dak/generate_releases.py

   1 #!/usr/bin/env python
   2
   3 """
   4 Create all the Release files
   5
   6 @contact: Debian FTPMaster <ftpmaster@debian.org>
   7 @copyright: 2011  Joerg Jaspert <joerg@debian.org>
   8 @copyright: 2011  Mark Hymers <mhy@debian.org>
   9 @license: GNU General Public License version 2 or later
  10
  11 """
  12
  13 # This program is free software; you can redistribute it and/or modify
  14 # it under the terms of the GNU General Public License as published by
  15 # the Free Software Foundation; either version 2 of the License, or
  16 # (at your option) any later version.
  17
  18 # This program is distributed in the hope that it will be useful,
  19 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  20 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  21 # GNU General Public License for more details.
  22
  23 # You should have received a copy of the GNU General Public License
  24 # along with this program; if not, write to the Free Software
  25 # Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  26
  27 ################################################################################
  28
  29 # <mhy> I wish they wouldnt leave biscuits out, thats just tempting. Damnit.
  30
  31 ################################################################################
  32
  33 import sys
  34 import os
  35 import os.path
  36 import stat
  37 import time
  38 import gzip
  39 import bz2
  40 import errno
  41 import apt_pkg
  42 import subprocess
  43 from tempfile import mkstemp, mkdtemp
  44 import commands
  45 from sqlalchemy.orm import object_session
  46
  47 from daklib import utils, daklog
  48 from daklib.regexes import re_gensubrelease, re_includeinrelease
  49 from daklib.dak_exceptions import *
  50 from daklib.dbconn import *
  51 from daklib.config import Config
  52 from daklib.dakmultiprocessing import DakProcessPool, PROC_STATUS_SUCCESS
  53 import daklib.daksubprocess
  54
  55 ################################################################################
  56 Logger = None                  #: Our logging object
  57
  58 ################################################################################
  59
  60 def usage (exit_code=0):
  61     """ Usage information"""
  62
  63     print """Usage: dak generate-releases [OPTIONS]
  64 Generate the Release files
  65
  66   -a, --archive=ARCHIVE      process suites in ARCHIVE
  67   -s, --suite=SUITE(s)       process this suite
  68                              Default: All suites not marked 'untouchable'
  69   -f, --force                Allow processing of untouchable suites
  70                              CAREFUL: Only to be used at (point) release time!
  71   -h, --help                 show this help and exit
  72   -q, --quiet                Don't output progress
  73
  74 SUITE can be a space separated list, e.g.
  75    --suite=unstable testing
  76   """
  77     sys.exit(exit_code)
  78
  79 ########################################################################
  80
  81 def sign_release_dir(suite, dirname):
  82     cnf = Config()
  83
  84     if cnf.has_key("Dinstall::SigningKeyring"):
  85         keyring = "--secret-keyring \"%s\"" % cnf["Dinstall::SigningKeyring"]
  86         if cnf.has_key("Dinstall::SigningPubKeyring"):
  87             keyring += " --keyring \"%s\"" % cnf["Dinstall::SigningPubKeyring"]
  88
  89         arguments = "--no-options --batch --no-tty --armour --personal-digest-preferences=SHA256"
  90
  91         relname = os.path.join(dirname, 'Release')
  92
  93         dest = os.path.join(dirname, 'Release.gpg')
  94         if os.path.exists(dest):
  95             os.unlink(dest)
  96
  97         inlinedest = os.path.join(dirname, 'InRelease')
  98         if os.path.exists(inlinedest):
  99             os.unlink(inlinedest)
 100
 101         defkeyid=""
 102         for keyid in suite.signingkeys or []:
 103             defkeyid += "--local-user %s " % keyid
 104
 105         os.system("gpg %s %s %s --detach-sign <%s >>%s" %
 106                   (keyring, defkeyid, arguments, relname, dest))
 107         os.system("gpg %s %s %s --clearsign <%s >>%s" %
 108                   (keyring, defkeyid, arguments, relname, inlinedest))
 109
 110 class XzFile(object):
 111     def __init__(self, filename, mode='r'):
 112         self.filename = filename
 113     def read(self):
 114         cmd = ("xz", "-d")
 115         with open(self.filename, 'r') as stdin:
 116             process = daklib.daksubprocess.Popen(cmd, stdin=stdin, stdout=subprocess.PIPE)
 117             (stdout, stderr) = process.communicate()
 118             return stdout
 119
 120
 121 class HashFunc(object):
 122     def __init__(self, release_field, func, db_name):
 123         self.release_field = release_field
 124         self.func = func
 125         self.db_name = db_name
 126
 127 RELEASE_HASHES = [
 128     HashFunc('MD5Sum', apt_pkg.md5sum, 'md5sum'),
 129     HashFunc('SHA1', apt_pkg.sha1sum, 'sha1'),
 130     HashFunc('SHA256', apt_pkg.sha256sum, 'sha256'),
 131 ]
 132
 133
 134 class ReleaseWriter(object):
 135     def __init__(self, suite):
 136         self.suite = suite
 137
 138     def suite_path(self):
 139         """
 140         Absolute path to the suite-specific files.
 141         """
 142         cnf = Config()
 143         suite_suffix = cnf.find("Dinstall::SuiteSuffix", "")
 144
 145         return os.path.join(self.suite.archive.path, 'dists',
 146                             self.suite.suite_name, suite_suffix)
 147
 148     def suite_release_path(self):
 149         """
 150         Absolute path where Release files are physically stored.
 151         This should be a path that sorts after the dists/ directory.
 152         """
 153         # TODO: Eventually always create Release in `zzz-dists` to avoid
 154         # special cases. However we don't want to move existing Release files
 155         # for released suites.
 156         # See `create_release_symlinks` below.
 157         if not self.suite.byhash:
 158             return self.suite_path()
 159
 160         cnf = Config()
 161         suite_suffix = cnf.find("Dinstall::SuiteSuffix", "")
 162
 163         return os.path.join(self.suite.archive.path, 'zzz-dists',
 164                             self.suite.suite_name, suite_suffix)
 165
 166     def create_release_symlinks(self):
 167         """
 168         Create symlinks for Release files.
 169         This creates the symlinks for Release files in the `suite_path`
 170         to the actual files in `suite_release_path`.
 171         """
 172         # TODO: Eventually always create the links.
 173         # See `suite_release_path` above.
 174         if not self.suite.byhash:
 175             return
 176
 177         relpath = os.path.relpath(self.suite_release_path(), self.suite_path())
 178         for f in ("Release", "Release.gpg", "InRelease"):
 179             source = os.path.join(relpath, f)
 180             dest = os.path.join(self.suite_path(), f)
 181             if not os.path.islink(dest):
 182                 os.unlink(dest)
 183             elif os.readlink(dest) == source:
 184                 continue
 185             else:
 186                 os.unlink(dest)
 187             os.symlink(source, dest)
 188
 189     def create_output_directories(self):
 190         for path in (self.suite_path(), self.suite_release_path()):
 191             try:
 192                 os.makedirs(path)
 193             except OSError as e:
 194                 if e.errno != errno.EEXIST:
 195                     raise
 196
 197     def _update_hashfile_table(self, session, fileinfo, hashes):
 198         # Mark all by-hash files as obsolete.  We will undo that for the ones
 199         # we still reference later.
 200         query = """
 201             UPDATE hashfile SET unreferenced = CURRENT_TIMESTAMP
 202             WHERE suite_id = :id AND unreferenced IS NULL"""
 203         session.execute(query, {'id': self.suite.suite_id})
 204
 205         if self.suite.byhash:
 206             query = "SELECT path FROM hashfile WHERE suite_id = :id"
 207             q = session.execute(query, {'id': self.suite.suite_id})
 208             known_hashfiles = set(row[0] for row in q)
 209             updated = []
 210             new = []
 211
 212             # Update the hashfile table with new or updated files
 213             for filename in fileinfo:
 214                 if not os.path.exists(filename):
 215                     # probably an uncompressed index we didn't generate
 216                     continue
 217                 byhashdir = os.path.join(os.path.dirname(filename), 'by-hash')
 218                 for h in hashes:
 219                     field = h.release_field
 220                     hashfile = os.path.join(byhashdir, field, fileinfo[filename][field])
 221                     if hashfile in known_hashfiles:
 222                         updated.append(hashfile)
 223                     else:
 224                         new.append(hashfile)
 225
 226             if updated:
 227                 session.execute("""
 228                     UPDATE hashfile SET unreferenced = NULL
 229                     WHERE path = ANY(:p) AND suite_id = :id""",
 230                     {'p': updated, 'id': self.suite.suite_id})
 231             if new:
 232                 session.execute("""
 233                     INSERT INTO hashfile (path, suite_id)
 234                     VALUES (:p, :id)""",
 235                     [{'p': hashfile, 'id': self.suite.suite_id} for hashfile in new])
 236
 237         session.commit()
 238
 239     def _make_byhash_links(self, fileinfo, hashes):
 240         # Create hardlinks in by-hash directories
 241         for filename in fileinfo:
 242             if not os.path.exists(filename):
 243                 # probably an uncompressed index we didn't generate
 244                 continue
 245
 246             for h in hashes:
 247                 field = h.release_field
 248                 hashfile = os.path.join(os.path.dirname(filename), 'by-hash', field, fileinfo[filename][field])
 249                 try:
 250                     os.makedirs(os.path.dirname(hashfile))
 251                 except OSError as exc:
 252                     if exc.errno != errno.EEXIST:
 253                         raise
 254                 try:
 255                     os.link(filename, hashfile)
 256                 except OSError as exc:
 257                     if exc.errno != errno.EEXIST:
 258                         raise
 259
 260     def generate_release_files(self):
 261         """
 262         Generate Release files for the given suite
 263
 264         @type suite: string
 265         @param suite: Suite name
 266         """
 267
 268         suite = self.suite
 269         session = object_session(suite)
 270
 271         architectures = get_suite_architectures(suite.suite_name, skipall=True, skipsrc=True, session=session)
 272
 273         # Attribs contains a tuple of field names and the database names to use to
 274         # fill them in
 275         attribs = ( ('Origin',      'origin'),
 276                     ('Label',       'label'),
 277                     ('Suite',       'release_suite_output'),
 278                     ('Version',     'version'),
 279                     ('Codename',    'codename'),
 280                     ('Changelogs',  'changelog_url'),
 281                   )
 282
 283         # A "Sub" Release file has slightly different fields
 284         subattribs = ( ('Archive',  'suite_name'),
 285                        ('Origin',   'origin'),
 286                        ('Label',    'label'),
 287                        ('Version',  'version') )
 288
 289         # Boolean stuff. If we find it true in database, write out "yes" into the release file
 290         boolattrs = ( ('NotAutomatic',         'notautomatic'),
 291                       ('ButAutomaticUpgrades', 'butautomaticupgrades'),
 292                       ('Acquire-By-Hash',      'byhash'),
 293                     )
 294
 295         cnf = Config()
 296
 297         suite_suffix = cnf.find("Dinstall::SuiteSuffix", "")
 298
 299         self.create_output_directories()
 300         self.create_release_symlinks()
 301
 302         outfile = os.path.join(self.suite_release_path(), "Release")
 303         out = open(outfile + ".new", "w")
 304
 305         for key, dbfield in attribs:
 306             # Hack to skip NULL Version fields as we used to do this
 307             # We should probably just always ignore anything which is None
 308             if key in ("Version", "Changelogs") and getattr(suite, dbfield) is None:
 309                 continue
 310
 311             out.write("%s: %s\n" % (key, getattr(suite, dbfield)))
 312
 313         out.write("Date: %s\n" % (time.strftime("%a, %d %b %Y %H:%M:%S UTC", time.gmtime(time.time()))))
 314
 315         if suite.validtime:
 316             validtime=float(suite.validtime)
 317             out.write("Valid-Until: %s\n" % (time.strftime("%a, %d %b %Y %H:%M:%S UTC", time.gmtime(time.time()+validtime))))
 318
 319         for key, dbfield in boolattrs:
 320             if getattr(suite, dbfield, False):
 321                 out.write("%s: yes\n" % (key))
 322
 323         out.write("Architectures: %s\n" % (" ".join([a.arch_string for a in architectures])))
 324
 325         components = [ c.component_name for c in suite.components ]
 326
 327         out.write("Components: %s\n" % (" ".join(components)))
 328
 329         # For exact compatibility with old g-r, write out Description here instead
 330         # of with the rest of the DB fields above
 331         if getattr(suite, 'description') is not None:
 332             out.write("Description: %s\n" % suite.description)
 333
 334         for comp in components:
 335             for dirpath, dirnames, filenames in os.walk(os.path.join(self.suite_path(), comp), topdown=True):
 336                 if not re_gensubrelease.match(dirpath):
 337                     continue
 338
 339                 subfile = os.path.join(dirpath, "Release")
 340                 subrel = open(subfile + '.new', "w")
 341
 342                 for key, dbfield in subattribs:
 343                     if getattr(suite, dbfield) is not None:
 344                         subrel.write("%s: %s\n" % (key, getattr(suite, dbfield)))
 345
 346                 for key, dbfield in boolattrs:
 347                     if getattr(suite, dbfield, False):
 348                         subrel.write("%s: yes\n" % (key))
 349
 350                 subrel.write("Component: %s%s\n" % (suite_suffix, comp))
 351
 352                 # Urgh, but until we have all the suite/component/arch stuff in the DB,
 353                 # this'll have to do
 354                 arch = os.path.split(dirpath)[-1]
 355                 if arch.startswith('binary-'):
 356                     arch = arch[7:]
 357
 358                 subrel.write("Architecture: %s\n" % (arch))
 359                 subrel.close()
 360
 361                 os.rename(subfile + '.new', subfile)
 362
 363         # Now that we have done the groundwork, we want to get off and add the files with
 364         # their checksums to the main Release file
 365         oldcwd = os.getcwd()
 366
 367         os.chdir(self.suite_path())
 368
 369         hashes = [x for x in RELEASE_HASHES if x.db_name in suite.checksums]
 370
 371         fileinfo = {}
 372
 373         uncompnotseen = {}
 374
 375         for dirpath, dirnames, filenames in os.walk(".", followlinks=True, topdown=True):
 376             for entry in filenames:
 377                 # Skip things we don't want to include
 378                 if not re_includeinrelease.match(entry):
 379                     continue
 380
 381                 if dirpath == '.' and entry in ["Release", "Release.gpg", "InRelease"]:
 382                     continue
 383
 384                 filename = os.path.join(dirpath.lstrip('./'), entry)
 385                 fileinfo[filename] = {}
 386                 contents = open(filename, 'r').read()
 387
 388                 # If we find a file for which we have a compressed version and
 389                 # haven't yet seen the uncompressed one, store the possibility
 390                 # for future use
 391                 if entry.endswith(".gz") and filename[:-3] not in uncompnotseen:
 392                     uncompnotseen[filename[:-3]] = (gzip.GzipFile, filename)
 393                 elif entry.endswith(".bz2") and filename[:-4] not in uncompnotseen:
 394                     uncompnotseen[filename[:-4]] = (bz2.BZ2File, filename)
 395                 elif entry.endswith(".xz") and filename[:-3] not in uncompnotseen:
 396                     uncompnotseen[filename[:-3]] = (XzFile, filename)
 397
 398                 fileinfo[filename]['len'] = len(contents)
 399
 400                 for hf in hashes:
 401                     fileinfo[filename][hf.release_field] = hf.func(contents)
 402
 403         for filename, comp in uncompnotseen.items():
 404             # If we've already seen the uncompressed file, we don't
 405             # need to do anything again
 406             if filename in fileinfo:
 407                 continue
 408
 409             fileinfo[filename] = {}
 410
 411             # File handler is comp[0], filename of compressed file is comp[1]
 412             contents = comp[0](comp[1], 'r').read()
 413
 414             fileinfo[filename]['len'] = len(contents)
 415
 416             for hf in hashes:
 417                 fileinfo[filename][hf.release_field] = hf.func(contents)
 418
 419
 420         for field in sorted(h.release_field for h in hashes):
 421             out.write('%s:\n' % field)
 422             for filename in sorted(fileinfo.keys()):
 423                 out.write(" %s %8d %s\n" % (fileinfo[filename][field], fileinfo[filename]['len'], filename))
 424
 425         out.close()
 426         os.rename(outfile + '.new', outfile)
 427
 428         self._update_hashfile_table(session, fileinfo, hashes)
 429         if suite.byhash:
 430             self._make_byhash_links(fileinfo, hashes)
 431
 432         sign_release_dir(suite, os.path.dirname(outfile))
 433
 434         os.chdir(oldcwd)
 435
 436         return
 437
 438
 439 def main ():
 440     global Logger
 441
 442     cnf = Config()
 443
 444     for i in ["Help", "Suite", "Force", "Quiet"]:
 445         if not cnf.has_key("Generate-Releases::Options::%s" % (i)):
 446             cnf["Generate-Releases::Options::%s" % (i)] = ""
 447
 448     Arguments = [('h',"help","Generate-Releases::Options::Help"),
 449                  ('a','archive','Generate-Releases::Options::Archive','HasArg'),
 450                  ('s',"suite","Generate-Releases::Options::Suite"),
 451                  ('f',"force","Generate-Releases::Options::Force"),
 452                  ('q',"quiet","Generate-Releases::Options::Quiet"),
 453                  ('o','option','','ArbItem')]
 454
 455     suite_names = apt_pkg.parse_commandline(cnf.Cnf, Arguments, sys.argv)
 456     Options = cnf.subtree("Generate-Releases::Options")
 457
 458     if Options["Help"]:
 459         usage()
 460
 461     Logger = daklog.Logger('generate-releases')
 462     pool = DakProcessPool()
 463
 464     session = DBConn().session()
 465
 466     if Options["Suite"]:
 467         suites = []
 468         for s in suite_names:
 469             suite = get_suite(s.lower(), session)
 470             if suite:
 471                 suites.append(suite)
 472             else:
 473                 print "cannot find suite %s" % s
 474                 Logger.log(['cannot find suite %s' % s])
 475     else:
 476         query = session.query(Suite).filter(Suite.untouchable == False)
 477         if 'Archive' in Options:
 478             query = query.join(Suite.archive).filter(Archive.archive_name==Options['Archive'])
 479         suites = query.all()
 480
 481     broken=[]
 482
 483     for s in suites:
 484         # Setup a multiprocessing Pool. As many workers as we have CPU cores.
 485         if s.untouchable and not Options["Force"]:
 486             print "Skipping %s (untouchable)" % s.suite_name
 487             continue
 488
 489         if not Options["Quiet"]:
 490             print "Processing %s" % s.suite_name
 491         Logger.log(['Processing release file for Suite: %s' % (s.suite_name)])
 492         pool.apply_async(generate_helper, (s.suite_id, ))
 493
 494     # No more work will be added to our pool, close it and then wait for all to finish
 495     pool.close()
 496     pool.join()
 497
 498     retcode = pool.overall_status()
 499
 500     if retcode > 0:
 501         # TODO: CENTRAL FUNCTION FOR THIS / IMPROVE LOGGING
 502         Logger.log(['Release file generation broken: %s' % (','.join([str(x[1]) for x in pool.results]))])
 503
 504     Logger.close()
 505
 506     sys.exit(retcode)
 507
 508 def generate_helper(suite_id):
 509     '''
 510     This function is called in a new subprocess.
 511     '''
 512     session = DBConn().session()
 513     suite = Suite.get(suite_id, session)
 514
 515     # We allow the process handler to catch and deal with any exceptions
 516     rw = ReleaseWriter(suite)
 517     rw.generate_release_files()
 518
 519     return (PROC_STATUS_SUCCESS, 'Release file written for %s' % suite.suite_name)
 520
 521 #######################################################################################
 522
 523 if __name__ == '__main__':
 524     main()