11.3.2021, 9:00 - 11:00: Due to updates GitLab may be unavailable for some minutes between 09:00 and 11:00.

stringview.py 13.6 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
# stringview.py - a string class where slices are views not copies as
#                    with the standard Python strings.
#
#    stringview.pxd - declarations for the cython Python to C compiler
#                    to speed up handling of StringViews.
#
# Copyright 2016  by Eckhart Arnold (arnold@badw.de)
#                 Bavarian Academy of Sciences an Humanities (badw.de)
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
# implied.  See the License for the specific language governing
# permissions and limitations under the License.
di68kap's avatar
di68kap committed
21

22 23 24 25 26 27 28 29 30 31 32 33
"""
StringView provides string-slicing without copying.
Slicing Python-strings always yields copies of a segment of the original
string. See: https://mail.python.org/pipermail/python-dev/2008-May/079699.html
However, this becomes costly (in terms of space and as a consequence also
time) when parsing longer documents. Unfortunately, Python's `memoryview`
does not work for unicode strings. Hence, the StringView class.

It is recommended to compile this modules with the Cython-compiler for
speedup. The modules comes with a ``stringview.pxd`` that contains some type
declarations to fully exploit the potential of the Cython-compiler.
"""
34

35
import collections
di68kap's avatar
di68kap committed
36 37

from DHParser.toolkit import typing
38
from typing import Optional, Union, Iterable, Tuple
39

40 41 42 43 44 45 46
try:
    import cython
    cython_optimized = cython.compiled  # type: bool
except ImportError:
    # import DHParser.Shadow as cython
    cython_optimized = False            # type: bool
    import DHParser.shadow_cython as cython
47

48 49

__all__ = ('StringView', 'EMPTY_STRING_VIEW', 'cython_optimized')
50 51


52 53 54 55
def first_char(text, begin: int, end: int) -> int:
    """Returns the index of the first non-whitespace character in string
     `text` within the bounds [begin, end].
    """
56 57 58 59 60
    while begin < end and text[begin] in ' \n\t':
        begin += 1
    return begin


61 62 63 64
def last_char(text, begin: int, end: int) -> int:
    """Returns the index of the first non-whitespace character in string
    `text` within the bounds [begin, end].
    """
eckhart's avatar
eckhart committed
65
    while end > begin and text[end - 1] in ' \n\t':
66 67 68 69
        end -= 1
    return end


70 71 72 73 74 75 76 77 78 79 80 81 82
def pack_index(index: int, length: int) -> int:
    """Transforms `index` into a positive index counting from the beginning
    of the string, capping it at the boundaries [0, len].
    Examples:
    >>> pack_index(-1, 5)
    4
    >>> pack_index(6, 5)
    5
    >>> pack_index(-7, 5)
    0
    """
    # assert length >= 0
    index = index if index >= 0 else index + length
83 84 85
    # TODO: Test the following code for speedup
    # if index < 0:
    #     index += length
86
    return 0 if index < 0 else length if index > length else index
87 88


89 90
def real_indices(begin: Optional[int],
                 end: Optional[int],
91
                 length) -> Tuple[int, int]:
92 93 94
    """Returns the tuple of real (i.e. positive) indices from the slice
    indices `begin`,  `end`, assuming a string of size `length`.
    """
95
    cbegin = 0 if begin is None else begin
96 97
    cend = length if end is None else end
    return pack_index(cbegin, length), pack_index(cend, length)
98 99


100
class StringView:  # collections.abc.Sized
101 102
    """
    A rudimentary StringView class, just enough for the use cases
103
    in parse.py. The difference between a StringView and the python
104 105 106 107
    builtin strings is that StringView-objects do slicing without
    copying, i.e. slices are just a view on a section of the sliced
    string.
    """
108
    __slots__ = ['_text', '_begin', '_end', '_len', '_fullstring']
109 110

    def __init__(self, text: str, begin: Optional[int] = 0, end: Optional[int] = None) -> None:
111
        # assert isinstance(text, str)
112 113 114 115 116
        self._text = text  # type: str
        self._begin, self._end = real_indices(begin, end, len(text))
        self._len = max(self._end - self._begin, 0)  # type: int
        if (self._begin == 0 and self._len == len(self._text)):
            self._fullstring = self._text  # type: str
117
        else:
118
            self._fullstring = ''
119

120
    def __bool__(self) -> bool:
121
        return self._end > self._begin  # and bool(self.text)
122

123
    def __len__(self) -> int:
124
        return self._len
125

126
    def __str__(self) -> str:
127
        # PERFORMANCE WARNING: This creates a copy of the string-slice
128 129
        if self._fullstring:  # optimization: avoid slicing/copying
            return self._fullstring
130 131
        # since the slice is being copyied now, anyway, the copy might
        # as well be stored in the string view
di68kap's avatar
di68kap committed
132
        # return self.text[self.begin:self.end]  # use this for debugging!
133 134
        self._fullstring = self._text[self._begin:self._end]
        return self._fullstring
135

136
    def __eq__(self, other) -> bool:
137 138
        # PERFORMANCE WARNING: This creates copies of the strings
        return len(other) == len(self) and str(self) == str(other)
139

140
    def __hash__(self) -> int:
141 142
        # PERFORMANCE WARNING: This creates a copy of the string-slice
        return hash(str(self))
143

144
    def __add__(self, other) -> Union[str, 'StringView']:
145
        if isinstance(other, str):
146
            return str(self) + other
147 148 149
        else:
            return StringView(str(self) + str(other))

150
    def __radd__(self, other) -> Union[str, 'StringView']:
151
        if isinstance(other, str):
152
            return other + str(self)
153 154 155
        else:
            return StringView(str(other) + str(self))

156
    @cython.locals(start=cython.int, end=cython.int)
157
    def __getitem__(self, index: Union[slice, int]) -> 'StringView':
158 159 160
        # assert isinstance(index, slice), "As of now, StringView only allows slicing."
        # assert index.step is None or index.step == 1, \
        #     "Step sizes other than 1 are not yet supported by StringView"
161
        try:
162 163
            start, stop = real_indices(index.start, index.stop, self._len)
            return StringView(self._text, self._begin + start, self._begin + stop)
164
        except AttributeError:
165 166 167 168 169 170 171 172 173 174 175 176
            return StringView(self._text, self._begin + index, self._begin + index + 1)

    def get_begin(self) -> int:
        """Returns the offset of the StringView. This is needed to correct
        the absolute offsets that the match objects of regular expression
        objects return.
        """
        return self._begin

    def get_text(self) -> str:
        """Returns the underlying string."""
        return self._text
177

178
    def count(self, sub: str, start: Optional[int] = None, end: Optional[int] = None) -> int:
179 180 181 182
        """Returns the number of non-overlapping occurrences of substring
        `sub` in StringView S[start:end].  Optional arguments start and end
        are interpreted as in slice notation.
        """
183 184 185 186 187
        if self._fullstring:
            if cython_optimized:
                return self._fullstring.count(sub, start or 0, self._len if end is None else end)
            else:
                return self._fullstring.count(sub, start, end)
188
        elif start is None and end is None:
189
            return self._text.count(sub, self._begin, self._end)
190
        else:
191 192
            start, end = real_indices(start, end, self._len)
            return self._text.count(sub, self._begin + start, self._begin + end)
193

194
    def find(self, sub: str, start: Optional[int] = None, end: Optional[int] = None) -> int:
195 196 197 198 199
        """Returns the lowest index in S where substring `sub` is found,
        such that `sub` is contained within S[start:end].  Optional
        arguments `start` and `end` are interpreted as in slice notation.
        Returns -1 on failure.
        """
200 201 202 203 204
        if self._fullstring:
            if cython_optimized:
                return self._fullstring.find(sub, start or 0, self._len if end is None else end)
            else:
                return self._fullstring.find(sub, start, end)
205
        elif start is None and end is None:
206
            return self._text.find(sub, self._begin, self._end) - self._begin
207
        else:
208 209
            start, end = real_indices(start, end, self._len)
            return self._text.find(sub, self._begin + start, self._begin + end) - self._begin
210

211
    def rfind(self, sub: str, start: Optional[int] = None, end: Optional[int] = None) -> int:
212 213 214 215 216
        """Returns the highest index in S where substring `sub` is found,
        such that `sub` is contained within S[start:end].  Optional
        arguments `start` and `end` are interpreted as in slice notation.
        Returns -1 on failure.
        """
217 218 219 220 221
        if self._fullstring:
            if cython_optimized:
                return self._fullstring.rfind(sub, start or 0, self._len if end is None else end)
            else:
                return self._fullstring.rfind(sub, start, end)
222
        if start is None and end is None:
223
            return self._text.rfind(sub, self._begin, self._end) - self._begin
224
        else:
225 226
            start, end = real_indices(start, end, self._len)
            return self._text.rfind(sub, self._begin + start, self._begin + end) - self._begin
227

228
    def startswith(self,
229
                   prefix: str,
230 231 232 233 234 235
                   start: int = 0,
                   end: Optional[int] = None) -> bool:
        """Return True if S starts with the specified prefix, False otherwise.
        With optional `start`, test S beginning at that position.
        With optional `end`, stop comparing S at that position.
        """
236 237 238
        start += self._begin
        end = self._end if end is None else self._begin + end
        return self._text.startswith(prefix, start, end)
239

240 241 242 243 244 245 246 247
    def endswith(self,
                 suffix: str,
                 start: int = 0,
                 end: Optional[int] = None) -> bool:
        """Return True if S ends with the specified suufix, False otherwise.
        With optional `start`, test S beginning at that position.
        With optional `end`, stop comparing S at that position.
        """
248 249 250
        start += self._begin
        end = self._end if end is None else self._begin + end
        return self._text.endswith(suffix, start, end)
251

252
    def match(self, regex, flags: int = 0):
253
        """Executes `regex.match` on the StringView object and returns the
eckhart's avatar
eckhart committed
254 255 256
        result, which is either a match-object or None. Keep in mind that
        match.end(), match.span() etc. are mapped to the underlying text,
        not the StringView-object!!!
257
        """
258
        return regex.match(self._text, pos=self._begin, endpos=self._end)
259 260

    def index(self, absolute_index: int) -> int:
eckhart's avatar
eckhart committed
261 262 263 264 265 266 267 268 269 270
        """Converts an index for a string watched by a StringView object
        to an index relative to the string view object, e.g.::

            >>> import re
            >>> sv = StringView('xxIxx')[2:3]
            >>> match = sv.match(re.compile('I'))
            >>> match.end()
            3
            >>> sv.index(match.end())
            1
271
        """
272
        return absolute_index - self._begin
273 274 275 276 277

    def indices(self, absolute_indices: Iterable[int]) -> Tuple[int, ...]:
        """Converts indices for a string watched by a StringView object
        to indices relative to the string view object. See also: `sv_index()`
        """
278
        return tuple(index - self._begin for index in absolute_indices)
279 280

    def search(self, regex):
281
        """Executes regex.search on the StringView object and returns the
eckhart's avatar
eckhart committed
282 283 284
        result, which is either a match-object or None. Keep in mind that
        match.end(), match.span() etc. are mapped to the underlying text,
        not the StringView-object!!!
285
        """
286
        return regex.search(self._text, pos=self._begin, endpos=self._end)
287

288 289
    def finditer(self, regex):
        """Executes regex.finditer on the StringView object and returns the
eckhart's avatar
eckhart committed
290 291
        iterator of match objects. Keep in mind that match.end(), match.span()
        etc. are mapped to the underlying text, not the StringView-object!!!
292
        """
293
        return regex.finditer(self._text, pos=self._begin, endpos=self._end)
294

295
    def strip(self):
296 297 298
        """Returns a copy of the StringView `self` with leading and trailing
        whitespace removed.
        """
299 300 301
        begin = first_char(self._text, self._begin, self._end) - self._begin
        end = last_char(self._text, self._begin, self._end) - self._begin
        return self if begin == 0 and end == self._len else self[begin:end]
302 303 304

    def lstrip(self):
        """Returns a copy of `self` with leading whitespace removed."""
305
        begin = first_char(self._text, self._begin, self._end) - self._begin
306 307 308 309
        return self if begin == 0 else self[begin:]

    def rstrip(self):
        """Returns a copy of `self` with trailing whitespace removed."""
310 311
        end = last_char(self._text, self._begin, self._end) - self._begin
        return self if end == self._len else self[:end]
312 313

    def split(self, sep=None):
314 315 316 317 318
        """Returns a list of the words in `self`, using `sep` as the
        delimiter string.  If `sep` is not specified or is None, any
        whitespace string is a separator and empty strings are
        removed from the result.
        """
319 320
        if self._fullstring:
            return self._fullstring.split(sep)
321 322
        else:
            pieces = []
eckhart's avatar
eckhart committed
323
            length = len(sep)
324 325 326
            k = 0
            i = self.find(sep, k)
            while i >= 0:
327
                pieces.append(self._text[self._begin + k: self._begin + i])
eckhart's avatar
eckhart committed
328
                k = i + length
329
                i = self.find(sep, k)
330
            pieces.append(self._text[self._begin + k: self._end])
331 332
            return pieces

333 334 335 336
    def replace(self, old, new):
        """Returns a string where `old` is replaced by `new`."""
        return str(self).replace(old, new)

337

338
EMPTY_STRING_VIEW = StringView('')