support combining characters in streams of text

Fake screens can't handle them yet.

103glyph.subx

@@ -130,9 +130,9 @@ wide-code-point?:  # c: code-point -> _/eax: boolean
     89/<- %ebp 4/r32/esp
     # eax = c
     8b/-> *(ebp+8) 0/r32/eax
-    # if (c >= 128) return  # characters beyond ASCII currently not supported
-    3d/compare-eax-and 0x80/imm32
-    0f 8d/jump-if->= $wide-code-point?:end/disp32
+    # if (c >= 4352) return false
+    3d/compare-eax-and 0x1100/imm32
+    0f 8d/jump-if->= $wide-code-point?:return-false/disp32
     # var letter-bitmap/eax = font[c]
     69/multiply %eax 0x22/imm32/glyph-size 0/r32/eax
     05/add-to-eax 0x0010000c/imm32/Font  # see boot.subx
@@ -147,15 +147,22 @@ $wide-code-point?:end:
     5d/pop-to-ebp
     c3/return
 
+$wide-code-point?:return-false:
+    b8/copy-to-eax 0/imm32/false
+    # . epilogue
+    89/<- %esp 5/r32/ebp
+    5d/pop-to-ebp
+    c3/return
+
 combining-code-point?:  # c: code-point -> _/eax: boolean
     # . prologue
     55/push-ebp
     89/<- %ebp 4/r32/esp
     # eax = c
     8b/-> *(ebp+8) 0/r32/eax
-    # if (c >= 128) return  # characters beyond ASCII currently not supported
-    3d/compare-eax-and 0x80/imm32
-    0f 8d/jump-if->= $combining-code-point?:end/disp32
+    # if (c >= 4352) return false
+    3d/compare-eax-and 0x1100/imm32
+    0f 8d/jump-if->= $combining-code-point?:return-false/disp32
     # var letter-bitmap/eax = font[c]
     69/multiply %eax 0x22/imm32/glyph-size 0/r32/eax
     05/add-to-eax 0x0010000c/imm32/Font  # see boot.subx
@@ -168,6 +175,13 @@ $combining-code-point?:end:
     5d/pop-to-ebp
     c3/return
 
+$combining-code-point?:return-false:
+    b8/copy-to-eax 0/imm32/false
+    # . epilogue
+    89/<- %esp 5/r32/ebp
+    5d/pop-to-ebp
+    c3/return
+
 # buffer: naked address to raw screen RAM without a length
 # letter-bitmap: naked address to 8-pixel wide font glyph
 draw-narrow-code-point-on-screen-buffer:  # buffer: (addr byte), letter-bitmap: (addr byte), x: int, y: int, color: int, background-color: int, overlay?: boolean, screen-width: int, screen-height: int

500fake-screen.mu

@@ -23,7 +23,7 @@ type screen {
 }
 
 type screen-cell {
-  data: code-point
+  data: code-point  # TODO: support combining characters overlaid on another character
   color: int
   background-color: int
   unused?: boolean
@@ -109,6 +109,27 @@ fn draw-code-point _screen: (addr screen), c: code-point, x: int, y: int, color:
   return 1
 }
 
+fn overlay-code-point _screen: (addr screen), c: code-point, x: int, y: int, color: int, background-color: int -> _/eax: int {
+  var screen/esi: (addr screen) <- copy _screen
+  {
+    compare screen, 0
+    break-if-!=
+    var result/eax: int <- overlay-code-point-on-real-screen c, x, y, color, background-color
+    return result
+  }
+  # fake screen
+  # TODO: support overlays in fake screen
+  var wide?/eax: boolean <- wide-code-point? c
+  compare wide?, 0/false
+  {
+    break-if-=
+    draw-wide-code-point-on-fake-screen screen, c, x, y, color, background-color
+    return 2
+  }
+  draw-narrow-code-point-on-fake-screen screen, c, x, y, color, background-color
+  return 1
+}
+
 fn draw-narrow-code-point-on-fake-screen _screen: (addr screen), c: code-point, x: int, y: int, color: int, background-color: int {
   var screen/esi: (addr screen) <- copy _screen
   # ignore if out of bounds

501draw-text.mu

@@ -207,10 +207,21 @@ fn draw-stream-wrapping-right-then-down screen: (addr screen), stream: (addr str
   var xcurr/ecx: int <- copy x
   var ycurr/edx: int <- copy y
   var c/ebx: code-point <- copy 0
+  var next-c/esi: code-point <- copy 0
   $draw-stream-wrapping-right-then-down:loop: {
-    var g/eax: grapheme <- read-grapheme stream
-    var _c/eax: code-point <- to-code-point g
-    c <- copy _c
+    # read c from either next-c or stream
+    $draw-stream-wrapping-right-then-down:read-base: {
+      compare next-c, 0
+      {
+        break-if-=
+        c <- copy next-c
+        next-c <- copy 0
+        break $draw-stream-wrapping-right-then-down:read-base
+      }
+      var g/eax: grapheme <- read-grapheme stream
+      var _c/eax: code-point <- to-code-point g
+      c <- copy _c
+    }
     compare c, 0xffffffff/end-of-file
     break-if-=
     compare c, 0xa/newline
@@ -223,6 +234,29 @@ fn draw-stream-wrapping-right-then-down screen: (addr screen), stream: (addr str
       break $draw-stream-wrapping-right-then-down:loop
     }
     var offset/eax: int <- draw-code-point screen, c, xcurr, ycurr, color, background-color
+    # overlay a combining character if necessary
+    $draw-stream-wrapping-right-then-down:read-combiner: {
+      var done?/eax: boolean <- stream-empty? stream
+      compare done?, 0/false
+      break-if-!=
+      # read a character
+      var g/eax: grapheme <- read-grapheme stream
+      var c/eax: code-point <- to-code-point g
+      # if not a combining character, save for next iteration and loop
+      {
+        var combining-code-point?/eax: boolean <- combining-code-point? c
+        compare combining-code-point?, 0/false
+      }
+      {
+        break-if-!=
+        next-c <- copy c
+        break $draw-stream-wrapping-right-then-down:read-combiner
+      }
+      # otherwise overlay it without saving its width
+      # This means strange results if a base and its combiner have different
+      # widths. We'll always follow the base width.
+      var dummy/eax: int <- overlay-code-point screen, c, xcurr, ycurr, color, background-color
+    }
     xcurr <- add offset
     compare xcurr, xmax
     {

apps/ex15.mu

@@ -77,4 +77,59 @@ fn main screen: (addr screen), keyboard: (addr keyboard), data-disk: (addr disk)
   # kaha
   var dummy/eax: int <- draw-code-point-on-real-screen 0x0915/devanagari-letter-ka, 0x13/x 9/y, 3/fg 0/bg
   var dummy/eax: int <- overlay-code-point-on-real-screen 0x0903/devanagari-visarga, 0x13/x 9/y, 3/fg 0/bg
+
+  # render the same letters as a single stream of utf-8 graphemes rather than individual code-points.
+  var text-storage: (stream byte 0x200)
+  var text/esi: (addr stream byte) <- address text-storage
+  var g/eax: grapheme <- to-grapheme 0x0915/devanagari-letter-ka
+  var ka/ecx: grapheme <- copy g
+  # ka
+  write-grapheme text, ka
+  # kaa
+  write-grapheme text, ka
+  g <- to-grapheme 0x093e/devanagari-vowel-aa
+  write-grapheme text, g
+  # ki
+  write-grapheme text, ka
+  g <- to-grapheme 0x093f/devanagari-vowel-i
+  write-grapheme text, g
+  # kee
+  write-grapheme text, ka
+  g <- to-grapheme 0x0940/devanagari-vowel-ii
+  write-grapheme text, g
+  # ku
+  write-grapheme text, ka
+  g <- to-grapheme 0x0941/devanagari-vowel-u
+  write-grapheme text, g
+  # koo
+  write-grapheme text, ka
+  g <- to-grapheme 0x0942/devanagari-vowel-oo
+  write-grapheme text, g
+  # kay
+  write-grapheme text, ka
+  g <- to-grapheme 0x0947/devanagari-vowel-E
+  write-grapheme text, g
+  # kai
+  write-grapheme text, ka
+  g <- to-grapheme 0x0948/devanagari-vowel-ai
+  write-grapheme text, g
+  # ko
+  write-grapheme text, ka
+  g <- to-grapheme 0x094b/devanagari-vowel-o
+  write-grapheme text, g
+  # kow
+  write-grapheme text, ka
+  g <- to-grapheme 0x094f/devanagari-vowel-aw
+  write-grapheme text, g
+  # kan
+  write-grapheme text, ka
+  g <- to-grapheme 0x0902/devanagari-anusvara
+  write-grapheme text, g
+  # kaha
+  write-grapheme text, ka
+  g <- to-grapheme 0x0903/devanagari-visarga
+  write-grapheme text, g
+  # render everything
+  set-cursor-position screen, 4/x 0xe/y
+  draw-stream-wrapping-right-then-down-from-cursor-over-full-screen screen, text, 3/fg 0/bg
 }