path: root/org/source_files_read.org

-*- mode: org -*-
#+TITLE:       doc_reform markup source raw
#+DESCRIPTION: documents - structuring, publishing in multiple formats & search
#+FILETAGS:    :doc_reform:sourcefile:read:
#+AUTHOR:      Ralph Amissah
#+EMAIL:       [[mailto:ralph.amissah@gmail.com][ralph.amissah@gmail.com]]
#+COPYRIGHT:   Copyright (C) 2015 - 2019 Ralph Amissah
#+LANGUAGE:    en
#+STARTUP:     indent content hideblocks hidestars
#+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 :cache no :noweb yes
#+EXPORT_SELECT_TAGS:  export
#+EXPORT_EXCLUDE_TAGS: noexport
#+TAGS: assert(a) class(c) debug(d) mixin(m) doc_reform(s) tangle(T) template(t) WEB(W) noexport(n)

[[./doc_reform.org][doc_reform]]  [[./][org/]]
* imports

#+name: imports_std
#+BEGIN_SRC d
import
  doc_reform.meta,
  doc_reform.source.paths_source,
  std.file,
  std.path;
#+END_SRC

* A. get _config file_ (read in)
** 0. module template                                  :module:config_files:

#+BEGIN_SRC d :tangle "../src/doc_reform/source/read_config_files.d"
/++
  read configuration files<BR>
  - read config files<BR>
  meta_config_files.d
+/
module doc_reform.source.read_config_files;
<<meta_config_file_in>>
<<meta_config_file_toml>>
<<meta_config_file_hub>>
#+END_SRC

*** 0. read config files (config_local_site & sisu_document_make) toml
**** 1. site configuration

#+name: meta_config_file_hub
#+BEGIN_SRC d
static template readConfigSite() {
  <<imports_std>>
  final auto readConfigSite(M,E)(M _manifest, E _env) {
    string config_file_str;
    string conf_filename = "NONE";
    auto _conf_file_details = ConfigFilePaths!()(_manifest, _env);
    string[] possible_config_path_locations = _conf_file_details.possible_config_path_locations.config_local_site;
    foreach(conf_fn; [_conf_file_details.config_filename_site_toml]) {
      foreach(pth; possible_config_path_locations) {
        char[] conf_file = asNormalizedPath(chainPath(pth.to!string, conf_fn)).array;
        conf_filename = conf_fn;
        if (config_file_str.length > 0) {
          // conf_filename = conf_fn;
          break;
        }
        try {
          if (exists(conf_file)) {
            debug(io) {
              writeln("WARNING (io debug) in config file found: ", conf_file);
            }
            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 filetype() {
        return conf_filename.extension.chompPrefix(".");
      }
      string content() {
        return config_file_str;
      }
    }
    return _ConfContent();
  }
}
#+END_SRC

**** 2. document make/config

#+name: meta_config_file_hub
#+BEGIN_SRC d
static template readConfigDoc() {
  <<imports_std>>
  final auto readConfigDoc(M,E)(M _manifest, E _env) {
    string config_file_str;
    string conf_filename = "NONE";
    auto _conf_file_details = ConfigFilePaths!()(_manifest, _env);
    string[] possible_config_path_locations = _conf_file_details.possible_config_path_locations.sisu_document_make;
    foreach(conf_fn; [_conf_file_details.config_filename_document_toml]) {
      foreach(pth; possible_config_path_locations) {
        auto conf_file = asNormalizedPath(chainPath(pth.to!string, conf_fn)).array;
        conf_filename = conf_fn;
        if (config_file_str.length > 0) {
          // conf_filename = conf_fn;
          break;
        }
        try {
          if (exists(conf_file)) {
            debug(io) {
              writeln("WARNING (io debug) in config file found: ", conf_file);
            }
            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 filetype() {
        return conf_filename.extension.chompPrefix(".");
      }
      auto content() {
        return config_file_str;
      }
    }
    return _ConfContent();
  }
}
#+END_SRC

** A. TOML                                                            :toml:
*** 1. TOML read config files (config_local_site & sisu_document_make) :file:config:
**** TOML config_local_site

#+name: meta_config_file_in
#+BEGIN_SRC d
static template configReadInSiteTOML() {
  <<imports_std>>
  final string configReadInSiteTOML(M,E)(M manifest, E env) {
    auto conf_file_details = ConfigFilePaths!()(manifest, env);
    string conf_toml = conf_file_details.config_filename_site_toml;
    string[] possible_config_path_locations = conf_file_details.possible_config_path_locations.config_local_site;
    string config_file_str;
    debug(io) {
      writeln("WARNING (io debug) in config filename: ", conf_toml);
      writeln("WARNING (io debug) in config possible path locations: ", possible_config_path_locations);
    }
    foreach(pth; possible_config_path_locations) {
      auto conf_file = asNormalizedPath(chainPath(pth.to!string, conf_toml)).array;
      if (config_file_str.length > 0) {
        break;
      }
      try {
        if (exists(conf_file)) {
          debug(io) {
            writeln("WARNING (io debug) in config file found: ", conf_file);
          }
          config_file_str = conf_file.readText;
          break;
        }
      } catch (ErrnoException ex) {
      } catch (FileException ex) {
      }
    }
    return config_file_str;
  }
}
#+END_SRC

**** TOML sisu_document_make

#+name: meta_config_file_in
#+BEGIN_SRC d
static template configReadInDocTOML() {
  <<imports_std>>
  final string configReadInDocTOML(M,E)(M manifest, E env) {
    auto conf_file_details = ConfigFilePaths!()(manifest, env);
    string conf_toml = conf_file_details.config_filename_document_toml;
    string[] possible_config_path_locations = conf_file_details.possible_config_path_locations.sisu_document_make;
    string config_file_str;
    debug(io) {
      writeln("WARNING (io debug) in config filename: ", conf_toml);
      writeln("WARNING (io debug) in config possible path locations: ", possible_config_path_locations);
    }
    foreach(pth; possible_config_path_locations) {
      auto conf_file = asNormalizedPath(chainPath(pth.to!string, conf_toml)).array;
      if (config_file_str.length > 0) {
        break;
      }
      try {
        if (exists(conf_file)) {
          debug(io) {
            writeln("WARNING (io debug) in config file found: ", conf_file);
          }
          config_file_str = conf_file.readText;
          break;
        }
      } catch (ErrnoException ex) {
      } catch (FileException ex) {
      }
    }
    return config_file_str;
  }
}
#+END_SRC

*** 2. TOML config files get

#+name: meta_config_file_toml
#+BEGIN_SRC d
static template configTOML() {
  import toml; //
  <<imports_std>>
  TOMLDocument configTOML(string configuration, string conf_toml_filename) {
    TOMLDocument _toml_conf;
    try {
      _toml_conf = parseTOML(configuration); // parseTOML(cast(string)(configuration));
    } catch(ErrnoException e) {
      stderr.writeln("Toml problem with content for ", conf_toml_filename);
      stderr.writeln(e.msg);
    }
    return _toml_conf;
  }
}
#+END_SRC

*** 3. TOML config (config_local_site & sisu_document_make) :file:config:hub:

#+name: meta_config_file_hub
#+BEGIN_SRC d
static template configReadSiteTOML() {
  <<imports_std>>
  import toml;
  final TOMLDocument configReadSiteTOML(M,E)(M _manifest, E _env) {
    string _configuration = configReadInSiteTOML!()(_manifest, _env);
    auto _conf_file_details = ConfigFilePaths!()(_manifest, _env);
    string _conf_toml = _conf_file_details.config_filename_site_toml;
    TOMLDocument _toml_conf = configTOML!()(_configuration, _conf_toml);
    return _toml_conf;
  }
}
static template configReadDocTOML() {
  <<imports_std>>
  import toml;
  final TOMLDocument configReadDocTOML(M,E)(M _manifest, E _env) {
    string _configuration = configReadInDocTOML!()(_manifest, _env);
    auto _conf_file_details = ConfigFilePaths!()(_manifest, _env);
    string _conf_toml = _conf_file_details.config_filename_document_toml;
    TOMLDocument _toml_conf = configTOML!()(_configuration, _conf_toml);
    return _toml_conf;
  }
}
#+END_SRC

* B. get _markup source_, read file                       :module:source_files:
** 0. module template (includes tuple)

#+BEGIN_SRC d :tangle "../src/doc_reform/source/read_source_files.d"
/++
  module source_read_source_files;<BR>
  - open markup files<BR>
  - if master file scan for addional files to import/insert
+/
module doc_reform.source.read_source_files;
static template DocReformRawMarkupContent() {
  import
    doc_reform.meta.rgx;
  <<imports_std>>
  mixin DocReformRgxInit;
  static auto rgx = Rgx();
  string[] _images=[];
  auto _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();
  auto DocReformRawMarkupContent(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();
      auto 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;
      auto t
        = raw.markupSourceHeaderContentRawLineTupleArray(source_txt_str);
      auto header_raw = t[0];
      auto sourcefile_body_content = t[1];
      if (fn_src.match(rgx.src_fn_master)) { // filename with path needed if master file (.ssm) not otherwise
        auto ins = Inserts();
        auto tu
          = ins.scan_master_src_for_insert_files_and_import_content(_opt_action, sourcefile_body_content, fn_src);
        static assert(!isTypeTuple!(tu));
        sourcefile_body_content = tu[0];
        insert_file_list = tu[1].dup;
        images_list = tu[2].dup;
      } else if (_opt_action.source || _opt_action.pod) {
        auto ins = Inserts();
        auto tu
          = ins.scan_master_src_for_insert_files_and_import_content(_opt_action, sourcefile_body_content, fn_src);
        static assert(!isTypeTuple!(tu));
        images_list = tu[2].dup;
      }
      t = tuple(
        header_raw,
        sourcefile_body_content,
        insert_file_list,
        images_list
      );
      static assert(t.length==4);
      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 {
    auto scan_subdoc_source(O)(
      O        _opt_action,
      char[][] markup_sourcefile_insert_content,
      string   fn_src
    ) {
      mixin DocReformRgxInitFlags;
      <<meta_inserts_scan>>
      foreach (line; markup_sourcefile_insert_content) {
        <<meta_inserts_scan_loop>>
      } // end src subdoc (inserts) loop
      <<meta_inserts_scan_post>>
    }
    auto scan_master_src_for_insert_files_and_import_content(O)(
      O        _opt_action,
      char[][] sourcefile_body_content,
      string   fn_src
    ) {
      import std.algorithm;
      mixin DocReformRgxInitFlags;
      <<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)) {
      debug(io) {
        writeln("in src, markup source file found: ", fn_src);
      }
      source_txt_str = fn_src.readText;
    }
  } 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
final private char[][] header0Content1(in string src_text) {
  /+ 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
final private char[][] markupSourceLineArray(in char[] src_text) {
  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
auto markupSourceReadIn(in string fn_src) {
  static auto rgx = Rgx();
  enforce(
    fn_src.match(rgx.src_pth_sst_or_ssm),
    "not a sisu markup filename: «" ~
    fn_src ~ "»"
  );
  auto 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
auto markupSourceHeaderContentRawLineTupleArray(in string source_txt_str) {
  string[] file_insert_list = [];
  string[] images_list = [];
  auto hc = header0Content1(source_txt_str);
  auto header = hc[0];
  char[] source_txt = hc[1];
  auto source_line_arr = markupSourceLineArray(source_txt);
  auto 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 sisu markup filename: «" ~
    fn_src_insert  ~ "»"
  );
  auto source_txt_str = readInMarkupSource(fn_src_insert);
  auto 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/]]

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

#+name: meta_inserts_scan
#+BEGIN_SRC d
char[][] contents_insert;
int[string] type1 = flags_type_init;
auto fn_pth_full = fn_src.match(rgx.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 (type1["curly_code"] == 1) {
  type1["header_make"] = 0;
  type1["header_meta"] = 0;
  if (line.matchFirst(rgx.block_curly_code_close)) {
    type1["curly_code"] = 0;
  }
  contents_insert ~= line;
} else if (line.matchFirst(rgx.block_curly_code_open)) {
  type1["curly_code"] = 1;
  type1["header_make"] = 0;
  type1["header_meta"] = 0;
  contents_insert ~= line;
} else if (type1["tic_code"] == 1) {
  type1["header_make"] = 0;
  type1["header_meta"] = 0;
  if (line.matchFirst(rgx.block_tic_close)) {
    type1["tic_code"] = 0;
  }
  contents_insert ~= line;
} else if (line.matchFirst(rgx.block_tic_code_open)) {
  type1["tic_code"] = 1;
  type1["header_make"] = 0;
  type1["header_meta"] = 0;
  contents_insert ~= line;
} else if (auto m = line.match(rgx.insert_src_fn_ssi_or_sst)) {
  type1["header_make"] = 0;
  type1["header_meta"] = 0;
  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.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 {
  type1["header_make"] = 0;
  type1["header_meta"] = 0;
  contents_insert ~= line; // images to extract for image list?
  if (_opt_action.source || _opt_action.pod) {
    auto _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
auto 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[string] type = flags_type_init;
auto fn_pth_full = fn_src.match(rgx.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 (type["curly_code"] == 1) {
  if (line.matchFirst(rgx.block_curly_code_close)) {
    type["curly_code"] = 0;
  }
  contents ~= line;
} else if (line.matchFirst(rgx.block_curly_code_open)) {
  type["curly_code"] = 1;
  contents ~= line;
} else if (type["tic_code"] == 1) {
  if (line.matchFirst(rgx.block_tic_close)) {
    type["tic_code"] = 0;
  }
  contents ~= line;
} else if (line.matchFirst(rgx.block_tic_code_open)) {
  type["tic_code"] = 1;
  contents ~= line;
} else if (auto m = line.match(rgx.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.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();
  auto contents_insert_tu = ins.scan_subdoc_source(
    _opt_action,
    markup_sourcefile_insert_content,
    fn_src_insert.to!string
  );
  contents ~= contents_insert_tu[0]; // images to extract for image list?
  if (_opt_action.source || _opt_action.pod) {
    auto _image_linelist = _extract_images(contents_insert_tu[0]);
    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) {
    auto _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);
}
auto t = tuple(
  contents,
  insert_file_list,
  images
);
return t;
#+END_SRC

* __END__