Improve the support for replacing nonexistent characters

decoder/mf_justify.c

@@ -117,7 +117,7 @@ static void render_left(const struct mf_font_s *font,
         if (c2 == '\t')
         {
             x = mf_round_to_tab(font, x0, x);
-            c1 = c2;
+            c1 = ' ';
             continue;
         }
 
@@ -155,7 +155,7 @@ static void render_right(const struct mf_font_s *font,
         if (c1 == '\t')
         {
             x = mf_round_to_prev_tab(font, x0, x);
-            c2 = c1;
+            c2 = ' ';
             continue;
         }
 

encoder/c_export.cc

@@ -2,6 +2,7 @@
 #include <vector>
 #include <iomanip>
 #include <map>
+#include <set>
 #include <algorithm>
 #include <string>
 #include <cctype>
@@ -222,6 +223,33 @@ static void encode_character_range(std::ostream &out, const DataFile &datafile,
     write_table(out, offsets, "uint16_t", "glyph_offsets_" + std::to_string(range_index), 4);
 }
 
+// Select the character to use as a fallback.
+static int select_fallback_char(const DataFile &datafile)
+{
+    std::set<int> chars;
+
+    size_t i = 0;
+    for (const DataFile::glyphentry_t &g: datafile.GetGlyphTable())
+    {
+        for (size_t c: g.chars)
+        {
+            chars.insert(c);
+        }
+        i++;
+    }
+
+    if (chars.count(0xFFFD))
+        return 0xFFFD; // Unicode replacement character
+
+    if (chars.count(0))
+        return 0; // Used by many BDF fonts as replacement char
+
+    if (chars.count('?'))
+        return '?';
+
+    return ' ';
+}
+
 void write_source(std::ostream &out, std::string name, const DataFile &datafile)
 {
     name = to_identifier(name);
@@ -268,7 +296,7 @@ void write_source(std::ostream &out, std::string name, const DataFile &datafile)
     out << "    " << datafile.GetFontInfo().baseline_x << ", /* baseline x */" << std::endl;
     out << "    " << datafile.GetFontInfo().baseline_y << ", /* baseline y */" << std::endl;
     out << "    " << datafile.GetFontInfo().line_height << ", /* line height */" << std::endl;
-    out << "    " << "0, /* fallback character */" << std::endl;
+    out << "    " << select_fallback_char(datafile) << ", /* fallback character */" << std::endl;
     out << "    " << "&mf_rlefont_character_width," << std::endl;
     out << "    " << "&mf_rlefont_render_character," << std::endl;
     out << "    }," << std::endl;

tests/example_text.txt

@@ -16,3 +16,4 @@ Very_long_words_that_just_keep_going_and_going_for_many_lines_at_a_time_causing_
 
 Non breaking spaces — they will prevent line wrapping just as efficiently and cause cutoffs at random points of the line.
 
+Unknown characters: ㎔☠