use wcwidth (if available)

pyrepl/reader.py

@@ -36,10 +36,29 @@ def decode(x, enc = sys.stdout.encoding):
  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))
+try:
+ from wcwidth import width
+except ImportError:
+ try:
+ from wcwidth import wcwidth
+ except ImportError:
+ import unicodedata
+
+ def wcwidth(c):
+ if unicodedata.east_asian_width(c) in "FW":
+ return 2
+ if unicodedata.category(c)[0] == "C":
+ return 0
+ return 1
+
+ try:
+ from functools import lru_cache
+ except ImportError:
+ lru_cache = lambda *a, **k: lambda f: f
+
+ @lru_cache(maxsize=1000)
+ def width(s):
+ return sum(max(0, wcwidth(c)) for c in s)
 
 
 _r_csi_seq = re.compile(r"\033\[[ -@]*[A-~]")
@@ -303,7 +322,7 @@ def calc_screen(self):
  p -= ll + 1
  prompt, lp = self.process_prompt(prompt)
  l, l2 = disp_str(line)
- wrapcount = (wlen(l) + lp) // w
+ wrapcount = (width(l) + lp) // w
  if wrapcount == 0:
  screen.append(prompt + l)
  screeninfo.append((lp, l2 + [1]))
@@ -347,7 +366,7 @@ def process_prompt(self, prompt):
  # They are CSI (or ANSI) sequences ( ESC [ ... LETTER )
 
  out_prompt = ''
- l = wlen(prompt)
+ l = sum(max(1, width(c)) for c in prompt) # FIXME
  pos = 0
  while True:
  s = prompt.find('\x01', pos)

pyrepl/unix_console.py

@@ -21,7 +21,6 @@
 
 import termios, select, os, struct, errno
 import signal, re, time, sys
-import unicodedata
 from fcntl import ioctl
 from . import curses
 from .fancy_termios import tcgetattr, tcsetattr
@@ -44,10 +43,29 @@ class InvalidTerminal(RuntimeError):
  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))
+try:
+ from wcwidth import width
+except ImportError:
+ try:
+ from wcwidth import wcwidth
+ except ImportError:
+ import unicodedata
+
+ def wcwidth(c):
+ if unicodedata.east_asian_width(c) in "FW":
+ return 2
+ if unicodedata.category(c)[0] == "C":
+ return 0
+ return 1
+
+ try:
+ from functools import lru_cache
+ except ImportError:
+ lru_cache = lambda *a, **k: lambda f: f
+
+ @lru_cache(maxsize=1000)
+ def width(s):
+ return sum(max(0, wcwidth(c)) for c in s)
 
 
 _error = (termios.error, curses.error, InvalidTerminal)
@@ -266,7 +284,7 @@ def __write_changed_line(self, y, oldline, newline, px):
  # avoid writing code generators these days...)
  x = 0
  i = 0
- minlen = min(wlen(oldline), wlen(newline))
+ minlen = min(width(oldline), width(newline))
  pi = 0
  xx = 0
  for c in oldline:
@@ -295,8 +313,8 @@ def __write_changed_line(self, y, oldline, newline, px):
  self.__move(x, y)
  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
+ elif (self.dch1 and self.ich1 and width(newline) == self.width
+ and x < width(newline) - 2
  and newline[i+1:] == oldline[i:-1]):
  cw = width(newline[i])
  self.__hide_cursor()
@@ -310,10 +328,10 @@ def __write_changed_line(self, y, oldline, newline, px):
  else:
  self.__hide_cursor()
  self.__move(x, y)
- if wlen(oldline) > wlen(newline):
+ if width(oldline) > width(newline):
  self.__write_code(self._el)
  self.__write(newline[i:])
- self.__posxy = wlen(newline), y
+ self.__posxy = width(newline), y
 
  if '\x1b' in newline:
  # ANSI escape characters are present, so we can't assume