euscanwww: Moved business logic to djeuscan.processing

Moved stuff from management commands to djeuscan.processing

Signed-off-by: volpino <fox91@anche.no>
This commit is contained in:
volpino
2012-06-22 09:32:39 +02:00
parent 6cc892717b
commit 9bfd728d4f
13 changed files with 1110 additions and 940 deletions

View File

@ -0,0 +1,3 @@
class FakeLogger(object):
def __getattr__(self, key):
return lambda *x, **y: None

View File

@ -0,0 +1,31 @@
from djeuscan.models import HerdLog, MaintainerLog, CategoryLog, WorldLog
from djeuscan import charts
from djeuscan.processing import FakeLogger
def regen_rrds(logger=None):
"""
Regenerates the rrd database
"""
if logger is None:
logger = FakeLogger()
logger.info("Regenering RRDs for world")
for wlog in WorldLog.objects.all():
charts.rrd_update('world', wlog.datetime, wlog)
logger.info("Regenering RRDs for categories")
for clog in CategoryLog.objects.all():
charts.rrd_update('category-%s' % clog.category,
clog.datetime, clog)
logger.info("Regenering RRDs for herds")
for hlog in HerdLog.objects.all():
charts.rrd_update('herd-%d' % hlog.herd.id, hlog.datetime, hlog)
logger.info("Regenering RRDs for maintainers")
for mlog in MaintainerLog.objects.all():
charts.rrd_update('maintainer-%d' % mlog.maintainer.id,
mlog.datetime, mlog)

View File

@ -0,0 +1,148 @@
import sys
from gentoolkit.query import Query
from gentoolkit.errors import GentoolkitFatalError
from django.db.transaction import commit_on_success
from django.core.management.color import color_style
from django.core.exceptions import ValidationError
from djeuscan.models import Package, Herd, Maintainer
class ScanMetadata(object):
def __init__(self, quiet=False):
self.quiet = quiet
self.style = color_style()
@commit_on_success
def scan(self, query=None, obj=None):
matches = Query(query).find(
include_masked=True,
in_installed=False,
)
if not matches:
sys.stderr.write(
self.style.ERROR("Unknown package '%s'\n" % query)
)
return
matches = sorted(matches)
pkg = matches.pop()
if '9999' in pkg.version and len(matches):
pkg = matches.pop()
if not obj:
obj, created = Package.objects.get_or_create(
category=pkg.category, name=pkg.name
)
else:
created = False
try:
obj.homepage = pkg.environment("HOMEPAGE")
obj.description = pkg.environment("DESCRIPTION")
except GentoolkitFatalError, err:
sys.stderr.write(
self.style.ERROR(
"Gentoolkit fatal error: '%s'\n" % str(err)
)
)
if created and not self.quiet:
sys.stdout.write('+ [p] %s/%s\n' % (pkg.category, pkg.name))
if pkg.metadata:
herds = dict(
[(herd[0], herd) for herd in pkg.metadata.herds(True)]
)
maintainers = dict(
[(m.email, m) for m in pkg.metadata.maintainers()]
)
existing_herds = [h.herd for h in obj.herds.all()]
new_herds = set(herds.keys()).difference(existing_herds)
old_herds = set(existing_herds).difference(herds.keys())
existing_maintainers = [m.email for m in obj.maintainers.all()]
new_maintainers = set(
maintainers.keys()).difference(existing_maintainers
)
old_maintainers = set(
existing_maintainers).difference(maintainers.keys()
)
for herd in obj.herds.all():
if herd.herd in old_herds:
obj.herds.remove(herd)
for herd in new_herds:
herd = self.store_herd(*herds[herd])
obj.herds.add(herd)
for maintainer in obj.maintainers.all():
if maintainer.email in old_maintainers:
obj.maintainers.remove(maintainer)
for maintainer in new_maintainers:
maintainer = maintainers[maintainer]
try:
maintainer = self.store_maintainer(
maintainer.name, maintainer.email
)
obj.maintainers.add(maintainer)
except ValidationError:
sys.stderr.write(
self.style.ERROR("Bad maintainer: '%s' '%s'\n" % \
(maintainer.name, maintainer.email))
)
obj.save()
def store_herd(self, name, email):
if not name:
name = '{nil}'
name = name.strip("\r").strip("\n").strip("\t").strip()
herd, created = Herd.objects.get_or_create(
herd=name,
defaults={"email": email}
)
if created and not self.quiet:
sys.stdout.write('+ [h] %s <%s>\n' % (name, email))
herd.email = email
herd.save()
return herd
def store_maintainer(self, name, email):
if not name:
name = email
if not name:
name = '{nil}'
maintainer, created = Maintainer.objects.get_or_create(
email=email,
defaults={"name": name}
)
if created:
if not self.quiet:
sys.stdout.write(
'+ [m] %s <%s>\n' % (name.encode('utf-8'), email)
)
return maintainer
def scan_metadata(packages=None, quiet=False, logger=None):
scan_handler = ScanMetadata(quiet=quiet)
if packages is None:
packages = Package.objects.all()
for pkg in packages:
if isinstance(pkg, Package):
scan_handler.scan('%s/%s' % (pkg.category, pkg.name), pkg)
else:
scan_handler.scan(pkg)

View File

@ -0,0 +1,320 @@
import subprocess
import portage
import sys
import os
import re
from django.db.transaction import commit_on_success
from django.core.management.color import color_style
from djeuscan.models import Package, Version, VersionLog
class ScanPortage(object):
def __init__(self, stdout=None, no_log=False, purge_packages=False,
purge_versions=False, kill_versions=False, quiet=False):
self.stdout = sys.stdout if stdout is None else stdout
self.no_log = no_log
self.purge_packages = purge_packages
self.purge_versions = purge_versions
self.kill_versions = kill_versions
self.quiet = quiet
self.style = color_style()
self._cache = {'packages': {}, 'versions': {}}
self._overlays = None
def cache_hash_package(self, category, name):
return '%s/%s' % (category, name)
def cache_store_package(self, package):
key = self.cache_hash_package(package.category, package.name)
self._cache['packages'][key] = package
def cache_get_package(self, category, name):
return self._cache['packages'].get(
self.cache_hash_package(category, name)
)
def cache_hash_version(self, category, name, version, revision, slot,
overlay):
key = '%s/%s-%s-r%s %s %s' % (category, name,
version, revision,
slot, overlay)
return key
def cache_get_version(self, category, name, version, revision, slot,
overlay):
key = self.cache_hash_version(category, name, version, revision, slot,
overlay)
return self._cache['versions'].get(key)
def cache_store_version(self, version):
key = self.cache_hash_version(
version.package.category, version.package.name, version.version,
version.revision, version.slot, version.overlay
)
self._cache['versions'][key] = version
def overlays(self):
if self._overlays:
return self._overlays
env = os.environ
env['OVERLAYS_LIST'] = 'all'
env['PRINT_COUNT_ALWAYS'] = 'never'
cmd = ['eix', '-!']
output = subprocess.Popen(cmd, stdout=subprocess.PIPE, env=env).\
communicate()[0]
output = output.strip().strip('\n').split('\n')
overlay_re = re.compile(r'^\[(?P<key>\d+)] "(?P<name>.*?)"')
self._overlays = {}
for line in output:
match = overlay_re.match(line)
if not match:
continue
self._overlays[match.group('key')] = match.group('name')
return self._overlays
@commit_on_success
def scan(self, query=None):
env = os.environ
env['MY'] = "<category>/<name>-<version>:<slot> [<overlaynum>]\n"
cmd = ['eix', '--format', '<availableversions:MY>', '--pure-packages',
'-x']
if query:
cmd.extend(['--exact', query])
if self.kill_versions:
if not self.quiet:
self.stdout.write('Killing existing versions...')
self.stdout.flush()
Version.objects.filter(packaged=True).update(alive=False)
if not self.quiet:
self.stdout.write('done\n')
output = subprocess.Popen(cmd, stdout=subprocess.PIPE, env=env).\
communicate()[0]
output = output.strip().strip('\n')
if len(output) == 0:
if not query:
return
if self.purge_packages:
if not self.quiet:
sys.stdout.write('- [p] %s\n' % (query))
if '/' in query:
cat, pkg = portage.catsplit(query)
Package.objects.filter(category=cat, name=pkg).delete()
else:
Package.objects.filter(name=query).delete()
else:
sys.stderr.write(
self.style.ERROR(
"Unknown package '%s'\n" % query
)
)
return
output = output.split('\n')
packages = {}
line_re = re.compile(
r'^(?P<cpv>.*?):(?P<slot>.*?) \[(?P<overlay>.*?)\]$'
)
package = None
for line in output:
match = line_re.match(line)
if not match:
continue
cpv = match.group('cpv')
slot = match.group('slot')
overlay = match.group('overlay')
cat, pkg, ver, rev = portage.catpkgsplit(cpv)
packages['%s/%s' % (cat, pkg)] = True
if not package or not \
(cat == package.category and pkg == package.name):
package = self.store_package(cat, pkg)
self.store_version(package, cpv, slot, overlay)
if self.purge_packages and not query:
for package in Package.objects.all():
cp = "%s/%s" % (package.category, package.name)
if cp not in packages:
if not self.quiet:
sys.stdout.write('- [p] %s\n' % (package))
package.delete()
def store_package(self, cat, pkg):
created = False
obj = self.cache_get_package(cat, pkg)
if not obj:
obj, created = Package.objects.get_or_create(
category=cat,
name=pkg
)
self.cache_store_package(obj)
if created:
if not self.quiet:
sys.stdout.write('+ [p] %s/%s\n' % (cat, pkg))
# Set all versions dead, then set found versions alive and
# delete old versions
if not self.kill_versions:
Version.objects.filter(
package=obj,
packaged=True
).update(alive=False)
return obj
def store_version(self, package, cpv, slot, overlay):
cat, pkg, ver, rev = portage.catpkgsplit(cpv)
overlays = self.overlays()
if overlay in overlays:
overlay = overlays[overlay]
else:
overlay = 'gentoo'
created = False
obj = self.cache_get_version(
package.category, package.name, ver, rev, slot, overlay
)
if not obj:
obj, created = Version.objects.get_or_create(
package=package, slot=slot,
revision=rev, version=ver,
overlay=overlay,
defaults={"alive": True, "packaged": True}
)
if not created: # Created objects have defaults values
obj.alive = True
obj.packaged = True
obj.save()
if created:
self.cache_store_version(obj)
# nothing to do (note: it can't be an upstream version because
# overlay can't be empty here)
if not created:
return
if not self.quiet:
sys.stdout.write('+ [v] %s \n' % (obj))
if overlay == 'gentoo':
package.n_packaged += 1
else:
package.n_overlay += 1
package.n_versions += 1
package.save()
if self.no_log:
return
VersionLog.objects.create(
package=obj.package,
action=VersionLog.VERSION_ADDED,
slot=obj.slot,
revision=obj.revision,
version=obj.version,
overlay=obj.overlay
)
@commit_on_success
def purge_versions(quiet=False, nolog=False):
# For each dead versions
for version in Version.objects.filter(packaged=True, alive=False):
if version.overlay == 'gentoo':
version.package.n_packaged -= 1
else:
version.package.n_overlay -= 1
version.package.n_versions -= 1
version.package.save()
if not quiet:
sys.stdout.write('- [v] %s\n' % (version))
if nolog:
continue
VersionLog.objects.create(
package=version.package,
action=VersionLog.VERSION_REMOVED,
slot=version.slot,
revision=version.revision,
version=version.version,
overlay=version.overlay
)
Version.objects.filter(packaged=True, alive=False).delete()
def scan_portage(packages=None, no_log=False, purge_packages=False,
purge_versions=False, prefetch=False, logger=None,
quiet=False, stdout=None):
stdout = sys.stdout if stdout is None else stdout
kill_versions = False
if packages is None:
prefetch = True
kill_versions = True
scan_handler = ScanPortage(
stdout=stdout,
no_log=no_log,
purge_packages=purge_packages,
purge_versions=purge_versions,
kill_versions=kill_versions,
quiet=quiet,
)
if not quiet:
stdout.write('Scanning portage tree...\n')
if prefetch:
if quiet:
stdout.write('Prefetching objects...')
stdout.flush()
for package in Package.objects.all():
scan_handler.cache_store_package(package)
for version in Version.objects.select_related('package').all():
scan_handler.cache_store_version(version)
if quiet:
stdout.write('done\n')
if packages is None:
scan_handler.scan()
else:
for pkg in packages:
if isinstance(pkg, Package):
scan_handler.scan('%s/%s' % (pkg.category, pkg.name), pkg)
else:
scan_handler.scan(pkg)
if purge_versions:
purge_versions(quiet, no_log)
if not quiet:
stdout.write('Done.\n')

View File

@ -0,0 +1,139 @@
import portage
import sys
from django.utils import timezone
from django.db.transaction import commit_on_success
from euscan import CONFIG, output
from euscan.scan import scan_upstream as euscan_scan_upstream
from djeuscan.models import Package, Version, EuscanResult, VersionLog
class ScanUpstream(object):
def __init__(self, quiet=False):
self.quiet = quiet
def scan(self, package):
CONFIG["format"] = "dict"
output.set_query(package)
euscan_scan_upstream(package)
out = output.get_formatted_output()
out_json = output.get_formatted_output("json")
try:
cpv = out[package]["metadata"]["cpv"]
except KeyError:
return {}
obj = self.store_package(cpv)
for res in out[package]["result"]:
self.store_version(obj, res["version"], " ".join(res["urls"]))
self.store_result(obj, out_json)
return out
def store_result(self, package, log):
# Remove previous logs
EuscanResult.objects.filter(package=package).delete()
obj = EuscanResult()
obj.package = package
obj.result = log
obj.datetime = timezone.now()
obj.save()
def store_package(self, cpv):
cat, pkg, ver, rev = portage.catpkgsplit(cpv)
obj, created = Package.objects.get_or_create(category=cat, name=pkg)
if created and not self.quiet:
sys.stdout.write('+ [p] %s/%s\n' % (cat, pkg))
# Set all versions dead, then set found versions alive and
# delete old versions
Version.objects.filter(package=obj, packaged=False).update(alive=False)
return obj
def store_version(self, package, ver, url):
obj, created = Version.objects.get_or_create(
package=package, slot='', revision='r0', version=ver, overlay='',
defaults={"alive": True, "urls": url, "packaged": False}
)
if not created:
obj.alive = True
obj.urls = url
obj.packaged = False
obj.save()
# If it's not a new version, just update the object and continue
if not created:
return
if not self.quiet:
sys.stdout.write('+ [u] %s %s\n' % (obj, url))
VersionLog.objects.create(
package=package,
action=VersionLog.VERSION_ADDED,
slot='',
revision='r0',
version=ver,
overlay=''
)
package.n_versions += 1
package.save()
@commit_on_success
def purge_versions(quiet=False):
# For each dead versions
for version in Version.objects.filter(packaged=False, alive=False):
VersionLog.objects.create(
package=version.package,
action=VersionLog.VERSION_REMOVED,
slot=version.slot,
revision=version.revision,
version=version.version,
overlay=version.overlay
)
version.package.n_versions -= 1
version.package.save()
if not quiet:
sys.stdout.write('- [u] %s %s\n' % (version, version.urls))
Version.objects.filter(packaged=False, alive=False).delete()
def scan_upstream(packages=None, purge_versions=False, quiet=False,
logger=None, stdout=None):
stdout = sys.stdout if stdout is None else stdout
scan_handler = ScanUpstream(quiet)
if not quiet:
stdout.write('Scanning upstream...\n')
if packages is None:
packages = Package.objects.all()
for pkg in packages:
if isinstance(pkg, Package):
scan_handler.scan('%s/%s' % (pkg.category, pkg.name))
else:
scan_handler.scan(pkg)
if purge_versions:
purge_versions(quiet)
if not quiet:
stdout.write('Done.\n')

View File

@ -0,0 +1,185 @@
import sys
from django.db.transaction import commit_on_success
from django.utils import timezone
from djeuscan.models import Package, Herd, Maintainer, Version
from djeuscan.models import HerdLog, MaintainerLog, CategoryLog, WorldLog
from djeuscan import charts
from djeuscan.processing import FakeLogger
from distutils.version import StrictVersion, LooseVersion
def _compare_versions(version1, version2):
try:
return cmp(StrictVersion(version1), StrictVersion(version2))
# in case of abnormal version number, fall back to LooseVersion
except ValueError:
return cmp(LooseVersion(version1), LooseVersion(version2))
def _add_safe(storage, key):
if key not in storage:
storage[key] = 1
else:
storage[key] += 1
def _add_last_ver(storage, version):
key = version['package_id']
if key not in storage:
storage[key] = version
return
if version['version'].startswith('9999'):
return
if _compare_versions(storage[key]['version'],
version['version']) < 0:
storage[key] = version
@commit_on_success
def update_counters(stdout=None, fast=False, quiet=False, nolog=False,
logger=None):
if logger is None:
logger = FakeLogger() # TODO: write log messages
if stdout is None:
stdout = sys.stdout
now = timezone.now()
categories = {}
herds = {}
maintainers = {}
wlog = None
if not nolog:
wlog = WorldLog()
wlog.datetime = now
for cat in Package.objects.values('category').distinct():
clog = CategoryLog()
clog.datetime = now
clog.category = cat['category']
categories[clog.category] = clog
for herd in Herd.objects.all():
hlog = HerdLog()
hlog.datetime = now
hlog.herd = herd
herds[herd.id] = hlog
for maintainer in Maintainer.objects.all():
mlog = MaintainerLog()
mlog.datetime = now
mlog.maintainer = maintainer
maintainers[maintainer.id] = mlog
package_queryset = Package.objects.all()
n_versions = {}
n_packaged = {}
n_overlay = {}
last_versions_gentoo = {}
last_versions_overlay = {}
last_versions_upstream = {}
if not fast:
attrs = ['id', 'version', 'overlay', 'packaged', 'package_id']
for version in Version.objects.all().values(*attrs):
overlay, packaged = version['overlay'], version['packaged']
package_id = version['package_id']
_add_safe(n_versions, package_id)
if not packaged:
_add_last_ver(last_versions_upstream, version)
continue
if overlay == 'gentoo':
_add_safe(n_packaged, package_id)
_add_last_ver(last_versions_gentoo, version)
else:
_add_safe(n_overlay, package_id)
_add_last_ver(last_versions_overlay, version)
for package in package_queryset.select_related('herds', 'maintainers'):
if not fast:
package.n_versions = n_versions.get(package.id, 0)
package.n_packaged = n_packaged.get(package.id, 0)
package.n_overlay = n_overlay.get(package.id, 0)
default = {'id': None}
package.last_version_gentoo_id = last_versions_gentoo.get(
package.id, default
)['id']
package.last_version_overlay_id = last_versions_overlay.get(
package.id, default
)['id']
package.last_version_upstream_id = last_versions_upstream.get(
package.id, default
)['id']
package.save()
n_packages_gentoo = int(package.n_packaged == package.n_versions)
n_packages_overlay = int(package.n_overlay and package.n_packaged \
+ package.n_overlay == package.n_versions)
n_packages_outdated = int(package.n_packaged + package.n_overlay \
< package.n_versions)
def update_row(storage, key):
storage[key].n_packages_gentoo += n_packages_gentoo
storage[key].n_packages_overlay += n_packages_overlay
storage[key].n_packages_outdated += n_packages_outdated
storage[key].n_versions_gentoo += package.n_packaged
storage[key].n_versions_overlay += package.n_overlay
storage[key].n_versions_upstream += package.n_versions - \
package.n_packaged - \
package.n_overlay
def update_log(storage, qs):
for row in qs:
update_row(storage, row['id'])
if not nolog:
update_log(herds, package.herds.all().values('id'))
update_log(maintainers, package.maintainers.all().values('id'))
update_row(categories, package.category)
wlog.n_packages_gentoo += n_packages_gentoo
wlog.n_packages_overlay += n_packages_overlay
wlog.n_packages_outdated += n_packages_outdated
wlog.n_versions_gentoo += package.n_packaged
wlog.n_versions_overlay += package.n_overlay
wlog.n_versions_upstream += package.n_versions - \
package.n_packaged - \
package.n_overlay
if nolog:
return
for clog in categories.values():
if not quiet:
stdout.write('+ [cl] %s\n' % clog)
charts.rrd_update('category-%s' % clog.category, now, clog)
clog.save()
for hlog in herds.values():
if not quiet:
stdout.write('+ [hl] %s\n' % hlog)
charts.rrd_update('herd-%d' % hlog.herd.id, now, hlog)
hlog.save()
for mlog in maintainers.values():
if not quiet:
stdout.write('+ [ml] %s\n' % mlog)
charts.rrd_update('maintainer-%d' % mlog.maintainer.id, now, mlog)
mlog.save()
charts.rrd_update('world', now, wlog)
wlog.save()

View File

@ -0,0 +1,66 @@
import subprocess
from StringIO import StringIO
from django.conf import settings
from djeuscan.processing import FakeLogger
def _launch_command(cmd):
"""
Helper for launching shell commands inside tasks
"""
fp = subprocess.Popen(cmd, stdout=subprocess.PIPE,
stderr=subprocess.PIPE)
output = StringIO(fp.communicate()[0])
return output.getvalue()
def emerge_sync():
"""
Launches an emerge --sync
"""
cmd = ["emerge", "--sync", "--root", settings.PORTAGE_ROOT,
"--config-root", settings.PORTAGE_CONFIGROOT]
return _launch_command(cmd)
def layman_sync():
"""
Syncs Layman repos
"""
from layman import Layman
l = Layman(config=settings.LAYMAN_CONFIG)
return l.sync(l.get_installed(), output_results=False)
def emerge_regen():
"""
Launches emerge --regen
"""
cmd = [
"emerge", "--regen", "--jobs", settings.EMERGE_REGEN_JOBS, "--root",
settings.PORTAGE_ROOT, "--config-root", settings.PORTAGE_CONFIGROOT
]
return _launch_command(cmd)
def eix_update():
"""
Launches eix-update
"""
cmd = ["eix-update"]
return _launch_command(cmd)
def update_portage_trees(logger=None):
logger = logger or FakeLogger()
logger.info("Running emerge --sync")
emerge_sync()
logger.info("Running layman --sync")
layman_sync()
logger.info("Running emerge --regen")
emerge_regen()
logger.info("Running eix-update")
eix_update()
logger.info("Done!")