-*- mode: org -*-
#+TITLE:       sisudoc spine (doc_reform) markup source raw
#+DESCRIPTION: documents - structuring, publishing in multiple formats & search
#+FILETAGS:    :spine:sourcefile:read:
#+AUTHOR:      Ralph Amissah
#+EMAIL:       [[mailto:ralph.amissah@gmail.com][ralph.amissah@gmail.com]]
#+COPYRIGHT:   Copyright (C) 2015 - 2024 Ralph Amissah
#+LANGUAGE:    en
#+STARTUP:     content hideblocks hidestars noindent entitiespretty
#+PROPERTY:    header-args  :exports code
#+PROPERTY:    header-args+ :noweb yes
#+PROPERTY:    header-args+ :results no
#+PROPERTY:    header-args+ :cache no
#+PROPERTY:    header-args+ :padline no
#+PROPERTY:    header-args+ :mkdirp yes
#+OPTIONS:     H:3 num:nil toc:t \n:t ::t |:t ^:nil -:t f:t *:t

- [[./doc-reform.org][doc-reform.org]]  [[./][org/]]

* A. get _config file_, read in (.readText) [#A]
** _module template_ :module:config_files:

#+HEADER: :tangle "../src/sisudoc/io_in/read_config_files.d"
#+HEADER: :noweb yes
#+BEGIN_SRC d
<<doc_header_including_copyright_and_license>>
/++
  read configuration files<BR>
  - read config files<BR>
  meta_config_files.d
+/
module sisudoc.io_in.read_config_files;
@safe:
import
  std.file,
  std.path;
import
  sisudoc.meta,
  sisudoc.io_in.paths_source,
  sisudoc.meta.rgx_files,
  sisudoc.meta.rgx;
<<meta_config_file_hub_read_site_config>>
<<meta_config_file_hub_read_document_config>>
<<meta_config_file_hub_read_site_yaml>>
#+END_SRC

*** read config files (config local site & dr document make) (yaml)
**** site configuration

SEE NOTES on configuration hierarchy in spine.org

#+NAME: meta_config_file_hub_read_site_config
#+BEGIN_SRC d
template readConfigSite() {
  @system final auto readConfigSite(Cf,O,Cfg)(Cf _conf_file_details, O _opt_action, Cfg _cfg) {
    mixin spineRgxIn;
    static auto rgx = RgxI();
    string conf_filename = "NONE";
    string config_file_str;
    string default_config_file_str = format(q"┃
flag:
  act0:                  "--html"
  act1:                  "--html --epub"
output:
  path:                  "%s"
default:
  language:              "en"
  papersize:             "a4"
  text_wrap:             "80"
  digest:                "sha256"
webserv:
  http:                  "%s"
  host:                  "%s"
  data_http:             "%s"
  data_host:             "%s"
  data_root_url:         "%s"
  data_root_path:        "%s"
  data_root_part:        ""
  images_root_part:      "image"
  cgi_search_form_title: "%s"
  cgi_http:              "%s"
  cgi_host:              "%s"
  cgi_bin_url:           "%s"
  cgi_bin_subpath:       "%s"
  cgi_bin_path:          "%s"
  cgi_search_script:     "%s"
  cgi_port:              ""
  cgi_user:              ""
  cgi_action:            "%s"
  db_sqlite_path:        "%s"
  db_sqlite_filename:    "%s"
  db_pg_table:           ""
  db_pg_user:            ""
┃",
  _cfg.processing_path_doc_root, // doc root
  _cfg.http_request_type,        // http
  _cfg.http_host,                // host / domain
  _cfg.http_request_type,        // data "http" or "https"
  _cfg.http_host,                // data domain "localhost"
  _cfg.www_url_doc_root,         // data root url "http://locahost" "https://sisudoc.org"
  _cfg.processing_path_doc_root, // data root path
  _cfg.cgi_search_form_title,    // cgi title // e.g. "≅ SiSU Spine search"
  _cfg.http_request_type,        // cgi http
  _cfg.http_host,                // cgi host
  _cfg.cgi_url_root,             // cgi bin url
  _cfg.cgi_bin_subpath,          // cgi bin path
  _cfg.cgi_bin_root,             // cgi bin path
  _cfg.cgi_filename,             // cgi filename
  _cfg.cgi_url_action,           // cgi action
  _cfg.db_sqlite_path,           // sqlite db path
  _cfg.db_sqlite_filename,       // sqlite db filename
);
    foreach(conf_fn; [_conf_file_details.config_filename_site]) {
      foreach(pth; _conf_file_details.possible_config_path_locations.config_local_site) {
        char[] conf_file;
        conf_filename = conf_fn;
        if (exists(pth)) {
          auto f_attrib = pth.getLinkAttributes;
          if (
            _conf_file_details.possible_config_path_locations.config_local_site.length == 1
            && f_attrib.attrIsFile
          ) {
            conf_file = pth.to!(char[]);
            conf_filename = pth.baseName;
          } else if (f_attrib.attrIsDir) {
            conf_file = ((chainPath(pth.to!string, conf_fn)).asNormalizedPath).array;
            conf_filename = conf_fn;
          }
          try {
            if (exists(conf_file)) {
              if (conf_file.getLinkAttributes.attrIsFile) {
                if (_opt_action.vox_gt1 || _opt_action.debug_do) {
                  writeln("config file used: \"", conf_file, "\" (cli flag settings override config file's individual settings)");
                }
                config_file_str = conf_file.readText;
                break;
              }
            }
          } catch (ErrnoException ex) {
          } catch (FileException ex) {
          }
        }
      }
      if (config_file_str.length > 0) { break; }
    }
    if (config_file_str.length > 0) {
      import dyaml;
      Node yaml_root;
      try {
        yaml_root = Loader.fromString(config_file_str).load();
      } catch (Throwable) {
        import std.stdio;
        writeln("ERROR failed to read config file content, not parsed as yaml, program default used");
        conf_filename = "VIRTUAL";
        config_file_str = default_config_file_str;
      }
    }
    if (config_file_str.length == 0) { /+ use dummy default config file +/
      // writeln("WARNING config file NOT found, default provided");
      conf_filename = "VIRTUAL";
      config_file_str = default_config_file_str;
    }
    struct _ConfContent {
      string filename() {
        return conf_filename;
      }
      string filetype() {
        string _ft = "";
        if (content.match(rgx.yaml_config)) {
          _ft = "yaml";
        }
        return _ft;
      }
      string content() {
        return config_file_str;
      }
    }
    return _ConfContent();
  }
}
#+END_SRC

**** document make/config

#+NAME: meta_config_file_hub_read_document_config
#+HEADER: :noweb yes
#+BEGIN_SRC d
static template readConfigDoc() {
  import
    std.file,
    std.path;
  import
    sisudoc.meta,
    sisudoc.io_in.paths_source,
    sisudoc.meta.rgx_files,
    sisudoc.meta.rgx;
  @system final auto readConfigDoc(M,E)(M _manifested, E _env) {
    mixin spineRgxIn;
    static auto rgx = RgxI();
    mixin spineRgxFiles;
    static auto rgx_files = RgxFiles();
    string config_file_str;
    string conf_filename = "NONE";
    auto _conf_file_details = configFilePaths!()(_manifested, _env);
    string[] possible_config_path_locations = _conf_file_details.possible_config_path_locations.dr_document_make;
    foreach(conf_fn; [_conf_file_details.config_filename_document]) {
      foreach(pth; possible_config_path_locations) {
        char[] conf_file = ((chainPath(pth.to!string, conf_fn)).asNormalizedPath).array;
        conf_filename = conf_fn;
        if (config_file_str.length > 0) {
          break;
        }
        try {
          if (exists(conf_file)) {
            if (conf_file.getLinkAttributes.attrIsFile) {
              config_file_str = conf_file.readText;
              break;
            }
          }
        } catch (ErrnoException ex) {
        } catch (FileException ex) {
        }
      }
      if (config_file_str.length > 0) { break; }
    }
    struct _ConfContent {
      string filename() {
        return conf_filename;
      }
      string content() {
        return config_file_str;
      }
      string filetype() {
        string _ft = "";
        if (content.match(rgx.yaml_config)) {
          _ft = "yaml";
        }
        return _ft;
      }
    }
    return _ConfContent();
  }
}
#+END_SRC

*** YAML config (config local site & dr document make) :file:config:hub:

#+NAME: meta_config_file_hub_read_site_yaml
#+HEADER: :noweb yes
#+BEGIN_SRC d
static template configReadSiteYAML() {
  import
    std.file,
    std.path;
  import
    sisudoc.meta,
    sisudoc.io_in.paths_source,
    sisudoc.meta.rgx_files,
    sisudoc.meta.rgx;
  final YAMLDocument configReadSiteYAML(M,E)(M _manifested, E _env) {
    string _configuration = configReadInSiteYAML!()(_manifested, _env);
    auto _conf_file_details = configFilePaths!()(_manifested, _env);
    string _conf_yaml_fn = _conf_file_details.config_filename_site;
    YAMLDocument _yaml_conf = configYAML!()(_configuration, _conf_yaml_fn);
    return _yaml_conf;
  }
}
static template configReadDocYAML() {
  import
    std.file,
    std.path;
  import
    sisudoc.meta,
    sisudoc.io_in.paths_source;
  final YAMLDocument configReadDocYAML(M,E)(M _manifested, E _env) {
    string _configuration = configReadInDocYAML!()(_manifested, _env);
    auto _conf_file_details = configFilePaths!()(_manifested, _env);
    string _conf_yaml_fn = _conf_file_details.config_filename_document;
    YAMLDocument _yaml_conf = configYAML!()(_configuration, _conf_yaml_fn);
    return _yaml_conf;
  }
}
#+END_SRC

* B. get _markup source_, read file (.readText) [#A]:module:source_files:
** _module template_ (includes tuple)

#+HEADER: :tangle "../src/sisudoc/io_in/read_source_files.d"
#+HEADER: :noweb yes
#+BEGIN_SRC d
<<doc_header_including_copyright_and_license>>
/++
  module source_read_source_files;<BR>
  - open markup files<BR>
  - if master file scan for addional files to import/insert
+/
module sisudoc.io_in.read_source_files;
@safe:
template spineRawMarkupContent() {
  import
    std.file,
    std.path;
  import
    sisudoc.meta,
    sisudoc.io_in.paths_source,
    sisudoc.meta.rgx_files,
    sisudoc.meta.rgx;
  mixin spineRgxIn;
  static auto rgx = RgxI();
  mixin spineRgxFiles;
  static auto rgx_files = RgxFiles();
  string[] _images=[];
  string[] _extract_images(S)(S content_block) {
    string[] images_;
    string _content_block = content_block.to!string;
    if (auto m = _content_block.matchAll(rgx.image)) {
      images_ ~= m.captures[1].to!string;
    }
    return images_;
  }
  auto rawsrc = RawMarkupContent();
  alias ContentsInsertsImages = Tuple!(
    char[][], "contents",
    string[], "insert_files",
    string[], "images"
  );
  alias HeaderContentInsertsImages = Tuple!(
    char[],   "header",
    char[][], "src_txt",
    string[], "insert_files",
    string[], "images"
  );
  auto spineRawMarkupContent(O,Fn)(O _opt_action, Fn fn_src) {
    auto _0_header_1_body_content_2_insert_filelist_tuple
      = rawsrc.sourceContentSplitIntoHeaderAndBody(_opt_action, rawsrc.sourceContent(fn_src), fn_src);
    return _0_header_1_body_content_2_insert_filelist_tuple;
  }
  struct RawMarkupContent {
    final sourceContent(in string fn_src) {
      auto raw = MarkupRawUnit();
      string source_txt_str
        = raw.markupSourceReadIn(fn_src);
      return source_txt_str;
    }
    final auto sourceContentSplitIntoHeaderAndBody(O)(
      O         _opt_action,
      in string source_txt_str,
      in string fn_src=""
    ) {
      auto raw = MarkupRawUnit();
      string[] insert_file_list;
      string[] images_list;
      HeaderContentInsertsImages t
        = raw.markupSourceHeaderContentRawLineTupleArray(source_txt_str);
      char[] header_raw = t.header;
      char[][] sourcefile_body_content = t.src_txt;
      if (fn_src.match(rgx_files.src_fn_master)) { // filename with path needed if master file (.ssm) not otherwise
        auto ins = Inserts();
        ContentsInsertsImages tu
          = ins.scan_master_src_for_insert_files_and_import_content(_opt_action, sourcefile_body_content, fn_src);
        sourcefile_body_content = tu.contents;
        insert_file_list = tu.insert_files.dup;
        images_list = tu.images.dup;
      } else if (_opt_action.source || _opt_action.pod) {
        auto ins = Inserts();
        ContentsInsertsImages tu
          = ins.scan_master_src_for_insert_files_and_import_content(_opt_action, sourcefile_body_content, fn_src);
        images_list = tu.images.dup;
      }
      string header_type = "";
      t = tuple(
        header_raw,
        sourcefile_body_content,
        insert_file_list,
        images_list
      );
      return t;
    }
  }
  struct MarkupRawUnit {
    import std.file;
    <<meta_markup_source_raw_read_file_source_string>>
    <<meta_markup_source_raw_doc_header_and_content_split>>
    <<meta_markup_source_raw_source_line_array>>
    <<meta_markup_source_raw_read_in_file>>
    <<meta_markup_source_raw_tuple_of_header_and_body>>
    <<meta_markup_source_raw_get_insert_source_line_array>>
  }
  struct Inserts {
    alias ContentsAndImages = Tuple!(
      char[][], "insert_contents",
      string[], "images"
    );
    ContentsAndImages scan_subdoc_source(O)(
      O        _opt_action,
      char[][] markup_sourcefile_insert_content,
      string   fn_src
    ) {
      <<meta_inserts_scan>>
      foreach (line; markup_sourcefile_insert_content) {
        <<meta_inserts_scan_loop>>
      } // end src subdoc (inserts) loop
      <<meta_inserts_scan_post>>
    }
    ContentsInsertsImages scan_master_src_for_insert_files_and_import_content(O)(
      O        _opt_action,
      char[][] sourcefile_body_content,
      string   fn_src
    ) {
      import std.algorithm;
      <<meta_master_doc_scan_for_insert_filenames>>
      foreach (line; sourcefile_body_content) {
        <<meta_master_doc_scan_for_insert_filenames_loop>>
      } // end src doc loop
      <<meta_master_doc_scan_for_insert_filenames_post>>
    }
  }
}
#+END_SRC

** get markup source, read file :source:markup:
*** read file, source string [#A] :string:

#+NAME: meta_markup_source_raw_read_file_source_string
#+BEGIN_SRC d
final private string readInMarkupSource(in char[] fn_src) {
  enforce(
    exists(fn_src) != 0,
    "file not found: «" ~
    fn_src ~ "»"
  );
  string source_txt_str;
  try {
    if (exists(fn_src)) {
      if (fn_src.getLinkAttributes.attrIsFile) {
        source_txt_str = fn_src.readText;
      } else {
      }
    }
  } catch (ErrnoException ex) {
  } catch (UTFException ex) {
    // Handle validation errors
  } catch (FileException ex) {
    // Handle errors
  }
  std.utf.validate(source_txt_str);
  return source_txt_str;
}
#+END_SRC

*** document header & content, array.length == 2 [#A] :array:

here you split document header and body, an array.length == 2
split is on first match of level A~ (which is required)

#+NAME: meta_markup_source_raw_doc_header_and_content_split
#+BEGIN_SRC d
@trusted final private char[][] header0Content1(in string src_text) { // cast(char[])
  /+ split string on _first_ match of "^:?A~\s" into [header, content] array/tuple +/
  char[][] header_and_content;
  auto m = (cast(char[]) src_text).matchFirst(rgx.heading_a);
  header_and_content ~= m.pre;
  header_and_content ~= m.hit ~ m.post;
  assert(header_and_content.length == 2,
    "document markup is broken, header body split == "
    ~ header_and_content.length.to!string
    ~ "; (header / body array split should == 2 (split is on level A~))"
  );
  return header_and_content;
}
#+END_SRC

*** source line array :array:

#+NAME: meta_markup_source_raw_source_line_array
#+BEGIN_SRC d
@trusted final private char[][] markupSourceLineArray(in char[] src_text) { // cast(char[])
  char[][] source_line_arr
    = (cast(char[]) src_text).split(rgx.newline_eol_strip_preceding);
  return source_line_arr;
}
#+END_SRC

*** source content raw line array :array:
- used for regular .sst files; master .ssm files and; .ssi inserts
- regex is passed for relevant enforce match

**** read in file

#+NAME: meta_markup_source_raw_read_in_file
#+BEGIN_SRC d
string markupSourceReadIn(in string fn_src) {
  static auto rgx_files = RgxFiles();
  enforce(
    fn_src.match(rgx_files.src_pth_sst_or_ssm),
    "not a dr markup filename: «" ~
    fn_src ~ "»"
  );
  string source_txt_str = readInMarkupSource(fn_src);
  return source_txt_str;
}
#+END_SRC

**** tuple (a) header, (b) body content, (c) file insert list & (d) image list?

- header
- body content
- file insert list
- [image list?]

#+NAME: meta_markup_source_raw_tuple_of_header_and_body
#+BEGIN_SRC d
HeaderContentInsertsImages markupSourceHeaderContentRawLineTupleArray(in string source_txt_str) {
  string[] file_insert_list = [];
  string[] images_list = [];
  char[][] hc = header0Content1(source_txt_str);
  char[] header = hc[0];
  char[] source_txt = hc[1];
  char[][] source_line_arr = markupSourceLineArray(source_txt);
  HeaderContentInsertsImages t = tuple(
    header,
    source_line_arr,
    file_insert_list,
    images_list
  );
  return t;
}
#+END_SRC

**** get insert source line array

#+NAME: meta_markup_source_raw_get_insert_source_line_array
#+BEGIN_SRC d
final char[][] getInsertMarkupSourceContentRawLineArray(
  in char[]    fn_src_insert,
  Regex!(char) rgx_file
) {
  enforce(
    fn_src_insert.match(rgx_file),
    "not a dr markup filename: «" ~
    fn_src_insert  ~ "»"
  );
  string source_txt_str = readInMarkupSource(fn_src_insert);
  char[][] source_line_arr = markupSourceLineArray(source_txt_str);
  return source_line_arr;
}
#+END_SRC

** get markup source, master file & inserts :masterfile:inserts:

- [[./doc-reform.org][doc-reform.org]]  [[./][org/]]

*** scan inserts (sub-document) source :scan_insert_src:
**** scan subdoc source

#+NAME: meta_inserts_scan
#+BEGIN_SRC d
char[][] contents_insert;
int code_block_status     = 0;
enum codeBlock { off, curly, tic, }
auto fn_pth_full = fn_src.match(rgx_files.src_pth_sst_or_ssm);
auto markup_src_file_path = fn_pth_full.captures[1];
#+END_SRC

**** loop insert (sub-document)

#+NAME: meta_inserts_scan_loop
#+BEGIN_SRC d
if (code_block_status == codeBlock.curly) {
  if (line.matchFirst(rgx.block_curly_code_close)) {
    code_block_status = codeBlock.off;
  }
  contents_insert ~= line;
} else if (line.matchFirst(rgx.block_curly_code_open)) {
  code_block_status   = codeBlock.curly;
  contents_insert ~= line;
} else if (code_block_status == codeBlock.tic) {
  if (line.matchFirst(rgx.block_tic_close)) {
    code_block_status = codeBlock.off;
  }
  contents_insert ~= line;
} else if (line.matchFirst(rgx.block_tic_code_open)) {
  code_block_status   = codeBlock.tic;
  contents_insert ~= line;
} else if (auto m = line.match(rgx_files.insert_src_fn_ssi_or_sst)) {
  auto insert_fn = m.captures[2];
  auto insert_sub_pth = m.captures[1];
  auto fn_src_insert
    = chainPath(markup_src_file_path, insert_sub_pth ~ insert_fn).array;
  auto raw = MarkupRawUnit();
  auto markup_sourcesubfile_insert_content
    = raw.getInsertMarkupSourceContentRawLineArray(fn_src_insert, rgx_files.src_fn_find_inserts);
  debug(insert_file) {
    writeln(line);
    writeln(fn_src_insert);
    writeln(
      "  length contents insert array: ",
      markup_sourcesubfile_insert_content.length
    );
  }
  if (_opt_action.source || _opt_action.pod) {
    _images ~= _extract_images(markup_sourcesubfile_insert_content);
  }
  auto ins = Inserts();
  /+
    - 1. load file
    - 2. read lines
    - 3. scan lines
      - a. if filename insert, and insert filename
         - repeat 1
      - b. else
         - add line to new array;
         - build image list, search for any image files to add to image list
  +/
} else {
  contents_insert ~= line; // images to extract for image list?
  if (_opt_action.source || _opt_action.pod) {
    string[] _image_linelist = _extract_images(line);
    if (_image_linelist.length > 0) {
      _images ~= _image_linelist;
    }
  }
}
#+END_SRC

**** post loop

#+NAME: meta_inserts_scan_post
#+BEGIN_SRC d
ContentsAndImages t = tuple(
  contents_insert,
  _images
);
return t;
#+END_SRC

*** scan document source :scan_src:
**** scan doc source

#+NAME: meta_master_doc_scan_for_insert_filenames
#+BEGIN_SRC d
char[][] contents;
int code_block_status     = 0;
enum codeBlock { off, curly, tic, }
auto fn_pth_full = fn_src.match(rgx_files.src_pth_sst_or_ssm);
auto markup_src_file_path = fn_pth_full.captures[1];
char[][] contents_insert;
string[] _images          =[];
string[] insert_file_list =[];
#+END_SRC

**** include inserts: _loop master_ scan for inserts (insert documents)

#+NAME: meta_master_doc_scan_for_insert_filenames_loop
#+BEGIN_SRC d
if (code_block_status == codeBlock.curly) {
  if (line.matchFirst(rgx.block_curly_code_close)) {
    code_block_status = codeBlock.off;
  }
  contents ~= line;
} else if (line.matchFirst(rgx.block_curly_code_open)) {
  code_block_status = codeBlock.curly;
  contents ~= line;
} else if (code_block_status == codeBlock.tic) {
  if (line.matchFirst(rgx.block_tic_close)) {
    code_block_status = codeBlock.off;
  }
  contents ~= line;
} else if (line.matchFirst(rgx.block_tic_code_open)) {
  code_block_status = codeBlock.tic;
  contents ~= line;
} else if (auto m = line.match(rgx_files.insert_src_fn_ssi_or_sst)) {
  auto insert_fn      = m.captures[2];
  auto insert_sub_pth = m.captures[1];
  auto fn_src_insert
    = chainPath(markup_src_file_path, insert_sub_pth ~ insert_fn).array;
    insert_file_list ~= fn_src_insert.to!string;
  auto raw = MarkupRawUnit();
  /+ TODO +/
  auto markup_sourcefile_insert_content
    = raw.getInsertMarkupSourceContentRawLineArray(fn_src_insert, rgx_files.src_fn_find_inserts);
  debug(insert_file) {
    writeln(line);
    writeln(fn_src_insert);
    writeln(
      "  length contents insert array: ",
      markup_sourcefile_insert_content.length
    );
  }
  auto ins = Inserts();
  ContentsAndImages contents_insert_tu = ins.scan_subdoc_source(
    _opt_action,
    markup_sourcefile_insert_content,
    fn_src_insert.to!string
  );
  contents ~= contents_insert_tu.insert_contents;
  if (_opt_action.source || _opt_action.pod) {
    string[] _image_linelist = _extract_images(contents_insert_tu.images);
    if (_image_linelist.length > 0) {
      _images ~= _image_linelist;
    }
  }
  /+
    - 1. load file
    - 2. read lines
    - 3. scan lines
      - a. if filename insert, and insert filename
         - repeat 1
      - b. else
         - add line to new array;
         - build image list, search for any image files to add to image list
  +/
} else {
  contents ~= line;
  if (_opt_action.source || _opt_action.pod) {
    string[] _image_linelist = _extract_images(line);
    if (_image_linelist.length > 0) {
      _images ~= _image_linelist;
    }
  }
}
#+END_SRC

**** post loop

#+NAME: meta_master_doc_scan_for_insert_filenames_post
#+BEGIN_SRC d
string[] images = [];
foreach(i; uniq(_images.sort())) {
  images ~= i;
}
debug(insert_file) {
  writeln(__LINE__);
  writeln(contents.length);
}
ContentsInsertsImages t = tuple(
  contents,
  insert_file_list,
  images
);
return t;
#+END_SRC

* document header including copyright & license

#+NAME: doc_header_including_copyright_and_license
#+HEADER: :noweb yes
#+BEGIN_SRC emacs-lisp
<<./sisudoc_spine_version_info_and_doc_header_including_copyright_and_license.org:spine_doc_header_including_copyright_and_license()>>
#+END_SRC

* __END__