Special-case native Java version parsing
[zeroinstall/solver.git] / zeroinstall / injector / distro.py
blob252886e443ba9c662537b3c039a23465d5e66fce
1 """
2 Integration with native distribution package managers.
3 @since: 0.28
4 """
6 # Copyright (C) 2009, Thomas Leonard
7 # See the README file for details, or visit http://0install.net.
9 from zeroinstall import _
10 import os, platform, re, subprocess, sys
11 from logging import warn, info
12 from zeroinstall.injector import namespaces, model, arch
13 from zeroinstall.support import basedir
15 _dotted_ints = '[0-9]+(?:\.[0-9]+)*'
17 # This matches a version number that would be a valid Zero Install version without modification
18 _zeroinstall_regexp = '(?:%s)(?:-(?:pre|rc|post|)(?:%s))*' % (_dotted_ints, _dotted_ints)
20 # This matches the interesting bits of distribution version numbers
21 # (first bit is for Java-style 6b17 syntax)
22 _version_regexp = '({ints}b)?({zero})(-r{ints})?'.format(zero = _zeroinstall_regexp, ints = _dotted_ints)
24 # We try to do updates atomically without locking, but we don't worry too much about
25 # duplicate entries or being a little out of sync with the on-disk copy.
26 class Cache(object):
27 def __init__(self, cache_leaf, source, format):
28 """Maintain a cache file (e.g. ~/.cache/0install.net/injector/$name).
29 If the size or mtime of $source has changed, or the cache
30 format version if different, reset the cache first."""
31 self.cache_leaf = cache_leaf
32 self.source = source
33 self.format = format
34 self.cache_dir = basedir.save_cache_path(namespaces.config_site,
35 namespaces.config_prog)
36 self.cached_for = {} # Attributes of source when cache was created
37 try:
38 self._load_cache()
39 except Exception as ex:
40 info(_("Failed to load cache (%s). Flushing..."), ex)
41 self.flush()
43 def flush(self):
44 # Wipe the cache
45 try:
46 info = os.stat(self.source)
47 mtime = int(info.st_mtime)
48 size = info.st_size
49 except Exception as ex:
50 warn("Failed to stat %s: %s", self.source, ex)
51 mtime = size = 0
52 self.cache = {}
53 import tempfile
54 tmp, tmp_name = tempfile.mkstemp(dir = self.cache_dir)
55 data = "mtime=%d\nsize=%d\nformat=%d\n\n" % (mtime, size, self.format)
56 while data:
57 wrote = os.write(tmp, data)
58 data = data[wrote:]
59 os.rename(tmp_name, os.path.join(self.cache_dir, self.cache_leaf))
61 self._load_cache()
63 # Populate self.cache from our saved cache file.
64 # Throws an exception if the cache doesn't exist or has the wrong format.
65 def _load_cache(self):
66 self.cache = cache = {}
67 stream = file(os.path.join(self.cache_dir, self.cache_leaf))
68 try:
69 for line in stream:
70 line = line.strip()
71 if not line:
72 break
73 key, value = line.split('=', 1)
74 if key in ('mtime', 'size', 'format'):
75 self.cached_for[key] = int(value)
77 self._check_valid()
79 for line in stream:
80 key, value = line.split('=', 1)
81 cache[key] = value[:-1]
82 finally:
83 stream.close()
85 # Check the source file hasn't changed since we created the cache
86 def _check_valid(self):
87 info = os.stat(self.source)
88 if self.cached_for['mtime'] != int(info.st_mtime):
89 raise Exception("Modification time of %s has changed" % self.source)
90 if self.cached_for['size'] != info.st_size:
91 raise Exception("Size of %s has changed" % self.source)
92 if self.cached_for.get('format', None) != self.format:
93 raise Exception("Format of cache has changed")
95 def get(self, key):
96 try:
97 self._check_valid()
98 except Exception as ex:
99 info(_("Cache needs to be refreshed: %s"), ex)
100 self.flush()
101 return None
102 else:
103 return self.cache.get(key, None)
105 def put(self, key, value):
106 cache_path = os.path.join(self.cache_dir, self.cache_leaf)
107 self.cache[key] = value
108 try:
109 stream = file(cache_path, 'a')
110 try:
111 stream.write('%s=%s\n' % (key, value))
112 finally:
113 stream.close()
114 except Exception as ex:
115 warn("Failed to write to cache %s: %s=%s: %s", cache_path, key, value, ex)
117 def try_cleanup_distro_version(version):
118 """Try to turn a distribution version string into one readable by Zero Install.
119 We do this by stripping off anything we can't parse.
120 @return: the part we understood, or None if we couldn't parse anything
121 @rtype: str"""
122 if ':' in version:
123 version = version.split(':')[1] # Skip 'epoch'
124 version = version.replace('_', '-')
125 match = re.match(_version_regexp, version)
126 if match:
127 major, version, revision = match.groups()
128 if major is not None:
129 version = major[:-1] + '.' + version
130 if revision is None:
131 return version
132 else:
133 return '%s-%s' % (version, revision[2:])
134 return None
136 class Distribution(object):
137 """Represents a distribution with which we can integrate.
138 Sub-classes should specialise this to integrate with the package managers of
139 particular distributions. This base class ignores the native package manager.
140 @since: 0.28
142 _packagekit = None
144 def get_package_info(self, package, factory):
145 """Get information about the given package.
146 Add zero or more implementations using the factory (typically at most two
147 will be added; the currently installed version and the latest available).
148 @param package: package name (e.g. "gimp")
149 @type package: str
150 @param factory: function for creating new DistributionImplementation objects from IDs
151 @type factory: str -> L{model.DistributionImplementation}
153 return
155 def get_score(self, distribution):
156 """Indicate how closely the host distribution matches this one.
157 The <package-implementation> with the highest score is passed
158 to L{Distribution.get_package_info}. If several elements get
159 the same score, get_package_info is called for all of them.
160 @param distribution: a distribution name
161 @type distribution: str
162 @return: an integer, or None if there is no match at all
163 @rtype: int | None
165 return 0
167 def get_feed(self, master_feed):
168 """Generate a feed containing information about distribution packages.
169 This should immediately return a feed containing an implementation for the
170 package if it's already installed. Information about versions that could be
171 installed using the distribution's package manager can be added asynchronously
172 later (see L{fetch_candidates}).
173 @param master_feed: feed containing the <package-implementation> elements
174 @type master_feed: L{model.ZeroInstallFeed}
175 @rtype: L{model.ZeroInstallFeed}"""
177 feed = model.ZeroInstallFeed(None)
178 feed.url = 'distribution:' + master_feed.url
180 for item, item_attrs in master_feed.get_package_impls(self):
181 package = item_attrs.get('package', None)
182 if package is None:
183 raise model.InvalidInterface(_("Missing 'package' attribute on %s") % item)
185 def factory(id, only_if_missing = False, installed = True):
186 assert id.startswith('package:')
187 if id in feed.implementations:
188 if only_if_missing:
189 return None
190 warn(_("Duplicate ID '%s' for DistributionImplementation"), id)
191 impl = model.DistributionImplementation(feed, id, self)
192 feed.implementations[id] = impl
194 impl.installed = installed
195 impl.metadata = item_attrs
197 item_main = item_attrs.get('main', None)
198 if item_main and not item_main.startswith('/'):
199 raise model.InvalidInterface(_("'main' attribute must be absolute, but '%s' doesn't start with '/'!") %
200 item_main)
201 impl.main = item_main
202 impl.upstream_stability = model.packaged
204 return impl
206 self.get_package_info(package, factory)
208 if master_feed.url == 'http://repo.roscidus.com/python/python' and all(not impl.installed for impl in feed.implementations.values()):
209 # Hack: we can support Python on platforms with unsupported package managers
210 # by adding the implementation of Python running us now to the list.
211 python_version = '.'.join([str(v) for v in sys.version_info if isinstance(v, int)])
212 impl_id = 'package:host:python:' + python_version
213 assert impl_id not in feed.implementations
214 impl = model.DistributionImplementation(feed, impl_id, self)
215 impl.installed = True
216 impl.version = model.parse_version(python_version)
217 impl.main = sys.executable
218 impl.upstream_stability = model.packaged
219 impl.machine = host_machine # (hopefully)
220 feed.implementations[impl_id] = impl
222 return feed
224 def fetch_candidates(self, master_feed):
225 """Collect information about versions we could install using
226 the distribution's package manager. On success, the distribution
227 feed in iface_cache is updated.
228 @return: a L{tasks.Blocker} if the task is in progress, or None if not"""
229 if self.packagekit.available:
230 package_names = [item.getAttribute("package") for item, item_attrs in master_feed.get_package_impls(self)]
231 return self.packagekit.fetch_candidates(package_names)
233 @property
234 def packagekit(self):
235 """For use by subclasses.
236 @rtype: L{packagekit.PackageKit}"""
237 if not self._packagekit:
238 from zeroinstall.injector import packagekit
239 self._packagekit = packagekit.PackageKit()
240 return self._packagekit
242 class CachedDistribution(Distribution):
243 """For distributions where querying the package database is slow (e.g. requires running
244 an external command), we cache the results.
245 @since: 0.39
246 @deprecated: use Cache instead
249 def __init__(self, db_status_file):
250 """@param db_status_file: update the cache when the timestamp of this file changes"""
251 self._status_details = os.stat(db_status_file)
253 self.versions = {}
254 self.cache_dir = basedir.save_cache_path(namespaces.config_site,
255 namespaces.config_prog)
257 try:
258 self._load_cache()
259 except Exception as ex:
260 info(_("Failed to load distribution database cache (%s). Regenerating..."), ex)
261 try:
262 self.generate_cache()
263 self._load_cache()
264 except Exception as ex:
265 warn(_("Failed to regenerate distribution database cache: %s"), ex)
267 def _load_cache(self):
268 """Load {cache_leaf} cache file into self.versions if it is available and up-to-date.
269 Throws an exception if the cache should be (re)created."""
270 stream = file(os.path.join(self.cache_dir, self.cache_leaf))
272 cache_version = None
273 for line in stream:
274 if line == '\n':
275 break
276 name, value = line.split(': ')
277 if name == 'mtime' and int(value) != int(self._status_details.st_mtime):
278 raise Exception(_("Modification time of package database file has changed"))
279 if name == 'size' and int(value) != self._status_details.st_size:
280 raise Exception(_("Size of package database file has changed"))
281 if name == 'version':
282 cache_version = int(value)
283 else:
284 raise Exception(_('Invalid cache format (bad header)'))
286 if cache_version is None:
287 raise Exception(_('Old cache format'))
289 versions = self.versions
290 for line in stream:
291 package, version, zi_arch = line[:-1].split('\t')
292 versionarch = (version, intern(zi_arch))
293 if package not in versions:
294 versions[package] = [versionarch]
295 else:
296 versions[package].append(versionarch)
298 def _write_cache(self, cache):
299 #cache.sort() # Might be useful later; currently we don't care
300 import tempfile
301 fd, tmpname = tempfile.mkstemp(prefix = 'zeroinstall-cache-tmp',
302 dir = self.cache_dir)
303 try:
304 stream = os.fdopen(fd, 'wb')
305 stream.write('version: 2\n')
306 stream.write('mtime: %d\n' % int(self._status_details.st_mtime))
307 stream.write('size: %d\n' % self._status_details.st_size)
308 stream.write('\n')
309 for line in cache:
310 stream.write(line + '\n')
311 stream.close()
313 os.rename(tmpname,
314 os.path.join(self.cache_dir,
315 self.cache_leaf))
316 except:
317 os.unlink(tmpname)
318 raise
320 # Maps machine type names used in packages to their Zero Install versions
321 _canonical_machine = {
322 'all' : '*',
323 'any' : '*',
324 'noarch' : '*',
325 '(none)' : '*',
326 'amd64': 'x86_64',
327 'i386': 'i386',
328 'i486': 'i486',
329 'i586': 'i586',
330 'i686': 'i686',
331 'ppc64': 'ppc64',
332 'ppc': 'ppc',
335 host_machine = arch.canonicalize_machine(platform.uname()[4])
336 def canonical_machine(package_machine):
337 machine = _canonical_machine.get(package_machine, None)
338 if machine is None:
339 # Safe default if we can't understand the arch
340 return host_machine
341 return machine
343 class DebianDistribution(Distribution):
344 """A dpkg-based distribution."""
346 cache_leaf = 'dpkg-status.cache'
348 def __init__(self, dpkg_status, pkgcache):
349 self.dpkg_cache = Cache('dpkg-status.cache', dpkg_status, 2)
350 self.apt_cache = {}
352 def _query_installed_package(self, package):
353 null = os.open('/dev/null', os.O_WRONLY)
354 child = subprocess.Popen(["dpkg-query", "-W", "--showformat=${Version}\t${Architecture}\t${Status}\n", "--", package],
355 stdout = subprocess.PIPE, stderr = null)
356 os.close(null)
357 stdout, stderr = child.communicate()
358 child.wait()
359 for line in stdout.split('\n'):
360 if not line: continue
361 version, debarch, status = line.split('\t', 2)
362 if not status.endswith(' installed'): continue
363 clean_version = try_cleanup_distro_version(version)
364 if clean_version:
365 return '%s\t%s' % (clean_version, canonical_machine(debarch.strip()))
366 else:
367 warn(_("Can't parse distribution version '%(version)s' for package '%(package)s'"), {'version': version, 'package': package})
369 return '-'
371 def get_package_info(self, package, factory):
372 # Add any already-installed package...
373 installed_cached_info = self._get_dpkg_info(package)
375 if installed_cached_info != '-':
376 installed_version, machine = installed_cached_info.split('\t')
377 impl = factory('package:deb:%s:%s:%s' % (package, installed_version, machine))
378 impl.version = model.parse_version(installed_version)
379 if machine != '*':
380 impl.machine = machine
381 else:
382 installed_version = None
384 # Add any uninstalled candidates (note: only one of these two methods will add anything)
386 # From PackageKit...
387 self.packagekit.get_candidates(package, factory, 'package:deb')
389 # From apt-cache...
390 cached = self.apt_cache.get(package, None)
391 if cached:
392 candidate_version = cached['version']
393 candidate_arch = cached['arch']
394 if candidate_version and candidate_version != installed_version:
395 impl = factory('package:deb:%s:%s:%s' % (package, candidate_version, candidate_arch), installed = False)
396 impl.version = model.parse_version(candidate_version)
397 if candidate_arch != '*':
398 impl.machine = candidate_arch
399 def install(handler):
400 raise model.SafeException(_("This program depends on '%s', which is a package that is available through your distribution. "
401 "Please install it manually using your distribution's tools and try again.") % package)
402 impl.download_sources.append(model.DistributionSource(package, cached['size'], install, needs_confirmation = False))
404 def get_score(self, disto_name):
405 return int(disto_name == 'Debian')
407 def _get_dpkg_info(self, package):
408 installed_cached_info = self.dpkg_cache.get(package)
409 if installed_cached_info == None:
410 installed_cached_info = self._query_installed_package(package)
411 self.dpkg_cache.put(package, installed_cached_info)
413 return installed_cached_info
415 def fetch_candidates(self, master_feed):
416 package_names = [item.getAttribute("package") for item, item_attrs in master_feed.get_package_impls(self)]
418 if self.packagekit.available:
419 return self.packagekit.fetch_candidates(package_names)
421 # No PackageKit. Use apt-cache directly.
422 for package in package_names:
423 # Check to see whether we could get a newer version using apt-get
424 try:
425 null = os.open('/dev/null', os.O_WRONLY)
426 child = subprocess.Popen(['apt-cache', 'show', '--no-all-versions', '--', package], stdout = subprocess.PIPE, stderr = null)
427 os.close(null)
429 arch = version = size = None
430 for line in child.stdout:
431 line = line.strip()
432 if line.startswith('Version: '):
433 version = line[9:]
434 version = try_cleanup_distro_version(version)
435 elif line.startswith('Architecture: '):
436 arch = canonical_machine(line[14:].strip())
437 elif line.startswith('Size: '):
438 size = int(line[6:].strip())
439 if version and arch:
440 cached = {'version': version, 'arch': arch, 'size': size}
441 else:
442 cached = None
443 child.wait()
444 except Exception as ex:
445 warn("'apt-cache show %s' failed: %s", package, ex)
446 cached = None
447 # (multi-arch support? can there be multiple candidates?)
448 self.apt_cache[package] = cached
450 class RPMDistribution(CachedDistribution):
451 """An RPM-based distribution."""
453 cache_leaf = 'rpm-status.cache'
455 def generate_cache(self):
456 cache = []
458 for line in os.popen("rpm -qa --qf='%{NAME}\t%{VERSION}-%{RELEASE}\t%{ARCH}\n'"):
459 package, version, rpmarch = line.split('\t', 2)
460 if package == 'gpg-pubkey':
461 continue
462 zi_arch = canonical_machine(rpmarch.strip())
463 clean_version = try_cleanup_distro_version(version)
464 if clean_version:
465 cache.append('%s\t%s\t%s' % (package, clean_version, zi_arch))
466 else:
467 warn(_("Can't parse distribution version '%(version)s' for package '%(package)s'"), {'version': version, 'package': package})
469 self._write_cache(cache)
471 def get_package_info(self, package, factory):
472 # Add installed versions...
473 versions = self.versions.get(package, [])
475 for version, machine in versions:
476 impl = factory('package:rpm:%s:%s:%s' % (package, version, machine))
477 impl.version = model.parse_version(version)
478 if machine != '*':
479 impl.machine = machine
481 # Add any uninstalled candidates found by PackageKit
482 self.packagekit.get_candidates(package, factory, 'package:rpm')
484 def get_score(self, disto_name):
485 return int(disto_name == 'RPM')
487 class SlackDistribution(Distribution):
488 """A Slack-based distribution."""
490 def __init__(self, packages_dir):
491 self._packages_dir = packages_dir
493 def get_package_info(self, package, factory):
494 # Add installed versions...
495 for entry in os.listdir(self._packages_dir):
496 name, version, arch, build = entry.rsplit('-', 3)
497 if name == package:
498 zi_arch = canonical_machine(arch)
499 clean_version = try_cleanup_distro_version("%s-%s" % (version, build))
500 if not clean_version:
501 warn(_("Can't parse distribution version '%(version)s' for package '%(package)s'"), {'version': version, 'package': name})
502 continue
504 impl = factory('package:slack:%s:%s:%s' % \
505 (package, clean_version, zi_arch))
506 impl.version = model.parse_version(clean_version)
507 if zi_arch != '*':
508 impl.machine = zi_arch
510 # Add any uninstalled candidates found by PackageKit
511 self.packagekit.get_candidates(package, factory, 'package:slack')
513 def get_score(self, disto_name):
514 return int(disto_name == 'Slack')
516 class GentooDistribution(Distribution):
518 def __init__(self, pkgdir):
519 self._pkgdir = pkgdir
521 def get_package_info(self, package, factory):
522 # Add installed versions...
523 _version_start_reqexp = '-[0-9]'
525 if package.count('/') != 1: return
527 category, leafname = package.split('/')
528 category_dir = os.path.join(self._pkgdir, category)
529 match_prefix = leafname + '-'
531 if not os.path.isdir(category_dir): return
533 for filename in os.listdir(category_dir):
534 if filename.startswith(match_prefix) and filename[len(match_prefix)].isdigit():
535 name = file(os.path.join(category_dir, filename, 'PF')).readline().strip()
537 match = re.search(_version_start_reqexp, name)
538 if match is None:
539 warn(_('Cannot parse version from Gentoo package named "%(name)s"'), {'name': name})
540 continue
541 else:
542 version = try_cleanup_distro_version(name[match.start() + 1:])
544 if category == 'app-emulation' and name.startswith('emul-'):
545 __, __, machine, __ = name.split('-', 3)
546 else:
547 machine, __ = file(os.path.join(category_dir, filename, 'CHOST')).readline().split('-', 1)
548 machine = arch.canonicalize_machine(machine)
550 impl = factory('package:gentoo:%s:%s:%s' % \
551 (package, version, machine))
552 impl.version = model.parse_version(version)
553 impl.machine = machine
555 # Add any uninstalled candidates found by PackageKit
556 self.packagekit.get_candidates(package, factory, 'package:gentoo')
558 def get_score(self, disto_name):
559 return int(disto_name == 'Gentoo')
561 class PortsDistribution(Distribution):
563 def __init__(self, pkgdir):
564 self._pkgdir = pkgdir
566 def get_package_info(self, package, factory):
567 _name_version_regexp = '^(.+)-([^-]+)$'
569 nameversion = re.compile(_name_version_regexp)
570 for pkgname in os.listdir(self._pkgdir):
571 pkgdir = os.path.join(self._pkgdir, pkgname)
572 if not os.path.isdir(pkgdir): continue
574 #contents = file(os.path.join(pkgdir, '+CONTENTS')).readline().strip()
576 match = nameversion.search(pkgname)
577 if match is None:
578 warn(_('Cannot parse version from Ports package named "%(pkgname)s"'), {'pkgname': pkgname})
579 continue
580 else:
581 name = match.group(1)
582 if name != package:
583 continue
584 version = try_cleanup_distro_version(match.group(2))
586 machine = host_machine
588 impl = factory('package:ports:%s:%s:%s' % \
589 (package, version, machine))
590 impl.version = model.parse_version(version)
591 impl.machine = machine
593 def get_score(self, disto_name):
594 return int(disto_name == 'Ports')
596 _host_distribution = None
597 def get_host_distribution():
598 """Get a Distribution suitable for the host operating system.
599 Calling this twice will return the same object.
600 @rtype: L{Distribution}"""
601 global _host_distribution
602 if not _host_distribution:
603 dpkg_db_status = '/var/lib/dpkg/status'
604 pkgcache = '/var/cache/apt/pkgcache.bin'
605 _rpm_db = '/var/lib/rpm/Packages'
606 _slack_db = '/var/log/packages'
607 _pkg_db = '/var/db/pkg'
609 if os.path.isdir(_pkg_db):
610 if sys.platform.startswith("linux"):
611 _host_distribution = GentooDistribution(_pkg_db)
612 elif sys.platform.startswith("freebsd"):
613 _host_distribution = PortsDistribution(_pkg_db)
614 elif os.access(dpkg_db_status, os.R_OK):
615 _host_distribution = DebianDistribution(dpkg_db_status, pkgcache)
616 elif os.path.isfile(_rpm_db):
617 _host_distribution = RPMDistribution(_rpm_db)
618 elif os.path.isdir(_slack_db):
619 _host_distribution = SlackDistribution(_slack_db)
620 else:
621 _host_distribution = Distribution()
623 return _host_distribution