Display wide characters correctly.

- Made termstr's length match the on-screen width.
  - Achieved by adding a zero width character after each wide
    character.
  - This fixes ljust and rjust, so that there is correct padding,
    otherwise Portal appearances aren't the correct width.
- When getting a sub-string containing half a wide character,
  the half character is a space.
- Some wide characters are still broken. e.g. country flags

eris/termstr.py

@@ -8,6 +8,7 @@ import os
 import weakref
 
 import pygments.formatters.terminal256
+import cwcwidth
 
 import eris.ColorMap
 import eris.terminal as terminal
@@ -130,15 +131,26 @@ def _join_lists(lists):
     return list(itertools.chain.from_iterable(lists))
 
 
+_ZERO_WIDTH_SPACE = chr(0x00ad)
+
+
+def _pad_wide_chars(str_):
+    return "".join(f"{char}{_ZERO_WIDTH_SPACE}"
+                   if cwcwidth.wcwidth(char) == 2 else char for char in str_)
+
+
 class TermStr(collections.UserString):
 
     def __init__(self, data, style=CharStyle()):
-        try:
-            self.data, self.style = data.data, data.style
-        except AttributeError:
+        if isinstance(style, tuple):
             self.data = data
-            self.style = (style if isinstance(style, tuple)
-                          else (style,) * len(data))
+            self.style = style
+        else:
+            try:
+                self.data, self.style = data.data, data.style
+            except AttributeError:
+                self.data = _pad_wide_chars(data)
+                self.style = (style,) * len(self.data)
 
     @classmethod
     def from_term(cls, data):
@@ -249,7 +261,17 @@ class TermStr(collections.UserString):
     __rmul__ = __mul__
 
     def __getitem__(self, index):
-        return self.__class__(self.data[index], self.style[index])
+        data = self.data[index]
+        if len(data) == 0:
+            result = ""
+        else:
+            first_char = " " if data[0] == _ZERO_WIDTH_SPACE else data[0]
+            if len(data) == 1:
+                result = first_char
+            else:
+                end_char = " " if cwcwidth.wcwidth(data[-1]) == 2 else data[-1]
+                result = first_char + data[1:-1] + end_char
+        return self.__class__(result, self.style[index])
 
     def join(self, parts):
         parts = [TermStr(part) if isinstance(part, str) else part

install-dependencies

@@ -10,7 +10,7 @@ if [ $DIST_ID != "ubuntu" ]; then
     exit 1
 fi
 echo "Installing the dependencies of the eris script…"
-sudo apt --yes install python3-pip python3.9 util-linux
+sudo apt --yes install python3-pip python3.9 util-linux python3-cwcwidth
 python3.9 -m pip install pyinotify pygments docopt pillow toml
 echo
 echo "Installing all the tools eris may need…"

tests/termstr_test.py

@@ -101,6 +101,11 @@ class TermStrTests(unittest.TestCase):
         self.assertEqual(foo_bold.ljust(5), foo_bold + TermStr("  "))
         self.assertEqual(foo_bold.rjust(0), foo_bold)
         self.assertEqual(foo_bold.rjust(5), TermStr("  ") + foo_bold)
+        baz = TermStr("b👋z")
+        self.assertEqual(len(baz), 4)
+        self.assertEqual(baz[3:], TermStr("z"))
+        self.assertEqual(baz[:2], TermStr("b "))
+        self.assertEqual(baz[2:], TermStr(" z"))
 
     def test_from_term(self):
         def test_round_trip(term_str):