UTF-8 symbols in hexdump

__init__.py

@@ -6,7 +6,75 @@
 import sys
 import struct
 
-printable = 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ1234567890!@#$%^&*()[]{}`~/=-\\?+|\',."<>: _'
+stdch = (u'␀·········␊··␍··'
+         u'················'
+         u' !"#$%&\'()*+,-./'
+         u'0123456789:;<=>?'
+         u'@ABCDEFGHIJKLMNO'
+         u'PQRSTUVWXYZ[\]^_'
+         u'`abcdefghijklmno'
+         u'pqrstuvwxyz{|}~·'
+         u'················'
+         u'················'
+         u'················'
+         u'················'
+         u'················'
+         u'················'
+         u'················'
+         u'················')
+
+decch = (u'␀␁␂␃␄␅␆␇␈␉␊␋␌␍␎␏'
+         u'␐␑␒␓␔␕␖␗␘␙␚·····'
+         u'␠!"#$%&\'()*+,-./'
+         u'0123456789:;<=>?'
+         u'@ABCDEFGHIJKLMNO'
+         u'PQRSTUVWXYZ[\]^_'
+         u'`abcdefghijklmno'
+         u'pqrstuvwxyz{|}~␡'
+         u'················'
+         u'················'
+         u'················'
+         u'················'
+         u'················'
+         u'················'
+         u'················'
+         u'················')
+
+cgach = (u'␀☺☻♥♦♣♠•◘○◙♂♀♪♫☼'
+         u'►◄↕‼¶§▬↨↑↓→←∟↔▲▼'
+         u'␣!"#$%&\'()*+,-./'
+         u'0123456789:;<=>?'
+         u'@ABCDEFGHIJKLMNO'
+         u'PQRSTUVWXYZ[\]^_'
+         u'`abcdefghijklmno'
+         u'pqrstuvwxyz{|}~⌂'
+         u'ÇüéâäàåçêëèïîìÄÅ'
+         u'ÉæÆôöòûùÿÖÜ¢£¥₧ƒ'
+         u'áíóúñѪº¿⌐¬½¼¡«»'
+         u'░▒▓│┤╡╢╖╕╣║╗╝╜╛┐'
+         u'└┴┬├─┼╞╟╚╔╩╦╠═╬╧'
+         u'╨╤╥╙╘╒╓╫╪┘┌█▄▌▐▀'
+         u'αßΓπΣσµτΦΘΩδ∞φε∩'
+         u'≡±≥≤⌠⌡÷≈°∙·√ⁿ²■¤')
+
+shpch = (u'␀☺☻♥♦♣♠•◘○◙♂♀♪♫☼'
+         u'►◄↕‼¶§▬↨↑↓→←∟↔▲▼'
+         u'␣!"#$%&\'()*+,-./'
+         u'0123456789:;<=>?'
+         u'@ABCDEFGHIJKLMNO'
+         u'PQRSTUVWXYZ[\]^_'
+         u'`abcdefghijklmno'
+         u'pqrstuvwxyz{|}~⌂'
+         u'ÇüéâäàåçêëèïîìÄÅ'
+         u'ÉæÆôöòûùÿÖÜ¢£¥₧ƒ'
+         u'áíóúñѪº¿⌐¬½¼¡«»'
+         u'░▒▓│┤╡╢╖╕╣║╗╝╜╛┐'
+         u'└┴┬├─┼╞╟╚╔╩╦╠═╬╧'
+         u'╨╤╥╙╘╒╓╫╪┘┌█▄▌▐▀'
+         u'αßΓπΣσµτΦΘΩδ∞φε∩'
+         u'≡±≥≤⌠⌡÷≈°∙·√ⁿ²■¤')
+
+
 
 def unpack(fmt, buf):
     """Unpack buf based on fmt, return the rest as a string."""
@@ -24,19 +92,14 @@ class HexDumper:
 
     def _to_printable(self, c):
         if not c:
-            return '◆'
+            return u'◌'
-        elif c in printable:
-            return c
-        elif c == '\0':
-            return '␀'
-        elif c == '\r':
-            return '␍'
-        elif c == '\n':
-            return '␤'
         else:
-            return '·'
+            return cgach[ord(c)]
 
 
+    def write(self, what):
+        self.fd.write(what.encode('utf-8'))
+
     def _flush(self):
         if not self.buf:
             return
@@ -44,26 +107,26 @@ class HexDumper:
         o = []
         for c in self.buf:
             if c:
-                o.append('%02x' % ord(c))
+                o.append(u'%02x' % ord(c))
             else:
-                o.append('--')
+                o.append(u'--')
-        o +=  (['  '] * (16 - len(self.buf)))
+        o +=  ([u'  '] * (16 - len(self.buf)))
         p = [self._to_printable(c) for c in self.buf]
 
-        self.fd.write('%08x  ' % self.offset)
+        self.write(u'%08x  ' % self.offset)
 
-        self.fd.write(' '.join(o[:8]))
+        self.write(u' '.join(o[:8]))
-        self.fd.write('  ')
+        self.write(u'  ')
-        self.fd.write(' '.join(o[8:]))
+        self.write(u' '.join(o[8:]))
 
-        self.fd.write('  ║')
+        self.write(u'  ┆')
 
-        self.fd.write(''.join(p))
+        self.write(u''.join(p))
 
-        self.fd.write('║\n')
+        self.write(u'┆\n')
 
+        self.offset += len(self.buf)
         self.buf = []
-        self.offset += 16
 
     def dump_chr(self, c):
         self.buf.append(c)
@@ -77,7 +140,7 @@ class HexDumper:
 
     def finish(self):
         self._flush()
-        self.fd.write('%08x\n' % self.offset)
+        self.write('%08x\n' % self.offset)
 
 
 def hexdump(buf, f=sys.stdout):

gapstr.py

@@ -2,9 +2,9 @@
 
 ## 2008 Massive Blowout
 
-"""Functions to treat a list as a string with gaps.
+"""Functions to treat a list as a byte array with gaps.
 
-Lists should have only string and integer items.
+Lists should have only byte and numeric items.
 
 """
 
@@ -56,7 +56,9 @@ class GapString:
 
     def hasgaps(self):
         for i in self.contents:
-            if isinstance(i, int):
+            try:
+                len(i)
+            except TypeError:
                 return True
         return False
 
@@ -66,7 +68,7 @@ class GapString:
         d = __init__.HexDumper(fd)
         for i in self.contents:
             try:
-                for j in range(i):
+                for j in xrange(i):
                     d.dump_drop()
             except TypeError:
                 for c in i:
@@ -117,8 +119,26 @@ class GapString:
         return new
 
     def __getitem__(self, idx):
-        # XXX: speed up
+        if False:
-        return str(self)[idx]
+            c = self[idx:idx+1]
+            if c.hasgaps():
+                return self.drop[0]
+            else:
+                return c.contents[0][0]
+        else:
+            l = 0
+            for i in self.contents:
+                try:
+                    l += len(i)
+                except TypeError:
+                    l += i
+                if l > idx:
+                    offs = idx - l
+                    try:
+                        return i[offs]
+                    except:
+                        return self.drop[0]
+            raise IndexError('Out of bounds')
 
     def __add__(self, other):
         if isinstance(other, str):
@@ -133,8 +153,12 @@ class GapString:
         try:
             mask = [ord(c) for c in mask]
         except TypeError:
-            mask = [mask]
+            pass
+        try:
             masklen = len(mask)
+        except TypeError:
+            masklen = 1
+            mask = [mask]
 
         new = self.__class__(drop=self.drop)
         for i in self.contents:
@@ -147,6 +171,7 @@ class GapString:
                     offset = (offset + 1) % masklen
                 new.append(''.join(r))
             except TypeError:
+                print("type error!")
                 new.append(i)
         return new