is_unicodepoint

aheui · Apr 5, 2024 · 7158c37 · 7158c37
1 parent ea40031
commit 7158c37
Show file tree

Hide file tree

Showing 3 changed files with 31 additions and 10 deletions.
diff --git a/aheui/aheui.py b/aheui/aheui.py
@@ -283,18 +283,26 @@ def read_number(input_buffer=input_buffer):
     return num
 
 
-@jit.dont_look_inside
-def write_number(value):
-    os.write(outfp, value)
+def write_number(value_str):
+    os.write(outfp, value_str)
 
 
-@jit.dont_look_inside
 def write_utf8(warnings, value):
-    if not (0 <= value < 0x110000):
-        warnings.warn(b'write-utf8-range', value)
-        value = 0xfffd
-    os.write(outfp, unichr(value).encode('utf-8'))
+    REPLACE_CHAR = unichr(0xfffd).encode('utf-8')
+
+    if bigint.is_unicodepoint(value):
+        codepoint = bigint.toint(value)
+        unicode_char = unichr(codepoint)
+        bytes = unicode_char.encode('utf-8')
+    else:
+        bytes = REPLACE_CHAR
 
+    os.write(outfp, bytes)
+
+
+def warn_utf8_range(warnings, value):
+    warnings.warn(b'write-utf8-range', value)
+    os.write(outfp, unichr(0xfffd).encode('utf-8'))
 
 class Program(object):
     _immutable_fields_ = ['labels[**]', 'opcodes[*]', 'values[*]', 'size']
@@ -417,7 +425,7 @@ def mainloop(program, debug):
             r = selected.pop()
             write_number(bigint.str(r))
         elif op == c.OP_POPCHAR:
-            r = selected.pop_longlong()
+            r = selected.pop()
             write_utf8(warnings, r)
         elif op == c.OP_PUSHNUM:
             num = read_number()

diff --git a/aheui/int/bigint.py b/aheui/int/bigint.py
@@ -67,3 +67,8 @@ def ge(r1, r2):
 def is_zero(r):
     # return r.sign == 0
     return r._size == 0  # pypy 7.3.15
+
+
+@jit.elidable
+def is_unicodepoint(r):
+    return 0 <= r._size and r.int_le(0x110000)
diff --git a/aheui/int/smallint.py b/aheui/int/smallint.py
@@ -2,7 +2,7 @@
 
 try:
     import builtins
-except ImportError:  # python2
+except ImportError:
     builtins = __builtins__
 from aheui._compat import _bytestr
 
@@ -37,6 +37,10 @@ def str(r):
     return _bytestr(r)
 
 
+def hex(r):
+    return hex(r)
+
+
 def add(r1, r2):
     return r1 + r2
 
@@ -63,3 +67,7 @@ def ge(r1, r2):
 
 def is_zero(r):
     return r == 0
+
+
+def is_unicodepoint(r):
+    return 0 < r <= 0x110000