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, 'md5'),
 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 generate_release_files(self):
 198         """
 199         Generate Release files for the given suite
 200
 201         @type suite: string
 202         @param suite: Suite name
 203         """
 204
 205         suite = self.suite
 206         session = object_session(suite)
 207
 208         architectures = get_suite_architectures(suite.suite_name, skipall=True, skipsrc=True, session=session)
 209
 210         # Attribs contains a tuple of field names and the database names to use to
 211         # fill them in
 212         attribs = ( ('Origin',      'origin'),
 213                     ('Label',       'label'),
 214                     ('Suite',       'release_suite_output'),
 215                     ('Version',     'version'),
 216                     ('Codename',    'codename'),
 217                     ('Changelogs',  'changelog_url'),
 218                   )
 219
 220         # A "Sub" Release file has slightly different fields
 221         subattribs = ( ('Archive',  'suite_name'),
 222                        ('Origin',   'origin'),
 223                        ('Label',    'label'),
 224                        ('Version',  'version') )
 225
 226         # Boolean stuff. If we find it true in database, write out "yes" into the release file
 227         boolattrs = ( ('NotAutomatic',         'notautomatic'),
 228                       ('ButAutomaticUpgrades', 'butautomaticupgrades'),
 229                       ('Acquire-By-Hash',      'byhash'),
 230                     )
 231
 232         cnf = Config()
 233
 234         suite_suffix = cnf.find("Dinstall::SuiteSuffix", "")
 235
 236         self.create_output_directories()
 237         self.create_release_symlinks()
 238
 239         outfile = os.path.join(self.suite_release_path(), "Release")
 240         out = open(outfile + ".new", "w")
 241
 242         for key, dbfield in attribs:
 243             # Hack to skip NULL Version fields as we used to do this
 244             # We should probably just always ignore anything which is None
 245             if key in ("Version", "Changelogs") and getattr(suite, dbfield) is None:
 246                 continue
 247
 248             out.write("%s: %s\n" % (key, getattr(suite, dbfield)))
 249
 250         out.write("Date: %s\n" % (time.strftime("%a, %d %b %Y %H:%M:%S UTC", time.gmtime(time.time()))))
 251
 252         if suite.validtime:
 253             validtime=float(suite.validtime)
 254             out.write("Valid-Until: %s\n" % (time.strftime("%a, %d %b %Y %H:%M:%S UTC", time.gmtime(time.time()+validtime))))
 255
 256         for key, dbfield in boolattrs:
 257             if getattr(suite, dbfield, False):
 258                 out.write("%s: yes\n" % (key))
 259
 260         out.write("Architectures: %s\n" % (" ".join([a.arch_string for a in architectures])))
 261
 262         components = [ c.component_name for c in suite.components ]
 263
 264         out.write("Components: %s\n" % (" ".join(components)))
 265
 266         # For exact compatibility with old g-r, write out Description here instead
 267         # of with the rest of the DB fields above
 268         if getattr(suite, 'description') is not None:
 269             out.write("Description: %s\n" % suite.description)
 270
 271         for comp in components:
 272             for dirpath, dirnames, filenames in os.walk(os.path.join(self.suite_path(), comp), topdown=True):
 273                 if not re_gensubrelease.match(dirpath):
 274                     continue
 275
 276                 subfile = os.path.join(dirpath, "Release")
 277                 subrel = open(subfile + '.new', "w")
 278
 279                 for key, dbfield in subattribs:
 280                     if getattr(suite, dbfield) is not None:
 281                         subrel.write("%s: %s\n" % (key, getattr(suite, dbfield)))
 282
 283                 for key, dbfield in boolattrs:
 284                     if getattr(suite, dbfield, False):
 285                         subrel.write("%s: yes\n" % (key))
 286
 287                 subrel.write("Component: %s%s\n" % (suite_suffix, comp))
 288
 289                 # Urgh, but until we have all the suite/component/arch stuff in the DB,
 290                 # this'll have to do
 291                 arch = os.path.split(dirpath)[-1]
 292                 if arch.startswith('binary-'):
 293                     arch = arch[7:]
 294
 295                 subrel.write("Architecture: %s\n" % (arch))
 296                 subrel.close()
 297
 298                 os.rename(subfile + '.new', subfile)
 299
 300         # Now that we have done the groundwork, we want to get off and add the files with
 301         # their checksums to the main Release file
 302         oldcwd = os.getcwd()
 303
 304         os.chdir(self.suite_path())
 305
 306         hashes = [x for x in RELEASE_HASHES if x.db_name in suite.checksums]
 307
 308         fileinfo = {}
 309
 310         uncompnotseen = {}
 311
 312         for dirpath, dirnames, filenames in os.walk(".", followlinks=True, topdown=True):
 313             for entry in filenames:
 314                 # Skip things we don't want to include
 315                 if not re_includeinrelease.match(entry):
 316                     continue
 317
 318                 if dirpath == '.' and entry in ["Release", "Release.gpg", "InRelease"]:
 319                     continue
 320
 321                 filename = os.path.join(dirpath.lstrip('./'), entry)
 322                 fileinfo[filename] = {}
 323                 contents = open(filename, 'r').read()
 324
 325                 # If we find a file for which we have a compressed version and
 326                 # haven't yet seen the uncompressed one, store the possibility
 327                 # for future use
 328                 if entry.endswith(".gz") and filename[:-3] not in uncompnotseen:
 329                     uncompnotseen[filename[:-3]] = (gzip.GzipFile, filename)
 330                 elif entry.endswith(".bz2") and filename[:-4] not in uncompnotseen:
 331                     uncompnotseen[filename[:-4]] = (bz2.BZ2File, filename)
 332                 elif entry.endswith(".xz") and filename[:-3] not in uncompnotseen:
 333                     uncompnotseen[filename[:-3]] = (XzFile, filename)
 334
 335                 fileinfo[filename]['len'] = len(contents)
 336
 337                 for hf in hashes:
 338                     fileinfo[filename][hf.release_field] = hf.func(contents)
 339
 340         for filename, comp in uncompnotseen.items():
 341             # If we've already seen the uncompressed file, we don't
 342             # need to do anything again
 343             if filename in fileinfo:
 344                 continue
 345
 346             fileinfo[filename] = {}
 347
 348             # File handler is comp[0], filename of compressed file is comp[1]
 349             contents = comp[0](comp[1], 'r').read()
 350
 351             fileinfo[filename]['len'] = len(contents)
 352
 353             for hf in hashes:
 354                 fileinfo[filename][hf.release_field] = hf.func(contents)
 355
 356
 357         for field in sorted(h.release_field for h in hashes):
 358             out.write('%s:\n' % field)
 359             for filename in sorted(fileinfo.keys()):
 360                 out.write(" %s %8d %s\n" % (fileinfo[filename][field], fileinfo[filename]['len'], filename))
 361
 362         out.close()
 363         os.rename(outfile + '.new', outfile)
 364
 365         # Mark all by-hash files as obsolete.  We will undo that for the ones
 366         # we still reference later.
 367         query = """
 368             UPDATE hashfile SET unreferenced = CURRENT_TIMESTAMP
 369             WHERE suite_id = :id AND unreferenced IS NULL"""
 370         session.execute(query, {'id': suite.suite_id})
 371
 372         if suite.byhash:
 373             query = "SELECT path FROM hashfile WHERE suite_id = :id"
 374             q = session.execute(query, {'id': suite.suite_id})
 375             known_hashfiles = set(row[0] for row in q)
 376             updated = []
 377             new = []
 378
 379             # Update the hashfile table with new or updated files
 380             for filename in fileinfo:
 381                 if not os.path.exists(filename):
 382                     # probably an uncompressed index we didn't generate
 383                     continue
 384                 byhashdir = os.path.join(os.path.dirname(filename), 'by-hash')
 385                 for h in hashes:
 386                     field = h.release_field
 387                     hashfile = os.path.join(byhashdir, field, fileinfo[filename][field])
 388                     if hashfile in known_hashfiles:
 389                         updated.append(hashfile)
 390                     else:
 391                         new.append(hashfile)
 392
 393             if updated:
 394                 session.execute("""
 395                     UPDATE hashfile SET unreferenced = NULL
 396                     WHERE path = ANY(:p) AND suite_id = :id""",
 397                     {'p': updated, 'id': suite.suite_id})
 398             if new:
 399                 session.execute("""
 400                     INSERT INTO hashfile (path, suite_id)
 401                     VALUES (:p, :id)""",
 402                     [{'p': hashfile, 'id': suite.suite_id} for hashfile in new])
 403
 404         session.commit()
 405
 406         if suite.byhash:
 407             # Create hardlinks in by-hash directories
 408             for filename in fileinfo:
 409                 if not os.path.exists(filename):
 410                     # probably an uncompressed index we didn't generate
 411                     continue
 412
 413                 for h in hashes:
 414                     field = h.release_field
 415                     hashfile = os.path.join(os.path.dirname(filename), 'by-hash', field, fileinfo[filename][field])
 416                     try:
 417                         os.makedirs(os.path.dirname(hashfile))
 418                     except OSError as exc:
 419                         if exc.errno != errno.EEXIST:
 420                             raise
 421                     try:
 422                         os.link(filename, hashfile)
 423                     except OSError as exc:
 424                         if exc.errno != errno.EEXIST:
 425                             raise
 426
 427
 428         sign_release_dir(suite, os.path.dirname(outfile))
 429
 430         os.chdir(oldcwd)
 431
 432         return
 433
 434
 435 def main ():
 436     global Logger
 437
 438     cnf = Config()
 439
 440     for i in ["Help", "Suite", "Force", "Quiet"]:
 441         if not cnf.has_key("Generate-Releases::Options::%s" % (i)):
 442             cnf["Generate-Releases::Options::%s" % (i)] = ""
 443
 444     Arguments = [('h',"help","Generate-Releases::Options::Help"),
 445                  ('a','archive','Generate-Releases::Options::Archive','HasArg'),
 446                  ('s',"suite","Generate-Releases::Options::Suite"),
 447                  ('f',"force","Generate-Releases::Options::Force"),
 448                  ('q',"quiet","Generate-Releases::Options::Quiet"),
 449                  ('o','option','','ArbItem')]
 450
 451     suite_names = apt_pkg.parse_commandline(cnf.Cnf, Arguments, sys.argv)
 452     Options = cnf.subtree("Generate-Releases::Options")
 453
 454     if Options["Help"]:
 455         usage()
 456
 457     Logger = daklog.Logger('generate-releases')
 458     pool = DakProcessPool()
 459
 460     session = DBConn().session()
 461
 462     if Options["Suite"]:
 463         suites = []
 464         for s in suite_names:
 465             suite = get_suite(s.lower(), session)
 466             if suite:
 467                 suites.append(suite)
 468             else:
 469                 print "cannot find suite %s" % s
 470                 Logger.log(['cannot find suite %s' % s])
 471     else:
 472         query = session.query(Suite).filter(Suite.untouchable == False)
 473         if 'Archive' in Options:
 474             query = query.join(Suite.archive).filter(Archive.archive_name==Options['Archive'])
 475         suites = query.all()
 476
 477     broken=[]
 478
 479     for s in suites:
 480         # Setup a multiprocessing Pool. As many workers as we have CPU cores.
 481         if s.untouchable and not Options["Force"]:
 482             print "Skipping %s (untouchable)" % s.suite_name
 483             continue
 484
 485         if not Options["Quiet"]:
 486             print "Processing %s" % s.suite_name
 487         Logger.log(['Processing release file for Suite: %s' % (s.suite_name)])
 488         pool.apply_async(generate_helper, (s.suite_id, ))
 489
 490     # No more work will be added to our pool, close it and then wait for all to finish
 491     pool.close()
 492     pool.join()
 493
 494     retcode = pool.overall_status()
 495
 496     if retcode > 0:
 497         # TODO: CENTRAL FUNCTION FOR THIS / IMPROVE LOGGING
 498         Logger.log(['Release file generation broken: %s' % (','.join([str(x[1]) for x in pool.results]))])
 499
 500     Logger.close()
 501
 502     sys.exit(retcode)
 503
 504 def generate_helper(suite_id):
 505     '''
 506     This function is called in a new subprocess.
 507     '''
 508     session = DBConn().session()
 509     suite = Suite.get(suite_id, session)
 510
 511     # We allow the process handler to catch and deal with any exceptions
 512     rw = ReleaseWriter(suite)
 513     rw.generate_release_files()
 514
 515     return (PROC_STATUS_SUCCESS, 'Release file written for %s' % suite.suite_name)
 516
 517 #######################################################################################
 518
 519 if __name__ == '__main__':
 520     main()