curlyq: workspace.py annotate

annotate workspace.py @ 2:8884b0bf779d

Improve the efficiency.

author	David Barts <n5jrn@me.com>
date	Thu, 26 Dec 2019 13:18:53 -0800
parents	173e86601dbc
children	091c03f1b2e8

rev	line source
0 984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	1 #!/usr/bin/env python3
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	2 # -- coding: utf-8 --
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	3
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	4 # A class that implements a workspace for curly-quoting a text. This is enough
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	5 # like a string that it can be accessed via subscripts and ranges, and enough
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	6 # like a TextIOBase object that it can be written to much like a stream.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	7 # (However, a Workspace is neither a string nor a TextIOBase object.)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	8 #
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	9 # The advantage of using UTF-16 (as we do here) is that all quotation marks
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	10 # of interest are represented in a single 16-bit value, so changing straight
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	11 # quotes to curly ones can be accomplished most easily.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	12 #
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	13 # It was a deliberate design decision to return empty strings when reading
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	14 # out-of-range indices but to throw exceptions when attempting to write
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	15 # them, because both decisions made coding easier in other modules.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	16
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	17 # I m p o r t s
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	18
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	19 import os, sys
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	20 import io
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	21 import codecs
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	22
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	23 # V a r i a b l e s
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	24
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	25 # C l a s s e s
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	26
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	27 class Workspace(object):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	28 # The most efficient 16-bit one on this platform
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	29 encoding = "UTF-16" + sys.byteorder[0].upper() + "E"
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	30 codec = codecs.lookup(encoding)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	31 # Errors should never happen; UTF-16 can represent all Unicode characters
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	32 errors = 'strict'
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	33
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	34 def __init__(self, initial_data=None):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	35 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	36 Constructor.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	37 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	38 self._length = 0
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	39 if initial_data is not None:
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	40 data = initial_data.encode(self.encoding, self.errors)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	41 self._fp = io.BytesIO(data)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	42 else:
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	43 self._fp = io.BytesIO()
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	44
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	45 def close(self):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	46 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	47 Causes our buffer to be discarded and this workspace to become
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	48 unusable.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	49 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	50 self._fp.close()
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	51
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	52 def flush(self):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	53 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	54 Does nothing, but allowed.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	55 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	56 pass
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	57
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	58 def seek(self, offset, whence=io.SEEK_SET):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	59 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	60 Seeks to an absolute position.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	61 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	62 return self._fp.seek(offset, whence)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	63
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	64 def tell(self):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	65 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	66 Returns current position.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	67 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	68 return self._fp.tell()
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	69
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	70 def read(self, nchars=None):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	71 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	72 Read characters.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	73 XXX - might return replacement chars from surrogate fragments.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	74 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	75 if nchars is not None and nchars >= 0:
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	76 nchars *= 2
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	77 return self._fp.read(nchars).decode(self.encoding, "replace")
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	78
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	79 def write(self, string):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	80 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	81 Write characters.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	82 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	83 self._fp.write(string.encode(self.encoding, self.errors))
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	84
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	85 def __len__(self):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	86 """
1 173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	87 Length in characters.
0 984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	88 """
1 173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	89 return len(self._fp.getbuffer()) // 2
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	90
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	91 def _mapped(self, index):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	92 if index < 0 or index >= len(self):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	93 raise IndexError("index {0} out of range".format(index))
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	94 i2 = index * 2
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	95 return slice(i2, i2 + 2)
0 984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	96
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	97 def __getitem__(self, key):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	98 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	99 Direct access to a single character or range of characters. We do
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	100 not support negative indices. Return value is based on what's most
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	101 useful for curling quotes.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	102 XXX - might return replacement chars from surrogate fragments.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	103 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	104 if isinstance(key, int):
1 173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	105 try:
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	106 key = self._mapped(key)
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	107 except IndexError:
0 984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	108 return ""
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	109 elif isinstance(key, slice):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	110 if key.step is not None:
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	111 raise ValueError("__getitem__ does not support steps in slices")
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	112 length = len(self)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	113 start = 0 if key.start is None else key.start
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	114 stop = length if key.stop is None else key.stop
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	115 start = max(0, min(length - 1, start))
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	116 stop = max(0, min(length, stop))
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	117 if stop <= start:
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	118 return ""
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	119 key = slice(start * 2, stop * 2)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	120 else:
1 173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	121 raise TypeError("__getitem__ only supports integers and slices")
0 984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	122 return self.codec.decode(self._fp.getbuffer()[key], "replace")[0]
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	123
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	124 def __setitem__(self, key, value):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	125 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	126 Direct access to a single character. We do not support negative
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	127 indices or replacing more than a single character at a time.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	128 XXX - only works on characters in the BMP.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	129 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	130 if not isinstance(key, int):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	131 raise TypeError("__setitem__ only supports integers")
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	132 if not value:
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	133 return
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	134 encoded = value[0].encode(self.encoding, self.errors)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	135 if len(encoded) != 2:
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	136 raise ValueError("{0!r} not in BMP".format(value[0]))
1 173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	137 self._fp.getbuffer()[self._mapped(key)] = encoded
0 984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	138
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	139 def __del__(self):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	140 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	141 Equivalent to .close().
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	142 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	143 self.close()
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	144
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	145 def getvalue(self):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	146 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	147 Gets the string represented by this workspace.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	148 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	149 return self.codec.decode(self._fp.getbuffer(), self.errors)[0]
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	150
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	151 def __enter__(self):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	152 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	153 Context manager.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	154 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	155 return self
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	156
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	157 def __exit__(self, exc_type, exc_val, exc_tb):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	158 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	159 Context manager: close on exit.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	160 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	161 self.close()
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	162 return False
1 173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	163
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	164 class Bounds(object):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	165 def __init__(self, start, stop):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	166 if start > stop or start < 0 or stop < 0:
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	167 raise ValueError("invalid bounds")
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	168 self.start = int(start)
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	169 self.stop = int(stop)
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	170
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	171 @classmethod
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	172 def from_object(cls, obj):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	173 if isinstance(obj, slice):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	174 return self(slice.start, slice.stop)
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	175 return self(obj[0], obj[1])
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	176
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	177 def __lt__(self, other):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	178 return self.start < other.start
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	179
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	180 def __le__(self, other):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	181 return self.start <= other.start
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	182
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	183 def __eq__(self, other):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	184 return self.start == other.start
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	185
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	186 def __ne__(self, other):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	187 return self.start != other.start
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	188
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	189 def __gt__(self, other):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	190 return self.start > other.start
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	191
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	192 def __ge__(self, other):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	193 return self.start >= other.start
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	194
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	195 def __contains__(self, scalar):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	196 return self.start <= scalar < self.stop
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	197
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	198 def __repr__(self):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	199 return "{0}({1!r}, {2!r})".format(self.__class__.__name__, self.start, self.stop)
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	200
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	201 class Mapping(object):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	202 def __init__(self, bounds, offset):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	203 if not isinstance(bounds, Bounds):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	204 raise TypeError("bounds must be a Bounds object")
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	205 if not isinstance(offset, int):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	206 raise TypeError("offset must be an int")
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	207 self.bounds = bounds
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	208 self.offset = offset
2 8884b0bf779d Improve the efficiency. David Barts <n5jrn@me.com> parents: 1 diff changeset	209 self.delta = self.offset - self.bounds.start
1 173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	210
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	211 def __repr__(self):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	212 return "{0}({1!r}, {2!r})".format(self.__class__.__name__, self.bounds, self.offset)
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	213
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	214 class SegmentedView(object):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	215 """
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	216 Implements a view on a subscriptable object. The view is composed of
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	217 zero or more segments of the source object. Has the same idiosyncratic
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	218 behavior for out-of-bounds indices that Workspace has (and for the
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	219 same reason).
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	220 """
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	221 def __init__(self, indexable, bounds):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	222 self.indexable = indexable
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	223 self._mmap = [ Mapping(Bounds(0, 0), 0) ]
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	224 pos = 0
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	225 for r in sorted(bounds):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	226 if pos is not None and r.start <= pos and r.stop > pos:
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	227 # merge ranges
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	228 self._mmap[-1].bounds.stop = r.stop
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	229 pos = r.stop
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	230 continue
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	231 opos = pos
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	232 pos += r.stop - r.start
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	233 self._mmap.append(Mapping(Bounds(opos, pos), r.start))
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	234 self._length = pos
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	235
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	236 def _mapped(self, index):
2 8884b0bf779d Improve the efficiency. David Barts <n5jrn@me.com> parents: 1 diff changeset	237 mmap_index = self._binsch(index)
8884b0bf779d Improve the efficiency. David Barts <n5jrn@me.com> parents: 1 diff changeset	238 if mmap_index is None:
1 173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	239 raise IndexError("index {0} out of range".format(index))
2 8884b0bf779d Improve the efficiency. David Barts <n5jrn@me.com> parents: 1 diff changeset	240 return index + self._mmap[mmap_index].delta
1 173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	241
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	242 def _binsch(self, index):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	243 a = 0
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	244 z = len(self._mmap) - 1
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	245 while a <= z:
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	246 m = (a + z) // 2
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	247 if index in self._mmap[m].bounds:
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	248 return m
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	249 if index < self._mmap[m].bounds.start:
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	250 z = m - 1
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	251 else:
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	252 assert index >= self._mmap[m].bounds.stop
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	253 a = m + 1
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	254 return None
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	255
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	256 def __setitem__(self, key, value):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	257 if not isinstance(key, int):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	258 raise TypeError("__setitem__ only supports integers")
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	259 self.indexable[self._mapped(key)] = value
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	260
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	261 # XXX - this is sorta brute-forced and could be more efficient
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	262 def __getitem__(self, key):
2 8884b0bf779d Improve the efficiency. David Barts <n5jrn@me.com> parents: 1 diff changeset	263 # Trivial cases
1 173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	264 if isinstance(key, int):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	265 return self._get1(key)
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	266 if not isinstance(key, slice):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	267 raise TypeError("expecting int or slice")
2 8884b0bf779d Improve the efficiency. David Barts <n5jrn@me.com> parents: 1 diff changeset	268 if key.step is not None:
8884b0bf779d Improve the efficiency. David Barts <n5jrn@me.com> parents: 1 diff changeset	269 raise ValueError("__getitem__ does not support steps in slices")
8884b0bf779d Improve the efficiency. David Barts <n5jrn@me.com> parents: 1 diff changeset	270
8884b0bf779d Improve the efficiency. David Barts <n5jrn@me.com> parents: 1 diff changeset	271 # Loop up the starting segment.
8884b0bf779d Improve the efficiency. David Barts <n5jrn@me.com> parents: 1 diff changeset	272 mi = self._binsch(key.start)
8884b0bf779d Improve the efficiency. David Barts <n5jrn@me.com> parents: 1 diff changeset	273 if mi is None:
8884b0bf779d Improve the efficiency. David Barts <n5jrn@me.com> parents: 1 diff changeset	274 return ""
8884b0bf779d Improve the efficiency. David Barts <n5jrn@me.com> parents: 1 diff changeset	275 m = self._mmap[mi]
8884b0bf779d Improve the efficiency. David Barts <n5jrn@me.com> parents: 1 diff changeset	276
8884b0bf779d Improve the efficiency. David Barts <n5jrn@me.com> parents: 1 diff changeset	277 # Horray! There's only one segment, so we can optimize.
8884b0bf779d Improve the efficiency. David Barts <n5jrn@me.com> parents: 1 diff changeset	278 if key.stop <= m.bounds.stop:
8884b0bf779d Improve the efficiency. David Barts <n5jrn@me.com> parents: 1 diff changeset	279 start = key.start + m.delta
8884b0bf779d Improve the efficiency. David Barts <n5jrn@me.com> parents: 1 diff changeset	280 stop = key.stop + m.delta
8884b0bf779d Improve the efficiency. David Barts <n5jrn@me.com> parents: 1 diff changeset	281 return self.indexable[start:stop]
8884b0bf779d Improve the efficiency. David Barts <n5jrn@me.com> parents: 1 diff changeset	282
8884b0bf779d Improve the efficiency. David Barts <n5jrn@me.com> parents: 1 diff changeset	283 # The most involved (multi-segment) case.
1 173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	284 with io.StringIO() as buf:
2 8884b0bf779d Improve the efficiency. David Barts <n5jrn@me.com> parents: 1 diff changeset	285 for m in self._mmap[mi:]:
8884b0bf779d Improve the efficiency. David Barts <n5jrn@me.com> parents: 1 diff changeset	286 if m.bounds.start >= key.stop:
8884b0bf779d Improve the efficiency. David Barts <n5jrn@me.com> parents: 1 diff changeset	287 break
8884b0bf779d Improve the efficiency. David Barts <n5jrn@me.com> parents: 1 diff changeset	288 start = max(key.start, m.bounds.start) + m.delta
8884b0bf779d Improve the efficiency. David Barts <n5jrn@me.com> parents: 1 diff changeset	289 stop = min(key.stop, m.bounds.stop) + m.delta
8884b0bf779d Improve the efficiency. David Barts <n5jrn@me.com> parents: 1 diff changeset	290 buf.write(self.indexable[start:stop])
1 173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	291 return buf.getvalue()
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	292
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	293 def __len__(self):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	294 return self._length
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	295
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	296 def _get1(self, index):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	297 try:
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	298 return self.indexable[self._mapped(index)]
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	299 except IndexError:
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	300 return ""
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	301
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	302 def getvalue(self):
173e86601dbc Add views. David Barts <n5jrn@me.com> parents: 0 diff changeset	303 return self[0:len(self)]

Mercurial > cgi-bin > hgweb.cgi > curlyq

annotate workspace.py @ 2:8884b0bf779d