uniblocks: avoid some problematic glyphs #1285

TERMS.md

@@ -135,3 +135,15 @@ implementing `rgb` use the 3x8bpc model; XTerm for instance:
 
 Thus emitting `setaf` with an RGB value close to black can result, when
 using `xterm-direct`'s `setaf` and `rgb` definitions, in a bright ANSI color.
+
+## Problematic characters
+
+Some characters seem to cause problems with one terminal or another. These
+are best avoided until the problems are better understood:
+
+* '­' U+00AD SOFT HYPHEN (some terminals allocate it a cell, some don't)
+* '܏' U+070F SYRIAC ABBREVIATION MARK
+* '۝' U+06DD ARABIC END OF AYAH
+* '࣢' U+08E2 ARABIC DISPUTED END OF AYAH
+* '﷽' U+FDFD ARABIC LIGATURE BISMILLAH AR-RAHMAN AR-RAHEEM
+    '

src/demo/unicodeblocks.c

@@ -103,7 +103,10 @@ draw_block(struct ncplane* nn, uint32_t blockstart){
     for(z = 0 ; z < CHUNKSIZE ; ++z){
       wchar_t w = blockstart + chunk * CHUNKSIZE + z;
       char utf8arr[MB_CUR_MAX * 3 + 5];
-      if(wcwidth(w) >= 1 && iswgraph(w)){
+      // problematic characters FIXME (see TERMS.md)
+      if(w == 0x070f || w == 0x08e2 || w == 0x06dd){
+        strcpy(utf8arr, "  ");
+      }else if(wcwidth(w) >= 1 && iswgraph(w)){
         mbstate_t ps;
         memset(&ps, 0, sizeof(ps));
         int bwc = wcrtomb(utf8arr, w, &ps);