stuff

TODO.md

@@ -1,5 +1,6 @@
 # Parity with Vanilla OC (only the stuff that makes sense for an emulator)
 
+- make the `unicode` library in testLuaArch support invalid UTF-8 (WHY IS IT OK WITH THAT)
 - in-memory version of `filesystem`
 - complete the GPU implementation (screen buffers and missing methods)
 - complete the screen implementation (bunch of missing methods)

src/components/gpu.c

@@ -131,22 +131,23 @@ void nni_gpu_set(nni_gpu *gpu, void *_, nn_component *component, nn_computer *co
         return;
     }
 
-    if(!nn_unicode_validate(s)) {
-        nn_setCError(computer, "invalid utf-8");
-        return;
-    }
-
     int current = 0;
     int len = 0;
     while(s[current] != 0) {
+        if(nn_unicode_isValidCodepoint(s + current)) {
             int codepoint = nn_unicode_codepointAt(s, current);
             nn_setPixel(gpu->currentScreen, x, y, nni_gpu_makePixel(gpu, s + current));
+            current += nn_unicode_codepointSize(codepoint);
+        } else {
+            unsigned int codepoint = (unsigned char)s[current];
+            nn_setPixel(gpu->currentScreen, x, y, nni_gpu_makePixel(gpu, s + current));
+            current++;
+        }
         if(isVertical) {
             y++;
         } else {
             x++;
         }
-        current += nn_unicode_codepointSize(codepoint);
         len++;
     }
 

src/neonucleus.h

@@ -300,6 +300,8 @@ nn_bool_t nn_decRef(nn_refc *refc);
 // Unicode (more specifically, UTF-8) stuff
 
 nn_bool_t nn_unicode_validate(const char *s);
+// expects NULL terminator
+nn_bool_t nn_unicode_isValidCodepoint(const char *s);
 // returned string must be nn_deallocStr()'d
 char *nn_unicode_char(nn_Alloc *alloc, unsigned int *codepoints, nn_size_t codepointCount);
 // returned array must be nn_dealloc()'d

src/sandbox.lua

@@ -17,6 +17,14 @@ local function copy(v, p)
     end
 end
 
+local function spcall(f, ...)
+    local t = {pcall(f, ...)}
+    if t[1] then
+        return table.unpack(t, 2)
+    end
+    return nil, t[2]
+end
+
 local function nextDeadline()
     return computer.uptime() + 5
 end
@@ -392,40 +400,9 @@ sandbox = {
 
     utf8 = copy(utf8),
     unicode = copy(unicode, {
-        wtrunc = function (str,space)
-			space = space - 1
-			return str:sub(1,(space >= utf8.len(str)) and (#str) or (utf8.offset(str,space+1)-1))
-		end,
         isWide = function(s) return unicode.wlen(s) > unicode.len(s) end,
         upper = string.upper,
         lower = string.lower,
-        --[[
-        sub = function (str,a,b)
-			if not b then b = utf8.len(str) end
-			if not a then a = 1 end
-			-- a = math.max(a,1)
-			
-			if a < 0 then
-				-- negative
-				
-				a = utf8.len(str) + a + 1
-			end
-			
-			if b < 0 then
-				b = utf8.len(str) + b + 1
-			end
-			
-			if a > b then return "" end
-			
-			if b >= utf8.len(str) then b = #str else b = utf8.offset(str,b+1)-1 end
-			
-			if a > utf8.len(str) then return "" end
-			a = utf8.offset(str,a)
-			
-			return str:sub(a,b)
-			-- return str:sub(a, b)
-		end,
-        ]]
     }),
     checkArg = checkArg,
     component = libcomponent,

src/unicode.c

@@ -162,6 +162,36 @@ static nn_bool_t nn_unicode_is_continuation(unsigned char byte) {
     return (byte >> 6) == 0b10;
 }
 
+nn_bool_t nn_unicode_isValidCodepoint(const char *s) {
+    if(s[0] <= 0x7F) {
+        return true;
+    } else if((s[0] >> 5) == 0b110) {
+        if (!nn_unicode_is_continuation(s[1])) {
+            return false;
+        }
+    } else if((s[0] >> 4) == 0b1110) {
+        if (!nn_unicode_is_continuation(s[1])) {
+            return false;
+        }
+        if (!nn_unicode_is_continuation(s[2])) {
+            return false;
+        }
+    } else if((s[0] >> 3) == 0b11110) {
+        if (!nn_unicode_is_continuation(s[1])) {
+            return false;
+        }
+        if (!nn_unicode_is_continuation(s[2])) {
+            return false;
+        }
+        if (!nn_unicode_is_continuation(s[3])) {
+            return false;
+        }
+    } else {
+        return false;
+    }
+    return true;
+}
+
 nn_bool_t nn_unicode_validate(const char *b) {
     const unsigned char* s = (const unsigned char*)b;
     while (*s) {