4 Check for obsolete binary packages
6 @contact: Debian FTP Master <ftpmaster@debian.org>
7 @copyright: 2000-2006 James Troup <james@nocrew.org>
8 @copyright: 2009 Torsten Werner <twerner@debian.org>
9 @license: GNU General Public License version 2 or later
12 # This program is free software; you can redistribute it and/or modify
13 # it under the terms of the GNU General Public License as published by
14 # the Free Software Foundation; either version 2 of the License, or
15 # (at your option) any later version.
17 # This program is distributed in the hope that it will be useful,
18 # but WITHOUT ANY WARRANTY; without even the implied warranty of
19 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
20 # GNU General Public License for more details.
22 # You should have received a copy of the GNU General Public License
23 # along with this program; if not, write to the Free Software
24 # Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
26 ################################################################################
28 # ``If you're claiming that's a "problem" that needs to be "fixed",
29 # you might as well write some letters to God about how unfair entropy
30 # is while you're at it.'' -- 20020802143104.GA5628@azure.humbug.org.au
32 ## TODO: fix NBS looping for version, implement Dubious NBS, fix up output of
33 ## duplicate source package stuff, improve experimental ?, add overrides,
34 ## avoid ANAIS for duplicated packages
36 ################################################################################
38 import commands, os, sys, re
41 from daklib.config import Config
42 from daklib.dbconn import *
43 from daklib import utils
44 from daklib.regexes import re_extract_src_version
45 from daklib.cruft import *
47 ################################################################################
49 no_longer_in_suite = {}; # Really should be static to add_nbs, but I'm lazy
54 ################################################################################
56 def usage(exit_code=0):
57 print """Usage: dak cruft-report
58 Check for obsolete or duplicated packages.
60 -h, --help show this help and exit.
61 -m, --mode=MODE chose the MODE to run in (full or daily).
62 -s, --suite=SUITE check suite SUITE.
63 -w, --wanna-build-dump where to find the copies of http://buildd.debian.org/stats/*.txt"""
66 ################################################################################
68 def add_nbs(nbs_d, source, version, package, suite_id, session):
69 # Ensure the package is still in the suite (someone may have already removed it)
70 if no_longer_in_suite.has_key(package):
73 q = session.execute("""SELECT b.id FROM binaries b, bin_associations ba
74 WHERE ba.bin = b.id AND ba.suite = :suite_id
75 AND b.package = :package LIMIT 1""", {'suite_id': suite_id,
78 no_longer_in_suite[package] = ""
81 nbs_d.setdefault(source, {})
82 nbs_d[source].setdefault(version, {})
83 nbs_d[source][version][package] = ""
85 ################################################################################
87 # Check for packages built on architectures they shouldn't be.
88 def do_anais(architecture, binaries_list, source, session):
89 if architecture == "any" or architecture == "all":
94 for arch in architecture.split():
95 architectures[arch.strip()] = ""
96 for binary in binaries_list:
97 q = session.execute("""SELECT a.arch_string, b.version
98 FROM binaries b, bin_associations ba, architecture a
99 WHERE ba.suite = :suiteid AND ba.bin = b.id
100 AND b.architecture = a.id AND b.package = :package""",
101 {'suiteid': suite_id, 'package': binary})
107 if architectures.has_key(arch):
108 versions.append(version)
109 versions.sort(apt_pkg.VersionCompare)
111 latest_version = versions.pop()
113 latest_version = None
114 # Check for 'invalid' architectures
119 if not architectures.has_key(arch):
120 versions_d.setdefault(version, [])
121 versions_d[version].append(arch)
124 anais_output += "\n (*) %s_%s [%s]: %s\n" % (binary, latest_version, source, architecture)
125 versions = versions_d.keys()
126 versions.sort(apt_pkg.VersionCompare)
127 for version in versions:
128 arches = versions_d[version]
130 anais_output += " o %s: %s\n" % (version, ", ".join(arches))
134 ################################################################################
136 # Check for out-of-date binaries on architectures that do not want to build that
137 # package any more, and have them listed as Not-For-Us
138 def do_nfu(nfu_packages):
143 for architecture in nfu_packages:
144 a2p[architecture] = []
145 for (package,bver,sver) in nfu_packages[architecture]:
146 output += " * [%s] does not want %s (binary %s, source %s)\n" % (architecture, package, bver, sver)
147 a2p[architecture].append(package)
151 print "Obsolete by Not-For-Us"
152 print "----------------------"
156 print "Suggested commands:"
157 for architecture in a2p:
158 if a2p[architecture]:
159 print (" dak rm -m \"[auto-cruft] NFU\" -s %s -a %s -b %s" %
160 (suite.suite_name, architecture, " ".join(a2p[architecture])))
163 def parse_nfu(architecture):
165 # utils/hpodder_1.1.5.0: Not-For-Us [optional:out-of-date]
166 r = re.compile("^\w+/([^_]+)_.*: Not-For-Us")
170 filename = "%s/%s-all.txt" % (cnf["Cruft-Report::Options::Wanna-Build-Dump"], architecture)
172 # Not all architectures may have a wanna-build dump, so we want to ignore missin
174 if os.path.exists(filename):
175 f = utils.open_file(filename)
186 utils.warn("No wanna-build dump file for architecture %s" % architecture)
189 ################################################################################
191 def do_newer_version(lowersuite_name, highersuite_name, code, session):
192 list = newer_version(lowersuite_name, highersuite_name, session)
195 title = "Newer version in %s" % lowersuite_name
197 print "-" * len(title)
200 (source, higher_version, lower_version) = i
201 print " o %s (%s, %s)" % (source, higher_version, lower_version)
202 nv_to_remove.append(source)
204 print "Suggested command:"
205 print " dak rm -m \"[auto-cruft] %s\" -s %s %s" % (code, highersuite_name,
206 " ".join(nv_to_remove))
209 ################################################################################
211 def queryWithoutSource(suite_id, session):
212 """searches for arch: all packages from suite that do no longer
213 reference a source package in the same suite
215 subquery unique_binaries: selects all packages with only 1 version
216 in suite since 'dak rm' does not allow to specify version numbers"""
219 with unique_binaries as
220 (select package, max(version) as version, max(source) as source
221 from bin_associations_binaries
222 where architecture = 2 and suite = :suite_id
223 group by package having count(*) = 1)
224 select ub.package, ub.version
225 from unique_binaries ub
226 left join src_associations_src sas
227 on ub.source = sas.src and sas.suite = :suite_id
229 order by ub.package"""
230 return session.execute(query, { 'suite_id': suite_id })
232 def reportWithoutSource(suite_name, suite_id, session):
233 rows = queryWithoutSource(suite_id, session)
234 title = 'packages without source in suite %s' % suite_name
235 if rows.rowcount > 0:
236 print '%s\n%s\n' % (title, '-' * len(title))
237 message = '"[auto-cruft] no longer built from source"'
239 (package, version) = row
240 print "* package %s in version %s is no longer built from source" % \
242 print " - suggested command:"
243 print " dak rm -m %s -s %s -a all -p -R -b %s\n" % \
244 (message, suite_name, package)
246 def queryNewerAll(suite_name, session):
247 """searches for arch != all packages that have an arch == all
248 package with a higher version in the same suite"""
251 select bab1.package, bab1.version as oldver,
252 array_to_string(array_agg(a.arch_string), ',') as oldarch,
253 bab2.version as newver
254 from bin_associations_binaries bab1
255 join bin_associations_binaries bab2
256 on bab1.package = bab2.package and bab1.version < bab2.version and
257 bab1.suite = bab2.suite and bab1.architecture > 2 and
258 bab2.architecture = 2
259 join architecture a on bab1.architecture = a.id
260 join suite s on bab1.suite = s.id
261 where s.suite_name = :suite_name
262 group by bab1.package, oldver, bab1.suite, newver"""
263 return session.execute(query, { 'suite_name': suite_name })
265 def reportNewerAll(suite_name, session):
266 rows = queryNewerAll(suite_name, session)
267 title = 'obsolete arch any packages in suite %s' % suite_name
268 if rows.rowcount > 0:
269 print '%s\n%s\n' % (title, '-' * len(title))
270 message = '"[auto-cruft] obsolete arch any package"'
272 (package, oldver, oldarch, newver) = row
273 print "* package %s is arch any in version %s but arch all in version %s" % \
274 (package, oldver, newver)
275 print " - suggested command:"
276 print " dak rm -m %s -s %s -a %s -p -b %s\n" % \
277 (message, suite_name, oldarch, package)
279 def queryNBS(suite_id, session):
280 """This one is really complex. It searches arch != all packages that
281 are no longer built from current source packages in suite.
283 temp table unique_binaries: will be populated with packages that
284 have only one version in suite because 'dak rm' does not allow
285 specifying version numbers
287 temp table newest_binaries: will be populated with packages that are
288 built from current sources
290 subquery uptodate_arch: returns all architectures built from current
293 subquery unique_binaries_uptodate_arch: returns all packages in
294 architectures from uptodate_arch
296 subquery unique_binaries_uptodate_arch_agg: same as
297 unique_binaries_uptodate_arch but with column architecture
300 subquery uptodate_packages: similar to uptodate_arch but returns all
301 packages built from current sources
303 subquery outdated_packages: returns all packages with architectures
304 no longer built from current source
308 create temp table unique_binaries (
309 package text not null,
310 architecture integer not null,
311 source integer not null);
313 insert into unique_binaries
314 select bab.package, bab.architecture, max(bab.source)
315 from bin_associations_binaries bab
316 where bab.suite = :suite_id and bab.architecture > 2
317 group by package, architecture having count(*) = 1;
319 create temp table newest_binaries (
320 package text not null,
321 architecture integer not null,
322 source text not null,
323 version debversion not null);
325 insert into newest_binaries
326 select ub.package, ub.architecture, nsa.source, nsa.version
327 from unique_binaries ub
328 join newest_src_association nsa
329 on ub.source = nsa.src and nsa.suite = :suite_id;
331 with uptodate_arch as
332 (select architecture, source, version
334 group by architecture, source, version),
335 unique_binaries_uptodate_arch as
336 (select ub.package, ub.architecture, ua.source, ua.version
337 from unique_binaries ub
340 join uptodate_arch ua
341 on ub.architecture = ua.architecture and s.source = ua.source),
342 unique_binaries_uptodate_arch_agg as
343 (select ubua.package,
344 array(select unnest(array_agg(a.arch_string)) order by 1) as arch_list,
345 ubua.source, ubua.version
346 from unique_binaries_uptodate_arch ubua
348 on ubua.architecture = a.id
349 group by ubua.source, ubua.version, ubua.package),
351 (select package, source, version
353 group by package, source, version),
355 (select array(select unnest(array_agg(package)) order by 1) as pkg_list,
356 arch_list, source, version
357 from unique_binaries_uptodate_arch_agg
359 (select package from uptodate_packages)
360 group by arch_list, source, version)
361 select * from outdated_packages order by source"""
362 return session.execute(query, { 'suite_id': suite_id })
364 def reportNBS(suite_name, suite_id):
365 session = DBConn().session()
366 nbsRows = queryNBS(suite_id, session)
367 title = 'NBS packages in suite %s' % suite_name
368 if nbsRows.rowcount > 0:
369 print '%s\n%s\n' % (title, '-' * len(title))
371 (pkg_list, arch_list, source, version) = row
372 pkg_string = ' '.join(pkg_list)
373 arch_string = ','.join(arch_list)
374 print "* source package %s version %s no longer builds" % \
376 print " binary package(s): %s" % pkg_string
377 print " on %s" % arch_string
378 print " - suggested command:"
379 message = '"[auto-cruft] NBS (no longer built by %s)"' % source
380 print " dak rm -m %s -s %s -a %s -p -R -b %s\n" % \
381 (message, suite_name, arch_string, pkg_string)
384 def reportAllNBS(suite_name, suite_id, session):
385 reportWithoutSource(suite_name, suite_id, session)
386 reportNewerAll(suite_name, session)
387 reportNBS(suite_name, suite_id)
389 ################################################################################
391 def do_dubious_nbs(dubious_nbs):
396 dubious_nbs_keys = dubious_nbs.keys()
397 dubious_nbs_keys.sort()
398 for source in dubious_nbs_keys:
399 print " * %s_%s builds: %s" % (source,
400 source_versions.get(source, "??"),
401 source_binaries.get(source, "(source does not exist)"))
402 print " won't admit to building:"
403 versions = dubious_nbs[source].keys()
404 versions.sort(apt_pkg.VersionCompare)
405 for version in versions:
406 packages = dubious_nbs[source][version].keys()
408 print " o %s: %s" % (version, ", ".join(packages))
412 ################################################################################
414 def obsolete_source(suite_name, session):
415 """returns obsolete source packages for suite_name without binaries
416 in the same suite sorted by install_date; install_date should help
417 detecting source only (or binary throw away) uploads; duplicates in
418 the suite are skipped
420 subquery 'source_suite_unique' returns source package names from
421 suite without duplicates; the rationale behind is that neither
422 cruft-report nor rm cannot handle duplicates (yet)"""
425 WITH source_suite_unique AS
426 (SELECT source, suite
427 FROM source_suite GROUP BY source, suite HAVING count(*) = 1)
428 SELECT ss.src, ss.source, ss.version,
429 to_char(ss.install_date, 'YYYY-MM-DD') AS install_date
431 JOIN source_suite_unique ssu
432 ON ss.source = ssu.source AND ss.suite = ssu.suite
433 JOIN suite s ON s.id = ss.suite
434 LEFT JOIN bin_associations_binaries bab
435 ON ss.src = bab.source AND ss.suite = bab.suite
436 WHERE s.suite_name = :suite_name AND bab.id IS NULL
437 ORDER BY install_date"""
438 args = { 'suite_name': suite_name }
439 return session.execute(query, args)
441 def source_bin(source, session):
442 """returns binaries built by source for all or no suite grouped and
443 ordered by package name"""
448 JOIN src_associations_src sas ON b.source = sas.src
449 WHERE sas.source = :source
451 ORDER BY b.package"""
452 args = { 'source': source }
453 return session.execute(query, args)
455 def newest_source_bab(suite_name, package, session):
456 """returns newest source that builds binary package in suite grouped
457 and sorted by source and package name"""
460 SELECT sas.source, MAX(sas.version) AS srcver
461 FROM src_associations_src sas
462 JOIN bin_associations_binaries bab ON sas.src = bab.source
463 JOIN suite s on s.id = bab.suite
464 WHERE s.suite_name = :suite_name AND bab.package = :package
465 GROUP BY sas.source, bab.package
466 ORDER BY sas.source, bab.package"""
467 args = { 'suite_name': suite_name, 'package': package }
468 return session.execute(query, args)
470 def report_obsolete_source(suite_name, session):
471 rows = obsolete_source(suite_name, session)
472 if rows.rowcount == 0:
475 """Obsolete source packages in suite %s
476 ----------------------------------%s\n""" % \
477 (suite_name, '-' * len(suite_name))
478 for os_row in rows.fetchall():
479 (src, old_source, version, install_date) = os_row
480 print " * obsolete source %s version %s installed at %s" % \
481 (old_source, version, install_date)
482 for sb_row in source_bin(old_source, session):
484 print " - has built binary %s" % package
485 for nsb_row in newest_source_bab(suite_name, package, session):
486 (new_source, srcver) = nsb_row
487 print " currently built by source %s version %s" % \
489 print " - suggested command:"
490 rm_opts = "-S -p -m \"[auto-cruft] obsolete source package\""
491 print " dak rm -s %s %s %s\n" % (suite_name, rm_opts, old_source)
493 def get_suite_binaries(suite, session):
494 # Initalize a large hash table of all binary packages
497 print "Getting a list of binary packages in %s..." % suite.suite_name
498 q = session.execute("""SELECT distinct b.package
499 FROM binaries b, bin_associations ba
500 WHERE ba.suite = :suiteid AND ba.bin = b.id""",
501 {'suiteid': suite.suite_id})
502 for i in q.fetchall():
507 ################################################################################
510 global suite, suite_id, source_binaries, source_versions
514 Arguments = [('h',"help","Cruft-Report::Options::Help"),
515 ('m',"mode","Cruft-Report::Options::Mode", "HasArg"),
516 ('s',"suite","Cruft-Report::Options::Suite","HasArg"),
517 ('w',"wanna-build-dump","Cruft-Report::Options::Wanna-Build-Dump","HasArg")]
519 if not cnf.has_key("Cruft-Report::Options::%s" % (i)):
520 cnf["Cruft-Report::Options::%s" % (i)] = ""
521 cnf["Cruft-Report::Options::Suite"] = cnf["Dinstall::DefaultSuite"]
523 if not cnf.has_key("Cruft-Report::Options::Mode"):
524 cnf["Cruft-Report::Options::Mode"] = "daily"
526 if not cnf.has_key("Cruft-Report::Options::Wanna-Build-Dump"):
527 cnf["Cruft-Report::Options::Wanna-Build-Dump"] = "/srv/ftp-master.debian.org/scripts/nfu"
529 apt_pkg.ParseCommandLine(cnf.Cnf, Arguments, sys.argv)
531 Options = cnf.SubTree("Cruft-Report::Options")
535 # Set up checks based on mode
536 if Options["Mode"] == "daily":
537 checks = [ "nbs", "nviu", "nvit", "obsolete source", "nfu" ]
538 elif Options["Mode"] == "full":
539 checks = [ "nbs", "nviu", "nvit", "obsolete source", "nfu", "dubious nbs", "bnb", "bms", "anais" ]
541 utils.warn("%s is not a recognised mode - only 'full' or 'daily' are understood." % (Options["Mode"]))
544 session = DBConn().session()
557 suite = get_suite(Options["Suite"].lower(), session)
558 suite_id = suite.suite_id
559 suite_name = suite.suite_name.lower()
561 if "obsolete source" in checks:
562 report_obsolete_source(suite_name, session)
565 reportAllNBS(suite_name, suite_id, session)
570 bins_in_suite = get_suite_binaries(suite, session)
572 # Checks based on the Sources files
573 components = cnf.ValueList("Suite::%s::Components" % (suite_name))
574 for component in components:
575 filename = "%s/dists/%s/%s/source/Sources.gz" % (cnf["Dir::Root"], suite_name, component)
576 # apt_pkg.ParseTagFile needs a real file handle and can't handle a GzipFile instance...
577 (fd, temp_filename) = utils.temp_filename()
578 (result, output) = commands.getstatusoutput("gunzip -c %s > %s" % (filename, temp_filename))
580 sys.stderr.write("Gunzip invocation failed!\n%s\n" % (output))
582 sources = utils.open_file(temp_filename)
583 Sources = apt_pkg.ParseTagFile(sources)
584 while Sources.Step():
585 source = Sources.Section.Find('Package')
586 source_version = Sources.Section.Find('Version')
587 architecture = Sources.Section.Find('Architecture')
588 binaries = Sources.Section.Find('Binary')
589 binaries_list = [ i.strip() for i in binaries.split(',') ]
592 # Check for binaries not built on any architecture.
593 for binary in binaries_list:
594 if not bins_in_suite.has_key(binary):
595 bin_not_built.setdefault(source, {})
596 bin_not_built[source][binary] = ""
598 if "anais" in checks:
599 anais_output += do_anais(architecture, binaries_list, source, session)
601 # build indices for checking "no source" later
602 source_index = component + '/' + source
603 src_pkgs[source] = source_index
604 for binary in binaries_list:
605 bin_pkgs[binary] = source
606 source_binaries[source] = binaries
607 source_versions[source] = source_version
610 os.unlink(temp_filename)
612 # Checks based on the Packages files
613 check_components = components[:]
614 if suite_name != "experimental":
615 check_components.append('main/debian-installer');
617 for component in check_components:
618 architectures = [ a.arch_string for a in get_suite_architectures(suite_name,
619 skipsrc=True, skipall=True,
621 for architecture in architectures:
622 if component == 'main/debian-installer' and re.match("kfreebsd", architecture):
624 filename = "%s/dists/%s/%s/binary-%s/Packages.gz" % (cnf["Dir::Root"], suite_name, component, architecture)
625 # apt_pkg.ParseTagFile needs a real file handle
626 (fd, temp_filename) = utils.temp_filename()
627 (result, output) = commands.getstatusoutput("gunzip -c %s > %s" % (filename, temp_filename))
629 sys.stderr.write("Gunzip invocation failed!\n%s\n" % (output))
633 nfu_packages.setdefault(architecture,[])
634 nfu_entries = parse_nfu(architecture)
636 packages = utils.open_file(temp_filename)
637 Packages = apt_pkg.ParseTagFile(packages)
638 while Packages.Step():
639 package = Packages.Section.Find('Package')
640 source = Packages.Section.Find('Source', "")
641 version = Packages.Section.Find('Version')
644 if bin2source.has_key(package) and \
645 apt_pkg.VersionCompare(version, bin2source[package]["version"]) > 0:
646 bin2source[package]["version"] = version
647 bin2source[package]["source"] = source
649 bin2source[package] = {}
650 bin2source[package]["version"] = version
651 bin2source[package]["source"] = source
652 if source.find("(") != -1:
653 m = re_extract_src_version.match(source)
656 if not bin_pkgs.has_key(package):
657 nbs.setdefault(source,{})
658 nbs[source].setdefault(package, {})
659 nbs[source][package][version] = ""
662 if package in nfu_entries and \
663 version != source_versions[source]: # only suggest to remove out-of-date packages
664 nfu_packages[architecture].append((package,version,source_versions[source]))
667 os.unlink(temp_filename)
669 # Distinguish dubious (version numbers match) and 'real' NBS (they don't)
671 for source in nbs.keys():
672 for package in nbs[source].keys():
673 versions = nbs[source][package].keys()
674 versions.sort(apt_pkg.VersionCompare)
675 latest_version = versions.pop()
676 source_version = source_versions.get(source,"0")
677 if apt_pkg.VersionCompare(latest_version, source_version) == 0:
678 add_nbs(dubious_nbs, source, latest_version, package, suite_id, session)
681 do_newer_version('unstable', 'experimental', 'NVIU', session)
684 do_newer_version('testing', 'testing-proposed-updates', 'NVIT', session)
688 if Options["Mode"] == "full":
696 print "Unbuilt binary packages"
697 print "-----------------------"
699 keys = bin_not_built.keys()
702 binaries = bin_not_built[source].keys()
704 print " o %s: %s" % (source, ", ".join(binaries))
708 report_multiple_source(suite)
710 if "anais" in checks:
711 print "Architecture Not Allowed In Source"
712 print "----------------------------------"
716 if "dubious nbs" in checks:
717 do_dubious_nbs(dubious_nbs)
720 ################################################################################
722 if __name__ == '__main__':