File: structuredfs.py

package info (click to toggle)
postgresql-multicorn 1.0.4-1
  • links: PTS, VCS
  • area: main
  • in suites: jessie, jessie-kfreebsd
  • size: 864 kB
  • ctags: 611
  • sloc: ansic: 2,690; python: 1,829; sql: 645; makefile: 93; sh: 29
file content (459 lines) | stat: -rw-r--r-- 16,120 bytes parent folder | download
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
"""

Handle nicely a set of files in a structured directory.

"""
import os
import sys
import io
import re
import errno
import string
import collections
import fcntl
from multicorn.compat import unicode_, basestring_

vformat = string.Formatter().vformat


try:
    str.isidentifier
except AttributeError:
    # Python 2
    # http://docs.python.org/py3k/reference/lexical_analysis.html#identifiers
    # the uppercase and lowercase letters A through Z, the underscore _
    # and, except for the first character, the digits 0 through 9.
    _IDENTIFIERS_RE = re.compile('^[a-zA-Z_][a-zA-Z_0-9]*$')

    def isidentifier(string):
        """
        Return whether the given string is a valid Python identifier.
        """
        return _IDENTIFIERS_RE.match(string) is not None
else:
    # Python 3
    def isidentifier(string):
        """
        Return whether the given string is a valid Python identifier.
        """
        return string.isidentifier()


def _tokenize_pattern(pattern):
    """
    Return an iterable of tokens from a string pattern.

    >>> list(_tokenize_pattern('{category}/{number}_{name}.txt'))
    [('property', 'category'),
     ('path separator', '/'),
     ('category', 'number'),
     ('literal', '_'),
     ('category', 'name'),
     ('path separator', '/')]

    """
    # We could re-purpose the parser for str.format() and use string.Formatter,
    # but we do not want to parse conversions and format specs.
    in_field = False
    field_name = None
    char_list = list(pattern)
    for prev_char, char, next_char in zip(
            [None] + char_list[:-1],
            char_list,
            char_list[1:] + [None]):
        if in_field:
            if char == '}':
                yield 'property', field_name
                field_name = None
                in_field = False
            else:
                field_name += char
        else:
            if char == '/':
                yield 'path separator', char
            elif char in '{}' and next_char == char:
                # Two brakets are parsed as one. Ignore the first one.
                pass
            elif char == '}' and prev_char != char:
                raise ValueError("Single '}' encountered in format string")
            elif char == '{' and prev_char != char:
                in_field = True
                field_name = ''
            else:
                # Includes normal chars but also an escaped bracket.
                yield 'literal', char
    if in_field:
        raise ValueError("Unmatched '{' in format string")

    # Artificially add this token to simplify the parser below
    yield 'path separator', '/'


def _parse_pattern(pattern):
    r"""
    Parse a string pattern and return (path_parts_re, path_parts_properties)

    >>> _parse_pattern('{category}/{number}_{name}.txt')
    (
        (
            <re object '^(?P<category>.*)$'>,
            <re object r'^(?P<number>.*)\_(?P<name>.*)\.txt$'>
        ), (
            ('category',)
            ('number', 'name')
        ),
    )
    """
    # A list of list of names
    path_parts_properties = []
    # The next list of names, being built
    properties = []
    # A set of all names
    all_properties = set()

    # A list of compiled re objects
    path_parts_re = []
    # The pattern being built for the next re
    next_re = ''

    for token_type, token in _tokenize_pattern(pattern):
        if token_type == 'path separator':
            if not next_re:
                raise ValueError('A slash-separated part is empty in %r' %
                                 pattern)
            path_parts_re.append(re.compile('^%s$' % next_re))
            next_re = ''
            path_parts_properties.append(tuple(properties))
            properties = []
        elif token_type == 'property':
            if not isidentifier(token):
                raise ValueError('Invalid property name for Filesystem: %r. '
                                 'Must be a valid identifier' % token)
            if token in all_properties:
                raise ValueError('Property name %r appears more than once '
                                 'in the pattern %r.' % (token, pattern))
            all_properties.add(token)
            properties.append(token)
            next_re += '(?P<%s>.*)' % token
        elif token_type == 'literal':
            next_re += re.escape(token)
        else:
            assert False, 'Unexpected token type: ' + token_type

    # Always end with an artificial '/' token so that the last regex is
    # in path_parts_re.
    assert token_type == 'path separator'

    return tuple(path_parts_re), tuple(path_parts_properties)


def strict_unicode(value):
    """
    Make sure that value is either unicode or (on Py 2.x) an ASCII string,
    and return it in unicode. Raise otherwise.
    """
    if not isinstance(value, basestring_):
        raise TypeError('Filename property values must be of type '
                        'unicode, got %r.' % value)
    return unicode_(value)


class Item(collections.Mapping):
    """
    Represents a single file in a :class:`StructuredDirectory`.

    Can be used as a mapping (dict-like) to access properties.

    Note that at a given point in time, the actual file for an Item may or
    may not exist in the filesystem.
    """
    def __init__(self, directory, properties, content=b''):
        properties = dict(properties)
        keys = set(properties)
        missing = directory.properties - keys
        if missing:
            raise ValueError('Missing properties: %s', ', '.join(missing))
        extra = keys - directory.properties
        if extra:
            raise ValueError('Unknown properties: %s', ', '.join(extra))
        self.directory = directory
        self._properties = {}
        self.content = content
        # TODO: check for ambiguities.
        # eg. with pattern = '{a}_{b}', values {'a': '1_2', 'b': '3'} and
        # {'a': '1', 'b': '2_3'} both give the same filename.
        for name, value in properties.items():
            value = strict_unicode(value)
            if '/' in value:
                raise ValueError('Property values can not contain a slash.')
            self._properties[name] = value

    @property
    def filename(self):
        """
        Return the normalized (slash-separated) filename for the item,
        relative to the root.
        """
        return vformat(self.directory.pattern, [], self)

    @property
    def full_filename(self):
        """
        Return the absolute filename for the item, in OS-specific form.
        """
        return self.directory._join(self.filename.split('/'))

    def open(self, shared_lock=True, fail_if=None):
        """Open the file underlying this item, if it is not in the cache.
        Shared_lock is a boolean indicating whether a shared or exclusive lock
        should be acquired.
        fail_if can be either None, "exists", or "missing".
        """
        self._fd, is_shared = self.directory.cache.get(self.full_filename,
                                                       (None, False))
        if shared_lock:
            if self._fd is None:
                # Open it with a shared lock
                self._fd = os.open(self.full_filename,
                                   os.O_RDONLY | os.O_SYNC)
                fcntl.flock(self._fd, fcntl.LOCK_SH)
                self.directory.cache[self.full_filename] = (self._fd,
                                                            shared_lock)
            # Do nothing if we already have a file descriptor
        else:
            if (self._fd is None or
                    not  (fcntl.fcntl(self._fd, fcntl.F_GETFL) & os.O_RDWR)):
                # Open it with an exclusive lock, sync mode, and fail if the
                # file already exists.
                dirname = os.path.dirname(self.full_filename)
                if not os.path.exists(dirname):
                    umask = os.umask(0)
                    os.makedirs(dirname, self.directory.file_mode)
                    os.umask(umask)
                flags = os.O_SYNC | os.O_RDWR
                if fail_if == 'exists':
                    flags = flags | os.O_CREAT | os.O_EXCL
                elif fail_if is None:
                    flags = flags | os.O_CREAT
                if self._fd is not None:
                    os.close(self._fd)

                umask = os.umask(0)
                self._fd = os.open(self.full_filename, flags,
                                   self.directory.file_mode)
                os.umask(umask)
            fcntl.flock(self._fd, fcntl.LOCK_EX)
            self.directory.cache[self.full_filename] = (self._fd, shared_lock)
        return self._fd

    def read(self):
        """
        Return the content of the file as a bytestring.

        :raises IOError: if the file does not exist in the filesystem.
        """
        fd = self.open(True, fail_if='missing')
        os.lseek(fd, 0, os.SEEK_SET)
        iof = io.open(fd, 'rb', closefd=False)
        content = iof.read()
        iof.close()
        return content

    def write(self, fd=None):
        if fd is None:
            fd = self.open(False)
        os.ftruncate(fd, 0)
        os.lseek(fd, 0, os.SEEK_SET)
        # Do not use a buffer, to ensure that the file is written in one
        # syscall.
        iof = io.open(fd, 'wb', closefd=False,
                      buffering=0)

        if isinstance(self.content, unicode_):
            self.content = self.content.encode(sys.getfilesystemencoding())

        if self.content is not None:
            iof.write(self.content)
        iof.close()

    def remove(self):
        os.unlink(self.full_filename)

    # collections.Mapping interface:

    def __len__(self):
        return len(self._properties)

    def __iter__(self):
        return iter(self._properties)

    def __getitem__(self, name):
        return self._properties[name]

    def __setitem__(self, name, value):
        self._properties[name] = value


class StructuredDirectory(object):
    """
    :param root_dir: Path to the root directory
    :param pattern: Pattern for files in this directory,
                    eg. '{category}/{number}_{name}.txt'
    """
    def __init__(self, root_dir, pattern, file_mode=0o700):
        self.root_dir = unicode_(root_dir)
        self.pattern = unicode_(pattern)
        # Cache for file descriptors.
        self.cache = {}
        parts_re, parts_properties = _parse_pattern(self.pattern)
        self.file_mode = file_mode
        self._path_parts_re = parts_re
        self._path_parts_properties = parts_properties
        self.properties = set(prop for part in parts_properties
                              for prop in part)

    def create(self, **values):
        """
        Return a new ``Item`` associated to this directory with the given
        ``values``.

        The file for this item may or may not exist.

        """
        return Item(self, values)

    def from_filename(self, filename):
        """
        Return an ``Item`` from a slash-separated ``filename`` relative
        to the root. Return ``None`` if ``filename`` does not match
        ``pattern``.

        Assuming a matching filename::

            f = 'a/b/c'
            directory.from_filename(f).filename == f

        The file for this item may or may not exist.

        """
        values = {}
        parts = filename.split('/')
        if len(parts) != len(self._path_parts_re):
            return None
        for part, part_re in zip(parts, self._path_parts_re):
            match = part_re.match(part)
            if match is None:
                return None
            values.update(match.groupdict())
        return Item(self, values)

    def get_items(self, **fixed_values):
        """
        Return an iterable of :class:`Item` objects for all files
        that match the given ``properties``.
        """
        keys = set(fixed_values)
        extra = keys - self.properties
        if extra:
            raise ValueError('Unknown properties: %s', ', '.join(extra))

        # Pre-compute everything we know about the request without looking
        # at the filesystem.

        # `fixed` is a list, one element for each "part" of the pattern.
        # Each element is a (fixed_part, fixed_part_values) tuple.
        #    fixed_part: the whole part if it is completly fixed, or None
        #    fixed_part_values: (name, value) pairs for$ fixed values
        #                       for this part.
        fixed = []
        for pattern_part, part_properties in zip(
                self.pattern.split('/'), self._path_parts_properties):
            fixed_part_values = tuple(
                (name, fixed_values[name]) for name in part_properties
                if name in fixed_values
            )
            if len(fixed_part_values) == len(part_properties):
                # All properties for this part are fixed
                fixed_part = vformat(pattern_part, [], dict(fixed_part_values))
            else:
                fixed_part = None
            fixed.append((fixed_part, fixed_part_values))

        return self._walk((), (), fixed)

    def clear_cache_entry(self, key):
        value, shared = self.cache.pop(key)
        os.close(value)

    def clear_cache(self, only_shared=False):
        for key, (value, shared) in list(self.cache.items()):
            if (not only_shared) or shared:
                self.clear_cache_entry(key)

    def _walk(self, previous_path_parts, previous_values, fixed):
        """
        Called for each directory or sub-directory.
        """
        # Empty previous_path_parts means look in root_dir, depth = 0
        depth = len(previous_path_parts)
        # If the pattern has N path parts, "leaf" files are at depth = N-1
        is_leaf = (depth == len(self._path_parts_re) - 1)

        for name, part_values in self._find_matching_names(
                previous_path_parts, fixed):
            path_parts = previous_path_parts + (name,)
            values = previous_values + tuple(part_values)
            filename = self._join(path_parts)
            if is_leaf:
                if os.path.isfile(filename):
                    yield Item(self, values)
            # Do not check if filename is a directory or even exists,
            # let listdir() raise later.
            else:
                for item in self._walk(path_parts, values, fixed):
                    yield item

    def _find_matching_names(self, previous_path_parts, fixed):
        """
        Yield names and parsed values that match the request in a directory.
        """
        depth = len(previous_path_parts)
        fixed_part, fixed_part_values = fixed[depth]
        if fixed_part is not None:
            yield fixed_part, fixed_part_values
            return

        try:
            names = self._listdir(previous_path_parts)
        except OSError as exc:
            if depth > 0 and exc.errno in [errno.ENOENT, errno.ENOTDIR]:
                # Does not exist or is not a directory, just return
                # without yielding any name.
                # If depth == 0, we're listing the root directory. Still raise
                # in that case.
                return
            else:
                # Re-raise other errors
                raise
        for name in names:
            match = self._path_parts_re[depth].match(name)
            if match is None:
                continue

            part_values = match.groupdict()
            if all(part_values[name] == value
                   for name, value in fixed_part_values):
                yield name, list(part_values.items())

    def _join(self, path_parts):
        """
        Return a full filesystem path from parts relative to the root.
        """
        # root_dir is unicode, so the join result should be unicode
        return os.path.join(self.root_dir, *path_parts)

    def _listdir(self, path_parts):
        """
        Wrap os.listdir to make it monkey-patchable in tests.
        """
        return os.listdir(self._join(path_parts))