openSUSE-release-tools/osc-check_repo.py

# Copyright (C) 2014 SUSE Linux Products GmbH
#
# This program is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation; either version 2 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License along
# with this program; if not, write to the Free Software Foundation, Inc.,
# 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.

# Copy this script to ~/.osc-plugins/ or /var/lib/osc-plugins .
# Then try to run 'osc check_repo --help' to see the usage.

from collections import defaultdict
from collections import namedtuple
import os
import shutil
import subprocess
import tempfile
import traceback
import sys

from osc import cmdln
from osc import conf
from osc import oscerr

# Expand sys.path to search modules inside the pluging directory
PLUGINDIR = os.path.dirname(os.path.realpath(__file__.replace('.pyc', '.py')))
sys.path.append(PLUGINDIR)
from osclib.conf import Config
from osclib.checkrepo import CheckRepo
from osclib.checkrepo import BINCACHE, DOWNLOADS
from osclib.cache import Cache
from osclib.cycle import CycleDetector
from osclib.memoize import CACHEDIR
from osclib.memoize import save_cache_path
from osclib.request_finder import RequestFinder

Cache.CACHE_DIR = save_cache_path('opensuse-repo-checker-http')


def _check_repo_download(self, request):
    toignore = set()
    request.downloads = defaultdict(list)

    # do not check package lists if the repository was excluded
    if request.build_excluded:
        return set()

    # XXX TODO - Rewrite the logic here, meanwhile set is to x86_64
    arch = 'x86_64'

    if request.src_package in request.i686_only:
        # Use imported binaries from x86_64 to check the requirements is fine,
        # but we can not get rpmlint.log from x86_64 repo if it was excluded,
        # i586 repo should be are build succeeded already as we have check it
        # in repositories_to_check(). Therefore, we have to download binary
        # binary files from i586 repo.
        arch = 'i586'

    # if not request.build_excluded:
    #     arch = 'x86_64'
    # else:
    #     arch = 'i586'

    ToDownload = namedtuple('ToDownload', ('project', 'repo', 'arch', 'package', 'size'))

    for i, goodrepo in enumerate(request.goodrepos):
        repo = goodrepo[1]

        # we assume x86_64 is there unless build is excluded
        pkglist = self.checkrepo.get_package_list_from_repository(
            request.shadow_src_project, repo, arch,
            request.src_package)
        todownload = [ToDownload(request.shadow_src_project, repo, arch,
                                 fn[0], fn[3]) for fn in pkglist]

        toignore.update(fn[1] for fn in pkglist)

        self.checkrepo._download(request, todownload)
        if request.error:
            return set()

    staging_prefix = '{}:'.format(self.checkrepo.staging.cstaging)
    if staging_prefix in str(request.group):
        pkglist = self.checkrepo.get_package_list_from_repository(
            request.group, 'standard', arch,
            request.src_package)
        todownload = [ToDownload(request.group, 'standard', arch,
                                 fn[0], fn[3]) for fn in pkglist]

        self.checkrepo._download(request, todownload)
        if request.error:
            return set()

        toignore.update(fn[1] for fn in pkglist)

        pkglist = self.checkrepo.get_package_list_from_repository(
            request.group + ':DVD', 'standard',
            'x86_64', request.src_package)
        todownload = [ToDownload(request.group + ':DVD', 'standard',
                                 'x86_64', fn[0], fn[3]) for fn in pkglist]

        toignore.update(fn[1] for fn in pkglist)

        self.checkrepo._download(request, todownload)
        if request.error:
            return set()

    # Update toignore with the names of the source project (here in
    # this method) and with the names of the target project (_toignore
    # method).
    toignore.update(self.checkrepo._toignore(request))
    return toignore


# Used in _check_repo_group only to cache error messages
_errors_printed = set()


def _check_repo_group(self, id_, requests, skip_cycle=None, debug=False):
    if skip_cycle is None:
        skip_cycle = []

    print '> Check group [%s]' % ', '.join(r.str_compact() for r in requests)

    # XXX TODO - If the requests comes from command line, the group is
    # still not there.

    # Do not continue if any of the packages do not successfully build.
    if not all(self.checkrepo.is_buildsuccess(r) for r in requests if r.action_type != 'delete'):
        return

    toignore = set()
    destdir = os.path.join(BINCACHE, str(requests[0].group))
    fetched = {r: False for r in self.checkrepo.groups.get(id_, [])}
    packs = []

    for request in requests:
        if request.action_type == 'delete':
            continue

        i = self._check_repo_download(request)
        if request.error and request.error not in _errors_printed:
            _errors_printed.add(request.error)
            if not request.updated:
                print ' - %s' % request.error
                self.checkrepo.change_review_state(request.request_id, 'new', message=request.error)
                request.updated = True
            else:
                print ' - %s' % request.error
            return
        toignore.update(i)
        fetched[request.request_id] = True

    packs.extend(requests)

    # Extend packs array with the packages and .spec files of the
    # not-fetched requests.  The not fetched ones are the requests of
    # the same group that are not listed as a paramater.
    for request_id, is_fetched in fetched.items():
        if not is_fetched:
            packs.extend(self.checkrepo.check_specs(request_id=request_id))

    # Download the repos from the request of the same group not
    # explicited in the command line.
    for rq in packs[:]:
        if rq.request_id in fetched and fetched[rq.request_id]:
            continue
        i = set()
        if rq.action_type == 'delete':
            # for delete requests we care for toignore
            i = self.checkrepo._toignore(rq)
            # We also check that nothing depends on the package and
            # that the request originates by the package maintainer
            error_delete = self.checkrepo.is_safe_to_delete(rq)
            if error_delete:
                rq.error = 'This delete request is not safe to accept yet, ' \
                           'please wait until the reasons dissapear in the ' \
                           'target project. %s' % error_delete
                print ' - %s' % rq.error
                self.checkrepo.change_review_state(request.request_id, 'new', message=request.error)
                rq.updated = True
            else:
                msg = 'Request is safe %s' % rq
                print 'ACCEPTED', msg
                self.checkrepo.change_review_state(rq.request_id, 'accepted', message=msg)

            # Remove it from the packs, so do not interfere with the
            # rest of the check.
            packs.remove(rq)
        else:
            # we need to call it to fetch the good repos to download
            # but the return value is of no interest right now.
            self.checkrepo.is_buildsuccess(rq)
            i = self._check_repo_download(rq)
            if rq.error:
                print 'ERROR (ALREADY ACEPTED?):', rq.error
                rq.updated = True

        toignore.update(i)

    # Detect cycles into the current Factory / openSUSE graph after we
    # update the links with the current list of request.
    cycle_detector = CycleDetector(self.checkrepo.staging)
    for (cycle, new_edges, new_pkgs) in cycle_detector.cycles(requests=packs):
        # If there are new cycles that also contains new packages,
        # mark all packages as updated, to avoid to be accepted
        if new_pkgs:
            print
            print ' - New cycle detected:', sorted(cycle)
            print ' - New edges:', new_edges

            if skip_cycle:
                print ' - Skiping this cycle and moving to the next check.'
                continue
            else:
                print ' - If you want to skip this cycle, run manually the ' \
                    'check repo with -c / --skipcycle option.'

            for request in requests:
                request.updated = True

    for rq in requests:
        # Check if there are any goodrepo without missing packages
        missing_repos = set(rq.missings.keys())
        if any(r not in missing_repos for r in rq.goodrepos):
            continue

        smissing = []
        all_missing_packages = {item for sublist in rq.missings.values() for item in sublist}
        for package in all_missing_packages:
            alreadyin = False
            for t in packs:
                if package == t.tgt_package:
                    alreadyin = True
            if alreadyin:
                continue
            request = self.checkrepo.find_request_id(rq.tgt_project, package)
            if request:
                greqs = self.checkrepo.groups.get(rq.group, [])
                if request in greqs:
                    continue
                package = '#[%s](%s)' % (request, package)
            smissing.append(package)
        if len(smissing):
            msg = 'Please make sure to wait before these depencencies are in %s: %s [%s]' % (
                rq.tgt_project, ', '.join(smissing), rq.tgt_package)
            if not rq.updated:
                self.checkrepo.change_review_state(rq.request_id, 'new', message=msg)
                print ' - %s' % msg
                rq.updated = True
            else:
                print ' - %s' % msg
            return

    # Create a temporal file for the params
    params_file = tempfile.NamedTemporaryFile(delete=False)
    params_file.write('\n'.join(f for f in toignore if f.strip()))
    params_file.close()

    # We want to test with the Perl script the binaries of one of the
    # repos, and if fail test the other repo.  The order of testing
    # will be stored in the execution_plan.

    execution_plan = defaultdict(list)

    DEBUG_PLAN = debug

    # Get all the (project, repo, disturl) where the disturl is
    # compatible with the request.  For the same package we can have
    # more than one good triplet, even with different MD5 DISTRUL.
    # The general strategy is collect that the different triplets and
    # provide some execution_plans where for the same (project, repo)
    # for every package, with a fallback to a different (project,
    # repo) in case that the original is not found.
    all_good_downloads = defaultdict(set)
    for rq in packs:
        for (prj, repo, disturl) in rq.downloads:
            if self.checkrepo.check_disturl(rq, md5_disturl=disturl):
                all_good_downloads[(prj, repo)].add(disturl)
                if DEBUG_PLAN:
                    print 'DEBUG Good DISTURL -', rq.str_compact(), (prj, repo), disturl
            elif DEBUG_PLAN:
                print 'DEBUG Bad DISTURL -', rq.str_compact(), (prj, repo), disturl

    if not all_good_downloads:
        print ' - No matching downloads for disturl found.'
        if len(packs) == 1 and packs[0].tgt_package in ('rpmlint-tests'):
            print ' - %s known to have no installable rpms, skipped' % packs[0].tgt_package
        return

    for project, repo in all_good_downloads:
        plan = (project, repo)
        valid_disturl = all_good_downloads[plan]
        if DEBUG_PLAN:
            print 'DEBUG Designing plan', plan, valid_disturl
        for rq in packs:
            if DEBUG_PLAN:
                print 'DEBUG In', rq
            # Find (project, repo) in rq.downloads.
            keys = [key for key in rq.downloads if key[0] == project and key[1] == repo and key[2] in valid_disturl]
            if DEBUG_PLAN:
                print 'DEBUG Keys', keys

            if keys:
                assert len(keys) == 1, 'Found more than one candidate for the plan'

                execution_plan[plan].append((rq, plan, rq.downloads[keys[0]]))
                if DEBUG_PLAN:
                    print 'DEBUG Downloads', rq.downloads[keys[0]]
            else:
                if DEBUG_PLAN:
                    print 'DEBUG Searching for a fallback!'
                fallbacks = [key for key in rq.downloads if (key[0], key[1]) in all_good_downloads and key[2] in all_good_downloads[(key[0], key[1])]]
                if fallbacks:
                    # XXX TODO - Recurse here to create combinations
                    # Meanwhile, I will priorize the one fallback that is in a staging project.
                    fallbacks_from_staging = [fb for fb in fallbacks if 'Staging' in fb[0]]
                    fallbacks = fallbacks_from_staging if fallbacks_from_staging else fallbacks
                    fallback = fallbacks.pop()
                    if DEBUG_PLAN:
                        print 'DEBUG Fallback found', fallback
                        print 'DEBUG Fallback downloads', rq.downloads[fallback]

                    alternative_plan = fallback[:2]
                    execution_plan[plan].append((rq, alternative_plan, rq.downloads[fallback]))
                # elif rq.status == 'succeeded':
                else:
                    print 'no fallback for', rq

    repo_checker_error = ''
    for project_repo in execution_plan:
        dirstolink = execution_plan[project_repo]

        if DEBUG_PLAN:
            print 'DEBUG Running plan', project_repo
            for rq, repo, downloads in dirstolink:
                print ' ', rq
                print ' ', repo
                for f in downloads:
                    print '   -', f

        # Makes sure to remove the directory is case of early exit.
        if os.path.exists(destdir):
            shutil.rmtree(destdir)

        os.makedirs(destdir)
        for rq, _, downloads in dirstolink:
            dir_ = destdir + '/%s' % rq.tgt_package
            for d in downloads:
                if not os.path.exists(dir_):
                    os.mkdir(dir_)
                target = os.path.join(dir_, os.path.basename(d))
                if os.path.exists(target):
                    print 'Warning, symlink already exists', d, target
                    os.unlink(target)
                os.symlink(d, target)

        repochecker = os.path.join(PLUGINDIR, 'repo-checker.pl')
        civs = "LC_ALL=C perl %s '%s' -r %s -f %s" % (repochecker, destdir, self.repo_dir, params_file.name)
        p = subprocess.Popen(civs, shell=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE, close_fds=True)
        stdoutdata, stderrdata = p.communicate()
        stdoutdata = stdoutdata.strip()
        ret = p.returncode

        # Clean the directory that contains all the symlinks
        shutil.rmtree(destdir)

        # There are several execution plans, each one can have its own
        # error message.
        if ret:
            print ' - Execution plan for %s failed' % str(project_repo)
        else:
            print ' - Successful plan', project_repo

        if stdoutdata:
            print '-' * 40
            print stdoutdata
            print '-' * 40
        if stderrdata:
            print '-' * 40
            print stderrdata
            print '-' * 40

        # Detect if this error message comes from a staging project.
        # Store it in the repo_checker_error, that is the text that
        # will be published in the error message.
        staging_prefix = '{}:'.format(self.checkrepo.staging.cstaging)
        if staging_prefix in project_repo[0]:
            repo_checker_error = stdoutdata
        if not any(staging_prefix in p_r[0] for p_r in execution_plan):
            repo_checker_error += '\nExecution plan: %s\n%s' % ('/'.join(project_repo), stdoutdata)

        # print ret, stdoutdata, stderrdata
        # raise Exception()

        if not ret:  # skip the others
            for p, gr, downloads in dirstolink:
                p.goodrepo = '%s/%s' % gr
            break

    os.unlink(params_file.name)

    updated = {}

    if ret:
        for rq in requests:
            if updated.get(rq.request_id, False) or rq.updated:
                continue
            if repo_checker_error not in _errors_printed:
                _errors_printed.add(repo_checker_error)
                print repo_checker_error
            self.checkrepo.change_review_state(rq.request_id, 'new', message=repo_checker_error)
            p.updated = True
            updated[rq.request_id] = 1
        return

    for rq in requests:
        if updated.get(rq.request_id, False) or rq.updated:
            continue
        if not hasattr(rq, 'goodrepo'):
            msg = 'Can not find a good repo for %s' % rq.str_compact()
            print 'NOT ACCEPTED - ', msg
            print 'Perhaps this request is not against i586/x86_64 build or i586 build only. For human to check!'
            continue
        msg = 'Builds for repo %s' % rq.goodrepo
        print 'ACCEPTED', msg
        self.checkrepo.change_review_state(rq.request_id, 'accepted', message=msg)
        self.checkrepo.remove_link_if_shadow_devel(rq)
        rq.updated = True
        updated[rq.request_id] = 1


def _mirror_full(self, plugin_dir, repo_dir):
    """Call bs_mirrorfull script to mirror packages."""
    url = 'https://api.opensuse.org/public/build/%s/%s/x86_64' % (self.checkrepo.project, 'standard')

    if not os.path.exists(repo_dir):
        os.mkdir(repo_dir)

    script = 'LC_ALL=C perl %s/bs_mirrorfull --nodebug %s %s' % (plugin_dir, url, repo_dir)
    os.system(script)


def _print_request_and_specs(self, request_and_specs):
    if not request_and_specs:
        return
    print request_and_specs[0]
    for spec in request_and_specs[1:]:
        print ' *', spec


@cmdln.alias('check', 'cr')
@cmdln.option('-p', '--project', dest='project', metavar='PROJECT', default='Factory',
              help='select a different project instead of openSUSE:Factory')
@cmdln.option('-s', '--skip', action='store_true', help='skip review')
@cmdln.option('-c', '--skipcycle', action='store_true', help='skip cycle check')
@cmdln.option('-n', '--dry', action='store_true', help='dry run, don\'t change review state')
@cmdln.option('-v', '--verbose', action='store_true', help='verbose output')
def do_check_repo(self, subcmd, opts, *args):
    """${cmd_name}: Checker review of submit requests.

    Usage:
       ${cmd_name} [SRID]...
           Shows pending review requests and their current state.
       ${cmd_name} PRJ
           Shows pending review requests in a specific project.
    ${cmd_option_list}
    """

    # Makes sure to remove the place where are the downloaded files
    # (is not the cache)
    try:
        shutil.rmtree(DOWNLOADS)
    except:
        pass
    os.makedirs(DOWNLOADS)

    Config('openSUSE:%s' % opts.project)
    self.checkrepo = CheckRepo(self.get_api_url(),
                               'openSUSE:%s' % opts.project,
                               readonly=opts.dry,
                               debug=opts.verbose)

    prjs_or_pkg = [arg for arg in args if not arg.isdigit()]
    ids = [arg for arg in args if arg.isdigit()]

    # Recover the requests that are for this project or package and
    # expand ids.
    for pop in prjs_or_pkg:
        # We try it as a project first
        as_prj = pop
        if ':' not in as_prj:
            as_prj = self.checkrepo.staging.prj_from_letter(as_prj)
        try:
            meta = self.checkrepo.staging.get_prj_pseudometa(as_prj)
            ids.extend(rq['id'] for rq in meta['requests'])
        except:
            # Now as a package
            as_pkg = pop
            srs = RequestFinder.find_sr([as_pkg], self.checkrepo.staging)
            ids.extend(srs.keys())

    if opts.skip:
        if not len(ids):
            raise oscerr.WrongArgs('Provide #IDs or package names to skip.')

        for request_id in ids:
            msg = 'skip review'
            print 'ACCEPTED', msg
            self.checkrepo.change_review_state(request_id, 'accepted', message=msg)
            _request = self.checkrepo.get_request(request_id, internal=True)
            self.checkrepo.remove_link_if_shadow_devel(_request)
        return

    # Store requests' package information and .spec files: store all
    # source containers involved.
    requests = []
    print 'Pending requests list:'
    print '----------------------'
    if not ids:
        # Return a list, we flat here with .extend()
        for request in self.checkrepo.pending_requests():
            request_and_specs = self.checkrepo.check_specs(request=request)
            self._print_request_and_specs(request_and_specs)
            requests.extend(request_and_specs)
    else:
        # We have a list, use them.
        for request_id in ids:
            request_and_specs = self.checkrepo.check_specs(request_id=request_id)
            self._print_request_and_specs(request_and_specs)
            requests.extend(request_and_specs)

    # Order the packs before grouping
    requests = sorted(requests, key=lambda p: p.request_id, reverse=True)

    # Group the requests into staging projects (or alone if is an
    # isolated request)
    #
    # For example:
    # {
    #     'openSUSE:Factory:Staging:J': [235851, 235753],
    #     235856: [235856],
    # }
    #
    # * The list of requests is not the full list of requests in this
    #   group / staging project, but only the ones listed as a
    #   paramenter.
    #
    # * The full list of requests can be found in
    #   self.checkrepo.groups['openSUSE:Factory:Staging:J']
    #
    groups = {}
    for request in requests:
        rqs = groups.get(request.group, [])
        rqs.append(request)
        groups[request.group] = rqs

    # Mirror the packages locally in the CACHEDIR
    self.repo_dir = '%s/repo-%s-%s-x86_64' % (CACHEDIR, 'openSUSE:{}'.format(opts.project), 'standard')
    self._mirror_full(PLUGINDIR, self.repo_dir)

    print
    print 'Analysis results'
    print '----------------'
    print

    # Sort the groups, from high to low. This put first the stating
    # projects also
    for id_, reqs in sorted(groups.items(), reverse=True):
        try:
            self._check_repo_group(id_, reqs,
                                   skip_cycle=opts.skipcycle,
                                   debug=opts.verbose)
        except Exception as e:
            print 'ERROR -- An exception happends while checking a group [%s]' % e
            if conf.config['debug']:
                print traceback.format_exc()
        print
        print