ocda: clean heading text used for navigation

heading text used for navigation is normalised, and | escaped

(assisted by Claude-Code)

1 files changed, 34 insertions, 9 deletions

diff --git a/src/sisudoc/ocda/meta/metadoc_from_src_functions.d b/src/sisudoc/ocda/meta/metadoc_from_src_functions.d
index 9e9f8ba..f51f6fe 100644
--- a/src/sisudoc/ocda/meta/metadoc_from_src_functions.d
+++ b/src/sisudoc/ocda/meta/metadoc_from_src_functions.d
@@ -144,41 +144,65 @@ template docAbstractionFunctions() {
     invariant() {
     }
   }
-  pure ObjGenericComposite obj_heading_ancestors()(
+  /+ ↓ a heading's text as it is used for navigation, in
+       tags.heading_ancestors_text and in the lev4 subtoc entries.
+       .
+       these are one line labels, so what is dropped is what cannot be shown
+       on one line and means nothing there: the note itself (the reference
+       marker goes with it), the line break markers, and a line break that
+       came from a heading wrapped over two lines in the source. what is
+       kept is the text and its inline markup.
+  +/
+  string heading_text_for_navigation()(string _txt) {
+    mixin InternalMarkup;
+    static auto mkup = InlineMarkup();
+    static auto rgx = RgxI();
+    string _out = _txt
+      .replaceAll(rgx.inline_notes_al, "")            // 【...】 and its marker
+      .replaceAll(rgx.inline_notes_curly_gen, "")     // ~{...}~ if any remains
+      .replace(mkup.br_line,        " ")
+      .replace(mkup.br_line_inline, " ")
+      .replace(mkup.br_line_spaced, " ")
+      .replace("\n", " ")
+      .replace("\r", " ");
+    while (_out.canFind("  ")) { _out = _out.replace("  ", " "); }
+    return _out.strip;
+  }
+  ObjGenericComposite obj_heading_ancestors()(
     ObjGenericComposite  obj,
     string[]             lv_ancestors_txt,
   ) {
     switch (obj.metainfo.heading_lev_markup) {
     case 0:
-      lv_ancestors_txt[0] = obj.text.to!string;
+      lv_ancestors_txt[0] = heading_text_for_navigation(obj.text.to!string);
       foreach(k; 1..8) { lv_ancestors_txt[k] = ""; }
       goto default;
     case 1:
-      lv_ancestors_txt[1] = obj.text.to!string;
+      lv_ancestors_txt[1] = heading_text_for_navigation(obj.text.to!string);
       foreach(k; 2..8) { lv_ancestors_txt[k] = ""; }
       goto default;
     case 2:
-      lv_ancestors_txt[2] = obj.text.to!string;
+      lv_ancestors_txt[2] = heading_text_for_navigation(obj.text.to!string);
       foreach(k; 3..8) { lv_ancestors_txt[k] = ""; }
       goto default;
     case 3:
-      lv_ancestors_txt[3] = obj.text.to!string;
+      lv_ancestors_txt[3] = heading_text_for_navigation(obj.text.to!string);
       foreach(k; 4..8) { lv_ancestors_txt[k] = ""; }
       goto default;
     case 4:
-      lv_ancestors_txt[4] = obj.text.to!string;
+      lv_ancestors_txt[4] = heading_text_for_navigation(obj.text.to!string);
       foreach(k; 5..8) { lv_ancestors_txt[k] = ""; }
       goto default;
     case 5:
-      lv_ancestors_txt[5] = obj.text.to!string;
+      lv_ancestors_txt[5] = heading_text_for_navigation(obj.text.to!string);
       foreach(k; 6..8) { lv_ancestors_txt[k] = ""; }
       goto default;
     case 6:
-      lv_ancestors_txt[6] = obj.text.to!string;
+      lv_ancestors_txt[6] = heading_text_for_navigation(obj.text.to!string);
       lv_ancestors_txt[7] = "";
       goto default;
     case 7:
-      lv_ancestors_txt[7] = obj.text.to!string;
+      lv_ancestors_txt[7] = heading_text_for_navigation(obj.text.to!string);
       goto default;
     default:
       obj.tags.heading_ancestors_text = lv_ancestors_txt.dup;
@@ -2671,6 +2695,7 @@ template docAbstractionFunctions() {
       char[] heading_toc_                             = (obj_["substantive"].dup.strip.to!(char[]))
         .replaceAll(rgx.inline_notes_al, "");
       heading_toc_                                    = _clean_heading_toc_(heading_toc_);
+      heading_toc_ = heading_text_for_navigation(heading_toc_.to!string).to!(char[]);
       auto attrib = "";
       string toc_txt_, subtoc_txt_;
       int[string] indent;