtractSDLang() {
  private import
    std.regex;
  private import
    ao_rgx;
  struct HeaderExtractSDL {
    mixin SiSUregisters;
    mixin RgxInit;
    auto rgx = Rgx();
    <<ao_conf_make_meta_sdl>>
    <<ao_conf_make_meta_sdl_to_aa>>
  }
}
#+END_SRC