@contact: Debian FTPMaster <ftpmaster@debian.org>
@copyright: 2008, 2009 Michael Casadevall <mcasadevall@debian.org>
@copyright: 2009 Mike O'Connor <stew@debian.org>
+@copyright: 2011 Torsten Werner <twerner@debian.org>
@license: GNU General Public License version 2 or later
"""
################################################################################
import sys
-import os
-import logging
-import math
-import gzip
-import threading
-import Queue
import apt_pkg
-from daklib import utils
-from daklib.binary import Binary
+
from daklib.config import Config
from daklib.dbconn import *
+from daklib.contents import BinaryContentsScanner, ContentsWriter, \
+ SourceContentsScanner
+from daklib import daklog
+from daklib import utils
################################################################################
def usage (exit_code=0):
- print """Usage: dak contents [options] command [arguments]
+ print """Usage: dak contents [options] subcommand
-COMMANDS
+SUBCOMMANDS
generate
generate Contents-$arch.gz files
- bootstrap
- scan the debs in the existing pool and load contents in the the database
+ scan-source
+ scan the source packages in the existing pool and load contents into
+ the src_contents table
- cruft
- remove files/paths which are no longer referenced by a binary
+ scan-binary
+ scan the (u)debs in the existing pool and load contents into the
+ bin_contents table
OPTIONS
-h, --help
show this help and exit
- -v, --verbose
- show verbose information messages
-
- -q, --quiet
- supress all output but errors
+OPTIONS for generate
+ -a, --archive=ARCHIVE
+ only operate on suites in the specified archive
-s, --suite={stable,testing,unstable,...}
- only operate on a single suite
-"""
- sys.exit(exit_code)
-
-################################################################################
+ only operate on specified suite names
-# where in dak.conf all of our configuration will be stowed
+ -c, --component={main,contrib,non-free}
+ only operate on specified components
-options_prefix = "Contents"
-options_prefix = "%s::Options" % options_prefix
+ -f, --force
+ write Contents files for suites marked as untouchable, too
-log = logging.getLogger()
+OPTIONS for scan-source and scan-binary
+ -l, --limit=NUMBER
+ maximum number of packages to scan
+"""
+ sys.exit(exit_code)
################################################################################
-# find me all of the contents for a given .deb
-contents_q = """PREPARE contents_q(int,int) as
- SELECT (p.path||'/'||n.file) AS fn,
- s.section,
- b.package,
- b.architecture
- FROM content_associations c join content_file_paths p ON (c.filepath=p.id)
- JOIN content_file_names n ON (c.filename=n.id)
- JOIN binaries b ON (b.id=c.binary_pkg)
- JOIN override o ON (o.package=b.package)
- JOIN section s ON (s.id=o.section)
- WHERE o.suite = $1 AND o.type = $2
- AND b.type='deb'
- ORDER BY fn"""
-
-# find me all of the contents for a given .udeb
-udeb_contents_q = """PREPARE udeb_contents_q(int,int,int) as
- SELECT (p.path||'/'||n.file) AS fn,
- s.section,
- b.package,
- b.architecture
- FROM content_associations c join content_file_paths p ON (c.filepath=p.id)
- JOIN content_file_names n ON (c.filename=n.id)
- JOIN binaries b ON (b.id=c.binary_pkg)
- JOIN override o ON (o.package=b.package)
- JOIN section s ON (s.id=o.section)
- WHERE o.suite = $1 AND o.type = $2
- AND s.id = $3
- AND b.type='udeb'
- ORDER BY fn"""
-
-# FROM content_file_paths p join content_associations c ON (c.filepath=p.id)
-# JOIN content_file_names n ON (c.filename=n.id)
-# JOIN binaries b ON (b.id=c.binary_pkg)
-# JOIN override o ON (o.package=b.package)
-# JOIN section s ON (s.id=o.section)
-# WHERE o.suite = $1 AND o.type = $2
-# AND s.id = $3
-# AND b.id in (SELECT ba.bin from bin_associations ba join binaries b on b.id=ba.bin where (b.architecture=$3 or b.architecture=$4)and ba.suite=$1 and b.type='udeb')
-# GROUP BY fn
-# ORDER BY fn;"""
-
-class EndOfContents(object):
- """
- A sentry object for the end of the filename stream
- """
- pass
-
-class GzippedContentWriter(object):
- """
- An object which will write contents out to a Contents-$arch.gz
- file on a separate thread
- """
-
- header = None # a class object holding the header section of contents file
-
- def __init__(self, filename):
- """
- @ptype filename: string
- @param filename: the name of the file to write to
- """
- self.queue = Queue.Queue()
- self.current_file = None
- self.first_package = True
- self.output = self.open_file(filename)
- self.thread = threading.Thread(target=self.write_thread,
- name='Contents writer')
- self.thread.start()
-
- def open_file(self, filename):
- """
- opens a gzip stream to the contents file
- """
- filepath = Config()["Contents::Root"] + filename
- filedir = os.path.dirname(filepath)
- if not os.path.isdir(filedir):
- os.makedirs(filedir)
- return gzip.open(filepath, "w")
-
- def write(self, filename, section, package):
- """
- enqueue content to be written to the file on a separate thread
- """
- self.queue.put((filename,section,package))
-
- def write_thread(self):
- """
- the target of a Thread which will do the actual writing
- """
- while True:
- next = self.queue.get()
- if isinstance(next, EndOfContents):
- self.output.write('\n')
- self.output.close()
- break
-
- (filename,section,package)=next
- if next != self.current_file:
- # this is the first file, so write the header first
- if not self.current_file:
- self.output.write(self._getHeader())
-
- self.output.write('\n%s\t' % filename)
- self.first_package = True
-
- self.current_file=filename
-
- if not self.first_package:
- self.output.write(',')
- else:
- self.first_package=False
- self.output.write('%s/%s' % (section,package))
-
- def finish(self):
- """
- enqueue the sentry object so that writers will know to terminate
- """
- self.queue.put(EndOfContents())
-
- @classmethod
- def _getHeader(self):
- """
- Internal method to return the header for Contents.gz files
-
- This is boilerplate which explains the contents of the file and how
- it can be used.
- """
- if not GzippedContentWriter.header:
- if Config().has_key("Contents::Header"):
- try:
- h = open(os.path.join( Config()["Dir::Templates"],
- Config()["Contents::Header"] ), "r")
- GzippedContentWriter.header = h.read()
- h.close()
- except:
- log.error( "error opening header file: %d\n%s" % (Config()["Contents::Header"],
- traceback.format_exc() ))
- GzippedContentWriter.header = None
- else:
- GzippedContentWriter.header = None
-
- return GzippedContentWriter.header
-
-
-class Contents(object):
- """
- Class capable of generating Contents-$arch.gz files
-
- Usage GenerateContents().generateContents( ["main","contrib","non-free"] )
- """
-
- def __init__(self):
- self.header = None
-
- def reject(self, message):
- log.error("E: %s" % message)
-
- def cruft(self):
- """
- remove files/paths from the DB which are no longer referenced
- by binaries and clean the temporary table
- """
- s = DBConn().session()
-
- # clear out all of the temporarily stored content associations
- # this should be run only after p-a has run. after a p-a
- # run we should have either accepted or rejected every package
- # so there should no longer be anything in the queue
- s.query(PendingContentsAssociation).delete()
-
- # delete any filenames we are storing which have no binary associated
- # with them
- cafq = s.query(ContentAssociation.filename_id).distinct()
- cfq = s.query(ContentFilename)
- cfq = cfq.filter(~ContentFilename.cafilename_id.in_(cafq))
- cfq.delete()
-
- # delete any paths we are storing which have no binary associated with
- # them
- capq = s.query(ContentAssociation.filepath_id).distinct()
- cpq = s.query(ContentFilepath)
- cpq = cpq.filter(~ContentFilepath.cafilepath_id.in_(capq))
- cpq.delete()
-
- s.commit()
-
-
- def bootstrap(self):
- """
- scan the existing debs in the pool to populate the contents database tables
- """
- pooldir = Config()[ 'Dir::Pool' ]
-
- s = DBConn().session()
-
- for suite in s.query(Suite).all():
- for arch in get_suite_architectures(suite.suite_name, skipsrc=True, skipall=True, session=s):
- q = s.query(BinAssociation).join(Suite)
- q = q.join(Suite).filter_by(suite_name=suite.suite_name)
- q = q.join(DBBinary).join(Architecture).filter_by(arch.arch_string)
- for ba in q:
- filename = ba.binary.poolfile.filename
- # Check for existing contents
- existingq = s.query(ContentAssociations).filter_by(binary_pkg=ba.binary_id).limit(1)
- if existingq.count() > 0:
- log.debug( "already imported: %s" % (filename))
- else:
- # We don't have existing contents so import them
- log.debug( "scanning: %s" % (filename) )
- debfile = os.path.join(pooldir, filename)
- if os.path.exists(debfile):
- Binary(debfile, self.reject).scan_package(ba.binary_id, True)
- else:
- log.error("missing .deb: %s" % filename)
-
-
- def generate(self):
- """
- Generate Contents-$arch.gz files for every available arch in each given suite.
- """
- cursor = DBConn().cursor()
-
- DBConn().prepare("contents_q", contents_q)
- DBConn().prepare("udeb_contents_q", udeb_contents_q)
-
- debtype_id=DBConn().get_override_type_id("deb")
- udebtype_id=DBConn().get_override_type_id("udeb")
-
- arch_all_id = DBConn().get_architecture_id("all")
- suites = self._suites()
-
-
- # Get our suites, and the architectures
- for suite in [i.lower() for i in suites]:
- suite_id = DBConn().get_suite_id(suite)
- arch_list = self._arches(cursor, suite_id)
-
- file_writers = {}
-
- try:
- for arch_id in arch_list:
- file_writers[arch_id[0]] = GzippedContentWriter("dists/%s/Contents-%s.gz" % (suite, arch_id[1]))
-
- cursor.execute("EXECUTE contents_q(%d,%d);" % (suite_id, debtype_id))
-
- while True:
- r = cursor.fetchone()
- if not r:
- break
-
- filename, section, package, arch = r
-
- if not file_writers.has_key( arch ):
- continue
+def write_all(cnf, archive_names = [], suite_names = [], component_names = [], force = None):
+ Logger = daklog.Logger('contents generate')
+ ContentsWriter.write_all(Logger, archive_names, suite_names, component_names, force)
+ Logger.close()
- if arch == arch_all_id:
- ## its arch all, so all contents files get it
- for writer in file_writers.values():
- writer.write(filename, section, package)
-
- else:
- file_writers[arch].write(filename, section, package)
-
- finally:
- # close all the files
- for writer in file_writers.values():
- writer.finish()
-
-
- # The MORE fun part. Ok, udebs need their own contents files, udeb, and udeb-nf (not-free)
- # This is HORRIBLY debian specific :-/
- for section, fn_pattern in [("debian-installer","dists/%s/Contents-udeb-%s.gz"),
- ("non-free/debian-installer", "dists/%s/Contents-udeb-nf-%s.gz")]:
-
- section_id = DBConn().get_section_id(section) # all udebs should be here)
- if section_id != -1:
-
- # Get our suites, and the architectures
- for suite in [i.lower() for i in suites]:
- suite_id = DBConn().get_suite_id(suite)
- arch_list = self._arches(cursor, suite_id)
-
- file_writers = {}
-
- try:
- for arch_id in arch_list:
- file_writers[arch_id[0]] = GzippedContentWriter(fn_pattern % (suite, arch_id[1]))
-
- cursor.execute("EXECUTE udeb_contents_q(%d,%d,%d)" % (suite_id, udebtype_id, section_id))
-
- while True:
- r = cursor.fetchone()
- if not r:
- break
-
- filename, section, package, arch = r
-
- if not file_writers.has_key( arch ):
- continue
-
- if arch == arch_all_id:
- ## its arch all, so all contents files get it
- for writer in file_writers.values():
- writer.write(filename, section, package)
+################################################################################
- else:
- file_writers[arch].write(filename, section, package)
- finally:
- # close all the files
- for writer in file_writers.values():
- writer.finish()
+def binary_scan_all(cnf, limit):
+ Logger = daklog.Logger('contents scan-binary')
+ result = BinaryContentsScanner.scan_all(limit)
+ processed = '%(processed)d packages processed' % result
+ remaining = '%(remaining)d packages remaining' % result
+ Logger.log([processed, remaining])
+ Logger.close()
+################################################################################
+def source_scan_all(cnf, limit):
+ Logger = daklog.Logger('contents scan-source')
+ result = SourceContentsScanner.scan_all(limit)
+ processed = '%(processed)d packages processed' % result
+ remaining = '%(remaining)d packages remaining' % result
+ Logger.log([processed, remaining])
+ Logger.close()
################################################################################
def main():
cnf = Config()
-
- arguments = [('h',"help", "%s::%s" % (options_prefix,"Help")),
- ('s',"suite", "%s::%s" % (options_prefix,"Suite"),"HasArg"),
- ('q',"quiet", "%s::%s" % (options_prefix,"Quiet")),
- ('v',"verbose", "%s::%s" % (options_prefix,"Verbose")),
+ cnf['Contents::Options::Help'] = ''
+ cnf['Contents::Options::Suite'] = ''
+ cnf['Contents::Options::Component'] = ''
+ cnf['Contents::Options::Limit'] = ''
+ cnf['Contents::Options::Force'] = ''
+ arguments = [('h', "help", 'Contents::Options::Help'),
+ ('a', 'archive', 'Contents::Options::Archive', 'HasArg'),
+ ('s', "suite", 'Contents::Options::Suite', "HasArg"),
+ ('c', "component", 'Contents::Options::Component', "HasArg"),
+ ('l', "limit", 'Contents::Options::Limit', "HasArg"),
+ ('f', "force", 'Contents::Options::Force'),
]
+ args = apt_pkg.parse_commandline(cnf.Cnf, arguments, sys.argv)
+ options = cnf.subtree('Contents::Options')
- commands = {'generate' : Contents.generate,
- 'bootstrap' : Contents.bootstrap,
- 'cruft' : Contents.cruft,
- }
+ if (len(args) != 1) or options['Help']:
+ usage()
- args = apt_pkg.ParseCommandLine(cnf.Cnf, arguments,sys.argv)
+ limit = None
+ if len(options['Limit']) > 0:
+ limit = int(options['Limit'])
- if (len(args) < 1) or not commands.has_key(args[0]):
- usage()
+ if args[0] == 'scan-source':
+ source_scan_all(cnf, limit)
+ return
- if cnf.has_key("%s::%s" % (options_prefix,"Help")):
- usage()
+ if args[0] == 'scan-binary':
+ binary_scan_all(cnf, limit)
+ return
- level=logging.INFO
- if cnf.has_key("%s::%s" % (options_prefix,"Quiet")):
- level=logging.ERROR
+ archive_names = utils.split_args(options['Archive'])
+ suite_names = utils.split_args(options['Suite'])
+ component_names = utils.split_args(options['Component'])
- elif cnf.has_key("%s::%s" % (options_prefix,"Verbose")):
- level=logging.DEBUG
+ force = bool(options['Force'])
+ if args[0] == 'generate':
+ write_all(cnf, archive_names, suite_names, component_names, force)
+ return
- logging.basicConfig( level=level,
- format='%(asctime)s %(levelname)s %(message)s',
- stream = sys.stderr )
+ usage()
- commands[args[0]](Contents())
if __name__ == '__main__':
main()