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             from daklib.textutils import split_uploaders
 365             for u in split_uploaders(control['Uploaders']):
 366                 db_source.uploaders.append(get_or_set_maintainer(u, session))
 367         session.flush()
 368
 369         return db_source
 370
 371     def _copy_file(self, db_file, archive, component, allow_tainted=False):
 372         """Copy a file to the given archive and component
 373
 374         @type  db_file: L{daklib.dbconn.PoolFile}
 375         @param db_file: file to copy
 376
 377         @type  archive: L{daklib.dbconn.Archive}
 378         @param archive: target archive
 379
 380         @type  component: L{daklib.dbconn.Archive}
 381         @param component: target component
 382
 383         @type  allow_tainted: bool
 384         @param allow_tainted: allow to copy from tainted archives (such as NEW)
 385         """
 386         session = self.session
 387
 388         if session.query(ArchiveFile).filter_by(archive=archive, component=component, file=db_file).first() is None:
 389             query = session.query(ArchiveFile).filter_by(file=db_file, component=component)
 390             if not allow_tainted:
 391                 query = query.join(Archive).filter(Archive.tainted == False)
 392
 393             source_af = query.first()
 394             if source_af is None:
 395                 raise ArchiveException('cp: Could not find {0} in component {1} in any archive.'.format(db_file.filename, component.component_name))
 396             target_af = ArchiveFile(archive, component, db_file)
 397             session.add(target_af)
 398             session.flush()
 399             self.fs.copy(source_af.path, target_af.path, link=False, mode=archive.mode)
 400
 401     def copy_binary(self, db_binary, suite, component, allow_tainted=False, extra_archives=None):
 402         """Copy a binary package to the given suite and component
 403
 404         @type  db_binary: L{daklib.dbconn.DBBinary}
 405         @param db_binary: binary to copy
 406
 407         @type  suite: L{daklib.dbconn.Suite}
 408         @param suite: target suite
 409
 410         @type  component: L{daklib.dbconn.Component}
 411         @param component: target component
 412
 413         @type  allow_tainted: bool
 414         @param allow_tainted: allow to copy from tainted archives (such as NEW)
 415
 416         @type  extra_archives: list of L{daklib.dbconn.Archive}
 417         @param extra_archives: extra archives to copy Built-Using sources from
 418         """
 419         session = self.session
 420         archive = suite.archive
 421         if archive.tainted:
 422             allow_tainted = True
 423
 424         filename = db_binary.poolfile.filename
 425
 426         # make sure source is present in target archive
 427         db_source = db_binary.source
 428         if session.query(ArchiveFile).filter_by(archive=archive, file=db_source.poolfile).first() is None:
 429             raise ArchiveException('{0}: cannot copy to {1}: source is not present in target archive'.format(filename, suite.suite_name))
 430
 431         # make sure built-using packages are present in target archive
 432         for db_source in db_binary.extra_sources:
 433             self._ensure_extra_source_exists(filename, db_source, archive, extra_archives=extra_archives)
 434
 435         # copy binary
 436         db_file = db_binary.poolfile
 437         self._copy_file(db_file, suite.archive, component, allow_tainted=allow_tainted)
 438         if suite not in db_binary.suites:
 439             db_binary.suites.append(suite)
 440         self.session.flush()
 441
 442     def copy_source(self, db_source, suite, component, allow_tainted=False):
 443         """Copy a source package to the given suite and component
 444
 445         @type  db_source: L{daklib.dbconn.DBSource}
 446         @param db_source: source to copy
 447
 448         @type  suite: L{daklib.dbconn.Suite}
 449         @param suite: target suite
 450
 451         @type  component: L{daklib.dbconn.Component}
 452         @param component: target component
 453
 454         @type  allow_tainted: bool
 455         @param allow_tainted: allow to copy from tainted archives (such as NEW)
 456         """
 457         archive = suite.archive
 458         if archive.tainted:
 459             allow_tainted = True
 460         for db_dsc_file in db_source.srcfiles:
 461             self._copy_file(db_dsc_file.poolfile, archive, component, allow_tainted=allow_tainted)
 462         if suite not in db_source.suites:
 463             db_source.suites.append(suite)
 464         self.session.flush()
 465
 466     def remove_file(self, db_file, archive, component):
 467         """Remove a file from a given archive and component
 468
 469         @type  db_file: L{daklib.dbconn.PoolFile}
 470         @param db_file: file to remove
 471
 472         @type  archive: L{daklib.dbconn.Archive}
 473         @param archive: archive to remove the file from
 474
 475         @type  component: L{daklib.dbconn.Component}
 476         @param component: component to remove the file from
 477         """
 478         af = self.session.query(ArchiveFile).filter_by(file=db_file, archive=archive, component=component)
 479         self.fs.unlink(af.path)
 480         self.session.delete(af)
 481
 482     def remove_binary(self, binary, suite):
 483         """Remove a binary from a given suite and component
 484
 485         @type  binary: L{daklib.dbconn.DBBinary}
 486         @param binary: binary to remove
 487
 488         @type  suite: L{daklib.dbconn.Suite}
 489         @param suite: suite to remove the package from
 490         """
 491         binary.suites.remove(suite)
 492         self.session.flush()
 493
 494     def remove_source(self, source, suite):
 495         """Remove a source from a given suite and component
 496
 497         @type  source: L{daklib.dbconn.DBSource}
 498         @param source: source to remove
 499
 500         @type  suite: L{daklib.dbconn.Suite}
 501         @param suite: suite to remove the package from
 502
 503         @raise ArchiveException: source package is still referenced by other
 504                                  binaries in the suite
 505         """
 506         session = self.session
 507
 508         query = session.query(DBBinary).filter_by(source=source) \
 509             .filter(DBBinary.suites.contains(suite))
 510         if query.first() is not None:
 511             raise ArchiveException('src:{0} is still used by binaries in suite {1}'.format(source.source, suite.suite_name))
 512
 513         source.suites.remove(suite)
 514         session.flush()
 515
 516     def commit(self):
 517         """commit changes"""
 518         try:
 519             self.session.commit()
 520             self.fs.commit()
 521         finally:
 522             self.session.rollback()
 523             self.fs.rollback()
 524
 525     def rollback(self):
 526         """rollback changes"""
 527         self.session.rollback()
 528         self.fs.rollback()
 529
 530     def __enter__(self):
 531         return self
 532
 533     def __exit__(self, type, value, traceback):
 534         if type is None:
 535             self.commit()
 536         else:
 537             self.rollback()
 538         return None
 539
 540 class ArchiveUpload(object):
 541     """handle an upload
 542
 543     This class can be used in a with-statement::
 544
 545        with ArchiveUpload(...) as upload:
 546           ...
 547
 548     Doing so will automatically run any required cleanup and also rollback the
 549     transaction if it was not committed.
 550     """
 551     def __init__(self, directory, changes, keyrings):
 552         self.transaction = ArchiveTransaction()
 553         """transaction used to handle the upload
 554         @type: L{daklib.archive.ArchiveTransaction}
 555         """
 556
 557         self.session = self.transaction.session
 558         """database session"""
 559
 560         self.original_directory = directory
 561         self.original_changes = changes
 562
 563         self.changes = None
 564         """upload to process
 565         @type: L{daklib.upload.Changes}
 566         """
 567
 568         self.directory = None
 569         """directory with temporary copy of files. set by C{prepare}
 570         @type: str
 571         """
 572
 573         self.keyrings = keyrings
 574
 575         self.fingerprint = self.session.query(Fingerprint).filter_by(fingerprint=changes.primary_fingerprint).one()
 576         """fingerprint of the key used to sign the upload
 577         @type: L{daklib.dbconn.Fingerprint}
 578         """
 579
 580         self.reject_reasons = []
 581         """reasons why the upload cannot by accepted
 582         @type: list of str
 583         """
 584
 585         self.warnings = []
 586         """warnings
 587         @note: Not used yet.
 588         @type: list of str
 589         """
 590
 591         self.final_suites = None
 592
 593         self.new = False
 594         """upload is NEW. set by C{check}
 595         @type: bool
 596         """
 597
 598         self._checked = False
 599         """checks passes. set by C{check}
 600         @type: bool
 601         """
 602
 603         self._new_queue = self.session.query(PolicyQueue).filter_by(queue_name='new').one()
 604         self._new = self._new_queue.suite
 605
 606     def warn(self, message):
 607         """add a warning message
 608
 609         Adds a warning message that can later be seen in C{self.warnings}
 610
 611         @type  message: string
 612         @param message: warning message
 613         """
 614         self.warnings.append(message)
 615
 616     def prepare(self):
 617         """prepare upload for further processing
 618
 619         This copies the files involved to a temporary directory.  If you use
 620         this method directly, you have to remove the directory given by the
 621         C{directory} attribute later on your own.
 622
 623         Instead of using the method directly, you can also use a with-statement::
 624
 625            with ArchiveUpload(...) as upload:
 626               ...
 627
 628         This will automatically handle any required cleanup.
 629         """
 630         assert self.directory is None
 631         assert self.original_changes.valid_signature
 632
 633         cnf = Config()
 634         session = self.transaction.session
 635
 636         self.directory = utils.temp_dirname(parent=cnf.get('Dir::TempPath'),
 637                                             mode=0o2750, group=cnf.unprivgroup)
 638         with FilesystemTransaction() as fs:
 639             src = os.path.join(self.original_directory, self.original_changes.filename)
 640             dst = os.path.join(self.directory, self.original_changes.filename)
 641             fs.copy(src, dst, mode=0o640)
 642
 643             self.changes = upload.Changes(self.directory, self.original_changes.filename, self.keyrings)
 644
 645             for f in self.changes.files.itervalues():
 646                 src = os.path.join(self.original_directory, f.filename)
 647                 dst = os.path.join(self.directory, f.filename)
 648                 if not os.path.exists(src):
 649                     continue
 650                 fs.copy(src, dst, mode=0o640)
 651
 652             source = self.changes.source
 653             if source is not None:
 654                 for f in source.files.itervalues():
 655                     src = os.path.join(self.original_directory, f.filename)
 656                     dst = os.path.join(self.directory, f.filename)
 657                     if not os.path.exists(dst):
 658                         try:
 659                             db_file = self.transaction.get_file(f, source.dsc['Source'], check_hashes=False)
 660                             db_archive_file = session.query(ArchiveFile).filter_by(file=db_file).first()
 661                             fs.copy(db_archive_file.path, dst, symlink=True)
 662                         except KeyError:
 663                             # Ignore if get_file could not find it. Upload will
 664                             # probably be rejected later.
 665                             pass
 666
 667     def unpacked_source(self):
 668         """Path to unpacked source
 669
 670         Get path to the unpacked source. This method does unpack the source
 671         into a temporary directory under C{self.directory} if it has not
 672         been done so already.
 673
 674         @rtype:  str or C{None}
 675         @return: string giving the path to the unpacked source directory
 676                  or C{None} if no source was included in the upload.
 677         """
 678         assert self.directory is not None
 679
 680         source = self.changes.source
 681         if source is None:
 682             return None
 683         dsc_path = os.path.join(self.directory, source._dsc_file.filename)
 684
 685         sourcedir = os.path.join(self.directory, 'source')
 686         if not os.path.exists(sourcedir):
 687             devnull = open('/dev/null', 'w')
 688             subprocess.check_call(["dpkg-source", "--no-copy", "--no-check", "-x", dsc_path, sourcedir], shell=False, stdout=devnull)
 689         if not os.path.isdir(sourcedir):
 690             raise Exception("{0} is not a directory after extracting source package".format(sourcedir))
 691         return sourcedir
 692
 693     def _map_suite(self, suite_name):
 694         for rule in Config().value_list("SuiteMappings"):
 695             fields = rule.split()
 696             rtype = fields[0]
 697             if rtype == "map" or rtype == "silent-map":
 698                 (src, dst) = fields[1:3]
 699                 if src == suite_name:
 700                     suite_name = dst
 701                     if rtype != "silent-map":
 702                         self.warnings.append('Mapping {0} to {1}.'.format(src, dst))
 703             elif rtype == "ignore":
 704                 ignored = fields[1]
 705                 if suite_name == ignored:
 706                     self.warnings.append('Ignoring target suite {0}.'.format(ignored))
 707                     suite_name = None
 708             elif rtype == "reject":
 709                 rejected = fields[1]
 710                 if suite_name == rejected:
 711                     self.reject_reasons.append('Uploads to {0} are not accepted.'.format(suite))
 712             ## XXX: propup-version and map-unreleased not yet implemented
 713         return suite_name
 714
 715     def _mapped_suites(self):
 716         """Get target suites after mappings
 717
 718         @rtype:  list of L{daklib.dbconn.Suite}
 719         @return: list giving the mapped target suites of this upload
 720         """
 721         session = self.session
 722
 723         suite_names = []
 724         for dist in self.changes.distributions:
 725             suite_name = self._map_suite(dist)
 726             if suite_name is not None:
 727                 suite_names.append(suite_name)
 728
 729         suites = session.query(Suite).filter(Suite.suite_name.in_(suite_names))
 730         return suites
 731
 732     def _check_new(self, suite):
 733         """Check if upload is NEW
 734
 735         An upload is NEW if it has binary or source packages that do not have
 736         an override in C{suite} OR if it references files ONLY in a tainted
 737         archive (eg. when it references files in NEW).
 738
 739         @rtype:  bool
 740         @return: C{True} if the upload is NEW, C{False} otherwise
 741         """
 742         session = self.session
 743         new = False
 744
 745         # Check for missing overrides
 746         for b in self.changes.binaries:
 747             override = self._binary_override(suite, b)
 748             if override is None:
 749                 self.warnings.append('binary:{0} is NEW.'.format(b.control['Package']))
 750                 new = True
 751
 752         if self.changes.source is not None:
 753             override = self._source_override(suite, self.changes.source)
 754             if override is None:
 755                 self.warnings.append('source:{0} is NEW.'.format(self.changes.source.control['Source']))
 756                 new = True
 757
 758         # Check if we reference a file only in a tainted archive
 759         files = self.changes.files.values()
 760         if self.changes.source is not None:
 761             files.extend(self.changes.source.files.values())
 762         for f in files:
 763             query = session.query(ArchiveFile).join(PoolFile).filter(PoolFile.sha1sum == f.sha1sum)
 764             query_untainted = query.join(Archive).filter(Archive.tainted == False)
 765
 766             in_archive = (query.first() is not None)
 767             in_untainted_archive = (query_untainted.first() is not None)
 768
 769             if in_archive and not in_untainted_archive:
 770                 self.warnings.append('{0} is only available in NEW.'.format(f.filename))
 771                 new = True
 772
 773         return new
 774
 775     def _final_suites(self):
 776         session = self.session
 777
 778         mapped_suites = self._mapped_suites()
 779         final_suites = set()
 780
 781         for suite in mapped_suites:
 782             overridesuite = suite
 783             if suite.overridesuite is not None:
 784                 overridesuite = session.query(Suite).filter_by(suite_name=suite.overridesuite).one()
 785             if self._check_new(overridesuite):
 786                 self.new = True
 787             final_suites.add(suite)
 788
 789         return final_suites
 790
 791     def _binary_override(self, suite, binary):
 792         """Get override entry for a binary
 793
 794         @type  suite: L{daklib.dbconn.Suite}
 795         @param suite: suite to get override for
 796
 797         @type  binary: L{daklib.upload.Binary}
 798         @param binary: binary to get override for
 799
 800         @rtype:  L{daklib.dbconn.Override} or C{None}
 801         @return: override for the given binary or C{None}
 802         """
 803         if suite.overridesuite is not None:
 804             suite = self.session.query(Suite).filter_by(suite_name=suite.overridesuite).one()
 805
 806         mapped_component = get_mapped_component(binary.component)
 807         if mapped_component is None:
 808             return None
 809
 810         query = self.session.query(Override).filter_by(suite=suite, package=binary.control['Package']) \
 811                 .join(Component).filter(Component.component_name == mapped_component.component_name) \
 812                 .join(OverrideType).filter(OverrideType.overridetype == binary.type)
 813
 814         try:
 815             return query.one()
 816         except NoResultFound:
 817             return None
 818
 819     def _source_override(self, suite, source):
 820         """Get override entry for a source
 821
 822         @type  suite: L{daklib.dbconn.Suite}
 823         @param suite: suite to get override for
 824
 825         @type  source: L{daklib.upload.Source}
 826         @param source: source to get override for
 827
 828         @rtype:  L{daklib.dbconn.Override} or C{None}
 829         @return: override for the given source or C{None}
 830         """
 831         if suite.overridesuite is not None:
 832             suite = self.session.query(Suite).filter_by(suite_name=suite.overridesuite).one()
 833
 834         # XXX: component for source?
 835         query = self.session.query(Override).filter_by(suite=suite, package=source.dsc['Source']) \
 836                 .join(OverrideType).filter(OverrideType.overridetype == 'dsc')
 837
 838         try:
 839             return query.one()
 840         except NoResultFound:
 841             return None
 842
 843     def _binary_component(self, suite, binary, only_overrides=True):
 844         """get component for a binary
 845
 846         By default this will only look at overrides to get the right component;
 847         if C{only_overrides} is C{False} this method will also look at the
 848         Section field.
 849
 850         @type  suite: L{daklib.dbconn.Suite}
 851
 852         @type  binary: L{daklib.upload.Binary}
 853
 854         @type  only_overrides: bool
 855         @param only_overrides: only use overrides to get the right component
 856
 857         @rtype: L{daklib.dbconn.Component} or C{None}
 858         """
 859         override = self._binary_override(suite, binary)
 860         if override is not None:
 861             return override.component
 862         if only_overrides:
 863             return None
 864         return get_mapped_component(binary.component, self.session)
 865
 866     def check(self, force=False):
 867         """run checks against the upload
 868
 869         @type  force: bool
 870         @param force: ignore failing forcable checks
 871
 872         @rtype:  bool
 873         @return: C{True} if all checks passed, C{False} otherwise
 874         """
 875         # XXX: needs to be better structured.
 876         assert self.changes.valid_signature
 877
 878         try:
 879             # Validate signatures and hashes before we do any real work:
 880             for chk in (
 881                     checks.SignatureCheck,
 882                     checks.ChangesCheck,
 883                     checks.HashesCheck,
 884                     checks.ExternalHashesCheck,
 885                     checks.SourceCheck,
 886                     checks.BinaryCheck,
 887                     checks.BinaryTimestampCheck,
 888                     checks.SingleDistributionCheck,
 889                     ):
 890                 chk().check(self)
 891
 892             final_suites = self._final_suites()
 893             if len(final_suites) == 0:
 894                 self.reject_reasons.append('No target suite found. Please check your target distribution and that you uploaded to the right archive.')
 895                 return False
 896
 897             self.final_suites = final_suites
 898
 899             for chk in (
 900                     checks.TransitionCheck,
 901                     checks.ACLCheck,
 902                     checks.NoSourceOnlyCheck,
 903                     checks.LintianCheck,
 904                     ):
 905                 chk().check(self)
 906
 907             for chk in (
 908                     checks.ACLCheck,
 909                     checks.SourceFormatCheck,
 910                     checks.SuiteArchitectureCheck,
 911                     checks.VersionCheck,
 912                     ):
 913                 for suite in final_suites:
 914                     chk().per_suite_check(self, suite)
 915
 916             if len(self.reject_reasons) != 0:
 917                 return False
 918
 919             self._checked = True
 920             return True
 921         except checks.Reject as e:
 922             self.reject_reasons.append(unicode(e))
 923         except Exception as e:
 924             self.reject_reasons.append("Processing raised an exception: {0}.\n{1}".format(e, traceback.format_exc()))
 925         return False
 926
 927     def _install_to_suite(self, suite, source_component_func, binary_component_func, source_suites=None, extra_source_archives=None):
 928         """Install upload to the given suite
 929
 930         @type  suite: L{daklib.dbconn.Suite}
 931         @param suite: suite to install the package into. This is the real suite,
 932                       ie. after any redirection to NEW or a policy queue
 933
 934         @param source_component_func: function to get the L{daklib.dbconn.Component}
 935                                       for a L{daklib.upload.Source} object
 936
 937         @param binary_component_func: function to get the L{daklib.dbconn.Component}
 938                                       for a L{daklib.upload.Binary} object
 939
 940         @param source_suites: see L{daklib.archive.ArchiveTransaction.install_binary}
 941
 942         @param extra_source_archives: see L{daklib.archive.ArchiveTransaction.install_binary}
 943
 944         @return: tuple with two elements. The first is a L{daklib.dbconn.DBSource}
 945                  object for the install source or C{None} if no source was
 946                  included. The second is a list of L{daklib.dbconn.DBBinary}
 947                  objects for the installed binary packages.
 948         """
 949         # XXX: move this function to ArchiveTransaction?
 950
 951         control = self.changes.changes
 952         changed_by = get_or_set_maintainer(control.get('Changed-By', control['Maintainer']), self.session)
 953
 954         if source_suites is None:
 955             source_suites = self.session.query(Suite).join((VersionCheck, VersionCheck.reference_id == Suite.suite_id)).filter(VersionCheck.check == 'Enhances').filter(VersionCheck.suite == suite).subquery()
 956
 957         source = self.changes.source
 958         if source is not None:
 959             component = source_component_func(source)
 960             db_source = self.transaction.install_source(self.directory, source, suite, component, changed_by, fingerprint=self.fingerprint)
 961         else:
 962             db_source = None
 963
 964         db_binaries = []
 965         for binary in self.changes.binaries:
 966             component = binary_component_func(binary)
 967             db_binary = self.transaction.install_binary(self.directory, binary, suite, component, fingerprint=self.fingerprint, source_suites=source_suites, extra_source_archives=extra_source_archives)
 968             db_binaries.append(db_binary)
 969
 970         if suite.copychanges:
 971             src = os.path.join(self.directory, self.changes.filename)
 972             dst = os.path.join(suite.archive.path, 'dists', suite.suite_name, self.changes.filename)
 973             self.transaction.fs.copy(src, dst, mode=suite.archive.mode)
 974
 975         return (db_source, db_binaries)
 976
 977     def _install_changes(self):
 978         assert self.changes.valid_signature
 979         control = self.changes.changes
 980         session = self.transaction.session
 981         config = Config()
 982
 983         changelog_id = None
 984         # Only add changelog for sourceful uploads and binNMUs
 985         if 'source' in self.changes.architectures or re_bin_only_nmu.search(control['Version']):
 986             query = 'INSERT INTO changelogs_text (changelog) VALUES (:changelog) RETURNING id'
 987             changelog_id = session.execute(query, {'changelog': control['Changes']}).scalar()
 988             assert changelog_id is not None
 989
 990         db_changes = DBChange()
 991         db_changes.changesname = self.changes.filename
 992         db_changes.source = control['Source']
 993         db_changes.binaries = control.get('Binary', None)
 994         db_changes.architecture = control['Architecture']
 995         db_changes.version = control['Version']
 996         db_changes.distribution = control['Distribution']
 997         db_changes.urgency = control['Urgency']
 998         db_changes.maintainer = control['Maintainer']
 999         db_changes.changedby = control.get('Changed-By', control['Maintainer'])
1000         db_changes.date = control['Date']
1001         db_changes.fingerprint = self.fingerprint.fingerprint
1002         db_changes.changelog_id = changelog_id
1003         db_changes.closes = self.changes.closed_bugs
1004
1005         self.transaction.session.add(db_changes)
1006         self.transaction.session.flush()
1007
1008         return db_changes
1009
1010     def _install_policy(self, policy_queue, target_suite, db_changes, db_source, db_binaries):
1011         u = PolicyQueueUpload()
1012         u.policy_queue = policy_queue
1013         u.target_suite = target_suite
1014         u.changes = db_changes
1015         u.source = db_source
1016         u.binaries = db_binaries
1017         self.transaction.session.add(u)
1018         self.transaction.session.flush()
1019
1020         dst = os.path.join(policy_queue.path, self.changes.filename)
1021         self.transaction.fs.copy(self.changes.path, dst, mode=policy_queue.change_perms)
1022
1023         return u
1024
1025     def try_autobyhand(self):
1026         """Try AUTOBYHAND
1027
1028         Try to handle byhand packages automatically.
1029
1030         @rtype:  list of L{daklib.upload.HashedFile}
1031         @return: list of remaining byhand files
1032         """
1033         assert len(self.reject_reasons) == 0
1034         assert self.changes.valid_signature
1035         assert self.final_suites is not None
1036         assert self._checked
1037
1038         byhand = self.changes.byhand_files
1039         if len(byhand) == 0:
1040             return True
1041
1042         suites = list(self.final_suites)
1043         assert len(suites) == 1, "BYHAND uploads must be to a single suite"
1044         suite = suites[0]
1045
1046         cnf = Config()
1047         control = self.changes.changes
1048         automatic_byhand_packages = cnf.subtree("AutomaticByHandPackages")
1049
1050         remaining = []
1051         for f in byhand:
1052             parts = f.filename.split('_', 2)
1053             if len(parts) != 3:
1054                 print "W: unexpected byhand filename {0}. No automatic processing.".format(f.filename)
1055                 remaining.append(f)
1056                 continue
1057
1058             package, version, archext = parts
1059             arch, ext = archext.split('.', 1)
1060
1061             try:
1062                 rule = automatic_byhand_packages.subtree(package)
1063             except KeyError:
1064                 remaining.append(f)
1065                 continue
1066
1067             if rule['Source'] != self.changes.source_name or rule['Section'] != f.section or rule['Extension'] != ext:
1068                 remaining.append(f)
1069                 continue
1070
1071             script = rule['Script']
1072             retcode = subprocess.call([script, os.path.join(self.directory, f.filename), control['Version'], arch, os.path.join(self.directory, self.changes.filename)], shell=False)
1073             if retcode != 0:
1074                 print "W: error processing {0}.".format(f.filename)
1075                 remaining.append(f)
1076
1077         return len(remaining) == 0
1078
1079     def _install_byhand(self, policy_queue_upload, hashed_file):
1080         """install byhand file
1081
1082         @type  policy_queue_upload: L{daklib.dbconn.PolicyQueueUpload}
1083
1084         @type  hashed_file: L{daklib.upload.HashedFile}
1085         """
1086         fs = self.transaction.fs
1087         session = self.transaction.session
1088         policy_queue = policy_queue_upload.policy_queue
1089
1090         byhand_file = PolicyQueueByhandFile()
1091         byhand_file.upload = policy_queue_upload
1092         byhand_file.filename = hashed_file.filename
1093         session.add(byhand_file)
1094         session.flush()
1095
1096         src = os.path.join(self.directory, hashed_file.filename)
1097         dst = os.path.join(policy_queue.path, hashed_file.filename)
1098         fs.copy(src, dst, mode=policy_queue.change_perms)
1099
1100         return byhand_file
1101
1102     def _do_bts_versiontracking(self):
1103         cnf = Config()
1104         fs = self.transaction.fs
1105
1106         btsdir = cnf.get('Dir::BTSVersionTrack')
1107         if btsdir is None or btsdir == '':
1108             return
1109
1110         base = os.path.join(btsdir, self.changes.filename[:-8])
1111
1112         # version history
1113         sourcedir = self.unpacked_source()
1114         if sourcedir is not None:
1115             fh = open(os.path.join(sourcedir, 'debian', 'changelog'), 'r')
1116             versions = fs.create("{0}.versions".format(base), mode=0o644)
1117             for line in fh.readlines():
1118                 if re_changelog_versions.match(line):
1119                     versions.write(line)
1120             fh.close()
1121             versions.close()
1122
1123         # binary -> source mapping
1124         debinfo = fs.create("{0}.debinfo".format(base), mode=0o644)
1125         for binary in self.changes.binaries:
1126             control = binary.control
1127             source_package, source_version = binary.source
1128             line = " ".join([control['Package'], control['Version'], control['Architecture'], source_package, source_version])
1129             print >>debinfo, line
1130         debinfo.close()
1131
1132     def _policy_queue(self, suite):
1133         if suite.policy_queue is not None:
1134             return suite.policy_queue
1135         return None
1136
1137     def install(self):
1138         """install upload
1139
1140         Install upload to a suite or policy queue.  This method does B{not}
1141         handle uploads to NEW.
1142
1143         You need to have called the C{check} method before calling this method.
1144         """
1145         assert len(self.reject_reasons) == 0
1146         assert self.changes.valid_signature
1147         assert self.final_suites is not None
1148         assert self._checked
1149         assert not self.new
1150
1151         db_changes = self._install_changes()
1152
1153         for suite in self.final_suites:
1154             overridesuite = suite
1155             if suite.overridesuite is not None:
1156                 overridesuite = self.session.query(Suite).filter_by(suite_name=suite.overridesuite).one()
1157
1158             policy_queue = self._policy_queue(suite)
1159
1160             redirected_suite = suite
1161             if policy_queue is not None:
1162                 redirected_suite = policy_queue.suite
1163
1164             # source can be in the suite we install to or any suite we enhance
1165             source_suite_ids = set([suite.suite_id, redirected_suite.suite_id])
1166             for enhanced_suite_id, in self.session.query(VersionCheck.reference_id) \
1167                     .filter(VersionCheck.suite_id.in_(source_suite_ids)) \
1168                     .filter(VersionCheck.check == 'Enhances'):
1169                 source_suite_ids.add(enhanced_suite_id)
1170
1171             source_suites = self.session.query(Suite).filter(Suite.suite_id.in_(source_suite_ids)).subquery()
1172
1173             source_component_func = lambda source: self._source_override(overridesuite, source).component
1174             binary_component_func = lambda binary: self._binary_component(overridesuite, binary)
1175
1176             (db_source, db_binaries) = self._install_to_suite(redirected_suite, source_component_func, binary_component_func, source_suites=source_suites, extra_source_archives=[suite.archive])
1177
1178             if policy_queue is not None:
1179                 self._install_policy(policy_queue, suite, db_changes, db_source, db_binaries)
1180
1181             # copy to build queues
1182             if policy_queue is None or policy_queue.send_to_build_queues:
1183                 for build_queue in suite.copy_queues:
1184                     self._install_to_suite(build_queue.suite, source_component_func, binary_component_func, source_suites=source_suites, extra_source_archives=[suite.archive])
1185
1186         self._do_bts_versiontracking()
1187
1188     def install_to_new(self):
1189         """install upload to NEW
1190
1191         Install upload to NEW.  This method does B{not} handle regular uploads
1192         to suites or policy queues.
1193
1194         You need to have called the C{check} method before calling this method.
1195         """
1196         # Uploads to NEW are special as we don't have overrides.
1197         assert len(self.reject_reasons) == 0
1198         assert self.changes.valid_signature
1199         assert self.final_suites is not None
1200
1201         source = self.changes.source
1202         binaries = self.changes.binaries
1203         byhand = self.changes.byhand_files
1204
1205         # we need a suite to guess components
1206         suites = list(self.final_suites)
1207         assert len(suites) == 1, "NEW uploads must be to a single suite"
1208         suite = suites[0]
1209
1210         # decide which NEW queue to use
1211         if suite.new_queue is None:
1212             new_queue = self.transaction.session.query(PolicyQueue).filter_by(queue_name='new').one()
1213         else:
1214             new_queue = suite.new_queue
1215         if len(byhand) > 0:
1216             # There is only one global BYHAND queue
1217             new_queue = self.transaction.session.query(PolicyQueue).filter_by(queue_name='byhand').one()
1218         new_suite = new_queue.suite
1219
1220
1221         def binary_component_func(binary):
1222             return self._binary_component(suite, binary, only_overrides=False)
1223
1224         # guess source component
1225         # XXX: should be moved into an extra method
1226         binary_component_names = set()
1227         for binary in binaries:
1228             component = binary_component_func(binary)
1229             binary_component_names.add(component.component_name)
1230         source_component_name = None
1231         for c in self.session.query(Component).order_by(Component.component_id):
1232             guess = c.component_name
1233             if guess in binary_component_names:
1234                 source_component_name = guess
1235                 break
1236         if source_component_name is None:
1237             source_component = self.session.query(Component).order_by(Component.component_id).first()
1238         else:
1239             source_component = self.session.query(Component).filter_by(component_name=source_component_name).one()
1240         source_component_func = lambda source: source_component
1241
1242         db_changes = self._install_changes()
1243         (db_source, db_binaries) = self._install_to_suite(new_suite, source_component_func, binary_component_func, source_suites=True, extra_source_archives=[suite.archive])
1244         policy_upload = self._install_policy(new_queue, suite, db_changes, db_source, db_binaries)
1245
1246         for f in byhand:
1247             self._install_byhand(policy_upload, f)
1248
1249         self._do_bts_versiontracking()
1250
1251     def commit(self):
1252         """commit changes"""
1253         self.transaction.commit()
1254
1255     def rollback(self):
1256         """rollback changes"""
1257         self.transaction.rollback()
1258
1259     def __enter__(self):
1260         self.prepare()
1261         return self
1262
1263     def __exit__(self, type, value, traceback):
1264         if self.directory is not None:
1265             shutil.rmtree(self.directory)
1266             self.directory = None
1267         self.changes = None
1268         self.transaction.rollback()
1269         return None