allow more flexible constant specifications

fx92/lexer.py

@@ -90,6 +90,22 @@ class Token:
 
         return base + args
 
+#---
+# Utilities
+#---
+
+def str2float(integer, decimal, exponent, percent):
+    m1 = integer or "0"
+    m2 = ".0" if decimal == "." else (decimal or "")
+    m3 = exponent or ""
+
+    f = float(m1 + m2 + m3)
+
+    if percent == "%":
+        f /= 100
+
+    return f
+
 #---
 # Lexer base
 #---
@@ -184,18 +200,25 @@ class BitcodeLexer(LexerBase):
             return Token(T.CONST, math.pi)
 
         # Constants
-        if code in range(0x30, 0x39+1):
+        if code in range(0x30, 0x39+1) or code == 0x2E:
             # Never thought pointer arithmetic would beat Python. Grr!
-            re_const = rb'([0-9]+(?:\x2E[0-9]*)?(?:\x2D[0-9]+)?)(%?)'
-            match = re.match(re_const, h[p:])
+            re_const=rb'([0-9]*)(\x2E[0-9]*)?(\x2D[\xA6\xA7\xC0]?[0-9]+)?(%)?'
+            m = re.match(re_const, h[p:])
 
-            if match is not None:
-                text = match.group(1).replace(b'\x2E', b'.').replace(b'\x2D', b'e')
-                self.pos += len(text) - 1
+            if m is not None:
+                integer = (m[1] or b"").decode()
+                decimal = (m[2] or b".").replace(b'\x2E', b'.').decode()
+                exp     = (m[3] or b"")
+                percent = (m[4] or b"").decode()
 
-                f = float(text.decode('utf-8'))
-                if match.group(2) == "%":
-                    f /= 100
+                exp = exp.replace(b'\x2D', b'e')
+                exp = exp.replace(b'\xA6', b'+')
+                exp = exp.replace(b'\xA7', b'-')
+                exp = exp.replace(b'\xC0', b'-')
+                exp = exp.decode()
+
+                self.pos += len(m[0]) - 1
+                f = str2float(integer, decimal, exp, percent)
                 return Token(T.CONST, f)
 
         # Variables
@@ -302,7 +325,7 @@ class TextLexer(LexerBase):
         r"IFELSE_END|IFELSE|IF",
         re.IGNORECASE)
     RE_CONST = re.compile(
-        r"([0-9]+(?:\.[0-9]+)?(?:[eE][0-9]+)?)(%?)")
+        r"([0-9]*)(\.[0-9]*)?([eE][+-]?[0-9]+)?(%)?")
     RE_FUN = re.compile(
         r"([a-zA-Z]+)\(")
 
@@ -375,11 +398,10 @@ class TextLexer(LexerBase):
             return Token(punct[c[0]])
 
         # Constants
-        m = re.match(self.RE_CONST, c)
-        if m is not None:
-            f = float(m[1])
-            if m[2] == "%":
-                f /= 100
+        if c[0] in "0123456789.":
+            m = re.match(self.RE_CONST, c)
+            if m is not None:
+                f = str2float(m[1], m[2], m[3], m[4])
 
             self.code = c[len(m[0]):]
             self.pending_param = True

tests/constants.out

@@ -0,0 +1,11 @@
+173.0
+-374.0
+37.947
+913390.0
+9.454
+0.034
+1445.0
+0.0
+100000.0
+0.38
+-0.02434

tests/constants.txt

@@ -0,0 +1,12 @@
+print 173
+print -374
+print 37.947
+print 913.39e3
+print 945.4e-2
+print .034
+print 1445.
+print .
+print 1.e5
+print 38%
+print -24.34e-1%
+