step6 headers (&config), native & sdlang to json internally

1 files changed, 320 insertions, 0 deletions

diff --git a/src/sdp/ao_header_extract_native.d b/src/sdp/ao_header_extract_native.d
new file mode 100644
index 0000000..49555ef
--- /dev/null
+++ b/src/sdp/ao_header_extract_native.d
@@ -0,0 +1,320 @@
+/+
+  extract native/orig header return json
++/
+template SiSUheaderExtractNative() {
+  private import
+    std.exception,
+    std.regex,
+    std.utf,
+    std.conv : to;
+  private import
+    ao_rgx;
+  struct HeaderDocMetadataAndMakeNativeToJson {
+    mixin SiSUheaderRegister;
+    mixin SiSUrgxInitFlags;
+    mixin RgxInit;
+    auto rgx = Rgx();
+    enum State { off, on }
+    string hm, hs;
+    auto header_metadata_and_make_jsonstr(
+      string header,
+      JSONValue[string] dochead_meta,
+      JSONValue[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"].str =
+                    to!string(m.captures[2]);
+                } else if (hm == "title") {
+                  dochead_meta[hm]["main"].str =
+                    to!string(m.captures[2]);
+                } else if (hm == "publisher") {
+                  dochead_meta[hm]["name"].str =
+                    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].type() == JSON_TYPE.OBJECT)) {
+                  switch (hm) {
+                  case "make":
+                    if (match(hs, rgx.native_subhead_make)) {
+                      if (dochead_make[hm][hs].type() == JSON_TYPE.STRING) {
+                        dochead_make[hm][hs].str = 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].type() == JSON_TYPE.OBJECT) {
+                  switch (hm) {
+                  case "creator":
+                    if (match(hs, rgx.native_subhead_creator)) {
+                      if (dochead_meta[hm][hs].type() == JSON_TYPE.STRING) {
+                        dochead_meta[hm][hs].str =
+                          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"].type() == JSON_TYPE.STRING)) {
+                        dochead_meta[hm]["sub"].str =
+                          to!string(s.captures[2]);
+                      } else if (dochead_meta[hm][hs].type() == JSON_TYPE.STRING) {
+                        dochead_meta[hm][hs].str =
+                          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].type() == JSON_TYPE.STRING) {
+                        dochead_meta[hm][hs].str =
+                          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].type() == JSON_TYPE.STRING) {
+                        dochead_meta[hm][hs].str =
+                          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].type() == JSON_TYPE.STRING) {
+                        dochead_meta[hm][hs].str =
+                          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].type() == JSON_TYPE.STRING) {
+                        dochead_meta[hm][hs].str =
+                          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].type() == JSON_TYPE.STRING) {
+                        dochead_meta[hm][hs].str =
+                          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].type() == JSON_TYPE.STRING) {
+                        dochead_meta[hm][hs].str =
+                          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].type() == JSON_TYPE.STRING) {
+                        dochead_meta[hm][hs].str =
+                          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].type() == JSON_TYPE.STRING) {
+                    //     dochead_meta[hm][hs].str = 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;
+    }
+    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;
+    }
+    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");
+    }
+    private auto headerNativeToJSON(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 = parseJSON(header_make_jsonstr).object;
+      auto dochead_meta = parseJSON(header_meta_jsonstr).object;
+      auto set_header = HeaderDocMetadataAndMakeNativeToJson();
+      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_jsonstr(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;
+    }
+  }
+}