blob: 9d3f8abe712b96f492f49f6072eaa2b26a9b8ce0 [file] [log] [blame]
Tarek Ziade1231a4e2011-05-19 13:07:25 +02001"""Implementation of the Metadata for Python packages PEPs.
2
3Supports all metadata formats (1.0, 1.1, 1.2).
4"""
5
6import re
7import logging
8
9from io import StringIO
10from email import message_from_file
11from packaging import logger
12from packaging.markers import interpret
13from packaging.version import (is_valid_predicate, is_valid_version,
14 is_valid_versions)
15from packaging.errors import (MetadataMissingError,
16 MetadataConflictError,
17 MetadataUnrecognizedVersionError)
18
19try:
20 # docutils is installed
21 from docutils.utils import Reporter
22 from docutils.parsers.rst import Parser
23 from docutils import frontend
24 from docutils import nodes
25
26 class SilentReporter(Reporter):
27
28 def __init__(self, source, report_level, halt_level, stream=None,
29 debug=0, encoding='ascii', error_handler='replace'):
30 self.messages = []
Éric Araujo80223142011-10-14 17:04:39 +020031 super(SilentReporter, self).__init__(
32 source, report_level, halt_level, stream,
33 debug, encoding, error_handler)
Tarek Ziade1231a4e2011-05-19 13:07:25 +020034
35 def system_message(self, level, message, *children, **kwargs):
36 self.messages.append((level, message, children, kwargs))
37
38 _HAS_DOCUTILS = True
39except ImportError:
40 # docutils is not installed
41 _HAS_DOCUTILS = False
42
43# public API of this module
44__all__ = ['Metadata', 'PKG_INFO_ENCODING', 'PKG_INFO_PREFERRED_VERSION']
45
46# Encoding used for the PKG-INFO files
47PKG_INFO_ENCODING = 'utf-8'
48
49# preferred version. Hopefully will be changed
50# to 1.2 once PEP 345 is supported everywhere
51PKG_INFO_PREFERRED_VERSION = '1.0'
52
53_LINE_PREFIX = re.compile('\n \|')
54_241_FIELDS = ('Metadata-Version', 'Name', 'Version', 'Platform',
55 'Summary', 'Description',
56 'Keywords', 'Home-page', 'Author', 'Author-email',
57 'License')
58
59_314_FIELDS = ('Metadata-Version', 'Name', 'Version', 'Platform',
60 'Supported-Platform', 'Summary', 'Description',
61 'Keywords', 'Home-page', 'Author', 'Author-email',
62 'License', 'Classifier', 'Download-URL', 'Obsoletes',
63 'Provides', 'Requires')
64
Éric Araujoe6db7a32011-09-10 05:22:48 +020065_314_MARKERS = ('Obsoletes', 'Provides', 'Requires', 'Classifier',
66 'Download-URL')
Tarek Ziade1231a4e2011-05-19 13:07:25 +020067
68_345_FIELDS = ('Metadata-Version', 'Name', 'Version', 'Platform',
69 'Supported-Platform', 'Summary', 'Description',
70 'Keywords', 'Home-page', 'Author', 'Author-email',
71 'Maintainer', 'Maintainer-email', 'License',
72 'Classifier', 'Download-URL', 'Obsoletes-Dist',
73 'Project-URL', 'Provides-Dist', 'Requires-Dist',
74 'Requires-Python', 'Requires-External')
75
76_345_MARKERS = ('Provides-Dist', 'Requires-Dist', 'Requires-Python',
77 'Obsoletes-Dist', 'Requires-External', 'Maintainer',
78 'Maintainer-email', 'Project-URL')
79
80_ALL_FIELDS = set()
81_ALL_FIELDS.update(_241_FIELDS)
82_ALL_FIELDS.update(_314_FIELDS)
83_ALL_FIELDS.update(_345_FIELDS)
84
85
86def _version2fieldlist(version):
87 if version == '1.0':
88 return _241_FIELDS
89 elif version == '1.1':
90 return _314_FIELDS
91 elif version == '1.2':
92 return _345_FIELDS
93 raise MetadataUnrecognizedVersionError(version)
94
95
96def _best_version(fields):
97 """Detect the best version depending on the fields used."""
98 def _has_marker(keys, markers):
99 for marker in markers:
100 if marker in keys:
101 return True
102 return False
103
104 keys = list(fields)
105 possible_versions = ['1.0', '1.1', '1.2']
106
107 # first let's try to see if a field is not part of one of the version
108 for key in keys:
109 if key not in _241_FIELDS and '1.0' in possible_versions:
110 possible_versions.remove('1.0')
111 if key not in _314_FIELDS and '1.1' in possible_versions:
112 possible_versions.remove('1.1')
113 if key not in _345_FIELDS and '1.2' in possible_versions:
114 possible_versions.remove('1.2')
115
116 # possible_version contains qualified versions
117 if len(possible_versions) == 1:
118 return possible_versions[0] # found !
119 elif len(possible_versions) == 0:
120 raise MetadataConflictError('Unknown metadata set')
121
122 # let's see if one unique marker is found
123 is_1_1 = '1.1' in possible_versions and _has_marker(keys, _314_MARKERS)
124 is_1_2 = '1.2' in possible_versions and _has_marker(keys, _345_MARKERS)
125 if is_1_1 and is_1_2:
126 raise MetadataConflictError('You used incompatible 1.1 and 1.2 fields')
127
128 # we have the choice, either 1.0, or 1.2
129 # - 1.0 has a broken Summary field but works with all tools
130 # - 1.1 is to avoid
131 # - 1.2 fixes Summary but is not widespread yet
132 if not is_1_1 and not is_1_2:
133 # we couldn't find any specific marker
134 if PKG_INFO_PREFERRED_VERSION in possible_versions:
135 return PKG_INFO_PREFERRED_VERSION
136 if is_1_1:
137 return '1.1'
138
139 # default marker when 1.0 is disqualified
140 return '1.2'
141
142
143_ATTR2FIELD = {
144 'metadata_version': 'Metadata-Version',
145 'name': 'Name',
146 'version': 'Version',
147 'platform': 'Platform',
148 'supported_platform': 'Supported-Platform',
149 'summary': 'Summary',
150 'description': 'Description',
151 'keywords': 'Keywords',
152 'home_page': 'Home-page',
153 'author': 'Author',
154 'author_email': 'Author-email',
155 'maintainer': 'Maintainer',
156 'maintainer_email': 'Maintainer-email',
157 'license': 'License',
158 'classifier': 'Classifier',
159 'download_url': 'Download-URL',
160 'obsoletes_dist': 'Obsoletes-Dist',
161 'provides_dist': 'Provides-Dist',
162 'requires_dist': 'Requires-Dist',
163 'requires_python': 'Requires-Python',
164 'requires_external': 'Requires-External',
165 'requires': 'Requires',
166 'provides': 'Provides',
167 'obsoletes': 'Obsoletes',
168 'project_url': 'Project-URL',
169}
170
171_PREDICATE_FIELDS = ('Requires-Dist', 'Obsoletes-Dist', 'Provides-Dist')
172_VERSIONS_FIELDS = ('Requires-Python',)
173_VERSION_FIELDS = ('Version',)
174_LISTFIELDS = ('Platform', 'Classifier', 'Obsoletes',
175 'Requires', 'Provides', 'Obsoletes-Dist',
176 'Provides-Dist', 'Requires-Dist', 'Requires-External',
177 'Project-URL', 'Supported-Platform')
178_LISTTUPLEFIELDS = ('Project-URL',)
179
180_ELEMENTSFIELD = ('Keywords',)
181
182_UNICODEFIELDS = ('Author', 'Maintainer', 'Summary', 'Description')
183
184_MISSING = object()
185
Jeremy Klothaa2b4422011-09-12 11:12:42 -0600186_FILESAFE = re.compile('[^A-Za-z0-9.]+')
Tarek Ziade1231a4e2011-05-19 13:07:25 +0200187
Tarek Ziade1231a4e2011-05-19 13:07:25 +0200188class Metadata:
189 """The metadata of a release.
190
191 Supports versions 1.0, 1.1 and 1.2 (auto-detected). You can
192 instantiate the class with one of these arguments (or none):
193 - *path*, the path to a METADATA file
194 - *fileobj* give a file-like object with METADATA as content
195 - *mapping* is a dict-like object
196 """
197 # TODO document that execution_context and platform_dependent are used
198 # to filter on query, not when setting a key
199 # also document the mapping API and UNKNOWN default key
200
201 def __init__(self, path=None, platform_dependent=False,
202 execution_context=None, fileobj=None, mapping=None):
203 self._fields = {}
204 self.requires_files = []
205 self.docutils_support = _HAS_DOCUTILS
206 self.platform_dependent = platform_dependent
207 self.execution_context = execution_context
208 if [path, fileobj, mapping].count(None) < 2:
209 raise TypeError('path, fileobj and mapping are exclusive')
210 if path is not None:
211 self.read(path)
212 elif fileobj is not None:
213 self.read_file(fileobj)
214 elif mapping is not None:
215 self.update(mapping)
216
217 def _set_best_version(self):
218 self._fields['Metadata-Version'] = _best_version(self._fields)
219
220 def _write_field(self, file, name, value):
221 file.write('%s: %s\n' % (name, value))
222
223 def __getitem__(self, name):
224 return self.get(name)
225
226 def __setitem__(self, name, value):
227 return self.set(name, value)
228
229 def __delitem__(self, name):
230 field_name = self._convert_name(name)
231 try:
232 del self._fields[field_name]
233 except KeyError:
234 raise KeyError(name)
235 self._set_best_version()
236
237 def __contains__(self, name):
238 return (name in self._fields or
239 self._convert_name(name) in self._fields)
240
241 def _convert_name(self, name):
242 if name in _ALL_FIELDS:
243 return name
244 name = name.replace('-', '_').lower()
245 return _ATTR2FIELD.get(name, name)
246
247 def _default_value(self, name):
248 if name in _LISTFIELDS or name in _ELEMENTSFIELD:
249 return []
250 return 'UNKNOWN'
251
252 def _check_rst_data(self, data):
253 """Return warnings when the provided data has syntax errors."""
254 source_path = StringIO()
255 parser = Parser()
256 settings = frontend.OptionParser().get_default_values()
257 settings.tab_width = 4
258 settings.pep_references = None
259 settings.rfc_references = None
260 reporter = SilentReporter(source_path,
261 settings.report_level,
262 settings.halt_level,
263 stream=settings.warning_stream,
264 debug=settings.debug,
265 encoding=settings.error_encoding,
266 error_handler=settings.error_encoding_error_handler)
267
268 document = nodes.document(settings, reporter, source=source_path)
269 document.note_source(source_path, -1)
270 try:
271 parser.parse(data, document)
272 except AttributeError:
273 reporter.messages.append((-1, 'Could not finish the parsing.',
274 '', {}))
275
276 return reporter.messages
277
278 def _platform(self, value):
279 if not self.platform_dependent or ';' not in value:
280 return True, value
281 value, marker = value.split(';')
282 return interpret(marker, self.execution_context), value
283
284 def _remove_line_prefix(self, value):
285 return _LINE_PREFIX.sub('\n', value)
286
287 #
288 # Public API
289 #
Jeremy Klothaa2b4422011-09-12 11:12:42 -0600290 def get_fullname(self, filesafe=False):
291 """Return the distribution name with version.
292
293 If filesafe is true, return a filename-escaped form."""
294 name, version = self['Name'], self['Version']
295 if filesafe:
296 # For both name and version any runs of non-alphanumeric or '.'
297 # characters are replaced with a single '-'. Additionally any
298 # spaces in the version string become '.'
299 name = _FILESAFE.sub('-', name)
300 version = _FILESAFE.sub('-', version.replace(' ', '.'))
301 return '%s-%s' % (name, version)
Tarek Ziade1231a4e2011-05-19 13:07:25 +0200302
303 def is_metadata_field(self, name):
304 """return True if name is a valid metadata key"""
305 name = self._convert_name(name)
306 return name in _ALL_FIELDS
307
308 def is_multi_field(self, name):
309 name = self._convert_name(name)
310 return name in _LISTFIELDS
311
312 def read(self, filepath):
313 """Read the metadata values from a file path."""
Victor Stinnerc3364522011-05-19 18:49:56 +0200314 with open(filepath, 'r', encoding='utf-8') as fp:
Tarek Ziade1231a4e2011-05-19 13:07:25 +0200315 self.read_file(fp)
316
317 def read_file(self, fileob):
318 """Read the metadata values from a file object."""
319 msg = message_from_file(fileob)
320 self._fields['Metadata-Version'] = msg['metadata-version']
321
322 for field in _version2fieldlist(self['Metadata-Version']):
323 if field in _LISTFIELDS:
324 # we can have multiple lines
325 values = msg.get_all(field)
326 if field in _LISTTUPLEFIELDS and values is not None:
327 values = [tuple(value.split(',')) for value in values]
328 self.set(field, values)
329 else:
330 # single line
331 value = msg[field]
332 if value is not None and value != 'UNKNOWN':
333 self.set(field, value)
334
335 def write(self, filepath):
336 """Write the metadata fields to filepath."""
Victor Stinnerc3364522011-05-19 18:49:56 +0200337 with open(filepath, 'w', encoding='utf-8') as fp:
Tarek Ziade1231a4e2011-05-19 13:07:25 +0200338 self.write_file(fp)
339
340 def write_file(self, fileobject):
341 """Write the PKG-INFO format data to a file object."""
342 self._set_best_version()
343 for field in _version2fieldlist(self['Metadata-Version']):
344 values = self.get(field)
345 if field in _ELEMENTSFIELD:
346 self._write_field(fileobject, field, ','.join(values))
347 continue
348 if field not in _LISTFIELDS:
349 if field == 'Description':
350 values = values.replace('\n', '\n |')
351 values = [values]
352
353 if field in _LISTTUPLEFIELDS:
354 values = [','.join(value) for value in values]
355
356 for value in values:
357 self._write_field(fileobject, field, value)
358
359 def update(self, other=None, **kwargs):
360 """Set metadata values from the given iterable `other` and kwargs.
361
362 Behavior is like `dict.update`: If `other` has a ``keys`` method,
363 they are looped over and ``self[key]`` is assigned ``other[key]``.
364 Else, ``other`` is an iterable of ``(key, value)`` iterables.
365
366 Keys that don't match a metadata field or that have an empty value are
367 dropped.
368 """
Éric Araujo6bbd7752011-09-10 05:18:20 +0200369 # XXX the code should just use self.set, which does tbe same checks and
370 # conversions already, but that would break packaging.pypi: it uses the
371 # update method, which does not call _set_best_version (which set
372 # does), and thus allows having a Metadata object (as long as you don't
373 # modify or write it) with extra fields from PyPI that are not fields
374 # defined in Metadata PEPs. to solve it, the best_version system
375 # should be reworked so that it's called only for writing, or in a new
376 # strict mode, or with a new, more lax Metadata subclass in p7g.pypi
Tarek Ziade1231a4e2011-05-19 13:07:25 +0200377 def _set(key, value):
378 if key in _ATTR2FIELD and value:
379 self.set(self._convert_name(key), value)
380
Éric Araujo6bbd7752011-09-10 05:18:20 +0200381 if not other:
382 # other is None or empty container
Tarek Ziade1231a4e2011-05-19 13:07:25 +0200383 pass
384 elif hasattr(other, 'keys'):
385 for k in other.keys():
386 _set(k, other[k])
387 else:
388 for k, v in other:
389 _set(k, v)
390
391 if kwargs:
Éric Araujo6bbd7752011-09-10 05:18:20 +0200392 for k, v in kwargs.items():
393 _set(k, v)
Tarek Ziade1231a4e2011-05-19 13:07:25 +0200394
395 def set(self, name, value):
396 """Control then set a metadata field."""
397 name = self._convert_name(name)
398
399 if ((name in _ELEMENTSFIELD or name == 'Platform') and
400 not isinstance(value, (list, tuple))):
401 if isinstance(value, str):
402 value = [v.strip() for v in value.split(',')]
403 else:
404 value = []
405 elif (name in _LISTFIELDS and
406 not isinstance(value, (list, tuple))):
407 if isinstance(value, str):
408 value = [value]
409 else:
410 value = []
411
412 if logger.isEnabledFor(logging.WARNING):
Tarek Ziadeb9c09872011-05-30 12:25:38 +0200413 project_name = self['Name']
414
Tarek Ziade1231a4e2011-05-19 13:07:25 +0200415 if name in _PREDICATE_FIELDS and value is not None:
416 for v in value:
417 # check that the values are valid predicates
418 if not is_valid_predicate(v.split(';')[0]):
419 logger.warning(
Tarek Ziadeb9c09872011-05-30 12:25:38 +0200420 '%r: %r is not a valid predicate (field %r)',
421 project_name, v, name)
Tarek Ziade1231a4e2011-05-19 13:07:25 +0200422 # FIXME this rejects UNKNOWN, is that right?
423 elif name in _VERSIONS_FIELDS and value is not None:
424 if not is_valid_versions(value):
Tarek Ziadeb9c09872011-05-30 12:25:38 +0200425 logger.warning('%r: %r is not a valid version (field %r)',
426 project_name, value, name)
Tarek Ziade1231a4e2011-05-19 13:07:25 +0200427 elif name in _VERSION_FIELDS and value is not None:
428 if not is_valid_version(value):
Tarek Ziadeb9c09872011-05-30 12:25:38 +0200429 logger.warning('%r: %r is not a valid version (field %r)',
430 project_name, value, name)
Tarek Ziade1231a4e2011-05-19 13:07:25 +0200431
432 if name in _UNICODEFIELDS:
433 if name == 'Description':
434 value = self._remove_line_prefix(value)
435
436 self._fields[name] = value
437 self._set_best_version()
438
439 def get(self, name, default=_MISSING):
440 """Get a metadata field."""
441 name = self._convert_name(name)
442 if name not in self._fields:
443 if default is _MISSING:
444 default = self._default_value(name)
445 return default
446 if name in _UNICODEFIELDS:
447 value = self._fields[name]
448 return value
449 elif name in _LISTFIELDS:
450 value = self._fields[name]
451 if value is None:
452 return []
453 res = []
454 for val in value:
455 valid, val = self._platform(val)
456 if not valid:
457 continue
458 if name not in _LISTTUPLEFIELDS:
459 res.append(val)
460 else:
461 # That's for Project-URL
462 res.append((val[0], val[1]))
463 return res
464
465 elif name in _ELEMENTSFIELD:
466 valid, value = self._platform(self._fields[name])
467 if not valid:
468 return []
469 if isinstance(value, str):
470 return value.split(',')
471 valid, value = self._platform(self._fields[name])
472 if not valid:
473 return None
474 return value
475
476 def check(self, strict=False, restructuredtext=False):
477 """Check if the metadata is compliant. If strict is False then raise if
478 no Name or Version are provided"""
479 # XXX should check the versions (if the file was loaded)
480 missing, warnings = [], []
481
482 for attr in ('Name', 'Version'): # required by PEP 345
483 if attr not in self:
484 missing.append(attr)
485
486 if strict and missing != []:
487 msg = 'missing required metadata: %s' % ', '.join(missing)
488 raise MetadataMissingError(msg)
489
490 for attr in ('Home-page', 'Author'):
491 if attr not in self:
492 missing.append(attr)
493
494 if _HAS_DOCUTILS and restructuredtext:
495 warnings.extend(self._check_rst_data(self['Description']))
496
497 # checking metadata 1.2 (XXX needs to check 1.1, 1.0)
498 if self['Metadata-Version'] != '1.2':
499 return missing, warnings
500
501 def is_valid_predicates(value):
502 for v in value:
503 if not is_valid_predicate(v.split(';')[0]):
504 return False
505 return True
506
507 for fields, controller in ((_PREDICATE_FIELDS, is_valid_predicates),
508 (_VERSIONS_FIELDS, is_valid_versions),
509 (_VERSION_FIELDS, is_valid_version)):
510 for field in fields:
511 value = self.get(field, None)
512 if value is not None and not controller(value):
513 warnings.append('Wrong value for %r: %s' % (field, value))
514
515 return missing, warnings
516
517 def todict(self):
518 """Return fields as a dict.
519
520 Field names will be converted to use the underscore-lowercase style
521 instead of hyphen-mixed case (i.e. home_page instead of Home-page).
522 """
523 data = {
524 'metadata_version': self['Metadata-Version'],
525 'name': self['Name'],
526 'version': self['Version'],
527 'summary': self['Summary'],
528 'home_page': self['Home-page'],
529 'author': self['Author'],
530 'author_email': self['Author-email'],
531 'license': self['License'],
532 'description': self['Description'],
533 'keywords': self['Keywords'],
534 'platform': self['Platform'],
535 'classifier': self['Classifier'],
536 'download_url': self['Download-URL'],
537 }
538
539 if self['Metadata-Version'] == '1.2':
540 data['requires_dist'] = self['Requires-Dist']
541 data['requires_python'] = self['Requires-Python']
542 data['requires_external'] = self['Requires-External']
543 data['provides_dist'] = self['Provides-Dist']
544 data['obsoletes_dist'] = self['Obsoletes-Dist']
545 data['project_url'] = [','.join(url) for url in
546 self['Project-URL']]
547
548 elif self['Metadata-Version'] == '1.1':
549 data['provides'] = self['Provides']
550 data['requires'] = self['Requires']
551 data['obsoletes'] = self['Obsoletes']
552
553 return data
554
555 # Mapping API
Éric Araujo505f0eb2011-09-19 15:12:23 +0200556 # XXX these methods should return views or sets in 3.x
Tarek Ziade1231a4e2011-05-19 13:07:25 +0200557
558 def keys(self):
Éric Araujo505f0eb2011-09-19 15:12:23 +0200559 return list(_version2fieldlist(self['Metadata-Version']))
Tarek Ziade1231a4e2011-05-19 13:07:25 +0200560
561 def __iter__(self):
562 for key in self.keys():
563 yield key
564
565 def values(self):
Éric Araujo505f0eb2011-09-19 15:12:23 +0200566 return [self[key] for key in self.keys()]
Tarek Ziade1231a4e2011-05-19 13:07:25 +0200567
568 def items(self):
Éric Araujo505f0eb2011-09-19 15:12:23 +0200569 return [(key, self[key]) for key in self.keys()]