diff options
Diffstat (limited to 'devel/management/commands/reporead.py')
| -rw-r--r-- | devel/management/commands/reporead.py | 369 |
1 files changed, 197 insertions, 172 deletions
diff --git a/devel/management/commands/reporead.py b/devel/management/commands/reporead.py index a8875c7e..c444538b 100644 --- a/devel/management/commands/reporead.py +++ b/devel/management/commands/reporead.py @@ -13,12 +13,8 @@ Example: ./manage.py reporead i686 /tmp/core.db.tar.gz """ -from django.core.management.base import BaseCommand, CommandError -from django.contrib.auth.models import User -from django.db import transaction -from django.db.models import Q - -import codecs +from collections import defaultdict +import io import os import re import sys @@ -27,14 +23,12 @@ import logging from datetime import datetime from optparse import make_option -# New in 2.6, but fast (C implementation) in 2.7. We will use it over codecs if -# available. Eventually remove the codecs import completely. -io = None -try: - import io -except ImportError: - pass +from django.core.management.base import BaseCommand, CommandError +from django.contrib.auth.models import User +from django.db import connections, router, transaction +from django.db.utils import IntegrityError +from devel.utils import UserFinder from main.models import Arch, Package, PackageDepend, PackageFile, Repo from packages.models import Conflict, Provision, Replacement @@ -43,6 +37,8 @@ logging.basicConfig( format='%(asctime)s -> %(levelname)s: %(message)s', datefmt='%Y-%m-%d %H:%M:%S', stream=sys.stderr) +TRACE = 5 +logging.addLevelName(TRACE, 'TRACE') logger = logging.getLogger() class Command(BaseCommand): @@ -58,8 +54,6 @@ class Command(BaseCommand): def handle(self, arch=None, filename=None, **options): if not arch: raise CommandError('Architecture is required.') - if not validate_arch(arch): - raise CommandError('Specified architecture %s is not currently known.' % arch) if not filename: raise CommandError('Package database file is required.') filename = os.path.normpath(filename) @@ -74,18 +68,13 @@ class Command(BaseCommand): elif v == 2: logger.level = logging.DEBUG - import signal, traceback - handler = lambda sig, stack: traceback.print_stack(stack) - signal.signal(signal.SIGQUIT, handler) - signal.signal(signal.SIGUSR1, handler) - return read_repo(arch, filename, options) class Pkg(object): """An interim 'container' object for holding Arch package data.""" - bare = ( 'name', 'base', 'arch', 'desc', 'filename', - 'md5sum', 'url', 'builddate', 'packager' ) + bare = ( 'name', 'base', 'arch', 'filename', + 'md5sum', 'sha256sum', 'url', 'packager' ) number = ( 'csize', 'isize' ) collections = ( 'depends', 'optdepends', 'conflicts', 'provides', 'replaces', 'groups', 'license', 'files' ) @@ -97,12 +86,12 @@ class Pkg(object): self.ver = None self.rel = None self.epoch = 0 + self.pgpsig = None for k in self.bare + self.number: setattr(self, k, None) for k in self.collections: setattr(self, k, ()) - # So we can tell the diffence between a package with no files, and a DB - # without files entries + self.files = None self.has_files = False def populate(self, values): @@ -112,18 +101,31 @@ class Pkg(object): setattr(self, k, v[0][:254]) elif k in self.number: setattr(self, k, long(v[0])) + elif k in ('desc', 'pgpsig'): + # do NOT prune these values at all + setattr(self, k, v[0]) elif k == 'version': match = self.version_re.match(v[0]) self.ver = match.group(3) self.rel = match.group(4) if match.group(2): self.epoch = int(match.group(2)) + elif k == 'builddate': + try: + self.builddate = datetime.utcfromtimestamp(int(v[0])) + except ValueError: + try: + self.builddate = datetime.strptime(v[0], + '%a %b %d %H:%M:%S %Y') + except ValueError: + logger.warning('Package %s had unparsable build date %s', + self.name, v[0]) elif k == 'files': - self.files = v + self.files = tuple(v) self.has_files = True else: # anything left in collections - setattr(self, k, v) + setattr(self, k, tuple(v)) @property def full_version(self): @@ -133,55 +135,6 @@ class Pkg(object): return u'%s-%s' % (self.ver, self.rel) -def find_user(userstring): - ''' - Attempt to find the corresponding User object for a standard - packager string, e.g. something like - 'A. U. Thor <author@example.com>'. - We start by searching for a matching email address; we then move onto - matching by first/last name. If we cannot find a user, then return None. - ''' - if userstring in find_user.cache: - return find_user.cache[userstring] - matches = re.match(r'^([^<]+)? ?<([^>]*)>', userstring) - if not matches: - return None - - user = None - name = matches.group(1) - email = matches.group(2) - - def user_email(): - return User.objects.get(email=email) - def profile_email(): - return User.objects.get(userprofile__public_email=email) - def user_name(): - # yes, a bit odd but this is the easiest way since we can't always be - # sure how to split the name. Ensure every 'token' appears in at least - # one of the two name fields. - name_q = Q() - for token in name.split(): - # ignore quoted parts; e.g. nicknames in strings - if re.match(r'^[\'"].*[\'"]$', token): - continue - name_q &= (Q(first_name__icontains=token) | - Q(last_name__icontains=token)) - return User.objects.get(name_q) - - for matcher in (user_email, profile_email, user_name): - try: - user = matcher() - break - except (User.DoesNotExist, User.MultipleObjectsReturned): - pass - - find_user.cache[userstring] = user - return user - -# cached mappings of user strings -> User objects so we don't have to do the -# lookup more than strictly necessary. -find_user.cache = {} - DEPEND_RE = re.compile(r"^(.+?)((>=|<=|=|>|<)(.*))?$") def create_depend(package, dep_str, optional=False): @@ -234,6 +187,8 @@ def create_multivalued(dbpkg, repopkg, db_attr, repo_attr): for name in getattr(repopkg, repo_attr): collection.create(name=name) +finder = UserFinder() + def populate_pkg(dbpkg, repopkg, force=False, timestamp=None): if repopkg.base: dbpkg.pkgbase = repopkg.base @@ -247,18 +202,11 @@ def populate_pkg(dbpkg, repopkg, force=False, timestamp=None): dbpkg.filename = repopkg.filename dbpkg.compressed_size = repopkg.csize dbpkg.installed_size = repopkg.isize - try: - dbpkg.build_date = datetime.utcfromtimestamp(int(repopkg.builddate)) - except ValueError: - try: - dbpkg.build_date = datetime.strptime(repopkg.builddate, - '%a %b %d %H:%M:%S %Y') - except ValueError: - logger.warning('Package %s had unparsable build date %s', - repopkg.name, repopkg.builddate) + dbpkg.build_date = repopkg.builddate dbpkg.packager_str = repopkg.packager # attempt to find the corresponding django user for this string - dbpkg.packager = find_user(repopkg.packager) + dbpkg.packager = finder.find(repopkg.packager) + dbpkg.pgp_signature = repopkg.pgpsig if timestamp: dbpkg.flag_date = None @@ -287,10 +235,13 @@ def populate_pkg(dbpkg, repopkg, force=False, timestamp=None): create_multivalued(dbpkg, repopkg, 'licenses', 'license') +pkg_same_version = lambda pkg, dbpkg: pkg.ver == dbpkg.pkgver \ + and pkg.rel == dbpkg.pkgrel and pkg.epoch == dbpkg.epoch + + def populate_files(dbpkg, repopkg, force=False): if not force: - if dbpkg.pkgver != repopkg.ver or dbpkg.pkgrel != repopkg.rel \ - or dbpkg.epoch != repopkg.epoch: + if not pkg_same_version(repopkg, dbpkg): logger.info("DB version (%s) didn't match repo version " "(%s) for package %s, skipping file list addition", dbpkg.full_version, repopkg.full_version, dbpkg.pkgname) @@ -299,6 +250,7 @@ def populate_files(dbpkg, repopkg, force=False): pass elif dbpkg.files_last_update > dbpkg.last_update: return + # only delete files if we are reading a DB that contains them if repopkg.has_files: dbpkg.packagefile_set.all().delete() @@ -318,95 +270,155 @@ def populate_files(dbpkg, repopkg, force=False): dbpkg.files_last_update = datetime.utcnow() dbpkg.save() -@transaction.commit_on_success -def db_update(archname, reponame, pkgs, options): - """ - Parses a list and updates the Arch dev database accordingly. - Arguments: - pkgs -- A list of Pkg objects. +def select_pkg_for_update(dbpkg): + database = router.db_for_write(Package, instance=dbpkg) + connection = connections[database] + if 'sqlite' in connection.settings_dict['ENGINE'].lower(): + return dbpkg + new_pkg = Package.objects.raw( + 'SELECT * FROM packages WHERE id = %s FOR UPDATE', + [dbpkg.id]) + return list(new_pkg)[0] + + +def update_common(archname, reponame, pkgs, sanity_check=True): + with transaction.commit_manually(): + repository = Repo.objects.get(name__iexact=reponame) + architecture = Arch.objects.get(name__iexact=archname) + # no-arg order_by() removes even the default ordering; we don't need it + dbpkgs = Package.objects.filter( + arch=architecture, repo=repository).order_by() + + logger.info("%d packages in current web DB", len(dbpkgs)) + logger.info("%d packages in new updating DB", len(pkgs)) + + # Try to catch those random package deletions that make Eric so unhappy. + if len(dbpkgs): + dbpercent = 100.0 * len(pkgs) / len(dbpkgs) + else: + dbpercent = 0.0 + logger.info("DB package ratio: %.1f%%", dbpercent) + + # Fewer than 20 packages makes the percentage check unreliable, but it also + # means we expect the repo to fluctuate a lot. + msg = "Package database has %.1f%% the number of packages in the " \ + "web database" % dbpercent + if not sanity_check: + pass + elif repository.testing or repository.staging: + pass + elif len(dbpkgs) == 0 and len(pkgs) == 0: + pass + elif len(dbpkgs) > 20 and dbpercent < 50.0: + logger.error(msg) + raise Exception(msg) + elif dbpercent < 75.0: + logger.warning(msg) + + # If isolation level is repeatable-read, we need to ensure each package + # update starts a new transaction and re-queries the database as necessary + # to guard against simultaneous updates + transaction.commit() + return dbpkgs + +def db_update(archname, reponame, pkgs, force=False): """ - logger.info('Updating Arch: %s', archname) - force = options.get('force', False) - filesonly = options.get('filesonly', False) + Parses a list of packages and updates the packages database accordingly. + """ + logger.info('Updating %s (%s)', reponame, archname) + dbpkgs = update_common(archname, reponame, pkgs, True) repository = Repo.objects.get(name__iexact=reponame) architecture = Arch.objects.get(name__iexact=archname) - # no-arg order_by() removes even the default ordering; we don't need it - dbpkgs = Package.objects.filter( - arch=architecture, repo=repository).order_by() + # This makes our inner loop where we find packages by name *way* more # efficient by not having to go to the database for each package to # SELECT them by name. - dbdict = dict([(pkg.pkgname, pkg) for pkg in dbpkgs]) + dbdict = dict((dbpkg.pkgname, dbpkg) for dbpkg in dbpkgs) - logger.debug("Creating sets") dbset = set(dbdict.keys()) syncset = set([pkg.name for pkg in pkgs]) - logger.info("%d packages in current web DB", len(dbset)) - logger.info("%d packages in new updating db", len(syncset)) + in_sync_not_db = syncset - dbset logger.info("%d packages in sync not db", len(in_sync_not_db)) - - # Try to catch those random package deletions that make Eric so unhappy. - if len(dbset): - dbpercent = 100.0 * len(syncset) / len(dbset) - else: - dbpercent = 0.0 - logger.info("DB package ratio: %.1f%%", dbpercent) - - # Fewer than 20 packages makes the percentage check unreliable, but it also - # means we expect the repo to fluctuate a lot. - msg = "Package database has %.1f%% the number of packages in the " \ - "web database" % dbpercent - if len(dbset) == 0 and len(syncset) == 0: - pass - elif not filesonly and \ - len(dbset) > 20 and dbpercent < 50.0 and \ - not repository.testing and not repository.staging: - logger.error(msg) - raise Exception(msg) - elif dbpercent < 75.0: - logger.warning(msg) - - if not filesonly: - # packages in syncdb and not in database (add to database) - for p in [x for x in pkgs if x.name in in_sync_not_db]: - logger.info("Adding package %s", p.name) - pkg = Package(pkgname = p.name, arch = architecture, repo = repository) - populate_pkg(pkg, p, timestamp=datetime.utcnow()) - - # packages in database and not in syncdb (remove from database) - in_db_not_sync = dbset - syncset - for p in in_db_not_sync: - logger.info("Removing package %s from database", p) - dbp = dbdict[p] - dbp.delete() + # packages in syncdb and not in database (add to database) + for pkg in (pkg for pkg in pkgs if pkg.name in in_sync_not_db): + logger.info("Adding package %s", pkg.name) + dbpkg = Package(pkgname=pkg.name, arch=architecture, repo=repository) + try: + with transaction.commit_on_success(): + populate_pkg(dbpkg, pkg, timestamp=datetime.utcnow()) + except IntegrityError: + logger.warning("Could not add package %s; " + "not fatal if another thread beat us to it.", + pkg.name, exc_info=True) + + # packages in database and not in syncdb (remove from database) + for pkgname in (dbset - syncset): + logger.info("Removing package %s", pkgname) + dbpkg = dbdict[pkgname] + with transaction.commit_on_success(): + # no race condition here as long as simultaneous threads both + # issue deletes; second delete will be a no-op + dbpkg.delete() # packages in both database and in syncdb (update in database) pkg_in_both = syncset & dbset - for p in [x for x in pkgs if x.name in pkg_in_both]: - logger.debug("Looking for package updates") - dbp = dbdict[p.name] + for pkg in (x for x in pkgs if x.name in pkg_in_both): + logger.debug("Checking package %s", pkg.name) + dbpkg = dbdict[pkg.name] timestamp = None # for a force, we don't want to update the timestamp. # for a non-force, we don't want to do anything at all. - if filesonly: - pass - elif p.ver == dbp.pkgver and p.rel == dbp.pkgrel \ - and p.epoch == dbp.epoch: + if pkg_same_version(pkg, dbpkg): if not force: continue else: timestamp = datetime.utcnow() - if filesonly: - logger.debug("Checking files for package %s in database", p.name) - populate_files(dbp, p, force=force) - else: - logger.info("Updating package %s in database", p.name) - populate_pkg(dbp, p, force=force, timestamp=timestamp) - logger.info('Finished updating Arch: %s', archname) + # The odd select_for_update song and dance here are to ensure + # simultaneous updates don't happen on a package, causing + # files/depends/all related items to be double-imported. + with transaction.commit_on_success(): + # TODO Django 1.4 select_for_update() will work once released + dbpkg = select_pkg_for_update(dbpkg) + if pkg_same_version(pkg, dbpkg): + logger.debug("Package %s was already updated", pkg.name) + continue + logger.info("Updating package %s", pkg.name) + populate_pkg(dbpkg, pkg, force=force, timestamp=timestamp) + + logger.info('Finished updating arch: %s', archname) + + +def filesonly_update(archname, reponame, pkgs, force=False): + """ + Parses a list of packages and updates the packages database accordingly. + """ + logger.info('Updating files for %s (%s)', reponame, archname) + dbpkgs = update_common(archname, reponame, pkgs, False) + dbdict = dict((dbpkg.pkgname, dbpkg) for dbpkg in dbpkgs) + dbset = set(dbdict.keys()) + + for pkg in (pkg for pkg in pkgs if pkg.name in dbset): + dbpkg = dbdict[pkg.name] + + # The odd select_for_update song and dance here are to ensure + # simultaneous updates don't happen on a package, causing + # files to be double-imported. + with transaction.commit_on_success(): + if not dbpkg.files_last_update or not dbpkg.last_update: + pass + elif dbpkg.files_last_update > dbpkg.last_update: + logger.debug("Files for %s are up to date", pkg.name) + continue + # TODO Django 1.4 select_for_update() will work once released + dbpkg = select_pkg_for_update(dbpkg) + logger.debug("Checking files for package %s", pkg.name) + populate_files(dbpkg, pkg, force=force) + + logger.info('Finished updating arch: %s', archname) def parse_info(iofile): @@ -421,7 +433,7 @@ def parse_info(iofile): continue elif line.startswith('%') and line.endswith('%'): blockname = line[1:-1].lower() - logger.debug("Parsing package block %s", blockname) + logger.log(TRACE, "Parsing package block %s", blockname) store[blockname] = [] elif blockname: store[blockname].append(line) @@ -454,60 +466,73 @@ def parse_repo(repopath): repodb = tarfile.open(repopath, "r") logger.debug("Starting package parsing") dbfiles = ('desc', 'depends', 'files') - pkgs = {} + newpkg = lambda: Pkg(reponame) + pkgs = defaultdict(newpkg) for tarinfo in repodb.getmembers(): if tarinfo.isreg(): pkgid, fname = os.path.split(tarinfo.name) if fname not in dbfiles: continue data_file = repodb.extractfile(tarinfo) - if io is None: - data_file = codecs.EncodedFile(data_file, 'utf-8') - else: - data_file = io.TextIOWrapper(io.BytesIO(data_file.read()), - encoding='utf=8') + data_file = io.TextIOWrapper(io.BytesIO(data_file.read()), + encoding='UTF-8') try: - data = parse_info(data_file) - p = pkgs.setdefault(pkgid, Pkg(reponame)) - p.populate(data) - except UnicodeDecodeError, e: + pkgs[pkgid].populate(parse_info(data_file)) + except UnicodeDecodeError: logger.warn("Could not correctly decode %s, skipping file", tarinfo.name) data_file.close() + del data_file - logger.debug("Done parsing file %s", fname) + logger.debug("Done parsing file %s/%s", pkgid, fname) repodb.close() logger.info("Finished repo parsing, %d total packages", len(pkgs)) return (reponame, pkgs.values()) -def validate_arch(archname): +def locate_arch(arch): "Check if arch is valid." - return Arch.objects.filter(name__iexact=archname).exists() + if isinstance(arch, Arch): + return arch + try: + return Arch.objects.get(name__iexact=arch) + except Arch.DoesNotExist: + raise CommandError( + 'Specified architecture %s is not currently known.' % arch) + def read_repo(primary_arch, repo_file, options): """ Parses repo.db.tar.gz file and returns exit status. """ + # always returns an Arch object, regardless of what is passed in + primary_arch = locate_arch(primary_arch) + force = options.get('force', False) + filesonly = options.get('filesonly', False) + repo, packages = parse_repo(repo_file) # group packages by arch -- to handle noarch stuff packages_arches = {} for arch in Arch.objects.filter(agnostic=True): packages_arches[arch.name] = [] - packages_arches[primary_arch] = [] + packages_arches[primary_arch.name] = [] for package in packages: if package.arch in packages_arches: packages_arches[package.arch].append(package) else: # we don't include mis-arched packages - logger.warning("Package %s arch = %s", - package.name,package.arch) - logger.info('Starting database updates.') + logger.warning("Package %s arch = %s", package.name, package.arch) + del packages + + logger.info('Starting database updates for %s.', repo_file) for arch in sorted(packages_arches.keys()): - db_update(arch, repo, packages_arches[arch], options) - logger.info('Finished database updates.') + if filesonly: + filesonly_update(arch, repo, packages_arches[arch], force) + else: + db_update(arch, repo, packages_arches[arch], force) + logger.info('Finished database updates for %s.', repo_file) return 0 # vim: set ts=4 sw=4 et: |