Add support for DEC graphic set from NetHack.

NetHack allows you to use the DEC graphics set to render the dungeon
(https://nethackwiki.com/wiki/DECgraphics). Since DEC graphics mode is
toggled on and off with ESC ( 0, ESC ( B respectively, we add a new
'state' to tmt to indicate when ESC ( have been pressed. Normally
anything following this sequence is ignored, anyway.

Each character then saves whether it was written in DEC mode or not, in
attributes. While this would not work for a general DEC parser (since
some DEC characters include whitespace/new lines etc that would not have
been interpretted correctly by the terminal emulator - cf
https://en.wikipedia.org/wiki/DEC_Special_Graphics) in this case this
is fine because NetHack only uses DEC graphics for rendering particular
glyphs (as suggested in the NetHack Wiki, and as suggested in paxed/dgamelaunch
scriptfor stripping gfx).

A test is added to check rendering.

nle/tests/2020-10-16.00_11_28.frame.5.txt

@@ -0,0 +1,24 @@
+                                       It is written in the Book of Amaterasu Omikami:
+
+                                           After the Creation, the cruel god Moloch rebelled
+                                           against the authority of Marduk the Creator.
+                   --------------          Moloch stole from Marduk the most powerful of all
+                   .............|          the artifacts of the gods, the Amulet of Yendor,
+                   |d..........{|          and he hid it in the dark cavities of Gehennom, the
+                   |...........d|          Under World, where he now lurks, and bides his time.
+                   |..........@..
+                   ---------+----      Your goddess Amaterasu Omikami seeks to possess the Amulet, and with it
+                                       to gain deserved ascendance over the other gods.
+
+                                       You, a newly trained Hatamoto, have been heralded
+                                       from birth as the instrument of Amaterasu Omikami.  You are destined
+                                       to recover the Amulet for your deity, or die in the
+                                       attempt.  Your hour of destiny has come.  For the sake
+                                       of us all:  Go bravely with Amaterasu Omikami!
+                                       --More--
+
+
+
+
+[CodeMagic the Hatamoto        ] St:16 Dx:14 Co:17 In:11 Wi:9 Ch:7 Lawful
+Dlvl:1 $:0 HP:15(15) Pw:2(2) AC:4 Xp:1 T:1

nle/tests/test_converter.py

@@ -30,6 +30,7 @@
 #  https://alt.org/nethack/trd/?file=https://s3.amazonaws.com/altorg/ttyrec/CodeMagic/2020-10-16.00:11:28.ttyrec.bz2  # noqa: B950
 # This ttyrec uses DECGraphics (https://en.wikipedia.org/wiki/DEC_Special_Graphics)
 TTYREC_DECGRAPHICS = "2020-10-16.00_11_28.ttyrec.bz2"
+TTYREC_DECGRAPHICS_FRAME_5 = "2020-10-16.00_11_28.frame.5.txt"
 
 SEQ_LENGTH = 20
 ROWS = 25
@@ -237,7 +238,7 @@ def test_illegal_buffers(self):
         with pytest.raises(ValueError, match=r"Numpy array required"):
             converter.convert(chars, colors, cursors, timestamps, actions)
 
-    def _test_ibm_graphics(self):
+    def test_ibm_graphics(self):
         seq_length = 10
         converter = Converter(ROWS, COLUMNS)
 
@@ -254,3 +255,23 @@ def _test_ibm_graphics(self):
             for row, line in enumerate(f):
                 actual = chars[-1][row].tobytes().decode("utf-8").rstrip()
                 assert actual == line.rstrip()
+
+    def test_dec_graphics(self):
+        seq_length = 10
+        COLUMNS = 120
+        converter = Converter(ROWS, COLUMNS)
+
+        chars = np.zeros((seq_length, ROWS, COLUMNS), dtype=np.uint8)
+        colors = np.zeros((seq_length, ROWS, COLUMNS), dtype=np.int8)
+        cursors = np.zeros((seq_length, 2), dtype=np.uint16)
+        actions = np.zeros((seq_length), dtype=np.uint8)
+        timestamps = np.zeros((seq_length,), dtype=np.int64)
+
+        converter.load_ttyrec(getfilename(TTYREC_DECGRAPHICS))
+        assert converter.convert(chars, colors, cursors, timestamps, actions) == 0
+
+        with open(getfilename(TTYREC_DECGRAPHICS_FRAME_5)) as f:
+            for row, line in enumerate(f):
+                actual = chars[5][row].tobytes().decode("utf-8").rstrip()
+                # print(actual)
+                assert actual == line.rstrip()

third_party/converter/converter.c

@@ -285,7 +285,7 @@ void write_to_buffers(Conversion *conv) {
     for (size_t c = 0; c < conv->cols; ++c) {
       assert(conv->chars.cur < conv->chars.end);
       assert(scr->lines[r]->chars[c].c < 256);
-      *conv->chars.cur++ = strip_gfx(scr->lines[r]->chars[c].c);
+      *conv->chars.cur++ = strip_gfx(scr->lines[r]->chars[c].c, scr->lines[r]->chars[c].a.dec);
       *conv->colors.cur++ = vt_char_color_extract(&(scr->lines[r]->chars[c]));
     }
   }

third_party/converter/stripgfx.c

@@ -11,7 +11,6 @@
 
 unsigned char gfx_dec_map[256];
 unsigned char gfx_ibm_map[256];
-unsigned int state = 0;
 
 /* clang-format off */
 static unsigned char no_graphics[MAXPCHARS] = {
@@ -348,7 +347,6 @@ static unsigned char dec_graphics[MAXPCHARS] = {
 void populate_gfx_arrays() {
   int i;
 
-  state = 0;
 
   memset(gfx_ibm_map, 0, 256);
   memset(gfx_dec_map, 0, 256);
@@ -373,33 +371,11 @@ void populate_gfx_arrays() {
 }
 
 unsigned char
-strip_gfx(unsigned char inchar)
+strip_gfx(unsigned char inchar, int use_dec)
 {
+  if (use_dec && gfx_dec_map[inchar])
+    return gfx_dec_map[inchar];
   if (gfx_ibm_map[inchar])
     return gfx_ibm_map[inchar];
-
-  if ((inchar == 0x0E) && (state == 0))
-    {
-      state = 1;
-      return 0x00;
-    }
-
-  if ((inchar == 0x0F) && (state == 1))
-    {
-      state = 0;
-      return inchar;
-    }
-
-  if ((inchar == 0x1B) && (state == 1))
-    {
-      state = 0;
-      return inchar;
-    }
-
-  if (gfx_dec_map[inchar] && (state == 1))
-    {
-      return gfx_dec_map[inchar];
-    }
-
   return inchar;
 }

third_party/converter/stripgfx.h

@@ -2,6 +2,6 @@
 #define INCLUDED_stripgfx_h
 
 void populate_gfx_arrays();
-unsigned char strip_gfx(unsigned char inchar);
+unsigned char strip_gfx(unsigned char inchar, int use_dec);
 
 #endif /* !INCLUDED_stripgfx_h */

third_party/libtmt/tmt.c

@@ -70,10 +70,10 @@ struct TMT{
     size_t pars[PAR_MAX];
     size_t npar;
     size_t arg;
-    enum {S_NUL, S_ESC, S_ARG} state;
+    enum {S_NUL, S_ESC, S_ARG, S_DEC} state;
 };
 
-static TMTATTRS defattrs = {.fg = TMT_COLOR_DEFAULT, .bg = TMT_COLOR_DEFAULT};
+static TMTATTRS defattrs = {.fg = TMT_COLOR_DEFAULT, .bg = TMT_COLOR_DEFAULT, .dec = 0};
 static void writecharatcurs(TMT *vt, wchar_t w);
 
 static wchar_t
@@ -273,7 +273,8 @@ handlechar(TMT *vt, char i)
     DO(S_ESC, "H",          t[c->c].c = L'*')
     DO(S_ESC, "7",          vt->oldcurs = vt->curs; vt->oldattrs = vt->attrs)
     DO(S_ESC, "8",          vt->curs = vt->oldcurs; vt->attrs = vt->oldattrs)
-    ON(S_ESC, "+*()",       vt->ignored = true; vt->state = S_ARG)
+    ON(S_ESC, "(",          vt->state = S_DEC)
+    ON(S_ESC, "+*)",        vt->ignored = true; vt->state = S_ARG)
     DO(S_ESC, "c",          tmt_reset(vt))
     ON(S_ESC, "[",          vt->state = S_ARG)
     ON(S_ARG, "\x1b",       vt->state = S_ESC)
@@ -310,6 +311,9 @@ handlechar(TMT *vt, char i)
     DO(S_ARG, "s",          vt->oldcurs = vt->curs; vt->oldattrs = vt->attrs)
     DO(S_ARG, "u",          vt->curs = vt->oldcurs; vt->attrs = vt->oldattrs)
     DO(S_ARG, "@",          ich(vt))
+    ON(S_DEC, "\x1b",       vt->state = S_ESC)
+    DO(S_DEC, "0",          vt->attrs.dec = true)
+    DO(S_DEC, "B",          vt->attrs.dec = false)
 
     return resetparser(vt), false;
 }

third_party/libtmt/tmt.h

@@ -87,6 +87,7 @@ struct TMTATTRS{
     bool invisible;
     tmt_color_t fg;
     tmt_color_t bg;
+    bool dec;
 };
 
 typedef struct TMTCHAR TMTCHAR;