def clean_byhash(now_date, session): cnf = Config() Logger.log(["Cleaning out unused by-hash files..."]) q = session.execute(""" DELETE FROM hashfile h USING suite s, archive a WHERE s.id = h.suite_id AND a.id = s.archive_id AND h.unreferenced + a.stayofexecution < CURRENT_TIMESTAMP RETURNING a.path, s.suite_name, h.path""") count = q.rowcount if not Options["No-Action"]: for base, suite, path in q: suite_suffix = utils.suite_suffix(suite) filename = os.path.join(base, 'dists', suite, suite_suffix, path) try: os.unlink(filename) except OSError as exc: if exc.errno != errno.ENOENT: raise Logger.log( ['database referred to non-existing file', filename]) else: Logger.log(['delete hashfile', suite, path]) session.commit() if count > 0: Logger.log(["total", count])
def suite_path(self): """ Absolute path to the suite-specific files. """ suite_suffix = utils.suite_suffix(self.suite.suite_name) return os.path.join(self.suite.archive.path, 'dists', self.suite.suite_name, suite_suffix)
def suite_release_path(self): """ Absolute path where Release files are physically stored. This should be a path that sorts after the dists/ directory. """ cnf = Config() suite_suffix = utils.suite_suffix(self.suite.suite_name) return os.path.join(self.suite.archive.path, 'zzz-dists', self.suite.suite_name, suite_suffix)
def generate_release_files(self): """ Generate Release files for the given suite @type suite: string @param suite: Suite name """ suite = self.suite session = object_session(suite) architectures = get_suite_architectures(suite.suite_name, skipall=True, skipsrc=True, session=session) # Attribs contains a tuple of field names and the database names to use to # fill them in attribs = (('Origin', 'origin'), ('Label', 'label'), ('Suite', 'release_suite_output'), ('Version', 'version'), ('Codename', 'codename'), ('Changelogs', 'changelog_url'), ) # A "Sub" Release file has slightly different fields subattribs = (('Archive', 'suite_name'), ('Origin', 'origin'), ('Label', 'label'), ('Version', 'version')) # Boolean stuff. If we find it true in database, write out "yes" into the release file boolattrs = (('NotAutomatic', 'notautomatic'), ('ButAutomaticUpgrades', 'butautomaticupgrades'), ('Acquire-By-Hash', 'byhash'), ) cnf = Config() cnf_suite_suffix = cnf.get("Dinstall::SuiteSuffix", "").rstrip("/") suite_suffix = utils.suite_suffix(suite.suite_name) self.create_output_directories() self.create_release_symlinks() outfile = os.path.join(self.suite_release_path(), "Release") out = open(outfile + ".new", "w") for key, dbfield in attribs: # Hack to skip NULL Version fields as we used to do this # We should probably just always ignore anything which is None if key in ("Version", "Changelogs") and getattr(suite, dbfield) is None: continue out.write("%s: %s\n" % (key, getattr(suite, dbfield))) out.write("Date: %s\n" % (time.strftime("%a, %d %b %Y %H:%M:%S UTC", time.gmtime(time.time())))) if suite.validtime: validtime = float(suite.validtime) out.write("Valid-Until: %s\n" % (time.strftime("%a, %d %b %Y %H:%M:%S UTC", time.gmtime(time.time() + validtime)))) for key, dbfield in boolattrs: if getattr(suite, dbfield, False): out.write("%s: yes\n" % (key)) out.write("Architectures: %s\n" % (" ".join([a.arch_string for a in architectures]))) components = [c.component_name for c in suite.components] out.write("Components: %s\n" % (" ".join(components))) # For exact compatibility with old g-r, write out Description here instead # of with the rest of the DB fields above if getattr(suite, 'description') is not None: out.write("Description: %s\n" % suite.description) for comp in components: for dirpath, dirnames, filenames in os.walk(os.path.join(self.suite_path(), comp), topdown=True): if not re_gensubrelease.match(dirpath): continue subfile = os.path.join(dirpath, "Release") subrel = open(subfile + '.new', "w") for key, dbfield in subattribs: if getattr(suite, dbfield) is not None: subrel.write("%s: %s\n" % (key, getattr(suite, dbfield))) for key, dbfield in boolattrs: if getattr(suite, dbfield, False): subrel.write("%s: yes\n" % (key)) subrel.write("Component: %s%s\n" % (suite_suffix, comp)) # Urgh, but until we have all the suite/component/arch stuff in the DB, # this'll have to do arch = os.path.split(dirpath)[-1] if arch.startswith('binary-'): arch = arch[7:] subrel.write("Architecture: %s\n" % (arch)) subrel.close() os.rename(subfile + '.new', subfile) # Now that we have done the groundwork, we want to get off and add the files with # their checksums to the main Release file oldcwd = os.getcwd() os.chdir(self.suite_path()) hashes = [x for x in RELEASE_HASHES if x.db_name in suite.checksums] fileinfo = {} fileinfo_byhash = {} uncompnotseen = {} for dirpath, dirnames, filenames in os.walk(".", followlinks=True, topdown=True): # SuiteSuffix deprecation: # components on security-master are updates/{main,contrib,non-free}, but # we want dists/${suite}/main. Until we can rename the components, # we cheat by having an updates -> . symlink. This should not be visited. if cnf_suite_suffix: path = os.path.join(dirpath, cnf_suite_suffix) try: target = os.readlink(path) if target == ".": dirnames.remove(cnf_suite_suffix) except (OSError, ValueError): pass for entry in filenames: if dirpath == '.' and entry in ["Release", "Release.gpg", "InRelease"]: continue filename = os.path.join(dirpath.lstrip('./'), entry) if re_includeinrelease_byhash.match(entry): fileinfo[filename] = fileinfo_byhash[filename] = {} elif re_includeinrelease_plain.match(entry): fileinfo[filename] = {} # Skip things we don't want to include else: continue contents = open(filename, 'r').read() # If we find a file for which we have a compressed version and # haven't yet seen the uncompressed one, store the possibility # for future use if entry.endswith(".gz") and filename[:-3] not in uncompnotseen: uncompnotseen[filename[:-3]] = (gzip.GzipFile, filename) elif entry.endswith(".bz2") and filename[:-4] not in uncompnotseen: uncompnotseen[filename[:-4]] = (bz2.BZ2File, filename) elif entry.endswith(".xz") and filename[:-3] not in uncompnotseen: uncompnotseen[filename[:-3]] = (XzFile, filename) fileinfo[filename]['len'] = len(contents) for hf in hashes: fileinfo[filename][hf.release_field] = hf.func(contents) for filename, comp in uncompnotseen.items(): # If we've already seen the uncompressed file, we don't # need to do anything again if filename in fileinfo: continue fileinfo[filename] = {} # File handler is comp[0], filename of compressed file is comp[1] contents = comp[0](comp[1], 'r').read() fileinfo[filename]['len'] = len(contents) for hf in hashes: fileinfo[filename][hf.release_field] = hf.func(contents) for field in sorted(h.release_field for h in hashes): out.write('%s:\n' % field) for filename in sorted(fileinfo.keys()): out.write(" %s %8d %s\n" % (fileinfo[filename][field], fileinfo[filename]['len'], filename)) out.close() os.rename(outfile + '.new', outfile) self._update_hashfile_table(session, fileinfo_byhash, hashes) self._make_byhash_links(fileinfo_byhash, hashes) self._make_byhash_base_symlink(fileinfo_byhash, hashes) sign_release_dir(suite, os.path.dirname(outfile)) os.chdir(oldcwd) return
def main(): global Cnf, Options, Logger os.umask(0o002) Cnf = utils.get_conf() Arguments = [('h', "help", "Generate-Index-Diffs::Options::Help"), ('a', 'archive', 'Generate-Index-Diffs::Options::Archive', 'hasArg'), ('c', None, "Generate-Index-Diffs::Options::CanonicalPath", "hasArg"), ('p', "patchname", "Generate-Index-Diffs::Options::PatchName", "hasArg"), ('d', "tmpdir", "Generate-Index-Diffs::Options::TempDir", "hasArg"), ('m', "maxdiffs", "Generate-Index-Diffs::Options::MaxDiffs", "hasArg"), ('n', "no-act", "Generate-Index-Diffs::Options::NoAct"), ('v', "verbose", "Generate-Index-Diffs::Options::Verbose"), ] suites = apt_pkg.parse_commandline(Cnf, Arguments, sys.argv) Options = Cnf.subtree("Generate-Index-Diffs::Options") if "Help" in Options: usage() maxdiffs = Options.get("MaxDiffs::Default", "56") maxpackages = Options.get("MaxDiffs::Packages", maxdiffs) maxcontents = Options.get("MaxDiffs::Contents", maxdiffs) maxsources = Options.get("MaxDiffs::Sources", maxdiffs) # can only be set via config at the moment max_parallel = int(Options.get("MaxParallel", "8")) if "PatchName" not in Options: format = "%Y-%m-%d-%H%M.%S" Options["PatchName"] = time.strftime(format) session = DBConn().session() pending_tasks = [] if not suites: query = session.query(Suite.suite_name) if Options.get('Archive'): archives = utils.split_args(Options['Archive']) query = query.join(Suite.archive).filter(Archive.archive_name.in_(archives)) suites = [s.suite_name for s in query] for suitename in suites: print("Processing: " + suitename) suiteobj = get_suite(suitename.lower(), session=session) # Use the canonical version of the suite name suite = suiteobj.suite_name if suiteobj.untouchable: print("Skipping: " + suite + " (untouchable)") continue skip_all = True if suiteobj.separate_contents_architecture_all or suiteobj.separate_packages_architecture_all: skip_all = False architectures = get_suite_architectures(suite, skipall=skip_all, session=session) components = [c.component_name for c in session.query(Component.component_name)] suite_suffix = utils.suite_suffix(suitename) if components and suite_suffix: longsuite = suite + "/" + suite_suffix else: longsuite = suite merged_pdiffs = suiteobj.merged_pdiffs tree = os.path.join(suiteobj.archive.path, 'dists', longsuite) # See if there are Translations which might need a new pdiff cwd = os.getcwd() for component in components: workpath = os.path.join(tree, component, "i18n") if os.path.isdir(workpath): os.chdir(workpath) for dirpath, dirnames, filenames in os.walk(".", followlinks=True, topdown=True): for entry in filenames: if not re_includeinpdiff.match(entry): continue (fname, fext) = os.path.splitext(entry) processfile = os.path.join(workpath, fname) storename = "%s/%s_%s_%s" % (Options["TempDir"], suite, component, fname) coroutine = genchanges(Options, processfile + ".diff", storename, processfile, maxdiffs, merged_pdiffs) pending_tasks.append(coroutine) os.chdir(cwd) for archobj in architectures: architecture = archobj.arch_string if architecture == "source": longarch = architecture packages = "Sources" maxsuite = maxsources else: longarch = "binary-%s" % architecture packages = "Packages" maxsuite = maxpackages for component in components: # Process Contents file = "%s/%s/Contents-%s" % (tree, component, architecture) storename = "%s/%s_%s_contents_%s" % (Options["TempDir"], suite, component, architecture) coroutine = genchanges(Options, file + ".diff", storename, file, maxcontents, merged_pdiffs) pending_tasks.append(coroutine) file = "%s/%s/%s/%s" % (tree, component, longarch, packages) storename = "%s/%s_%s_%s" % (Options["TempDir"], suite, component, architecture) coroutine = genchanges(Options, file + ".diff", storename, file, maxsuite, merged_pdiffs) pending_tasks.append(coroutine) asyncio.run(process_pdiff_tasks(pending_tasks, max_parallel))