Improve detection of suspiciously reordered text segments

* src/bidi.c (bidi_find_first_overridden): Detect reordered weak
and neutral characters as well.

* lisp/international/mule-cmds.el
(highlight-confusing-reorderings): Count and announce how many
instances were found.

* test/src/xdisp-tests.el
(xdisp-tests--find-directional-overrides-case-3): New test.
(xdisp-tests--find-directional-overrides-case-2): Adjust expected
result.

1 files changed, 10 insertions, 2 deletions

diff --git a/src/bidi.c b/src/bidi.c
index 511b4602fec..890a60acc43 100644
--- a/src/bidi.c
+++ b/src/bidi.c
@@ -3575,6 +3575,8 @@ bidi_find_first_overridden (struct bidi_it *bidi_it)
      overrides, and isolates, i.e. before resolving implicit levels.  */
   int max_l2r = bidi_it->paragraph_dir == L2R ? 0 : 2;
   int max_r2l = 1;
+  /* Same for WEAK and NEUTRAL_ON types.  */
+  int max_weak = bidi_it->paragraph_dir == L2R ? 1 : 2;
 
   do
     {
@@ -3582,6 +3584,8 @@ bidi_find_first_overridden (struct bidi_it *bidi_it)
          because the directional overrides are applied by the
          former.  */
       bidi_type_t type = bidi_resolve_weak (bidi_it);
+      unsigned level = bidi_it->level_stack[bidi_it->stack_idx].level;
+      bidi_category_t category = bidi_get_category (bidi_it->orig_type);
 
       /* Detect strong L or R types that have been overridden by
          explicit overrides.  */
@@ -3594,10 +3598,14 @@ bidi_find_first_overridden (struct bidi_it *bidi_it)
              reorder) by explicit embeddings and isolates.  */
           || ((bidi_it->orig_type == STRONG_L
                || bidi_it->orig_type == WEAK_EN)
-              && bidi_it->level_stack[bidi_it->stack_idx].level > max_l2r)
+              && level > max_l2r)
           || ((bidi_it->orig_type == STRONG_R
                || bidi_it->orig_type == STRONG_AL)
-              && bidi_it->level_stack[bidi_it->stack_idx].level > max_r2l))
+              && level > max_r2l)
+          /* Detect other weak or neutral types whose level was
+             tweaked by explicit embeddings and isolates.  */
+          || ((category == WEAK || bidi_it->orig_type == NEUTRAL_ON)
+              && level > max_weak))
         found_pos = bidi_it->charpos;
     } while (found_pos == ZV
              && bidi_it->charpos < ZV