David Caro has uploaded a new change for review.
Change subject: Added script to manage repositories
......................................................................
Added script to manage repositories
Quite big script, here's what it can do right now:
* Remove duplicated rpms and create hard links instead
* Sign packages in batch mode (passing passphrase as parameter)
* Create/update rpm repos using multiple sources, passed as arguments or in a
configuration file:
· From jenkins url (multijob or simple job)
· From koji url
· From a url with rpm links on it, recursively or not
· From a directory, all packages, optionally filtering
· From a directory only the latest packages, optionally filtering
* Remove the old rpm versions and keep only the N newest
* Run createrepo in a multidistro repository (runs on each distribution subrepo)
* Generate the sources tree from the source rpms:
· Including the patches or not
· Generating detached signatures or not
I tried to keep the output informative but human readable, but mostly to be run
on jenkins jobs.
Change-Id: Ib630eb1e3b701eaf224451f02692723447aa1f8b
Signed-off-by: David Caro <dcaro...@redhat.com>
---
A repoman.py
1 file changed, 983 insertions(+), 0 deletions(-)
git pull ssh://gerrit.ovirt.org:29418/jenkins refs/changes/99/33299/1
diff --git a/repoman.py b/repoman.py
new file mode 100755
index 0000000..5aa0fa2
--- /dev/null
+++ b/repoman.py
@@ -0,0 +1,983 @@
+#!/usr/bin/env python
+"""
+This program is a helper to rpm repo management, some common tasks
+"""
+import os
+import sys
+import shutil
+import rpm
+import hashlib
+import argparse
+import logging
+import subprocess
+import re
+import requests
+import gnupg
+import tempfile
+import atexit
+import pexpect
+from urllib3 import connectionpool
+from getpass import getpass
+
+DISTROS = (
+ re.compile(r'\.fc\d+'),
+ re.compile(r'\.el\d+'),
+)
+
+BASE_REPOS_PATH = '/var/www/html/pub'
+# To be set on main, if set at all
+TEMP_DIR = None
+
+
+class NotSamePackage(Exception):
+ """Thrown when trying to compare different packages"""
+ pass
+
+
+def cleanup(temp_dir):
+ if os.path.isdir(temp_dir):
+ shutil.rmtree(temp_dir)
+ logging.info('Cleaning up temporary dir %s', temp_dir)
+
+
+def tryint(mayint):
+ try:
+ return int(mayint)
+ except ValueError:
+ return mayint
+
+
+def cmpver(ver1, ver2):
+ ver1 = '.' in ver1 and ver1.split('.') or (ver1,)
+ ver2 = '.' in ver2 and ver2.split('.') or (ver2,)
+ ver1 = [tryint(i) for i in ver1]
+ ver2 = [tryint(i) for i in ver2]
+ if ver1 > ver2:
+ return -1
+ if ver1 == ver2:
+ return 0
+ else:
+ return 1
+
+
+def cmpfullver(fullver1, fullver2):
+ ver1, rel1 = fullver1.split('-', 1)
+ ver2, rel2 = fullver2.split('-', 1)
+ ver_res = cmpver(ver1, ver2)
+ if ver_res != 0:
+ return ver_res
+ return cmpver(rel1, rel2)
+
+
+def print_busy(prev_pos=0):
+ sys.stdout.write('\r')
+ if prev_pos == 0:
+ sys.stdout.write('-')
+ elif prev_pos == 1:
+ sys.stdout.write('/')
+ elif prev_pos == 2:
+ sys.stdout.write('|')
+ else:
+ sys.stdout.write('\\')
+ sys.stdout.flush()
+ return (prev_pos + 1) % 4
+
+
+def get_distro(release):
+ for distro in DISTROS:
+ match = distro.search(release)
+ if match:
+ return match.group()[1:]
+ raise Exception('Unknown distro for %s' % release)
+
+
+def to_human_size(fsize):
+ mb = fsize/(1024*1024)
+ if mb >= 1:
+ return '%dM' % mb
+ kb = fsize / 1024
+ if kb >= 1:
+ return '%dK' % kb
+ return '%dB' % fsize
+
+
+def download(path, dest_path):
+ headers = requests.head(path)
+ chunk_size = 4096
+ length = int(headers.headers.get('content-length', 0)) or 0
+ logging.info('Downloading %s, length %s ...',
+ path,
+ length and to_human_size(length) or 'unknown')
+ num_dots = 100
+ dot_frec = length/num_dots or 1
+ stream = requests.get(path, stream=True)
+ prev_percent = 0
+ progress = 0
+ if length:
+ sys.stdout.write(' %[')
+ sys.stdout.flush()
+ with open(dest_path, 'w') as rpm_fd:
+ for chunk in stream.iter_content(chunk_size=chunk_size):
+ if chunk:
+ rpm_fd.write(chunk)
+ progress += len(chunk)
+ cur_percent = int(progress / dot_frec)
+ if length and cur_percent > prev_percent:
+ for _ in xrange(cur_percent - prev_percent):
+ sys.stdout.write('=')
+ sys.stdout.flush()
+ prev_percent = cur_percent
+ elif not length:
+ prev_percent = print_busy(prev_percent)
+ if length:
+ if cur_percent < num_dots:
+ sys.stdout.write('=')
+ sys.stdout.write(']')
+ sys.stdout.write('\n')
+ if not length:
+ logging.info(' Done')
+
+
+class RPM(object):
+ def __init__(self, path):
+ trans = rpm.TransactionSet()
+ # Ignore unsigned rpms
+ trans.setVSFlags(rpm._RPMVSF_NOSIGNATURES)
+ if path.startswith('http:') or path.startswith('https:'):
+ name = path.rsplit('/', 1)[-1]
+ if not name:
+ raise Exception('Passed trailing slash in path %s, '
+ 'unable to guess package name'
+ % path)
+ fpath = TEMP_DIR + '/' + name
+ download(path, fpath)
+ path = fpath
+ self.path = path
+ with open(path) as fdno:
+ try:
+ hdr = trans.hdrFromFdno(fdno)
+ except Exception as exc:
+ print "Failed to parse header for %s" % path
+ raise
+ self.inode = os.fstat(fdno.fileno()).st_ino
+ self.is_source = hdr[rpm.RPMTAG_SOURCEPACKAGE] and True or False
+ self.sourcerpm = hdr[rpm.RPMTAG_SOURCERPM]
+ self.name = hdr[rpm.RPMTAG_NAME]
+ self.version = hdr[rpm.RPMTAG_VERSION]
+ self.release = hdr[rpm.RPMTAG_RELEASE]
+ self.signature = hdr[rpm.RPMTAG_SIGPGP]
+ self._raw_hdr = hdr
+ # will be calculated if needed
+ self._md5 = None
+ # ovirt release rnd guest tools pms must go to all the distros
+ if self.name.startswith('ovirt-release') \
+ or self.name.startswith('ovirt-guest-tools'):
+ self.distro = 'all'
+ else:
+ self.distro = get_distro(self.release)
+ self.arch = hdr[rpm.RPMTAG_ARCH] or 'none'
+ self.full_name = 'rpm(%s %s %s %s)' % (
+ self.name, self.distro, self.arch,
+ self.is_source and 'src' or 'bin',
+ )
+ # remove the distro from the release for the version string
+ if self.distro:
+ self.ver_release = re.sub(
+ r'\.%s[^.]*' % self.distro,
+ '',
+ self.release,
+ 1
+ )
+ else:
+ self.ver_release = self.release
+ self.full_version = '%s-%s' % (self.version, self.ver_release)
+
+ @property
+ def md5(self):
+ if self._md5 is None:
+ with open(self.path) as fdno:
+ self._md5 = hashlib.md5(fdno.read()).hexdigest()
+ return self._md5
+
+ def generate_path(self, distro=None):
+ if distro is None:
+ distro = self.distro
+ if self.is_source:
+ arch_path = 'SRPMS'
+ arch_name = 'src'
+ else:
+ arch_path = self.arch
+ arch_name = self.arch
+ return 'rpm/%s/%s/%s-%s-%s.%s.rpm' % (
+ distro,
+ arch_path,
+ self.name,
+ self.version,
+ self.release,
+ arch_name,
+ )
+
+ def sign(self, keyuid, passwd):
+ logging.info("SIGNING: %s", self.path)
+ child = pexpect.spawn(
+ 'rpmsign',
+ [
+ '--resign',
+ '-D', '_signature gpg',
+ '-D', '_gpg_name %s' % keyuid,
+ self.path,
+ ],
+ )
+ child.expect('Enter pass phrase: ')
+ child.sendline(passwd)
+ child.expect(pexpect.EOF)
+ child.close()
+ if child.exitstatus != 0:
+ raise Exception("Failed to sign package.")
+ self.__init__(self.path)
+ if not self.signature:
+ raise Exception("Failed to sign rpm %s with key '%s'"
+ % (self.path, keyuid))
+
+ def __str__(self):
+ return 'rpm(%s %s %s %s %s %s)' % (
+ self.name, self.version,
+ self.release, self.arch,
+ self.is_source and 'src' or 'bin',
+ self.signature and 'signed' or 'unsigned',
+ )
+
+ def __repr__(self):
+ return self.__str__()
+
+
+class RPMinode(list):
+ """
+ Simple list, abstracts a set of rpm inodes referencing the same
+ name-version
+ """
+ def __init__(self, inode):
+ self.inode = inode
+ super(RPMinode, self).__init__(self)
+
+ def delete(self, noop=False):
+ for pkg in self:
+ if noop:
+ os.remove(pkg.path)
+ else:
+ logging.info('NOOP::%s would have been removed', pkg.path)
+
+ def get_rpms(self, regmatch=None, fmatch=None):
+ pkgs = self
+ if regmatch:
+ pkgs = [pkg for pkg in self if regmatch.search(pkg.path)]
+ if fmatch:
+ pkgs = [pkg for pkg in pkgs if fmatch(pkg)]
+ return pkgs
+
+
+class RPMEntry(dict):
+ """Abstracsts a set of rpm inodes for a version"""
+ def __init__(self, version):
+ self.version = version
+ super(RPMEntry, self).__init__(self)
+
+ def add_pkg(self, pkg):
+ if pkg.inode not in self:
+ self[pkg.inode] = RPMinode(pkg.inode)
+ self[pkg.inode].append(pkg)
+ return True
+
+ def del_inode(self, inode, noop=False):
+ if inode in self:
+ self[inode].delete(noop)
+ self.pop(inode)
+
+ def get_rpms(self, regmatch=None, fmatch=None):
+ pkgs = []
+ for inode in self.itervalues():
+ pkgs.extend(inode.get_rpms(regmatch=regmatch,
+ fmatch=fmatch))
+ return pkgs
+
+
+class RPMVersionList(dict):
+ """List of available versions for a package name"""
+ def add_pkg(self, pkg, onlyifnewer):
+ if onlyifnewer and (
+ pkg.full_version in self
+ or next((ver for ver in self.keys()
+ if cmpfullver(ver, pkg.full_version) >= 0), None)):
+ return False
+ elif pkg.full_version not in self:
+ self[pkg.full_version] = RPMEntry(pkg.full_version)
+ return self[pkg.full_version].add_pkg(pkg)
+
+ def get_latest(self, num=1):
+ """
+ Returns the list of available inodes for the latest version
+ if any
+ """
+ if not self:
+ return None
+ if not num:
+ num = len(self)
+ sorted_list = self.keys()
+ sorted_list.sort(cmp=cmpfullver)
+ latest = {}
+ for pos in xrange(num):
+ latest[sorted_list[pos]] = self.get(sorted_list[pos])
+ return latest
+
+ def del_version(self, version, noop=False):
+ if version in self:
+ for inode in self[version].keys():
+ self[version].del_inode(inode, noop=noop)
+ self.pop(version)
+
+ def get_rpms(self, regmatch=None, fmatch=None, latest=0):
+ pkgs = []
+ for entry in self.get_latest(latest).itervalues():
+ pkgs.extend(entry.get_rpms(regmatch=regmatch,
+ fmatch=fmatch))
+ return pkgs
+
+
+class RPMList(dict):
+ """List of rpms, separated by name
+ The shape of the rpms list is:
+ {
+ name1:{
+ version1:{
+ inode1: [rpm_object1, ...],
+ inode2: [...],
+ },
+ version2:{...}
+ },
+ name2: {...}
+ }
+
+ Where the top level is abstracted by RPMList(dict)
+ The second level by RPMVersionList(dict)
+ The third by RPMEntry(dict)
+ The third by RPMinode(list)
+ And the last level by RPM
+
+ So:
+
+ RPMList{
+ name1: RPMVersionList{
+ version1: RPMEntry{
+ inode1: RPMinode[RPM, RPM, ...]
+ inode2: RPMinode[...]
+ },
+ version2: RPMEntry{...}
+ },
+ name2: RPMVersionList{...}
+ }
+ """
+ def add_pkg(self, pkg, onlyifnewer=False):
+ if pkg.full_name not in self:
+ self[pkg.full_name] = RPMVersionList()
+ return self[pkg.full_name].add_pkg(pkg, onlyifnewer)
+
+ def del_pkg(self, name):
+ if name in self:
+ for version in self[name].keys():
+ self[name].del_version(version)
+ self.pop(name)
+
+ def get_rpms(self, regmatch=None, fmatch=None, latest=0):
+ rpms = []
+ for version in self.itervalues():
+ rpms.extend(version.get_rpms(
+ regmatch=regmatch,
+ fmatch=fmatch,
+ latest=latest))
+ return rpms
+
+
+class RPMRepo(object):
+ """Represents the repository sctructure, does not include metadata"""
+ def __init__(self, path):
+ self.rpms = RPMList()
+ self.path = path
+ self.to_copy = []
+ self.distros = set()
+ if path:
+ logging.info('Loading repo %s', path)
+ for pkg in list_rpms(path):
+ self.add_pkg(RPM(pkg), to_copy=False, hidelog=True)
+ logging.info('Repo %s loaded', path)
+
+ def add_pkg(self, pkg, onlyifnewer=False, to_copy=True, hidelog=False):
+ if self.rpms.add_pkg(pkg, onlyifnewer):
+ if to_copy:
+ self.to_copy.append(pkg)
+ if not hidelog:
+ logging.info('Adding package %s to repo %s', pkg.path,
+ self.path)
+ else:
+ if not hidelog:
+ logging.info("Not adding %s, there's already an equal or "
+ "newer version", pkg)
+ if pkg.distro != 'all':
+ self.distros.add(pkg.distro)
+
+ def save(self, onlylatest=False):
+ logging.info('Saving new added rpms into %s', self.path)
+ for pkg in self.to_copy:
+ if onlylatest and not self.is_latest_version(pkg):
+ logging.info('Skipping %s a newer version is already '
+ 'in the repo.', pkg)
+ continue
+ if pkg.distro == 'all':
+ if not self.distros:
+ raise Exception('No distros found in the repo and no '
+ 'packages with any distros added.')
+ dst_distros = self.distros
+ else:
+ dst_distros = [pkg.distro]
+ for distro in dst_distros:
+ dst_path = self.path + '/' + pkg.generate_path(distro)
+ pkg_copy(pkg.path, dst_path)
+ logging.info('Saved %s\n', self.path)
+
+ def is_latest_version(self, pkg):
+ verlist = self.rpms.get(pkg.full_name, {})
+ if not verlist or pkg.full_version in verlist.get_latest():
+ return True
+ return False
+
+ def generate_sources(self, with_patches=False, key=None, passphrase=None):
+ logging.info("Generating src directory from srpms")
+ for name, versions in self.rpms.iteritems():
+ if not name.endswith('src)'):
+ continue
+ for version in versions.itervalues():
+ pkg = version.popitem()[1][0]
+ logging.info("Parsing srpm %s", pkg)
+ dst_dir = '%s/src/%s' % (self.path, pkg.name)
+ extract_sources(pkg.path, dst_dir, with_patches)
+ if key:
+ sign_sources(dst_dir, key, passphrase)
+ logging.info('src dir generated')
+
+ def createrepo(self):
+ for distro in self.distros:
+ logging.info('Creating metadata for %s', distro)
+ subprocess.call(['createrepo', self.path + '/rpm/' + distro])
+
+ def delete_old(self, keep=1, noop=False):
+ new_rpms = RPMList(self.rpms)
+ for name, versions in self.rpms.iteritems():
+ if len(versions) <= keep:
+ continue
+ to_keep = RPMVersionList()
+ for _ in range(keep):
+ latest = versions.get_latest()
+ to_keep.update(latest)
+ versions.pop(latest.keys()[0])
+ new_rpms[name] = to_keep
+ for version in versions.keys():
+ logging.info('Deleting %s version %s', name, version)
+ versions.del_version(version, noop)
+ self.rpms = new_rpms
+
+ def get_rpms(self, regmatch=None, fmatch=None, latest=0):
+ return self.rpms.get_rpms(
+ regmatch=regmatch,
+ fmatch=fmatch,
+ latest=latest)
+
+ def sign_rpms(self, key, passwd):
+ gpg = gnupg.GPG(gnupghome=os.path.expanduser('~/.gnupg'))
+ with open(key) as key_fd:
+ skey = gpg.import_keys(key_fd.read())
+ fprint = skey.results[0]['fingerprint']
+ keyuid = None
+ for key in gpg.list_keys(True):
+ if key['fingerprint'] == fprint:
+ keyuid = key['uids'][0]
+ for pkg in self.get_rpms(
+ fmatch=lambda pkg: not pkg.signature):
+ pkg.sign(keyuid, passwd)
+ logging.info("Done signing")
+
+
+def copy(what, where):
+ """Try to link, try to copy if cross-device"""
+ try:
+ os.link(what, where)
+ except OSError as oerror:
+ if oerror.errno == 18:
+ shutil.copy2(what, where)
+ else:
+ raise
+
+
+def extract_sources(rpm_path, dst_dir, with_patches=False):
+ if not os.path.isdir(dst_dir):
+ os.makedirs(dst_dir)
+ oldpath = os.getcwd()
+ if not dst_dir.startswith('/'):
+ dst_dir = oldpath + '/' + dst_dir
+ if not rpm_path.startswith('/'):
+ rpm_path = oldpath + '/' + rpm_path
+ dst_path = dst_dir + '/' + rpm_path.rsplit('/', 1)[-1]
+ copy(rpm_path, dst_path)
+ os.chdir(dst_dir)
+ try:
+ rpm2cpio = subprocess.Popen(['rpm2cpio', dst_path],
+ stdout=subprocess.PIPE)
+ cpio_cmd = ['cpio', '-iv', '*gz', '*.zip', '*.7z']
+ if with_patches:
+ cpio_cmd.append('*.patch')
+ with open(os.devnull, 'w') as devnull:
+ cpio = subprocess.Popen(
+ cpio_cmd,
+ stdin=rpm2cpio.stdout,
+ stdout=devnull,
+ stderr=devnull,
+ )
+ rpm2cpio.stdout.close()
+ cpio.communicate()
+ finally:
+ os.chdir(oldpath)
+ os.remove(dst_path)
+
+
+def sign_sources(src_dir, key, passphrase=None):
+ oldpath = os.getcwd()
+ if not src_dir.startswith('/'):
+ src_dir = oldpath + '/' + src_dir
+ gpg = gnupg.GPG(gnupghome=os.path.expanduser('~/.gnupg'))
+ with open(key) as key_fd:
+ skey = gpg.import_keys(key_fd.read())
+ fprint = skey.results[0]['fingerprint']
+ keyid = None
+ for key in gpg.list_keys(True):
+ if key['fingerprint'] == fprint:
+ keyid = key['keyid']
+ for dname, _, files in os.walk(src_dir):
+ for fname in files:
+ if fname.endswith('.sig'):
+ continue
+ fname = os.path.join(dname, fname)
+ with open(fname) as fd:
+ signature = gpg.sign_file(
+ fd,
+ keyid=keyid,
+ passphrase=passphrase,
+ detach=True,
+ )
+ if not signature.data:
+ raise Exception("Failed to sign file %s: \n%s",
+ file, signature.stderr)
+ with open(fname + '.sig', 'w') as sfd:
+ sfd.write(signature.data)
+
+
+def pkg_copy(src_path, dst_path):
+ if os.path.exists(dst_path):
+ logging.debug('Not saving %s, already exists', dst_path)
+ return
+ logging.info('Saving package %s', dst_path)
+ if not os.path.exists(dst_path.rsplit('/', 1)[0]):
+ os.makedirs(dst_path.rsplit('/', 1)[0])
+ copy(src_path, dst_path)
+
+
+def list_rpms(path):
+ '''Find all the rpms under the given dir'''
+ rpms = []
+ for root, _, files in os.walk(path):
+ for fname in files:
+ if fname.endswith('.rpm'):
+ rpms.append(root + '/' + fname)
+ return rpms
+
+
+def rm_dups(path, noop=False):
+ '''Remove all the duplicated rpms from a directory tree, favoring
+ the signed ones'''
+ rpms = RPMList()
+ for rpm_path in list_rpms(path):
+ rpms.add_pkg(RPM(rpm_path))
+ # create a backup just in case
+ if noop:
+ logging.info('Not creating backup, as nothing will be changed (NOOP)')
+ else:
+ logging.info('Creating backup at %s, delete if happy with the '
+ 'results', path + '.bkp')
+ subprocess.call(['cp', '-la', path, path + '.bkp'])
+ for pkg_name, versions in rpms.iteritems():
+ for version, inodes in versions.iteritems():
+ # skip if no duplicates
+ if len(inodes) <= 1:
+ logging.debug('No dupes for %s.%s', pkg_name, version)
+ continue
+ logging.info('Dupes found for %s.%s!', pkg_name, version)
+ # get a base package, with signature if able
+ base_inode = None
+ for inode, pkg_files in inodes.iteritems():
+ if pkg_files and pkg_files[0].signature is not None:
+ base_inode = inode
+ break
+ if base_inode is not None:
+ base = inodes.pop(base_inode)[0]
+ else:
+ # we do not have any signed copy for that package
+ base = inodes.popitem()[1][0]
+ # Now replace all the other with hard links to the base
+ logging.info(' Using %s as base', base)
+ for inode in inodes.itervalues():
+ for pkg in inode:
+ logging.info(' Linking %s -> %s', pkg.path, base.path)
+ if pkg.md5 != base.md5:
+ logging.warn(' MD5 checksums do not match '
+ '(probably mixed signed and unsigned '
+ 'copies)')
+ else:
+ logging.debug(' MD5 checksums match, ok')
+ if noop:
+ logging.info(' NOOP, not doing anything')
+ continue
+ os.remove(pkg.path)
+ copy(base.path, pkg.path)
+ logging.info('DONE')
+
+
+def read_conf(conf_file):
+ with open(conf_file) as conf_fd:
+ for line in conf_fd:
+ line = line.strip()
+ if not line or line.startswith('#'):
+ continue
+ yield line
+
+
+def expand_koji(koji_lvl1_url):
+ pkg_list = []
+ url_base = koji_lvl1_url.rsplit('/', 1)[0]
+ lvl1_page = requests.get(koji_lvl1_url).text
+ lvl2_reg = re.compile(r'(?<=href=")[^"]+(?=.*(buildArch|buildSRPM))')
+ logging.info('Parsing Koji URL: %s', koji_lvl1_url)
+ lvl2_urls = [
+ get_link(koji_lvl1_url, match.group())
+ for match in (lvl2_reg.search(i) for i in lvl1_page.splitlines())
+ if match
+ ]
+ for url in lvl2_urls:
+ logging.info(' Got 2nd level URL: %s', url)
+ pkg_list.extend(expand_page(url))
+ if not pkg_list:
+ logging.warn(' No packages found')
+ logging.info(' Done')
+ return pkg_list
+
+
+def expand_jenkins(jenkins_lvl1_url):
+ pkg_list = []
+ lvl1_page = requests.get(jenkins_lvl1_url + '/api/json?depth=3').json()
+ url = lvl1_page['url']
+ logging.info('Parsing jenkins URL: %s', jenkins_lvl1_url)
+ if url.endswith('/'):
+ url = url[:-1]
+ # handle multicongif jobs
+ for run in lvl1_page.get('runs', (lvl1_page,)):
+ if run['number'] != lvl1_page['number']:
+ continue
+ for artifact in run['artifacts']:
+ if not artifact['relativePath'].endswith('.rpm'):
+ continue
+ new_url = '%s/artifact/%s' % (run['url'], artifact['relativePath'])
+ pkg_list.append(new_url)
+ logging.info(' Got URL: %s', new_url)
+ if not pkg_list:
+ logging.warn(' No packages found')
+ logging.info(' Done')
+ return pkg_list
+
+
+def expand_page(page_url):
+ logging.info('Parsing URL: %s', page_url)
+ data = requests.get(page_url).text
+ pkg_reg = re.compile(r'(?<=href=")[^"]+\.rpm')
+ pkg_list = [
+ get_link(page_url, match.group())
+ for match in (pkg_reg.search(i) for i in data.splitlines())
+ if match
+ ]
+ for pkg_url in pkg_list:
+ logging.info(' Got package URL: %s', pkg_url)
+ return pkg_list
+
+
+def get_link(page_url, link, internal=False):
+ page_url = page_url.rsplit('?', 1)[0]
+ if page_url.endswith('/'):
+ page_url = page_url[:-1]
+ if link.startswith(page_url):
+ return link
+ elif link.startswith('/'):
+ base_url = '/'.join(page_url.split('/', 3)[:-1])
+ return base_url + link
+ elif re.match('https?://', link):
+ if internal:
+ return False
+ else:
+ return link
+ else:
+ return page_url + '/' + link
+
+
+def expand_recursive(page_url, level=0):
+ if level > 0:
+ logging.debug('Recursively fetching URL (level %d): %s',
+ level, page_url)
+ else:
+ logging.info('Recursively fetching URL (level %d): %s',
+ level, page_url)
+ pkg_list = []
+ data = requests.get(page_url).text
+ url_reg = re.compile(
+ r'(?<=href=")(/|%s|(?![^:]+?://))[^"]+/(?=")' % page_url)
+ next_urls = (
+ get_link(page_url, match.group(), internal=True)
+ for match in url_reg.finditer(data)
+ if match and (match.group().startswith(page_url)
+ or not match.group().startswith('http'))
+ and match.group() != page_url
+ )
+ for next_url in next_urls:
+ pkg_list.extend(expand_recursive(next_url, level + 1))
+ pkg_list.extend(expand_page(page_url))
+ return pkg_list
+
+
+def expand_pkg_url(pkg_url):
+ pkg_list = []
+ if 'koji' in pkg_url:
+ pkg_list.extend(expand_koji(pkg_url))
+ elif 'jenkins' in pkg_url:
+ pkg_list.extend(expand_jenkins(pkg_url))
+ elif not pkg_url.endswith('.rpm'):
+ pkg_list.extend(expand_page(pkg_url))
+ else:
+ pkg_list.append(pkg_url)
+ return pkg_list
+
+
+def expand_dir(pkg_dir, latest=0, name_reg='.*'):
+ if not pkg_dir.startswith('/'):
+ pkg_dir += '/' + BASE_REPOS_PATH
+ repo = RPMRepo(pkg_dir)
+ match = re.compile(name_reg)
+ return [pkg.path
+ for pkg in repo.get_rpms(
+ regmatch=match,
+ latest=latest)
+ ]
+
+
+def expand_source(pkg_src):
+ """Return a list of direct urls and absolute paths to the rpms
+ from the given rpm source"""
+ pkg_list = set()
+ if pkg_src.startswith('conf:'):
+ for conf_line in read_conf(pkg_src.split(':', 1)[1]):
+ # TODO: avoid conf loops
+ pkg_list = pkg_list.union(expand_source(conf_line))
+ elif pkg_src.startswith('pub:'):
+ for job in get_publisher_jobs(pkg_src):
+ pkg_list = pkg_list.union(expand_source(job))
+ elif pkg_src.startswith('rec:'):
+ pkg_list = pkg_list.union(expand_recursive(pkg_src.split(':', 1)[1]))
+ elif pkg_src.startswith('http'):
+ pkg_list = pkg_list.union(expand_pkg_url(pkg_src))
+ elif pkg_src.startswith('dir:'):
+ _, pkg_src = pkg_src.split(':', 1)
+ if ':' in pkg_src:
+ dirname, filter = pkg_src.split(':', 1)
+ else:
+ dirname, filter = pkg_src, '.*'
+ pkg_list = pkg_list.union(expand_dir(
+ pkg_dir=dirname,
+ name_reg=filter))
+ elif pkg_src.startswith('latest:'):
+ _, pkg_src = pkg_src.split(':', 1)
+ if ':' in pkg_src:
+ dirname, filter = pkg_src.split(':', 1)
+ else:
+ dirname, filter = pkg_src, '.*'
+ pkgs = set(expand_dir(
+ pkg_dir=dirname,
+ name_reg=filter,
+ latest=True))
+ for pkg in pkgs:
+ logging.info(" Adding rpm %s" % pkg)
+ pkg_list = pkg_list.union(pkgs)
+ elif ':' in pkg_src:
+ repo_name, pkg_name = pkg_src.split(':', 1)
+ pkg_list = pkg_list.union(expand_dir(
+ repo_name,
+ name_reg=pkg_name))
+ else:
+ logging.error('Unknown source %s', pkg_src)
+ return 1
+ return pkg_list
+
+
+def get_publisher_jobs(pub_url):
+ console_log = requests.get(pub_url + '/consoleText').text
+ entry_reg = re.compile(r'Copied.*(?<=")\w+[^"]*".*number [0-9]+')
+ pub_jobs = [
+ match.group() for match
+ in (entry_reg.search(i) for i in console_log.splitlines())
+ if match
+ ]
+ return pub_jobs
+
+
+def parse_args():
+ parser = argparse.ArgumentParser()
+ parser.add_argument('-v', '--verbose', action='store_true')
+ parser.add_argument('-n', '--noop', action='store_true')
+ subparsers = parser.add_subparsers(dest='action')
+
+ par_rm_dups = subparsers.add_parser('rm-dups', help='Remove duplicated '
+ 'packages and create hard links.')
+ par_rm_dups.add_argument('-d', '--dir', required=True,
+ help='Directory to add the package to.')
+
+ par_repo = subparsers.add_parser('repo', help='Repository management.')
+ par_repo.add_argument('-d', '--dir', required=True,
+ help='Directory of the repo.')
+ repo_subparser = par_repo.add_subparsers(dest='repoaction')
+ add_rpm = repo_subparser.add_parser('add', help='Add a package')
+ add_rpm.add_argument('-t', '--temp-dir', action='store', default=None,
+ help='Temporary dir to use when downloading packages')
+ add_rpm.add_argument(
+ 'rpm_source', nargs='+',
+ help='An rpm source to add, it can be one of: '
+ 'a path to a rpm/tarball, '
+ 'a path to another dir prepended with "dir:" or "latest:"'
+ ' with an optional regexp filter, like dir:path[:regexp], '
+ 'an http(s) url to an rpm/tarball, '
+ 'an http(s) url to a page containing rpms/tarballs, '
+ 'a link to a jenkins job, '
+ 'a link to a koji job, '
+ 'a link to a http(s) page to recurse prepended with "rec:", '
+ 'a repopath:regexp pair, where repopath is the path to a repo and '
+ 'regexp a filter for the rpms in it.'
+ ' a path to a configuration file prepended with "conf:" '
+ 'containing a plain text list of rpm sources, one by line (comments '
+ 'with # and empty lines allowed). '
+ 'For any repository path specified with dir: latest: or the '
+ 'repopath:regexp options, if it\'s not an absolute path, '
+ + BASE_REPOS_PATH + ' will be prepended.'
+
+ )
+
+ generate_src = repo_subparser.add_parser(
+ 'generate-src',
+ help='Populate the src dir with the tarballs from the src.rpm '
+ 'files in the repo')
+ generate_src.add_argument('-p', '--with-patches', action='store_true',
+ help='Include the patch files')
+ generate_src.add_argument('-k', '--key', action='store',
+ default=None,
+ help='Key to sign the sources with')
+ generate_src.add_argument('--passphrase', action='store',
+ default='ask',
+ help='Passphrase to use when signing')
+
+ rpms_sign = repo_subparser.add_parser(
+ 'sign-rpms',
+ help='sign the unsigned rpms.')
+ rpms_sign.add_argument('-k', '--key', action='store',
+ default=None, required=True,
+ help='Key to sign the rpms with')
+ rpms_sign.add_argument('--passphrase', action='store',
+ default='ask',
+ help='Passphrase to use when signing')
+
+ repo_subparser.add_parser(
+ 'createrepo',
+ help='Run createrepo on each distro repository.')
+
+ remove_old = repo_subparser.add_parser(
+ 'remove-old',
+ help='Remove old versions of packages.')
+ remove_old.add_argument('-k', '--keep', action='store',
+ default=1, help='Number of versions to '
+ 'keep')
+ return parser.parse_args()
+
+
+def main():
+ global TEMP_DIR
+ args = parse_args()
+
+ if args.verbose:
+ logging.root.level = logging.DEBUG
+ # we want connectionpool debug logs
+ connectionpool.log.setLevel(logging.DEBUG)
+ else:
+ logging.root.level = logging.INFO
+ # we don't want connectionpool info logs
+ connectionpool.log.setLevel(logging.WARN)
+
+ if args.action == 'rm-dups':
+ if args.dir.endswith('/'):
+ path = args.dir[:-1]
+ else:
+ path = args.dir
+ rm_dups(path=path, noop=args.noop)
+ elif args.action == 'repo':
+ if args.dir.endswith('/'):
+ path = args.dir[:-1]
+ else:
+ path = args.dir
+ repo = RPMRepo(path)
+ logging.info('')
+ if args.repoaction == 'add':
+ if args.temp_dir is None:
+ TEMP_DIR = tempfile.mkdtemp()
+ atexit.register(cleanup, TEMP_DIR)
+ else:
+ TEMP_DIR = args.temp_dir
+ if not os.path.exists(TEMP_DIR):
+ os.makedirs(TEMP_DIR)
+ pkg_list = []
+ logging.info('Resolving all the package sources')
+ for pkg_src in args.rpm_source:
+ pkg_list.extend(expand_source(pkg_src))
+ logging.info('All sources resolved')
+ logging.info('')
+ logging.info('Adding packages to the repo %s', repo.path)
+ for pkg in pkg_list:
+ pkg = RPM(pkg)
+ repo.add_pkg(pkg)
+ logging.info('')
+ repo.save()
+ elif args.repoaction == 'generate-src':
+ if args.key and args.passphrase == 'ask':
+ passphrase = getpass('Key passphrase: ')
+ else:
+ passphrase = args.passphrase
+ repo.generate_sources(args.with_patches, args.key, passphrase)
+ elif args.repoaction == 'createrepo':
+ repo.createrepo()
+ elif args.repoaction == 'remove-old':
+ repo.delete_old(keep=int(args.keep), noop=args.noop)
+ elif args.repoaction == 'sign-rpms':
+ if args.passphrase == 'ask':
+ passphrase = getpass('Key passphrase: ')
+ else:
+ passphrase = args.passphrase
+ repo.sign_rpms(key=args.key, passwd=passphrase)
+
+if __name__ == '__main__':
+ main()
--
To view, visit http://gerrit.ovirt.org/33299
To unsubscribe, visit http://gerrit.ovirt.org/settings
Gerrit-MessageType: newchange
Gerrit-Change-Id: Ib630eb1e3b701eaf224451f02692723447aa1f8b
Gerrit-PatchSet: 1
Gerrit-Project: jenkins
Gerrit-Branch: master
Gerrit-Owner: David Caro <dcaro...@redhat.com>
_______________________________________________
Engine-patches mailing list
Engine-patches@ovirt.org
http://lists.ovirt.org/mailman/listinfo/engine-patches
