path: root/src/sdp/ao_read_source_files.d

/++
  module ao_read_source_files;<BR>
  - open markup files<BR>
  - if master file scan for addional files to import/insert
+/
template SiSUmarkupRaw() {
  private import
    std.exception,
    std.regex,
    std.stdio,
    std.utf,
    std.conv : to;
  private import
    ao_rgx;       // ao_defaults.d
  mixin RgxInit;
  auto rgx = Rgx();
  struct MarkupRaw {
    final sourceContent(in string fn_src) {
      auto raw = MarkupRawUnit();
      auto source_txt_str =
        raw.markupSourceReadIn(fn_src);
      return source_txt_str;
    }
    final auto sourceContentSplitIntoHeaderAndBody(in string source_txt_str, in string fn_src="") {
      auto raw = MarkupRawUnit();
      string[] insert_file_list;
      auto t =
        raw.markupSourceHeaderContentRawLineTupleArray(source_txt_str);
      auto header_raw = t[0];
      auto sourcefile_body_content = t[1];
      if (match(fn_src, 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(sourcefile_body_content, fn_src);
        static assert(!isTypeTuple!(tu));
        sourcefile_body_content = tu[0];
        insert_file_list = tu[1].dup;
      }
      t = tuple(
        header_raw,
        sourcefile_body_content,
        insert_file_list
      );
      static assert(t.length==3);
      return t;
    }
  }
  private
  struct MarkupRawUnit {
    private import std.file;
    final private string readInMarkupSource(in char[] fn_src) {
      enforce(
        exists(fn_src)!=0,
        "file not found"
      );
      string source_txt_str;
      try {
        if (exists(fn_src)) {
          source_txt_str = readText(fn_src);
        }
      }
      catch (ErrnoException ex) {
      }
      catch (UTFException ex) {
        // Handle validation errors
      }
      catch (FileException ex) {
        // Handle errors
      }
      std.utf.validate(source_txt_str);
      return source_txt_str;
    }
    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 = matchFirst(cast(char[]) src_text, 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 == " ~ to!string(header_and_content.length) ~
        "; (header / body array split should == 2 (split is on level A~))"
      );
      return header_and_content;
    }
    final private char[][] markupSourceLineArray(in char[] src_text) {
      char[][] source_line_arr =
        split(cast(char[]) src_text, rgx.newline_eol_strip_preceeding);
      return source_line_arr;
    }
    auto markupSourceReadIn(in string fn_src) {
      auto rgx = Rgx();
      enforce(
        match(fn_src, rgx.src_pth),
        "not a sisu markup filename"
      );
      auto source_txt_str = readInMarkupSource(fn_src);
      return source_txt_str;
    }
    auto markupSourceHeaderContentRawLineTupleArray(in string source_txt_str) {
      string[] file_insert_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
      );
      return t;
    }
    final char[][] getInsertMarkupSourceContentRawLineArray(
      in char[] fn_src,
      Regex!(char) rgx_file
    ) {
      enforce(
        match(fn_src, rgx_file),
        "not a sisu markup filename"
      );
      auto source_txt_str = readInMarkupSource(fn_src);
      auto source_line_arr = markupSourceLineArray(source_txt_str);
      return source_line_arr;
    }
  }
  struct Inserts {
    private import ao_defaults;    // ao_defaults.d
    auto scan_subdoc_source(
      char[][] markup_sourcefile_insert_content,
      string fn_src
    ) {
      mixin SiSUrgxInitFlags;
      char[][] contents_insert;
      auto type1 = flags_type_init;
      auto fn_pth_full = match(fn_src, rgx.src_pth);
      auto markup_src_file_path = fn_pth_full.captures[1];
      foreach (line; markup_sourcefile_insert_content) {
        if (type1["curly_code"] == 1) {
          type1["header_make"] = 0;
          type1["header_meta"] = 0;
          if (matchFirst(line, rgx.block_curly_code_close)) {
            type1["curly_code"] = 0;
          }
          contents_insert ~= line;
        } else if (matchFirst(line, 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 (matchFirst(line, rgx.block_tic_close)) {
            type1["tic_code"] = 0;
          }
          contents_insert ~= line;
        } else if (matchFirst(line, rgx.block_tic_code_open)) {
          type1["tic_code"] = 1;
          type1["header_make"] = 0;
          type1["header_meta"] = 0;
          contents_insert ~= line;
        } else if (
          (type1["header_make"] == 1)
          && matchFirst(line, rgx.native_header_sub)
        ) {
            type1["header_make"] = 1;
            type1["header_meta"] = 0;
        } else if (
          (type1["header_meta"] == 1)
          && matchFirst(line, rgx.native_header_sub)
        ) {
            type1["header_meta"] = 1;
            type1["header_make"] = 0;
        } else if (auto m = match(line, 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) {                              // insert file
            tell_l("red", line);
            tell_l("red", fn_src_insert);
            tell_l("fuchsia", "ERROR");
            writeln(
              "  length contents insert array: ",
              markup_sourcesubfile_insert_content.length
            );
          }
          auto ins = Inserts();
          /+
            1. load file,
            2. read lines;
            3. scan lines,
            4. if filename insert, and insert filename
            5.   repeat 1
            6. else
            7.   add line to new array;
          +/
        } else {
          type1["header_make"] = 0;
          type1["header_meta"] = 0;
          contents_insert ~= line;
        }
      } // end src subdoc (inserts) loop
      return contents_insert;
    }
    auto scan_master_src_for_insert_files_and_import_content(
      char[][] sourcefile_body_content,
      string fn_src
    ) {
      mixin SiSUrgxInitFlags;
      char[][] contents;
      auto type = flags_type_init;
      auto fn_pth_full = match(fn_src, rgx.src_pth);
      auto markup_src_file_path = fn_pth_full.captures[1];
      string[] insert_file_list =[];
      foreach (line; sourcefile_body_content) {
        if (type["curly_code"] == 1) {
          if (matchFirst(line, rgx.block_curly_code_close)) {
            type["curly_code"] = 0;
          }
          contents ~= line;
        } else if (matchFirst(line, rgx.block_curly_code_open)) {
          type["curly_code"] = 1;
          contents ~= line;
        } else if (type["tic_code"] == 1) {
          if (matchFirst(line, rgx.block_tic_close)) {
            type["tic_code"] = 0;
          }
          contents ~= line;
        } else if (matchFirst(line, rgx.block_tic_code_open)) {
          type["tic_code"] = 1;
          contents ~= line;
        } else if (auto m = match(line, 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 ~= to!string(fn_src_insert);
          auto raw = MarkupRawUnit();
          /+ TODO +/
          if (auto ma = match(line, rgx.src_fn_text)) {
            /+ .sst when inserted, not used: headers and heading level ^:?A~ so remove +/
            writeln(__LINE__); writeln(ma);
          }
          auto markup_sourcefile_insert_content =
            raw.getInsertMarkupSourceContentRawLineArray(fn_src_insert, rgx.src_fn_find_inserts);
          debug(insert) {                              // insert file
            tell_l("red", line);
            tell_l("red", fn_src_insert);
            writeln(
              "  length contents insert array: ",
              markup_sourcefile_insert_content.length
            );
          }
          auto ins = Inserts();
          auto contents_insert = ins.scan_subdoc_source(
            markup_sourcefile_insert_content,
            to!string(fn_src_insert)
          );
          contents ~= contents_insert;
          /+
            1. load file,
            2. read lines;
            3. scan lines,
            4. if filename insert, and insert filename
            5.   repeat 1
            6. else
            7.   add line to new array;
          +/
        } else {
          contents ~= line;
        }
      } // end src doc loop
      debug(insert) {                              // insert file
        writeln(__LINE__);
        writeln(contents.length);
      }
      auto t = tuple(
        contents,
        insert_file_list
      );
      return t;
    }
  }
}