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 ################################################################################
39 warnings.filterwarnings('ignore', \
40 "apt_pkg\.ParseTagFile\(\) is deprecated\. Please see apt_pkg\.TagFile\(\) for the replacement\.", \
42 warnings.filterwarnings('ignore', \
43 "Attribute '.*' of the 'apt_pkg\.TagFile' object is deprecated, use '.*' instead\.", \
46 ################################################################################
48 import commands, os, sys, re
51 from daklib.config import Config
52 from daklib.dbconn import *
53 from daklib import utils
54 from daklib.regexes import re_extract_src_version
55 from daklib.cruft import *
57 ################################################################################
59 no_longer_in_suite = {}; # Really should be static to add_nbs, but I'm lazy
64 ################################################################################
66 def usage(exit_code=0):
67 print """Usage: dak cruft-report
68 Check for obsolete or duplicated packages.
70 -h, --help show this help and exit.
71 -m, --mode=MODE chose the MODE to run in (full or daily).
72 -s, --suite=SUITE check suite SUITE.
73 -w, --wanna-build-dump where to find the copies of http://buildd.debian.org/stats/*.txt"""
76 ################################################################################
78 def add_nbs(nbs_d, source, version, package, suite_id, session):
79 # Ensure the package is still in the suite (someone may have already removed it)
80 if no_longer_in_suite.has_key(package):
83 q = session.execute("""SELECT b.id FROM binaries b, bin_associations ba
84 WHERE ba.bin = b.id AND ba.suite = :suite_id
85 AND b.package = :package LIMIT 1""", {'suite_id': suite_id,
88 no_longer_in_suite[package] = ""
91 nbs_d.setdefault(source, {})
92 nbs_d[source].setdefault(version, {})
93 nbs_d[source][version][package] = ""
95 ################################################################################
97 # Check for packages built on architectures they shouldn't be.
98 def do_anais(architecture, binaries_list, source, session):
99 if architecture == "any" or architecture == "all":
104 for arch in architecture.split():
105 architectures[arch.strip()] = ""
106 for binary in binaries_list:
107 q = session.execute("""SELECT a.arch_string, b.version
108 FROM binaries b, bin_associations ba, architecture a
109 WHERE ba.suite = :suiteid AND ba.bin = b.id
110 AND b.architecture = a.id AND b.package = :package""",
111 {'suiteid': suite_id, 'package': binary})
117 if architectures.has_key(arch):
118 versions.append(version)
119 versions.sort(apt_pkg.VersionCompare)
121 latest_version = versions.pop()
123 latest_version = None
124 # Check for 'invalid' architectures
129 if not architectures.has_key(arch):
130 versions_d.setdefault(version, [])
131 versions_d[version].append(arch)
134 anais_output += "\n (*) %s_%s [%s]: %s\n" % (binary, latest_version, source, architecture)
135 versions = versions_d.keys()
136 versions.sort(apt_pkg.VersionCompare)
137 for version in versions:
138 arches = versions_d[version]
140 anais_output += " o %s: %s\n" % (version, ", ".join(arches))
144 ################################################################################
146 # Check for out-of-date binaries on architectures that do not want to build that
147 # package any more, and have them listed as Not-For-Us
148 def do_nfu(nfu_packages):
153 for architecture in nfu_packages:
154 a2p[architecture] = []
155 for (package,bver,sver) in nfu_packages[architecture]:
156 output += " * [%s] does not want %s (binary %s, source %s)\n" % (architecture, package, bver, sver)
157 a2p[architecture].append(package)
161 print "Obsolete by Not-For-Us"
162 print "----------------------"
166 print "Suggested commands:"
167 for architecture in a2p:
168 if a2p[architecture]:
169 print (" dak rm -m \"[auto-cruft] NFU\" -s %s -a %s -b %s" %
170 (suite.suite_name, architecture, " ".join(a2p[architecture])))
173 def parse_nfu(architecture):
175 # utils/hpodder_1.1.5.0: Not-For-Us [optional:out-of-date]
176 r = re.compile("^\w+/([^_]+)_.*: Not-For-Us")
180 filename = "%s/%s-all.txt" % (cnf["Cruft-Report::Options::Wanna-Build-Dump"], architecture)
182 # Not all architectures may have a wanna-build dump, so we want to ignore missin
184 if os.path.exists(filename):
185 f = utils.open_file(filename)
196 utils.warn("No wanna-build dump file for architecture %s" % architecture)
199 ################################################################################
201 def do_newer_version(lowersuite_name, highersuite_name, code, session):
202 list = newer_version(lowersuite_name, highersuite_name, session)
205 title = "Newer version in %s" % lowersuite_name
207 print "-" * len(title)
210 (source, higher_version, lower_version) = i
211 print " o %s (%s, %s)" % (source, higher_version, lower_version)
212 nv_to_remove.append(source)
214 print "Suggested command:"
215 print " dak rm -m \"[auto-cruft] %s\" -s %s %s" % (code, highersuite_name,
216 " ".join(nv_to_remove))
219 ################################################################################
221 def queryWithoutSource(suite_id, session):
222 """searches for arch: all packages from suite that do no longer
223 reference a source package in the same suite
225 subquery unique_binaries: selects all packages with only 1 version
226 in suite since 'dak rm' does not allow to specify version numbers"""
229 with unique_binaries as
230 (select package, max(version) as version, max(source) as source
231 from bin_associations_binaries
232 where architecture = 2 and suite = :suite_id
233 group by package having count(*) = 1)
234 select ub.package, ub.version
235 from unique_binaries ub
236 left join src_associations_src sas
237 on ub.source = sas.src and sas.suite = :suite_id
239 order by ub.package"""
240 return session.execute(query, { 'suite_id': suite_id })
242 def reportWithoutSource(suite_name, suite_id, session):
243 rows = queryWithoutSource(suite_id, session)
244 title = 'packages without source in suite %s' % suite_name
245 if rows.rowcount > 0:
246 print '%s\n%s\n' % (title, '-' * len(title))
247 message = '"[auto-cruft] no longer built from source"'
249 (package, version) = row
250 print "* package %s in version %s is no longer built from source" % \
252 print " - suggested command:"
253 print " dak rm -m %s -s %s -a all -p -R -b %s\n" % \
254 (message, suite_name, package)
256 def queryNewerAll(suite_name, session):
257 """searches for arch != all packages that have an arch == all
258 package with a higher version in the same suite"""
261 select bab1.package, bab1.version as oldver,
262 array_to_string(array_agg(a.arch_string), ',') as oldarch,
263 bab2.version as newver
264 from bin_associations_binaries bab1
265 join bin_associations_binaries bab2
266 on bab1.package = bab2.package and bab1.version < bab2.version and
267 bab1.suite = bab2.suite and bab1.architecture > 2 and
268 bab2.architecture = 2
269 join architecture a on bab1.architecture = a.id
270 join suite s on bab1.suite = s.id
271 where s.suite_name = :suite_name
272 group by bab1.package, oldver, bab1.suite, newver"""
273 return session.execute(query, { 'suite_name': suite_name })
275 def reportNewerAll(suite_name, session):
276 rows = queryNewerAll(suite_name, session)
277 title = 'obsolete arch any packages in suite %s' % suite_name
278 if rows.rowcount > 0:
279 print '%s\n%s\n' % (title, '-' * len(title))
280 message = '"[auto-cruft] obsolete arch any package"'
282 (package, oldver, oldarch, newver) = row
283 print "* package %s is arch any in version %s but arch all in version %s" % \
284 (package, oldver, newver)
285 print " - suggested command:"
286 print " dak rm -m %s -s %s -a %s -p -b %s\n" % \
287 (message, suite_name, oldarch, package)
289 def queryNBS(suite_id, session):
290 """This one is really complex. It searches arch != all packages that
291 are no longer built from current source packages in suite.
293 temp table unique_binaries: will be populated with packages that
294 have only one version in suite because 'dak rm' does not allow
295 specifying version numbers
297 temp table newest_binaries: will be populated with packages that are
298 built from current sources
300 subquery uptodate_arch: returns all architectures built from current
303 subquery unique_binaries_uptodate_arch: returns all packages in
304 architectures from uptodate_arch
306 subquery unique_binaries_uptodate_arch_agg: same as
307 unique_binaries_uptodate_arch but with column architecture
310 subquery uptodate_packages: similar to uptodate_arch but returns all
311 packages built from current sources
313 subquery outdated_packages: returns all packages with architectures
314 no longer built from current source
318 create temp table unique_binaries (
319 package text not null,
320 architecture integer not null,
321 source integer not null);
323 insert into unique_binaries
324 select bab.package, bab.architecture, max(bab.source)
325 from bin_associations_binaries bab
326 where bab.suite = :suite_id and bab.architecture > 2
327 group by package, architecture having count(*) = 1;
329 create temp table newest_binaries (
330 package text not null,
331 architecture integer not null,
332 source text not null,
333 version debversion not null);
335 insert into newest_binaries
336 select ub.package, ub.architecture, nsa.source, nsa.version
337 from unique_binaries ub
338 join newest_src_association nsa
339 on ub.source = nsa.src and nsa.suite = :suite_id;
341 with uptodate_arch as
342 (select architecture, source, version
344 group by architecture, source, version),
345 unique_binaries_uptodate_arch as
346 (select ub.package, ub.architecture, ua.source, ua.version
347 from unique_binaries ub
350 join uptodate_arch ua
351 on ub.architecture = ua.architecture and s.source = ua.source),
352 unique_binaries_uptodate_arch_agg as
353 (select ubua.package,
354 array(select unnest(array_agg(a.arch_string)) order by 1) as arch_list,
355 ubua.source, ubua.version
356 from unique_binaries_uptodate_arch ubua
358 on ubua.architecture = a.id
359 group by ubua.source, ubua.version, ubua.package),
361 (select package, source, version
363 group by package, source, version),
365 (select array(select unnest(array_agg(package)) order by 1) as pkg_list,
366 arch_list, source, version
367 from unique_binaries_uptodate_arch_agg
369 (select package from uptodate_packages)
370 group by arch_list, source, version)
371 select * from outdated_packages order by source"""
372 return session.execute(query, { 'suite_id': suite_id })
374 def reportNBS(suite_name, suite_id):
375 session = DBConn().session()
376 nbsRows = queryNBS(suite_id, session)
377 title = 'NBS packages in suite %s' % suite_name
378 if nbsRows.rowcount > 0:
379 print '%s\n%s\n' % (title, '-' * len(title))
381 (pkg_list, arch_list, source, version) = row
382 pkg_string = ' '.join(pkg_list)
383 arch_string = ','.join(arch_list)
384 print "* source package %s version %s no longer builds" % \
386 print " binary package(s): %s" % pkg_string
387 print " on %s" % arch_string
388 print " - suggested command:"
389 message = '"[auto-cruft] NBS (no longer built by %s)"' % source
390 print " dak rm -m %s -s %s -a %s -p -R -b %s\n" % \
391 (message, suite_name, arch_string, pkg_string)
394 def reportAllNBS(suite_name, suite_id, session):
395 reportWithoutSource(suite_name, suite_id, session)
396 reportNewerAll(suite_name, session)
397 reportNBS(suite_name, suite_id)
399 ################################################################################
401 def do_dubious_nbs(dubious_nbs):
406 dubious_nbs_keys = dubious_nbs.keys()
407 dubious_nbs_keys.sort()
408 for source in dubious_nbs_keys:
409 print " * %s_%s builds: %s" % (source,
410 source_versions.get(source, "??"),
411 source_binaries.get(source, "(source does not exist)"))
412 print " won't admit to building:"
413 versions = dubious_nbs[source].keys()
414 versions.sort(apt_pkg.VersionCompare)
415 for version in versions:
416 packages = dubious_nbs[source][version].keys()
418 print " o %s: %s" % (version, ", ".join(packages))
422 ################################################################################
424 def obsolete_source(suite_name, session):
425 """returns obsolete source packages for suite_name without binaries
426 in the same suite sorted by install_date; install_date should help
427 detecting source only (or binary throw away) uploads; duplicates in
428 the suite are skipped
430 subquery 'source_suite_unique' returns source package names from
431 suite without duplicates; the rationale behind is that neither
432 cruft-report nor rm cannot handle duplicates (yet)"""
435 WITH source_suite_unique AS
436 (SELECT source, suite
437 FROM source_suite GROUP BY source, suite HAVING count(*) = 1)
438 SELECT ss.src, ss.source, ss.version,
439 to_char(ss.install_date, 'YYYY-MM-DD') AS install_date
441 JOIN source_suite_unique ssu
442 ON ss.source = ssu.source AND ss.suite = ssu.suite
443 JOIN suite s ON s.id = ss.suite
444 LEFT JOIN bin_associations_binaries bab
445 ON ss.src = bab.source AND ss.suite = bab.suite
446 WHERE s.suite_name = :suite_name AND bab.id IS NULL
447 ORDER BY install_date"""
448 args = { 'suite_name': suite_name }
449 return session.execute(query, args)
451 def source_bin(source, session):
452 """returns binaries built by source for all or no suite grouped and
453 ordered by package name"""
458 JOIN src_associations_src sas ON b.source = sas.src
459 WHERE sas.source = :source
461 ORDER BY b.package"""
462 args = { 'source': source }
463 return session.execute(query, args)
465 def newest_source_bab(suite_name, package, session):
466 """returns newest source that builds binary package in suite grouped
467 and sorted by source and package name"""
470 SELECT sas.source, MAX(sas.version) AS srcver
471 FROM src_associations_src sas
472 JOIN bin_associations_binaries bab ON sas.src = bab.source
473 JOIN suite s on s.id = bab.suite
474 WHERE s.suite_name = :suite_name AND bab.package = :package
475 GROUP BY sas.source, bab.package
476 ORDER BY sas.source, bab.package"""
477 args = { 'suite_name': suite_name, 'package': package }
478 return session.execute(query, args)
480 def report_obsolete_source(suite_name, session):
481 rows = obsolete_source(suite_name, session)
482 if rows.rowcount == 0:
485 """Obsolete source packages in suite %s
486 ----------------------------------%s\n""" % \
487 (suite_name, '-' * len(suite_name))
488 for os_row in rows.fetchall():
489 (src, old_source, version, install_date) = os_row
490 print " * obsolete source %s version %s installed at %s" % \
491 (old_source, version, install_date)
492 for sb_row in source_bin(old_source, session):
494 print " - has built binary %s" % package
495 for nsb_row in newest_source_bab(suite_name, package, session):
496 (new_source, srcver) = nsb_row
497 print " currently built by source %s version %s" % \
499 print " - suggested command:"
500 rm_opts = "-S -p -m \"[auto-cruft] obsolete source package\""
501 print " dak rm -s %s %s %s\n" % (suite_name, rm_opts, old_source)
503 def get_suite_binaries(suite, session):
504 # Initalize a large hash table of all binary packages
507 print "Getting a list of binary packages in %s..." % suite.suite_name
508 q = session.execute("""SELECT distinct b.package
509 FROM binaries b, bin_associations ba
510 WHERE ba.suite = :suiteid AND ba.bin = b.id""",
511 {'suiteid': suite.suite_id})
512 for i in q.fetchall():
517 ################################################################################
520 global suite, suite_id, source_binaries, source_versions
524 Arguments = [('h',"help","Cruft-Report::Options::Help"),
525 ('m',"mode","Cruft-Report::Options::Mode", "HasArg"),
526 ('s',"suite","Cruft-Report::Options::Suite","HasArg"),
527 ('w',"wanna-build-dump","Cruft-Report::Options::Wanna-Build-Dump","HasArg")]
529 if not cnf.has_key("Cruft-Report::Options::%s" % (i)):
530 cnf["Cruft-Report::Options::%s" % (i)] = ""
531 cnf["Cruft-Report::Options::Suite"] = cnf["Dinstall::DefaultSuite"]
533 if not cnf.has_key("Cruft-Report::Options::Mode"):
534 cnf["Cruft-Report::Options::Mode"] = "daily"
536 if not cnf.has_key("Cruft-Report::Options::Wanna-Build-Dump"):
537 cnf["Cruft-Report::Options::Wanna-Build-Dump"] = "/srv/ftp-master.debian.org/scripts/nfu"
539 apt_pkg.ParseCommandLine(cnf.Cnf, Arguments, sys.argv)
541 Options = cnf.SubTree("Cruft-Report::Options")
545 # Set up checks based on mode
546 if Options["Mode"] == "daily":
547 checks = [ "nbs", "nviu", "nvit", "obsolete source", "nfu" ]
548 elif Options["Mode"] == "full":
549 checks = [ "nbs", "nviu", "nvit", "obsolete source", "nfu", "dubious nbs", "bnb", "bms", "anais" ]
551 utils.warn("%s is not a recognised mode - only 'full' or 'daily' are understood." % (Options["Mode"]))
554 session = DBConn().session()
567 suite = get_suite(Options["Suite"].lower(), session)
568 suite_id = suite.suite_id
569 suite_name = suite.suite_name.lower()
571 if "obsolete source" in checks:
572 report_obsolete_source(suite_name, session)
575 reportAllNBS(suite_name, suite_id, session)
580 bins_in_suite = get_suite_binaries(suite, session)
582 # Checks based on the Sources files
583 components = cnf.ValueList("Suite::%s::Components" % (suite_name))
584 for component in components:
585 filename = "%s/dists/%s/%s/source/Sources.gz" % (cnf["Dir::Root"], suite_name, component)
586 # apt_pkg.ParseTagFile needs a real file handle and can't handle a GzipFile instance...
587 (fd, temp_filename) = utils.temp_filename()
588 (result, output) = commands.getstatusoutput("gunzip -c %s > %s" % (filename, temp_filename))
590 sys.stderr.write("Gunzip invocation failed!\n%s\n" % (output))
592 sources = utils.open_file(temp_filename)
593 Sources = apt_pkg.ParseTagFile(sources)
594 while Sources.Step():
595 source = Sources.Section.Find('Package')
596 source_version = Sources.Section.Find('Version')
597 architecture = Sources.Section.Find('Architecture')
598 binaries = Sources.Section.Find('Binary')
599 binaries_list = [ i.strip() for i in binaries.split(',') ]
602 # Check for binaries not built on any architecture.
603 for binary in binaries_list:
604 if not bins_in_suite.has_key(binary):
605 bin_not_built.setdefault(source, {})
606 bin_not_built[source][binary] = ""
608 if "anais" in checks:
609 anais_output += do_anais(architecture, binaries_list, source, session)
611 # build indices for checking "no source" later
612 source_index = component + '/' + source
613 src_pkgs[source] = source_index
614 for binary in binaries_list:
615 bin_pkgs[binary] = source
616 source_binaries[source] = binaries
617 source_versions[source] = source_version
620 os.unlink(temp_filename)
622 # Checks based on the Packages files
623 check_components = components[:]
624 if suite_name != "experimental":
625 check_components.append('main/debian-installer');
627 for component in check_components:
628 architectures = [ a.arch_string for a in get_suite_architectures(suite_name,
629 skipsrc=True, skipall=True,
631 for architecture in architectures:
632 if component == 'main/debian-installer' and re.match("kfreebsd", architecture):
634 filename = "%s/dists/%s/%s/binary-%s/Packages.gz" % (cnf["Dir::Root"], suite_name, component, architecture)
635 # apt_pkg.ParseTagFile needs a real file handle
636 (fd, temp_filename) = utils.temp_filename()
637 (result, output) = commands.getstatusoutput("gunzip -c %s > %s" % (filename, temp_filename))
639 sys.stderr.write("Gunzip invocation failed!\n%s\n" % (output))
643 nfu_packages.setdefault(architecture,[])
644 nfu_entries = parse_nfu(architecture)
646 packages = utils.open_file(temp_filename)
647 Packages = apt_pkg.ParseTagFile(packages)
648 while Packages.Step():
649 package = Packages.Section.Find('Package')
650 source = Packages.Section.Find('Source', "")
651 version = Packages.Section.Find('Version')
654 if bin2source.has_key(package) and \
655 apt_pkg.VersionCompare(version, bin2source[package]["version"]) > 0:
656 bin2source[package]["version"] = version
657 bin2source[package]["source"] = source
659 bin2source[package] = {}
660 bin2source[package]["version"] = version
661 bin2source[package]["source"] = source
662 if source.find("(") != -1:
663 m = re_extract_src_version.match(source)
666 if not bin_pkgs.has_key(package):
667 nbs.setdefault(source,{})
668 nbs[source].setdefault(package, {})
669 nbs[source][package][version] = ""
672 if package in nfu_entries and \
673 version != source_versions[source]: # only suggest to remove out-of-date packages
674 nfu_packages[architecture].append((package,version,source_versions[source]))
677 os.unlink(temp_filename)
679 # Distinguish dubious (version numbers match) and 'real' NBS (they don't)
681 for source in nbs.keys():
682 for package in nbs[source].keys():
683 versions = nbs[source][package].keys()
684 versions.sort(apt_pkg.VersionCompare)
685 latest_version = versions.pop()
686 source_version = source_versions.get(source,"0")
687 if apt_pkg.VersionCompare(latest_version, source_version) == 0:
688 add_nbs(dubious_nbs, source, latest_version, package, suite_id, session)
691 do_newer_version('unstable', 'experimental', 'NVIU', session)
694 do_newer_version('testing', 'testing-proposed-updates', 'NVIT', session)
698 if Options["Mode"] == "full":
706 print "Unbuilt binary packages"
707 print "-----------------------"
709 keys = bin_not_built.keys()
712 binaries = bin_not_built[source].keys()
714 print " o %s: %s" % (source, ", ".join(binaries))
718 report_multiple_source(suite)
720 if "anais" in checks:
721 print "Architecture Not Allowed In Source"
722 print "----------------------------------"
726 if "dubious nbs" in checks:
727 do_dubious_nbs(dubious_nbs)
730 ################################################################################
732 if __name__ == '__main__':