curlyq: runes.py annotate

annotate runes.py @ 29:d5bf9985b5c4 default tip

Add degree symbol, fix bug in HTML curler.

author	David Barts <n5jrn@me.com>
date	Thu, 07 Oct 2021 11:55:46 -0700
parents	a771878f6cf4
children

rev	line source
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	1 #!/usr/bin/env python3
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	2 # -- coding: utf-8 --
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	3
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	4 # Something like Java's StringBuilder, but for Python. It needs more
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	5 # thorough testing. This runs about 3x slower than io.StringIO, but
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	6 # unlike that class allows for easy and meaningful random access via
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	7 # subscripts.
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	8 #
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	9 # This implements two objects: Runes and Workspace. The former is a
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	10 # relatively low-level object that deals mostly in UTF-16 rune values;
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	11 # however, you can create a Runes object from a Python string, and you
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	12 # can create a Python string from a Runes object. Aside from that, Runes
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	13 # deals in numeric values, not Python strings. Workspace methods tend
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	14 # to accept and return Python strings, which makes them more programmer-
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	15 # friendly, at the cost of often having to fire up a codec to convert
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	16 # things back and forth between Python strings and UTF-16 runes.
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	17
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	18 # I m p o r t s
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	19
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	20 import array
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	21 import codecs
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	22 import sys
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	23
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	24 # C l a s s e s
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	25
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	26 class Runes(object):
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	27 """
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	28 A mutable sequence of UTF-16 runes. The attributes encoding and
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	29 codec contain the name of the encoding and the codec used to
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	30 generate the UTF-16. The attribute buffer contains the buffer (an
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	31 array of 16-bit unsigned integers) used to back this object;
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	32 modifications to that array will be reflected in this object.
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	33 """
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	34 # The most efficient 16-bit one on this platform
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	35 encoding = "UTF-16" + sys.byteorder[0].upper() + "E"
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	36 codec = codecs.lookup(encoding)
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	37 _ERRORS = 'surrogatepass'
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	38 _MIN_S = 0xd800 # lowest possible surrogate
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	39 _MID_S = 0xdc00 # high surrogate if <, low if >=
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	40 _MAX_S = 0xdfff # highest possible surrogate
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	41
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	42 def __init__(self, based_on=None):
20 be0fd5c8121d Can't use memoryview to base Runes on; different methods inside. David Barts <n5jrn@me.com> parents: 17 diff changeset	43 if isinstance(based_on, array.array):
be0fd5c8121d Can't use memoryview to base Runes on; different methods inside. David Barts <n5jrn@me.com> parents: 17 diff changeset	44 if based_on.typecode == 'H':
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	45 self.buffer = based_on
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	46 else:
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	47 self.buffer = array.array('H', based_on)
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	48 elif isinstance(based_on, str):
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	49 self.buffer = array.array('H', self.codec.encode(based_on, self._ERRORS)[0])
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	50 elif based_on is None:
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	51 self.buffer = array.array('H')
12 ab7d6e908034 Allow cloning. David Barts <n5jrn@me.com> parents: 10 diff changeset	52 elif isinstance(based_on, Runes):
ab7d6e908034 Allow cloning. David Barts <n5jrn@me.com> parents: 10 diff changeset	53 self.buffer = array.array('H', based_on.buffer)
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	54 else:
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	55 self.buffer = array.array('H', based_on)
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	56
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	57 def __str__(self):
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	58 """
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	59 Convert this object to a string. We deliberately do not have a
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	60 __repr__ method, to underscore that runes are not strings.
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	61 """
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	62 return self.codec.decode(self.buffer, self._ERRORS)[0]
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	63
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	64 def __bytes__(self):
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	65 return bytes(self.buffer)
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	66
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	67 def __len__(self):
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	68 return len(self.buffer)
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	69
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	70 def _checkindex(self, index, allow_equal=False):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	71 ok = 0 <= index <= len(self) if allow_equal else 0 <= index < len(self)
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	72 if not ok:
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	73 raise IndexError("index {0} out of range".format(index))
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	74
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	75 def _checktype(self, other):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	76 if not isinstance(other, Runes):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	77 raise TypeError("Runes required")
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	78
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	79 def __lt__(self, other):
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	80 self._checktype(other)
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	81 return self.buffer < other.buffer
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	82
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	83 def __le__(self, other):
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	84 self._checktype(other)
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	85 return self.buffer <= other.buffer
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	86
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	87 def __gt__(self, other):
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	88 self._checktype(other)
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	89 return self.buffer > other.buffer
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	90
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	91 def __ge__(self, other):
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	92 self._checktype(other)
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	93 return self.buffer >= other.buffer
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	94
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	95 def __eq__(self, other):
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	96 self._checktype(other)
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	97 return self.buffer == other.buffer
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	98
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	99 def __ne__(self, other):
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	100 self._checktype(other)
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	101 return self.buffer != other.buffer
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	102
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	103 def __hash__(self):
13 6acda841a690 Add support for shared substrings. David Barts <n5jrn@me.com> parents: 12 diff changeset	104 raise TypeError("unhashable type")
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	105
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	106 def __bool__(self):
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	107 return bool(self.buffer)
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	108
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	109 def __getitem__(self, key):
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	110 if isinstance(key, int):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	111 return self.buffer[key]
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	112 elif isinstance(key, slice):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	113 return Runes(self.buffer[key])
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	114 else:
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	115 raise AssertionError("this shouldn't happen")
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	116
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	117 def __setitem__(self, key, value):
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	118 if isinstance(key, int):
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	119 if isinstance(value, int):
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	120 self.buffer[key] = value
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	121 else:
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	122 raise TypeError("integer required")
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	123 elif isinstance(value, Runes):
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	124 self.buffer[key] = value.buffer
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	125 else:
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	126 raise TypeError("Runes required")
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	127
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	128 def __delitem__(self, key):
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	129 del self.buffer[key]
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	130
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	131 def clear(self):
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	132 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	133 Remove all data from our buffer. This merely marks the buffer as
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	134 empty; it does nothing to destroy its contents by overwriting.
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	135 """
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	136 del self[:]
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	137
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	138 def zero(self):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	139 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	140 Overwrite our buffer with zeroes.
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	141 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	142 for i in range(len(self.buffer)):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	143 self.buffer[i] = 0
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	144
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	145 def __iter__(self):
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	146 return iter(self.buffer)
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	147
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	148 def __reversed__(self):
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	149 return reversed(self.buffer)
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	150
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	151 def append(self, value):
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	152 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	153 Append data to our buffer.
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	154 """
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	155 if isinstance(value, int):
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	156 self.buffer.append(value)
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	157 elif isinstance(value, Runes):
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	158 self.buffer.extend(value.buffer)
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	159 else:
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	160 raise TypeError("integer or Runes required")
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	161
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	162 def __contains__(self, value):
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	163 return value in self.buffer
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	164
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	165 @classmethod
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	166 def is_high_surrogate(cls, value):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	167 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	168 Is value in the UTF-16 high surrogate range?
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	169 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	170 return cls._MIN_S <= value < cls._MID_S
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	171
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	172 @classmethod
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	173 def is_low_surrogate(cls, value):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	174 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	175 Is value in the UTF-16 low surrogate range?
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	176 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	177 return cls._MIN_S <= value <= cls._MAX_S
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	178
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	179 @classmethod
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	180 def is_surrogate(cls, value):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	181 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	182 Is value in the UTF-16 surrogate range?
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	183 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	184 return cls._MIN_S <= value <= cls._MAX_S
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	185
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	186 def index(self, value, from_index=0):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	187 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	188 Substring index, throws exception if not found.
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	189 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	190 self._checktype(value)
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	191 slimit = len(self)
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	192 rlimit = len(value)
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	193 for i in range(from_index, len(self)):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	194 match = True
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	195 for j in range(rlimit):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	196 k = i + j
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	197 if k >= slimit or value.buffer[j] != self.buffer[k]:
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	198 match = False
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	199 break
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	200 if match:
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	201 return i
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	202 raise ValueError("substring not found")
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	203
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	204 def find(self, value, from_index=0):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	205 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	206 Substring index, returns -1 if not found.
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	207 """
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	208 try:
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	209 return self.index(value, from_index)
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	210 except ValueError:
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	211 return -1
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	212
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	213 def rindex(self, value, from_index=None):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	214 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	215 Reverse substring index, throws exception if not found.
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	216 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	217 self._checktype(value)
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	218 if from_index is None:
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	219 from_index = len(self) - 1
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	220 rfrom = len(value) - 1
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	221 for i in range(from_index, -1, -1):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	222 match = True
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	223 for j in range(rfrom, -1, -1):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	224 k = i - (rfrom - j)
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	225 if k < 0 or value.buffer[j] != self.buffer[k]:
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	226 match = False
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	227 break
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	228 if match:
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	229 return i - rfrom
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	230 raise ValueError("substring not found")
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	231
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	232 def rfind(self, value, from_index=None):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	233 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	234 Reverse substring index, returns -1 if not found.
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	235 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	236 try:
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	237 return self.rindex(value, from_index)
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	238 except ValueError:
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	239 return -1
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	240
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	241 class Workspace(Runes):
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	242 """
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	243 A Runes object (q.v.) that acts a bit more string-like.
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	244 """
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	245
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	246 def __setitem__(self, key, value):
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	247 if isinstance(value, str):
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	248 if isinstance(key, int):
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	249 value = ord(value)
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	250 if value > 0xffff:
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	251 raise ValueError("character not in BMP")
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	252 super().__setitem__(key, value)
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	253 else:
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	254 super().__setitem__(key, Runes(value))
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	255 else:
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	256 super().__setitem__(key, value)
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	257
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	258 def __getitem__(self, key):
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	259 if isinstance(key, int):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	260 return chr(self.buffer[key])
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	261 elif isinstance(key, slice):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	262 view = memoryview(self.buffer)
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	263 try:
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	264 result = view[key]
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	265 if not isinstance(result, memoryview):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	266 assert isinstance(result, int)
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	267 return chr(result)
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	268 ret = self.codec.decode(result, self._ERRORS)[0]
15 0be0586104b7 Plug the leak properly (I hope). David Barts <n5jrn@me.com> parents: 14 diff changeset	269 result.release()
0be0586104b7 Plug the leak properly (I hope). David Barts <n5jrn@me.com> parents: 14 diff changeset	270 return ret
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	271 finally:
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	272 view.release()
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	273 else:
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	274 raise AssertionError("this shouldn't happen")
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	275
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	276 def __contains__(self, value):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	277 if isinstance(value, int):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	278 return value in self.buffer
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	279 return self.find(value) != -1
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	280
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	281 def __iter__(self):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	282 for i in range(len(self)):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	283 yield self[i]
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	284
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	285 def __reversed__(self):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	286 for i in range(len(self)-1, -1, -1):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	287 yield self[i]
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	288
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	289 def append(self, value):
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	290 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	291 Append string or runes to this item.
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	292 """
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	293 if isinstance(value, str):
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	294 value = Runes(value)
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	295 elif not isinstance(value, (int, Runes)):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	296 raise TypeError("integer, string, or Runes required")
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	297 super().append(value)
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	298
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	299 def _runify(self, value):
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	300 if isinstance(value, str):
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	301 return Runes(value)
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	302 elif isinstance(value, Runes):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	303 return value
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	304 else:
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	305 raise TypeError("Runes or string required")
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	306
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	307 def index(self, value, from_index=0):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	308 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	309 Substring index, throws exception if not found.
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	310 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	311 return super().index(self._runify(value), from_index)
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	312
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	313 def find(self, value, from_index=0):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	314 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	315 Substring index, returns -1 if not found.
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	316 """
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	317 try:
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	318 return self.index(value, from_index)
10 397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	319 except ValueError:
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	320 return -1
397c178c5b98 Make it array-based. David Barts <n5jrn@me.com> parents: diff changeset	321
22 a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	322 def rindex(self, value, from_index=None):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	323 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	324 Reverse substring index, throws exception if not found.
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	325 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	326 return super().rindex(self._runify(value), from_index)
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	327
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	328 def rfind(self, value, from_index=None):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	329 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	330 Reverse substring index, returns -1 if not found.
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	331 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	332 try:
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	333 return self.rindex(value, from_index)
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	334 except ValueError:
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	335 return -1
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	336
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	337 def _code_point_at(self, index):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	338 self._checkindex(index, allow_equal=False)
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	339 v0 = self.buffer[index]
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	340 i1 = index + 1
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	341 v1 = None if i1 >= len(self.buffer) else self.buffer[i1]
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	342 if Runes.is_high_surrogate(v0) and v1 is not None and Runes.is_low_surrogate(v1):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	343 return slice(index, index+2)
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	344 else:
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	345 return slice(index, i1)
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	346
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	347 def code_point_at(self, index):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	348 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	349 Similar to java.lang.String.codePointAt, but returns a 1-character
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	350 string, not an int.
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	351 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	352 return self[self._code_point_at(index)]
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	353
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	354 def _code_point_before(self, index):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	355 self._checkindex(index - 1, allow_equal=True)
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	356 i1 = index - 1
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	357 v1 = self.buffer[i1]
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	358 i2 = index - 2
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	359 v2 = None if i2 < 0 else self.buffer[i2]
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	360 if Runes.is_low_surrogate(v1) and v2 is not None and Runes.is_high_surrogate(v2):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	361 return slice(i2, index)
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	362 else:
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	363 return slice(i1, index)
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	364
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	365 def code_point_before(self, index):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	366 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	367 Similar to java.lang.String.codePointBefore but returns a 1-character
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	368 string, not an int.
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	369 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	370 return self[self._code_point_before(index)]
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	371
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	372 def code_point_count(self, begin=None, end=None):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	373 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	374 Same behavior as java.lang.String.codePointCount (q.v.).
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	375 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	376 if begin is None: begin = 0
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	377 if end is None: end = len(self.buffer)
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	378 self._checkindex(begin, allow_equal=False)
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	379 self._checkindex(end, allow_equal=True)
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	380 if begin > end:
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	381 raise IndexError("invalid range (begin > end)")
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	382 i = begin
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	383 ret = 0
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	384 while i < end:
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	385 i = self._code_point_at(i).stop
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	386 ret += 1
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	387 return ret
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	388
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	389 def offset_by_code_points(self, index, offset):
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	390 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	391 Same behavior as java.lang.String.offsetByCodePoints (q.v.).
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	392 """
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	393 self._checkindex(index, allow_equal=True)
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	394 if offset < 0:
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	395 next = lambda i: self._code_point_before(i).start
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	396 else:
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	397 next = lambda i: self._code_point_at(i).stop
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	398 seen = 0
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	399 limit = abs(offset)
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	400 while seen < limit:
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	401 index = next(index)
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	402 seen += 1
a771878f6cf4 Remove deadwood, update runes.py. David Barts <n5jrn@me.com> parents: 21 diff changeset	403 return index

Mercurial > cgi-bin > hgweb.cgi > curlyq

annotate runes.py @ 29:d5bf9985b5c4 default tip