daklib/archive.py

   1 # Copyright (C) 2012, Ansgar Burchardt <ansgar@debian.org>
   2 #
   3 # This program is free software; you can redistribute it and/or modify
   4 # it under the terms of the GNU General Public License as published by
   5 # the Free Software Foundation; either version 2 of the License, or
   6 # (at your option) any later version.
   7 #
   8 # This program is distributed in the hope that it will be useful,
   9 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  10 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  11 # GNU General Public License for more details.
  12 #
  13 # You should have received a copy of the GNU General Public License along
  14 # with this program; if not, write to the Free Software Foundation, Inc.,
  15 # 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
  16
  17 """module to manipulate the archive
  18
  19 This module provides classes to manipulate the archive.
  20 """
  21
  22 from daklib.dbconn import *
  23 import daklib.checks as checks
  24 from daklib.config import Config
  25 import daklib.upload as upload
  26 import daklib.utils as utils
  27 from daklib.fstransactions import FilesystemTransaction
  28 from daklib.regexes import re_changelog_versions, re_bin_only_nmu
  29
  30 import apt_pkg
  31 from datetime import datetime
  32 import os
  33 import shutil
  34 import subprocess
  35 from sqlalchemy.orm.exc import NoResultFound
  36 import tempfile
  37 import traceback
  38
  39 class ArchiveException(Exception):
  40     pass
  41
  42 class HashMismatchException(ArchiveException):
  43     pass
  44
  45 class ArchiveTransaction(object):
  46     """manipulate the archive in a transaction
  47     """
  48     def __init__(self):
  49         self.fs = FilesystemTransaction()
  50         self.session = DBConn().session()
  51
  52     def get_file(self, hashed_file, source_name, check_hashes=True):
  53         """Look for file C{hashed_file} in database
  54
  55         @type  hashed_file: L{daklib.upload.HashedFile}
  56         @param hashed_file: file to look for in the database
  57
  58         @type  source_name: str
  59         @param source_name: source package name
  60
  61         @type  check_hashes: bool
  62         @param check_hashes: check size and hashes match
  63
  64         @raise KeyError: file was not found in the database
  65         @raise HashMismatchException: hash mismatch
  66
  67         @rtype:  L{daklib.dbconn.PoolFile}
  68         @return: database entry for the file
  69         """
  70         poolname = os.path.join(utils.poolify(source_name), hashed_file.filename)
  71         try:
  72             poolfile = self.session.query(PoolFile).filter_by(filename=poolname).one()
  73             if check_hashes and (poolfile.filesize != hashed_file.size
  74                                  or poolfile.md5sum != hashed_file.md5sum
  75                                  or poolfile.sha1sum != hashed_file.sha1sum
  76                                  or poolfile.sha256sum != hashed_file.sha256sum):
  77                 raise HashMismatchException('{0}: Does not match file already existing in the pool.'.format(hashed_file.filename))
  78             return poolfile
  79         except NoResultFound:
  80             raise KeyError('{0} not found in database.'.format(poolname))
  81
  82     def _install_file(self, directory, hashed_file, archive, component, source_name):
  83         """Install a file
  84
  85         Will not give an error when the file is already present.
  86
  87         @rtype:  L{daklib.dbconn.PoolFile}
  88         @return: batabase object for the new file
  89         """
  90         session = self.session
  91
  92         poolname = os.path.join(utils.poolify(source_name), hashed_file.filename)
  93         try:
  94             poolfile = self.get_file(hashed_file, source_name)
  95         except KeyError:
  96             poolfile = PoolFile(filename=poolname, filesize=hashed_file.size)
  97             poolfile.md5sum = hashed_file.md5sum
  98             poolfile.sha1sum = hashed_file.sha1sum
  99             poolfile.sha256sum = hashed_file.sha256sum
 100             session.add(poolfile)
 101             session.flush()
 102
 103         try:
 104             session.query(ArchiveFile).filter_by(archive=archive, component=component, file=poolfile).one()
 105         except NoResultFound:
 106             archive_file = ArchiveFile(archive, component, poolfile)
 107             session.add(archive_file)
 108             session.flush()
 109
 110             path = os.path.join(archive.path, 'pool', component.component_name, poolname)
 111             hashed_file_path = os.path.join(directory, hashed_file.filename)
 112             self.fs.copy(hashed_file_path, path, link=False, mode=archive.mode)
 113
 114         return poolfile
 115
 116     def install_binary(self, directory, binary, suite, component, allow_tainted=False, fingerprint=None, source_suites=None, extra_source_archives=None):
 117         """Install a binary package
 118
 119         @type  directory: str
 120         @param directory: directory the binary package is located in
 121
 122         @type  binary: L{daklib.upload.Binary}
 123         @param binary: binary package to install
 124
 125         @type  suite: L{daklib.dbconn.Suite}
 126         @param suite: target suite
 127
 128         @type  component: L{daklib.dbconn.Component}
 129         @param component: target component
 130
 131         @type  allow_tainted: bool
 132         @param allow_tainted: allow to copy additional files from tainted archives
 133
 134         @type  fingerprint: L{daklib.dbconn.Fingerprint}
 135         @param fingerprint: optional fingerprint
 136
 137         @type  source_suites: SQLAlchemy subquery for C{daklib.dbconn.Suite} or C{True}
 138         @param source_suites: suites to copy the source from if they are not
 139                               in C{suite} or C{True} to allow copying from any
 140                               suite.
 141
 142         @type  extra_source_archives: list of L{daklib.dbconn.Archive}
 143         @param extra_source_archives: extra archives to copy Built-Using sources from
 144
 145         @rtype:  L{daklib.dbconn.DBBinary}
 146         @return: databse object for the new package
 147         """
 148         session = self.session
 149         control = binary.control
 150         maintainer = get_or_set_maintainer(control['Maintainer'], session)
 151         architecture = get_architecture(control['Architecture'], session)
 152
 153         (source_name, source_version) = binary.source
 154         source_query = session.query(DBSource).filter_by(source=source_name, version=source_version)
 155         source = source_query.filter(DBSource.suites.contains(suite)).first()
 156         if source is None:
 157             if source_suites != True:
 158                 source_query = source_query.join(DBSource.suites) \
 159                     .filter(Suite.suite_id == source_suites.c.id)
 160             source = source_query.first()
 161             if source is None:
 162                 raise ArchiveException('{0}: trying to install to {1}, but could not find source'.format(binary.hashed_file.filename, suite.suite_name))
 163             self.copy_source(source, suite, component)
 164
 165         db_file = self._install_file(directory, binary.hashed_file, suite.archive, component, source_name)
 166
 167         unique = dict(
 168             package=control['Package'],
 169             version=control['Version'],
 170             architecture=architecture,
 171             )
 172         rest = dict(
 173             source=source,
 174             maintainer=maintainer,
 175             poolfile=db_file,
 176             binarytype=binary.type,
 177             fingerprint=fingerprint,
 178             )
 179
 180         try:
 181             db_binary = session.query(DBBinary).filter_by(**unique).one()
 182             for key, value in rest.iteritems():
 183                 if getattr(db_binary, key) != value:
 184                     raise ArchiveException('{0}: Does not match binary in database.'.format(binary.hashed_file.filename))
 185         except NoResultFound:
 186             db_binary = DBBinary(**unique)
 187             for key, value in rest.iteritems():
 188                 setattr(db_binary, key, value)
 189             session.add(db_binary)
 190             session.flush()
 191             import_metadata_into_db(db_binary, session)
 192
 193             self._add_built_using(db_binary, binary.hashed_file.filename, control, suite, extra_archives=extra_source_archives)
 194
 195         if suite not in db_binary.suites:
 196             db_binary.suites.append(suite)
 197
 198         session.flush()
 199
 200         return db_binary
 201
 202     def _ensure_extra_source_exists(self, filename, source, archive, extra_archives=None):
 203         """ensure source exists in the given archive
 204
 205         This is intended to be used to check that Built-Using sources exist.
 206
 207         @type  filename: str
 208         @param filename: filename to use in error messages
 209
 210         @type  source: L{daklib.dbconn.DBSource}
 211         @param source: source to look for
 212
 213         @type  archive: L{daklib.dbconn.Archive}
 214         @param archive: archive to look in
 215
 216         @type  extra_archives: list of L{daklib.dbconn.Archive}
 217         @param extra_archives: list of archives to copy the source package from
 218                                if it is not yet present in C{archive}
 219         """
 220         session = self.session
 221         db_file = session.query(ArchiveFile).filter_by(file=source.poolfile, archive=archive).first()
 222         if db_file is not None:
 223             return True
 224
 225         # Try to copy file from one extra archive
 226         if extra_archives is None:
 227             extra_archives = []
 228         db_file = session.query(ArchiveFile).filter_by(file=source.poolfile).filter(ArchiveFile.archive_id.in_([ a.archive_id for a in extra_archives])).first()
 229         if db_file is None:
 230             raise ArchiveException('{0}: Built-Using refers to package {1} (= {2}) not in target archive {3}.'.format(filename, source.source, source.version, archive.archive_name))
 231
 232         source_archive = db_file.archive
 233         for dsc_file in source.srcfiles:
 234             af = session.query(ArchiveFile).filter_by(file=dsc_file.poolfile, archive=source_archive, component=db_file.component).one()
 235             # We were given an explicit list of archives so it is okay to copy from tainted archives.
 236             self._copy_file(af.file, archive, db_file.component, allow_tainted=True)
 237
 238     def _add_built_using(self, db_binary, filename, control, suite, extra_archives=None):
 239         """Add Built-Using sources to C{db_binary.extra_sources}
 240         """
 241         session = self.session
 242         built_using = control.get('Built-Using', None)
 243
 244         if built_using is not None:
 245             for dep in apt_pkg.parse_depends(built_using):
 246                 assert len(dep) == 1, 'Alternatives are not allowed in Built-Using field'
 247                 bu_source_name, bu_source_version, comp = dep[0]
 248                 assert comp == '=', 'Built-Using must contain strict dependencies'
 249
 250                 bu_source = session.query(DBSource).filter_by(source=bu_source_name, version=bu_source_version).first()
 251                 if bu_source is None:
 252                     raise ArchiveException('{0}: Built-Using refers to non-existing source package {1} (= {2})'.format(filename, bu_source_name, bu_source_version))
 253
 254                 self._ensure_extra_source_exists(filename, bu_source, suite.archive, extra_archives=extra_archives)
 255
 256                 db_binary.extra_sources.append(bu_source)
 257
 258     def install_source(self, directory, source, suite, component, changed_by, allow_tainted=False, fingerprint=None):
 259         """Install a source package
 260
 261         @type  directory: str
 262         @param directory: directory the source package is located in
 263
 264         @type  source: L{daklib.upload.Source}
 265         @param source: source package to install
 266
 267         @type  suite: L{daklib.dbconn.Suite}
 268         @param suite: target suite
 269
 270         @type  component: L{daklib.dbconn.Component}
 271         @param component: target component
 272
 273         @type  changed_by: L{daklib.dbconn.Maintainer}
 274         @param changed_by: person who prepared this version of the package
 275
 276         @type  allow_tainted: bool
 277         @param allow_tainted: allow to copy additional files from tainted archives
 278
 279         @type  fingerprint: L{daklib.dbconn.Fingerprint}
 280         @param fingerprint: optional fingerprint
 281
 282         @rtype:  L{daklib.dbconn.DBSource}
 283         @return: database object for the new source
 284         """
 285         session = self.session
 286         archive = suite.archive
 287         control = source.dsc
 288         maintainer = get_or_set_maintainer(control['Maintainer'], session)
 289         source_name = control['Source']
 290
 291         ### Add source package to database
 292
 293         # We need to install the .dsc first as the DBSource object refers to it.
 294         db_file_dsc = self._install_file(directory, source._dsc_file, archive, component, source_name)
 295
 296         unique = dict(
 297             source=source_name,
 298             version=control['Version'],
 299             )
 300         rest = dict(
 301             maintainer=maintainer,
 302             changedby=changed_by,
 303             #install_date=datetime.now().date(),
 304             poolfile=db_file_dsc,
 305             fingerprint=fingerprint,
 306             dm_upload_allowed=(control.get('DM-Upload-Allowed', 'no') == 'yes'),
 307             )
 308
 309         created = False
 310         try:
 311             db_source = session.query(DBSource).filter_by(**unique).one()
 312             for key, value in rest.iteritems():
 313                 if getattr(db_source, key) != value:
 314                     raise ArchiveException('{0}: Does not match source in database.'.format(source._dsc_file.filename))
 315         except NoResultFound:
 316             created = True
 317             db_source = DBSource(**unique)
 318             for key, value in rest.iteritems():
 319                 setattr(db_source, key, value)
 320             # XXX: set as default in postgres?
 321             db_source.install_date = datetime.now().date()
 322             session.add(db_source)
 323             session.flush()
 324
 325             # Add .dsc file. Other files will be added later.
 326             db_dsc_file = DSCFile()
 327             db_dsc_file.source = db_source
 328             db_dsc_file.poolfile = db_file_dsc
 329             session.add(db_dsc_file)
 330             session.flush()
 331
 332         if suite in db_source.suites:
 333             return db_source
 334
 335         db_source.suites.append(suite)
 336
 337         if not created:
 338             return db_source
 339
 340         ### Now add remaining files and copy them to the archive.
 341
 342         for hashed_file in source.files.itervalues():
 343             hashed_file_path = os.path.join(directory, hashed_file.filename)
 344             if os.path.exists(hashed_file_path):
 345                 db_file = self._install_file(directory, hashed_file, archive, component, source_name)
 346                 session.add(db_file)
 347             else:
 348                 db_file = self.get_file(hashed_file, source_name)
 349                 self._copy_file(db_file, archive, component, allow_tainted=allow_tainted)
 350
 351             db_dsc_file = DSCFile()
 352             db_dsc_file.source = db_source
 353             db_dsc_file.poolfile = db_file
 354             session.add(db_dsc_file)
 355
 356         session.flush()
 357
 358         # Importing is safe as we only arrive here when we did not find the source already installed earlier.
 359         import_metadata_into_db(db_source, session)
 360
 361         # Uploaders are the maintainer and co-maintainers from the Uploaders field
 362         db_source.uploaders.append(maintainer)
 363         if 'Uploaders' in control:
 364             def split_uploaders(field):
 365                 import re
 366                 for u in re.sub(">[ ]*,", ">\t", field).split("\t"):
 367                     yield u.strip()
 368
 369             for u in split_uploaders(control['Uploaders']):
 370                 db_source.uploaders.append(get_or_set_maintainer(u, session))
 371         session.flush()
 372
 373         return db_source
 374
 375     def _copy_file(self, db_file, archive, component, allow_tainted=False):
 376         """Copy a file to the given archive and component
 377
 378         @type  db_file: L{daklib.dbconn.PoolFile}
 379         @param db_file: file to copy
 380
 381         @type  archive: L{daklib.dbconn.Archive}
 382         @param archive: target archive
 383
 384         @type  component: L{daklib.dbconn.Archive}
 385         @param component: target component
 386
 387         @type  allow_tainted: bool
 388         @param allow_tainted: allow to copy from tainted archives (such as NEW)
 389         """
 390         session = self.session
 391
 392         if session.query(ArchiveFile).filter_by(archive=archive, component=component, file=db_file).first() is None:
 393             query = session.query(ArchiveFile).filter_by(file=db_file, component=component)
 394             if not allow_tainted:
 395                 query = query.join(Archive).filter(Archive.tainted == False)
 396
 397             source_af = query.first()
 398             if source_af is None:
 399                 raise ArchiveException('cp: Could not find {0} in component {1} in any archive.'.format(db_file.filename, component.component_name))
 400             target_af = ArchiveFile(archive, component, db_file)
 401             session.add(target_af)
 402             session.flush()
 403             self.fs.copy(source_af.path, target_af.path, link=False, mode=archive.mode)
 404
 405     def copy_binary(self, db_binary, suite, component, allow_tainted=False, extra_archives=None):
 406         """Copy a binary package to the given suite and component
 407
 408         @type  db_binary: L{daklib.dbconn.DBBinary}
 409         @param db_binary: binary to copy
 410
 411         @type  suite: L{daklib.dbconn.Suite}
 412         @param suite: target suite
 413
 414         @type  component: L{daklib.dbconn.Component}
 415         @param component: target component
 416
 417         @type  allow_tainted: bool
 418         @param allow_tainted: allow to copy from tainted archives (such as NEW)
 419
 420         @type  extra_archives: list of L{daklib.dbconn.Archive}
 421         @param extra_archives: extra archives to copy Built-Using sources from
 422         """
 423         session = self.session
 424         archive = suite.archive
 425         if archive.tainted:
 426             allow_tainted = True
 427
 428         filename = db_binary.poolfile.filename
 429
 430         # make sure source is present in target archive
 431         db_source = db_binary.source
 432         if session.query(ArchiveFile).filter_by(archive=archive, file=db_source.poolfile).first() is None:
 433             raise ArchiveException('{0}: cannot copy to {1}: source is not present in target archive'.format(filename, suite.suite_name))
 434
 435         # make sure built-using packages are present in target archive
 436         for db_source in db_binary.extra_sources:
 437             self._ensure_extra_source_exists(filename, db_source, archive, extra_archives=extra_archives)
 438
 439         # copy binary
 440         db_file = db_binary.poolfile
 441         self._copy_file(db_file, suite.archive, component, allow_tainted=allow_tainted)
 442         if suite not in db_binary.suites:
 443             db_binary.suites.append(suite)
 444         self.session.flush()
 445
 446     def copy_source(self, db_source, suite, component, allow_tainted=False):
 447         """Copy a source package to the given suite and component
 448
 449         @type  db_source: L{daklib.dbconn.DBSource}
 450         @param db_source: source to copy
 451
 452         @type  suite: L{daklib.dbconn.Suite}
 453         @param suite: target suite
 454
 455         @type  component: L{daklib.dbconn.Component}
 456         @param component: target component
 457
 458         @type  allow_tainted: bool
 459         @param allow_tainted: allow to copy from tainted archives (such as NEW)
 460         """
 461         archive = suite.archive
 462         if archive.tainted:
 463             allow_tainted = True
 464         for db_dsc_file in db_source.srcfiles:
 465             self._copy_file(db_dsc_file.poolfile, archive, component, allow_tainted=allow_tainted)
 466         if suite not in db_source.suites:
 467             db_source.suites.append(suite)
 468         self.session.flush()
 469
 470     def remove_file(self, db_file, archive, component):
 471         """Remove a file from a given archive and component
 472
 473         @type  db_file: L{daklib.dbconn.PoolFile}
 474         @param db_file: file to remove
 475
 476         @type  archive: L{daklib.dbconn.Archive}
 477         @param archive: archive to remove the file from
 478
 479         @type  component: L{daklib.dbconn.Component}
 480         @param component: component to remove the file from
 481         """
 482         af = self.session.query(ArchiveFile).filter_by(file=db_file, archive=archive, component=component)
 483         self.fs.unlink(af.path)
 484         self.session.delete(af)
 485
 486     def remove_binary(self, binary, suite):
 487         """Remove a binary from a given suite and component
 488
 489         @type  binary: L{daklib.dbconn.DBBinary}
 490         @param binary: binary to remove
 491
 492         @type  suite: L{daklib.dbconn.Suite}
 493         @param suite: suite to remove the package from
 494         """
 495         binary.suites.remove(suite)
 496         self.session.flush()
 497
 498     def remove_source(self, source, suite):
 499         """Remove a source from a given suite and component
 500
 501         @type  source: L{daklib.dbconn.DBSource}
 502         @param source: source to remove
 503
 504         @type  suite: L{daklib.dbconn.Suite}
 505         @param suite: suite to remove the package from
 506
 507         @raise ArchiveException: source package is still referenced by other
 508                                  binaries in the suite
 509         """
 510         session = self.session
 511
 512         query = session.query(DBBinary).filter_by(source=source) \
 513             .filter(DBBinary.suites.contains(suite))
 514         if query.first() is not None:
 515             raise ArchiveException('src:{0} is still used by binaries in suite {1}'.format(source.source, suite.suite_name))
 516
 517         source.suites.remove(suite)
 518         session.flush()
 519
 520     def commit(self):
 521         """commit changes"""
 522         try:
 523             self.session.commit()
 524             self.fs.commit()
 525         finally:
 526             self.session.rollback()
 527             self.fs.rollback()
 528
 529     def rollback(self):
 530         """rollback changes"""
 531         self.session.rollback()
 532         self.fs.rollback()
 533
 534     def __enter__(self):
 535         return self
 536
 537     def __exit__(self, type, value, traceback):
 538         if type is None:
 539             self.commit()
 540         else:
 541             self.rollback()
 542         return None
 543
 544 class ArchiveUpload(object):
 545     """handle an upload
 546
 547     This class can be used in a with-statement::
 548
 549        with ArchiveUpload(...) as upload:
 550           ...
 551
 552     Doing so will automatically run any required cleanup and also rollback the
 553     transaction if it was not committed.
 554     """
 555     def __init__(self, directory, changes, keyrings):
 556         self.transaction = ArchiveTransaction()
 557         """transaction used to handle the upload
 558         @type: L{daklib.archive.ArchiveTransaction}
 559         """
 560
 561         self.session = self.transaction.session
 562         """database session"""
 563
 564         self.original_directory = directory
 565         self.original_changes = changes
 566
 567         self.changes = None
 568         """upload to process
 569         @type: L{daklib.upload.Changes}
 570         """
 571
 572         self.directory = None
 573         """directory with temporary copy of files. set by C{prepare}
 574         @type: str
 575         """
 576
 577         self.keyrings = keyrings
 578
 579         self.fingerprint = self.session.query(Fingerprint).filter_by(fingerprint=changes.primary_fingerprint).one()
 580         """fingerprint of the key used to sign the upload
 581         @type: L{daklib.dbconn.Fingerprint}
 582         """
 583
 584         self.reject_reasons = []
 585         """reasons why the upload cannot by accepted
 586         @type: list of str
 587         """
 588
 589         self.warnings = []
 590         """warnings
 591         @note: Not used yet.
 592         @type: list of str
 593         """
 594
 595         self.final_suites = None
 596
 597         self.new = False
 598         """upload is NEW. set by C{check}
 599         @type: bool
 600         """
 601
 602         self._new_queue = self.session.query(PolicyQueue).filter_by(queue_name='new').one()
 603         self._new = self._new_queue.suite
 604
 605     def prepare(self):
 606         """prepare upload for further processing
 607
 608         This copies the files involved to a temporary directory.  If you use
 609         this method directly, you have to remove the directory given by the
 610         C{directory} attribute later on your own.
 611
 612         Instead of using the method directly, you can also use a with-statement::
 613
 614            with ArchiveUpload(...) as upload:
 615               ...
 616
 617         This will automatically handle any required cleanup.
 618         """
 619         assert self.directory is None
 620         assert self.original_changes.valid_signature
 621
 622         cnf = Config()
 623         session = self.transaction.session
 624
 625         self.directory = tempfile.mkdtemp(dir=cnf.get('Dir::TempPath'))
 626         with FilesystemTransaction() as fs:
 627             src = os.path.join(self.original_directory, self.original_changes.filename)
 628             dst = os.path.join(self.directory, self.original_changes.filename)
 629             fs.copy(src, dst)
 630
 631             self.changes = upload.Changes(self.directory, self.original_changes.filename, self.keyrings)
 632
 633             for f in self.changes.files.itervalues():
 634                 src = os.path.join(self.original_directory, f.filename)
 635                 dst = os.path.join(self.directory, f.filename)
 636                 if not os.path.exists(src):
 637                     continue
 638                 fs.copy(src, dst)
 639
 640             source = self.changes.source
 641             if source is not None:
 642                 for f in source.files.itervalues():
 643                     src = os.path.join(self.original_directory, f.filename)
 644                     dst = os.path.join(self.directory, f.filename)
 645                     if not os.path.exists(dst):
 646                         try:
 647                             db_file = self.transaction.get_file(f, source.dsc['Source'], check_hashes=False)
 648                             db_archive_file = session.query(ArchiveFile).filter_by(file=db_file).first()
 649                             fs.copy(db_archive_file.path, dst, symlink=True)
 650                         except KeyError:
 651                             # Ignore if get_file could not find it. Upload will
 652                             # probably be rejected later.
 653                             pass
 654
 655     def unpacked_source(self):
 656         """Path to unpacked source
 657
 658         Get path to the unpacked source. This method does unpack the source
 659         into a temporary directory under C{self.directory} if it has not
 660         been done so already.
 661
 662         @rtype:  str or C{None}
 663         @return: string giving the path to the unpacked source directory
 664                  or C{None} if no source was included in the upload.
 665         """
 666         assert self.directory is not None
 667
 668         source = self.changes.source
 669         if source is None:
 670             return None
 671         dsc_path = os.path.join(self.directory, source._dsc_file.filename)
 672
 673         sourcedir = os.path.join(self.directory, 'source')
 674         if not os.path.exists(sourcedir):
 675             subprocess.check_call(["dpkg-source", "--no-copy", "-x", dsc_path, sourcedir], shell=False)
 676         if not os.path.isdir(sourcedir):
 677             raise Exception("{0} is not a directory after extracting source package".format(sourcedir))
 678         return sourcedir
 679
 680     def _map_suite(self, suite_name):
 681         for rule in Config().value_list("SuiteMappings"):
 682             fields = rule.split()
 683             rtype = fields[0]
 684             if rtype == "map" or rtype == "silent-map":
 685                 (src, dst) = fields[1:3]
 686                 if src == suite_name:
 687                     suite_name = dst
 688                     if rtype != "silent-map":
 689                         self.warnings.append('Mapping {0} to {0}.'.format(src, dst))
 690             elif rtype == "ignore":
 691                 ignored = fields[1]
 692                 if suite_name == ignored:
 693                     self.warnings.append('Ignoring target suite {0}.'.format(ignored))
 694                     suite_name = None
 695             elif rtype == "reject":
 696                 rejected = fields[1]
 697                 if suite_name == rejected:
 698                     self.reject_reasons.append('Uploads to {0} are not accepted.'.format(suite))
 699             ## XXX: propup-version and map-unreleased not yet implemented
 700         return suite_name
 701
 702     def _mapped_suites(self):
 703         """Get target suites after mappings
 704
 705         @rtype:  list of L{daklib.dbconn.Suite}
 706         @return: list giving the mapped target suites of this upload
 707         """
 708         session = self.session
 709
 710         suite_names = []
 711         for dist in self.changes.distributions:
 712             suite_name = self._map_suite(dist)
 713             if suite_name is not None:
 714                 suite_names.append(suite_name)
 715
 716         suites = session.query(Suite).filter(Suite.suite_name.in_(suite_names))
 717         return suites
 718
 719     def _check_new(self, suite):
 720         """Check if upload is NEW
 721
 722         An upload is NEW if it has binary or source packages that do not have
 723         an override in C{suite} OR if it references files ONLY in a tainted
 724         archive (eg. when it references files in NEW).
 725
 726         @rtype:  bool
 727         @return: C{True} if the upload is NEW, C{False} otherwise
 728         """
 729         session = self.session
 730
 731         # Check for missing overrides
 732         for b in self.changes.binaries:
 733             override = self._binary_override(suite, b)
 734             if override is None:
 735                 return True
 736
 737         if self.changes.source is not None:
 738             override = self._source_override(suite, self.changes.source)
 739             if override is None:
 740                 return True
 741
 742         # Check if we reference a file only in a tainted archive
 743         files = self.changes.files.values()
 744         if self.changes.source is not None:
 745             files.extend(self.changes.source.files.values())
 746         for f in files:
 747             query = session.query(ArchiveFile).join(PoolFile).filter(PoolFile.sha1sum == f.sha1sum)
 748             query_untainted = query.join(Archive).filter(Archive.tainted == False)
 749
 750             in_archive = (query.first() is not None)
 751             in_untainted_archive = (query_untainted.first() is not None)
 752
 753             if in_archive and not in_untainted_archive:
 754                 return True
 755
 756     def _final_suites(self):
 757         session = self.session
 758
 759         mapped_suites = self._mapped_suites()
 760         final_suites = set()
 761
 762         for suite in mapped_suites:
 763             overridesuite = suite
 764             if suite.overridesuite is not None:
 765                 overridesuite = session.query(Suite).filter_by(suite_name=suite.overridesuite).one()
 766             if self._check_new(overridesuite):
 767                 self.new = True
 768             final_suites.add(suite)
 769
 770         return final_suites
 771
 772     def _binary_override(self, suite, binary):
 773         """Get override entry for a binary
 774
 775         @type  suite: L{daklib.dbconn.Suite}
 776         @param suite: suite to get override for
 777
 778         @type  binary: L{daklib.upload.Binary}
 779         @param binary: binary to get override for
 780
 781         @rtype:  L{daklib.dbconn.Override} or C{None}
 782         @return: override for the given binary or C{None}
 783         """
 784         if suite.overridesuite is not None:
 785             suite = self.session.query(Suite).filter_by(suite_name=suite.overridesuite).one()
 786
 787         query = self.session.query(Override).filter_by(suite=suite, package=binary.control['Package']) \
 788                 .join(Component).filter(Component.component_name == binary.component) \
 789                 .join(OverrideType).filter(OverrideType.overridetype == binary.type)
 790
 791         try:
 792             return query.one()
 793         except NoResultFound:
 794             return None
 795
 796     def _source_override(self, suite, source):
 797         """Get override entry for a source
 798
 799         @type  suite: L{daklib.dbconn.Suite}
 800         @param suite: suite to get override for
 801
 802         @type  source: L{daklib.upload.Source}
 803         @param source: source to get override for
 804
 805         @rtype:  L{daklib.dbconn.Override} or C{None}
 806         @return: override for the given source or C{None}
 807         """
 808         if suite.overridesuite is not None:
 809             suite = self.session.query(Suite).filter_by(suite_name=suite.overridesuite).one()
 810
 811         # XXX: component for source?
 812         query = self.session.query(Override).filter_by(suite=suite, package=source.dsc['Source']) \
 813                 .join(OverrideType).filter(OverrideType.overridetype == 'dsc')
 814
 815         try:
 816             return query.one()
 817         except NoResultFound:
 818             return None
 819
 820     def _binary_component(self, suite, binary, only_overrides=True):
 821         """get component for a binary
 822
 823         By default this will only look at overrides to get the right component;
 824         if C{only_overrides} is C{False} this method will also look at the
 825         Section field.
 826
 827         @type  suite: L{daklib.dbconn.Suite}
 828
 829         @type  binary: L{daklib.upload.Binary}
 830
 831         @type  only_overrides: bool
 832         @param only_overrides: only use overrides to get the right component
 833
 834         @rtype: L{daklib.dbconn.Component} or C{None}
 835         """
 836         override = self._binary_override(suite, binary)
 837         if override is not None:
 838             return override.component
 839         if only_overrides:
 840             return None
 841         return get_mapped_component(binary.component, self.session)
 842
 843     def check(self, force=False):
 844         """run checks against the upload
 845
 846         @type  force: bool
 847         @param force: ignore failing forcable checks
 848
 849         @rtype:  bool
 850         @return: C{True} if all checks passed, C{False} otherwise
 851         """
 852         # XXX: needs to be better structured.
 853         assert self.changes.valid_signature
 854
 855         try:
 856             for chk in (
 857                     checks.SignatureCheck,
 858                     checks.ChangesCheck,
 859                     checks.TransitionCheck,
 860                     checks.UploadBlockCheck,
 861                     checks.HashesCheck,
 862                     checks.SourceCheck,
 863                     checks.BinaryCheck,
 864                     checks.BinaryTimestampCheck,
 865                     checks.ACLCheck,
 866                     checks.SingleDistributionCheck,
 867                     checks.NoSourceOnlyCheck,
 868                     checks.LintianCheck,
 869                     ):
 870                 chk().check(self)
 871
 872             final_suites = self._final_suites()
 873             if len(final_suites) == 0:
 874                 self.reject_reasons.append('Ended with no suite to install to.')
 875                 return False
 876
 877             for chk in (
 878                     checks.SourceFormatCheck,
 879                     checks.SuiteArchitectureCheck,
 880                     checks.VersionCheck,
 881                     ):
 882                 for suite in final_suites:
 883                     chk().per_suite_check(self, suite)
 884
 885             if len(self.reject_reasons) != 0:
 886                 return False
 887
 888             self.final_suites = final_suites
 889             return True
 890         except checks.Reject as e:
 891             self.reject_reasons.append(unicode(e))
 892         except Exception as e:
 893             self.reject_reasons.append("Processing raised an exception: {0}.\n{1}".format(e, traceback.format_exc()))
 894         return False
 895
 896     def _install_to_suite(self, suite, source_component_func, binary_component_func, source_suites=None, extra_source_archives=None):
 897         """Install upload to the given suite
 898
 899         @type  suite: L{daklib.dbconn.Suite}
 900         @param suite: suite to install the package into. This is the real suite,
 901                       ie. after any redirection to NEW or a policy queue
 902
 903         @param source_component_func: function to get the L{daklib.dbconn.Component}
 904                                       for a L{daklib.upload.Source} object
 905
 906         @param binary_component_func: function to get the L{daklib.dbconn.Component}
 907                                       for a L{daklib.upload.Binary} object
 908
 909         @param source_suites: see L{daklib.archive.ArchiveTransaction.install_binary}
 910
 911         @param extra_source_archives: see L{daklib.archive.ArchiveTransaction.install_binary}
 912
 913         @return: tuple with two elements. The first is a L{daklib.dbconn.DBSource}
 914                  object for the install source or C{None} if no source was
 915                  included. The second is a list of L{daklib.dbconn.DBBinary}
 916                  objects for the installed binary packages.
 917         """
 918         # XXX: move this function to ArchiveTransaction?
 919
 920         control = self.changes.changes
 921         changed_by = get_or_set_maintainer(control.get('Changed-By', control['Maintainer']), self.session)
 922
 923         if source_suites is None:
 924             source_suites = self.session.query(Suite).join((VersionCheck, VersionCheck.reference_id == Suite.suite_id)).filter(VersionCheck.check == 'Enhances').filter(VersionCheck.suite == suite).subquery()
 925
 926         source = self.changes.source
 927         if source is not None:
 928             component = source_component_func(source)
 929             db_source = self.transaction.install_source(self.directory, source, suite, component, changed_by, fingerprint=self.fingerprint)
 930         else:
 931             db_source = None
 932
 933         db_binaries = []
 934         for binary in self.changes.binaries:
 935             component = binary_component_func(binary)
 936             db_binary = self.transaction.install_binary(self.directory, binary, suite, component, fingerprint=self.fingerprint, source_suites=source_suites, extra_source_archives=extra_source_archives)
 937             db_binaries.append(db_binary)
 938
 939         if suite.copychanges:
 940             src = os.path.join(self.directory, self.changes.filename)
 941             dst = os.path.join(suite.archive.path, 'dists', suite.suite_name, self.changes.filename)
 942             self.transaction.fs.copy(src, dst)
 943
 944         return (db_source, db_binaries)
 945
 946     def _install_changes(self):
 947         assert self.changes.valid_signature
 948         control = self.changes.changes
 949         session = self.transaction.session
 950         config = Config()
 951
 952         changelog_id = None
 953         # Only add changelog for sourceful uploads and binNMUs
 954         if 'source' in self.changes.architectures or re_bin_only_nmu.search(control['Version']):
 955             query = 'INSERT INTO changelogs_text (changelog) VALUES (:changelog) RETURNING id'
 956             changelog_id = session.execute(query, {'changelog': control['Changes']}).scalar()
 957             assert changelog_id is not None
 958
 959         db_changes = DBChange()
 960         db_changes.changesname = self.changes.filename
 961         db_changes.source = control['Source']
 962         db_changes.binaries = control.get('Binary', None)
 963         db_changes.architecture = control['Architecture']
 964         db_changes.version = control['Version']
 965         db_changes.distribution = control['Distribution']
 966         db_changes.urgency = control['Urgency']
 967         db_changes.maintainer = control['Maintainer']
 968         db_changes.changedby = control.get('Changed-By', control['Maintainer'])
 969         db_changes.date = control['Date']
 970         db_changes.fingerprint = self.fingerprint.fingerprint
 971         db_changes.changelog_id = changelog_id
 972         db_changes.closes = self.changes.closed_bugs
 973
 974         self.transaction.session.add(db_changes)
 975         self.transaction.session.flush()
 976
 977         return db_changes
 978
 979     def _install_policy(self, policy_queue, target_suite, db_changes, db_source, db_binaries):
 980         u = PolicyQueueUpload()
 981         u.policy_queue = policy_queue
 982         u.target_suite = target_suite
 983         u.changes = db_changes
 984         u.source = db_source
 985         u.binaries = db_binaries
 986         self.transaction.session.add(u)
 987         self.transaction.session.flush()
 988
 989         dst = os.path.join(policy_queue.path, self.changes.filename)
 990         self.transaction.fs.copy(self.changes.path, dst)
 991
 992         return u
 993
 994     def try_autobyhand(self):
 995         """Try AUTOBYHAND
 996
 997         Try to handle byhand packages automatically.
 998
 999         @rtype:  list of L{daklib.upload.HashedFile}
1000         @return: list of remaining byhand files
1001         """
1002         assert len(self.reject_reasons) == 0
1003         assert self.changes.valid_signature
1004         assert self.final_suites is not None
1005
1006         byhand = self.changes.byhand_files
1007         if len(byhand) == 0:
1008             return True
1009
1010         suites = list(self.final_suites)
1011         assert len(suites) == 1, "BYHAND uploads must be to a single suite"
1012         suite = suites[0]
1013
1014         cnf = Config()
1015         control = self.changes.changes
1016         automatic_byhand_packages = cnf.subtree("AutomaticByHandPackages")
1017
1018         remaining = []
1019         for f in byhand:
1020             parts = f.filename.split('_', 2)
1021             if len(parts) != 3:
1022                 print "W: unexpected byhand filename {0}. No automatic processing.".format(f.filename)
1023                 remaining.append(f)
1024                 continue
1025
1026             package, version, archext = parts
1027             arch, ext = archext.split('.', 1)
1028
1029             rule = automatic_byhand_packages.get(package)
1030             if rule is None:
1031                 remaining.append(f)
1032                 continue
1033
1034             if rule['Source'] != control['Source'] or rule['Section'] != f.section or rule['Extension'] != ext:
1035                 remaining.append(f)
1036                 continue
1037
1038             script = rule['Script']
1039             retcode = subprocess.call([script, os.path.join(self.directory, f.filename), control['Version'], arch, os.path.join(self.directory, self.changes.filename)], shell=False)
1040             if retcode != 0:
1041                 print "W: error processing {0}.".format(f.filename)
1042                 remaining.append(f)
1043
1044         return len(remaining) == 0
1045
1046     def _install_byhand(self, policy_queue_upload, hashed_file):
1047         """install byhand file
1048
1049         @type  policy_queue_upload: L{daklib.dbconn.PolicyQueueUpload}
1050
1051         @type  hashed_file: L{daklib.upload.HashedFile}
1052         """
1053         fs = self.transaction.fs
1054         session = self.transaction.session
1055         policy_queue = policy_queue_upload.policy_queue
1056
1057         byhand_file = PolicyQueueByhandFile()
1058         byhand_file.upload = policy_queue_upload
1059         byhand_file.filename = hashed_file.filename
1060         session.add(byhand_file)
1061         session.flush()
1062
1063         src = os.path.join(self.directory, hashed_file.filename)
1064         dst = os.path.join(policy_queue.path, hashed_file.filename)
1065         fs.copy(src, dst)
1066
1067         return byhand_file
1068
1069     def _do_bts_versiontracking(self):
1070         cnf = Config()
1071         fs = self.transaction.fs
1072
1073         btsdir = cnf.get('Dir::BTSVersionTrack')
1074         if btsdir is None or btsdir == '':
1075             return
1076
1077         base = os.path.join(btsdir, self.changes.filename[:-8])
1078
1079         # version history
1080         sourcedir = self.unpacked_source()
1081         if sourcedir is not None:
1082             fh = open(os.path.join(sourcedir, 'debian', 'changelog'), 'r')
1083             versions = fs.create("{0}.versions".format(base), mode=0o644)
1084             for line in fh.readlines():
1085                 if re_changelog_versions.match(line):
1086                     versions.write(line)
1087             fh.close()
1088             versions.close()
1089
1090         # binary -> source mapping
1091         debinfo = fs.create("{0}.debinfo".format(base), mode=0o644)
1092         for binary in self.changes.binaries:
1093             control = binary.control
1094             source_package, source_version = binary.source
1095             line = " ".join([control['Package'], control['Version'], source_package, source_version])
1096             print >>debinfo, line
1097         debinfo.close()
1098
1099     def _policy_queue(self, suite):
1100         if suite.policy_queue is not None:
1101             return suite.policy_queue
1102         return None
1103
1104     def install(self):
1105         """install upload
1106
1107         Install upload to a suite or policy queue.  This method does B{not}
1108         handle uploads to NEW.
1109
1110         You need to have called the C{check} method before calling this method.
1111         """
1112         assert len(self.reject_reasons) == 0
1113         assert self.changes.valid_signature
1114         assert self.final_suites is not None
1115         assert not self.new
1116
1117         db_changes = self._install_changes()
1118
1119         for suite in self.final_suites:
1120             overridesuite = suite
1121             if suite.overridesuite is not None:
1122                 overridesuite = self.session.query(Suite).filter_by(suite_name=suite.overridesuite).one()
1123
1124             policy_queue = self._policy_queue(suite)
1125
1126             redirected_suite = suite
1127             if policy_queue is not None:
1128                 redirected_suite = policy_queue.suite
1129
1130             source_component_func = lambda source: self._source_override(overridesuite, source).component
1131             binary_component_func = lambda binary: self._binary_component(overridesuite, binary)
1132
1133             (db_source, db_binaries) = self._install_to_suite(redirected_suite, source_component_func, binary_component_func, extra_source_archives=[suite.archive])
1134
1135             if policy_queue is not None:
1136                 self._install_policy(policy_queue, suite, db_changes, db_source, db_binaries)
1137
1138             # copy to build queues
1139             if policy_queue is None or policy_queue.send_to_build_queues:
1140                 source_suites = self.session.query(Suite).filter_by(suite_id=suite.suite_id).subquery()
1141                 for build_queue in suite.copy_queues:
1142                     self._install_to_suite(build_queue.suite, source_component_func, binary_component_func, source_suites=source_suites, extra_source_archives=[suite.archive])
1143
1144         self._do_bts_versiontracking()
1145
1146     def install_to_new(self):
1147         """install upload to NEW
1148
1149         Install upload to NEW.  This method does B{not} handle regular uploads
1150         to suites or policy queues.
1151
1152         You need to have called the C{check} method before calling this method.
1153         """
1154         # Uploads to NEW are special as we don't have overrides.
1155         assert len(self.reject_reasons) == 0
1156         assert self.changes.valid_signature
1157         assert self.final_suites is not None
1158
1159         source = self.changes.source
1160         binaries = self.changes.binaries
1161         byhand = self.changes.byhand_files
1162
1163         new_queue = self.transaction.session.query(PolicyQueue).filter_by(queue_name='new').one()
1164         if len(byhand) > 0:
1165             new_queue = self.transaction.session.query(PolicyQueue).filter_by(queue_name='byhand').one()
1166         new_suite = new_queue.suite
1167
1168         # we need a suite to guess components
1169         suites = list(self.final_suites)
1170         assert len(suites) == 1, "NEW uploads must be to a single suite"
1171         suite = suites[0]
1172
1173         def binary_component_func(binary):
1174             return self._binary_component(suite, binary, only_overrides=False)
1175
1176         # guess source component
1177         # XXX: should be moved into an extra method
1178         binary_component_names = set()
1179         for binary in binaries:
1180             component = binary_component_func(binary)
1181             binary_component_names.add(component.component_name)
1182         source_component_name = None
1183         for c in self.session.query(Component).order_by(Component.component_id):
1184             guess = c.component_name
1185             if guess in binary_component_names:
1186                 source_component_name = guess
1187                 break
1188         if source_component_name is None:
1189             raise Exception('Could not guess source component.')
1190         source_component = self.session.query(Component).filter_by(component_name=source_component_name).one()
1191         source_component_func = lambda source: source_component
1192
1193         db_changes = self._install_changes()
1194         (db_source, db_binaries) = self._install_to_suite(new_suite, source_component_func, binary_component_func, source_suites=True, extra_source_archives=[suite.archive])
1195         policy_upload = self._install_policy(new_queue, suite, db_changes, db_source, db_binaries)
1196
1197         for f in byhand:
1198             self._install_byhand(policy_upload, f)
1199
1200         self._do_bts_versiontracking()
1201
1202     def commit(self):
1203         """commit changes"""
1204         self.transaction.commit()
1205
1206     def rollback(self):
1207         """rollback changes"""
1208         self.transaction.rollback()
1209
1210     def __enter__(self):
1211         self.prepare()
1212         return self
1213
1214     def __exit__(self, type, value, traceback):
1215         if self.directory is not None:
1216             shutil.rmtree(self.directory)
1217             self.directory = None
1218         self.changes = None
1219         self.transaction.rollback()
1220         return None