[WIP] unicode support

Author: obfusk

pyrepl/reader.py

@@ -20,15 +20,24 @@
 # CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
 
 from __future__ import unicode_literals
+import sys
 import unicodedata
 from pyrepl import commands
 from pyrepl import input
 try:
     unicode
+    decode = lambda x, enc = sys.stdout.encoding: unicode(x, enc)
 except NameError:
     unicode = str
     unichr = chr
     basestring = bytes, str
+    decode = lambda x, _ = None: x
+
+
+def width(c):
+    return 2 if unicodedata.east_asian_width(c) in "FW" else 1
+def wlen(s):
+    return sum(map(width, s))
 
 
 def _make_unctrl_map():
@@ -39,8 +48,8 @@ def _make_unctrl_map():
     for i in range(32):
         c = unichr(i)
         uc_map[c] = '^' + unichr(ord('A') + i - 1)
-    uc_map[b'\t'] = '    '  # display TABs as 4 characters
-    uc_map[b'\177'] = unicode('^?')
+    uc_map['\t'] = '    '  # display TABs as 4 characters
+    uc_map['\177'] = unicode('^?')
     for i in range(256):
         c = unichr(i)
         if c not in uc_map:
@@ -53,7 +62,7 @@ def _my_unctrl(c, u=_make_unctrl_map()):
         return u[c]
     else:
         if unicodedata.category(c).startswith('C'):
-            return br'\u%04x' % ord(c)
+            return '\\u%04x' % ord(c)
         else:
             return c
 
@@ -75,7 +84,7 @@ def disp_str(buffer, join=''.join, uc=_my_unctrl):
     s = [uc(x) for x in buffer]
     b = []  # XXX: bytearray
     for x in s:
-        b.append(1)
+        b.append(width(x[0]))
         b.extend([0] * (len(x) - 1))
     return join(s), b
 
@@ -280,7 +289,7 @@ def calc_screen(self):
                     for mline in self.msg.split("\n"):
                         screen.append(mline)
                         screeninfo.append((0, []))
-                self.lxy = p, ln
+                # self.lxy = p, ln
             prompt = self.get_prompt(ln, ll >= p >= 0)
             while '\n' in prompt:
                 pre_prompt, _, prompt = prompt.partition('\n')
@@ -289,8 +298,8 @@ def calc_screen(self):
             p -= ll + 1
             prompt, lp = self.process_prompt(prompt)
             l, l2 = disp_str(line)
-            wrapcount = (len(l) + lp) // w
-            if wrapcount == 0:
+            wrapcount = (wlen(l) + lp) // w
+            if 1 or wrapcount == 0:                           #  FIXME
                 screen.append(prompt + l)
                 screeninfo.append((lp, l2 + [1]))
             else:
@@ -318,7 +327,7 @@ def process_prompt(self, prompt):
         is returned with these control characters removed.  """
 
         out_prompt = ''
-        l = len(prompt)
+        l = wlen(prompt)
         pos = 0
         while True:
             s = prompt.find('\x01', pos)
@@ -420,7 +429,7 @@ def get_prompt(self, lineno, cursor_on_line):
         # the object on which str() was called.  This ensures that even if the
         # same object is used e.g. for ps1 and ps2, str() is called only once.
         if res not in self._pscache:
-            self._pscache[res] = str(res)
+            self._pscache[res] = decode(res)
         return self._pscache[res]
 
     def push_input_trans(self, itrans):
@@ -438,23 +447,26 @@ def pos2xy(self, pos):
         if pos == len(self.buffer):
             y = len(self.screeninfo) - 1
             p, l2 = self.screeninfo[y]
-            return p + len(l2) - 1, y
+            return p + sum(l2) + l2.count(0) - 1, y
         else:
             for p, l2 in self.screeninfo:
-                l = l2.count(1)
+                l = len(l2) - l2.count(0)
                 if l > pos:
                     break
                 else:
                     pos -= l
                     y += 1
             c = 0
             i = 0
-            while c < pos:
-                c += l2[i]
+            j = 0
+            while j < pos:
+                j += 1 if l2[i] else 0
+                c += l2[i] or 1
                 i += 1
             while l2[i] == 0:
+                c += 1
                 i += 1
-            return p + i, y
+            return p + c, y
 
     def insert(self, text):
         """Insert 'text' at the insertion point."""

pyrepl/unix_console.py

@@ -28,6 +28,7 @@
 import re
 import time
 import sys
+import unicodedata
 from fcntl import ioctl
 from . import curses
 from .fancy_termios import tcgetattr, tcsetattr
@@ -44,6 +45,13 @@ class InvalidTerminal(RuntimeError):
 except NameError:
     unicode = str
 
+
+def width(c):
+    return 2 if unicodedata.east_asian_width(c) in "FW" else 1
+def wlen(s):
+    return sum(map(width, s))
+
+
 _error = (termios.error, curses.error, InvalidTerminal)
 
 # there are arguments for changing this to "refresh"
@@ -247,46 +255,56 @@ def __write_changed_line(self, y, oldline, newline, px):
         # structuring this function are equally painful (I'm trying to
         # avoid writing code generators these days...)
         x = 0
-        minlen = min(len(oldline), len(newline))
+        i = 0
+        minlen = min(wlen(oldline), wlen(newline))
+        pi = 0
+        xx = 0
+        for c in oldline:
+          xx += width(c)
+          pi += 1
+          if xx >= px: break
         #
         # reuse the oldline as much as possible, but stop as soon as we
         # encounter an ESCAPE, because it might be the start of an escape
         # sequene
-        #XXX unicode check!
-        while x < minlen and oldline[x] == newline[x] and newline[x] != '\x1b':
-            x += 1
-        if oldline[x:] == newline[x+1:] and self.ich1:
+        while x < minlen and oldline[i] == newline[i] and newline[i] != '\x1b':
+            x += width(newline[i])
+            i += 1
+        if oldline[i:] == newline[i+1:] and self.ich1:
             if (y == self.__posxy[1] and x > self.__posxy[0] and
-                    oldline[px:x] == newline[px+1:x+1]):
+                    oldline[pi:i] == newline[pi+1:i+1]):
+                i = pi
                 x = px
             self.__move(x, y)
-            self.__write_code(self.ich1)
-            self.__write(newline[x])
-            self.__posxy = x + 1, y
-        elif x < minlen and oldline[x + 1:] == newline[x + 1:]:
+            cw = width(newline[i])
+            self.__write_code(cw*self.ich1)
+            self.__write(newline[i])
+            self.__posxy = x + cw, y
+        elif (x < minlen and oldline[i + 1:] == newline[i + 1:]
+              and width(oldline[i]) == width(newline[i])):
             self.__move(x, y)
-            self.__write(newline[x])
-            self.__posxy = x + 1, y
-        elif (self.dch1 and self.ich1 and len(newline) == self.width
-              and x < len(newline) - 2
-              and newline[x+1:-1] == oldline[x:-2]):
+            self.__write(newline[i])
+            self.__posxy = x + width(newline[i]), y
+        elif (self.dch1 and self.ich1 and wlen(newline) == self.width
+              and x < wlen(newline) - 2
+              and newline[i+1:-1] == oldline[i:-2]):
+            raise NotImplementedError()                       #  FIXME
             self.__hide_cursor()
             self.__move(self.width - 2, y)
             self.__posxy = self.width - 2, y
             self.__write_code(self.dch1)
             self.__move(x, y)
             self.__write_code(self.ich1)
-            self.__write(newline[x])
-            self.__posxy = x + 1, y
+            self.__write(newline[i])
+            self.__posxy = x + width(newline[i]), y
         else:
             self.__hide_cursor()
             self.__move(x, y)
-            if len(oldline) > len(newline):
+            if wlen(oldline) > wlen(newline):
                 self.__write_code(self._el)
-            self.__write(newline[x:])
-            self.__posxy = len(newline), y
+            self.__write(newline[i:])
+            self.__posxy = wlen(newline), y
 
-        #XXX: check for unicode mess
         if '\x1b' in newline:
             # ANSI escape characters are present, so we can't assume
             # anything about the position of the cursor.  Moving the cursor