Annotation of /mga-gnome/trunk/mga-gnome

#!/usr/bin/python -u

# A lot of the code comes from ftpadmin, see
#   http://git.gnome.org/browse/sysadmin-bin/tree/ftpadmin
# Written by Olav Vitters

# basic modules:
import os
import os.path
import sys
import re
import subprocess

# command line parsing, error handling:
import argparse
import errno

# overwriting files by moving them (safer):
import tempfile
import shutil

# version comparison:
import rpm

# opening tarballs:
import tarfile
import gzip
import bz2
import lzma # pyliblzma

# getting links from HTML document:
from sgmllib import SGMLParser
import urllib2
import urlparse

# for checking hashes
import hashlib

# for parsing ftp-release-list emails
import email
from email.mime.text import MIMEText

# to be able to sleep for a while
import time

# version freeze
import datetime

MEDIA="Core Release Source"
URL="http://download.gnome.org/sources/"
PKGROOT='~/pkgs'
SLEEP_INITIAL=300

re_majmin = re.compile(r'^([0-9]+\.[0-9]+).*')
re_version = re.compile(r'([-.]|\d+|[^-.\d]+)')

def version_cmp(a, b):
    """Compares two versions

    Returns
      -1 if a < b
      0  if a == b
      1  if a > b
    """

    return rpm.labelCompare(('1', a, '1'), ('1', b, '1'))

def get_latest_version(versions, max_version=None):
    """Gets the latest version number

    if max_version is specified, gets the latest version number before
    max_version"""
    latest = None
    for version in versions:
        if ( latest is None or version_cmp(version, latest) > 0 ) \
           and ( max_version is None or version_cmp(version, max_version) < 0 ):
            latest = version
    return latest

def judge_version_increase(version_old, version_new):
        """Judge quality of version increase:

        Returns a tuple containing judgement and message

        Judgement:
        Less than 0: Error
        0 to 4: Better not
        5+: Ok"""
        versions = (version_old, version_new)

        # First do a basic version comparison to ensure version_new is actually newer
        compare = version_cmp(version_new, version_old)

        if compare == 0:
            # 1.0.0 -> 1.0.1
            return (-2, "Already at version %s!" % (version_old))

        if compare != 1:
            # 1.0.1 -> 1.0.0
            return (-3, "Version %s is older than current version %s!" % (version_new, version_old))

        # Version is newer, but we don't want to see if it follows the GNOME versioning scheme
        majmins = [re_majmin.sub(r'\1', ver) for ver in versions if re_majmin.match(ver) is not None]

        if len(majmins) == 1:
            return (-1, "Version number scheme changes: %s" % (", ".join(versions)))

        if len(majmins) == 0:
            return (0, "Unsupported version numbers: %s" % (", ".join(versions)))

        # Follows GNOME versioning scheme
        # Meaning: x.y.z
        #          x = major
        #          y = minor  : even if stable
        #          z = micro

        # Major+minor the same? Then go ahead and upgrade!
        if majmins[0] == majmins[1]:
            # Majmin of both versions are the same, looks good!
            # 1.1.x -> 1.1.x or 1.0.x -> 1.0.x
            return (10, None)

        # More detailed analysis needed, so figure out the numbers
        majmin_nrs = [map(long, ver.split('.')) for ver in majmins]

        # Check/ensure major version number is the same
        if majmin_nrs[0][0] != majmin_nrs[1][0]:
            # 1.0.x -> 2.0.x
            return (1, "Major version number increase")

        # Minor indicates stable/unstable
        devstate = (majmin_nrs[0][1] % 2 == 0, majmin_nrs[1][1] % 2 == 0)

        # Upgrading to unstable is weird
        if not devstate[1]:
            if devstate[0]:
                # 1.2.x -> 1.3.x
                return (1, "Stable to unstable increase")

            # 1.3.x -> 1.5.x
            return (4, "Unstable to unstable version increase")

        # Unstable => stable is always ok
        if not devstate[0]:
            # 1.1.x -> 1.2.x
            return (5, "Unstable to stable")

        # Can only be increase of minors from one stable to the next
        # 1.0.x -> 1.2.x
        return (6, "Stable version increase")

def line_input (file):
    for line in file:
        if line[-1] == '\n':
            yield line[:-1]
        else:
            yield line

def call_editor(filename):
    """Return a sequence of possible editor binaries for the current platform"""

    editors = []

    for varname in 'VISUAL', 'EDITOR':
        if varname in os.environ:
            editors.append(os.environ[varname])

    editors.extend(('/usr/bin/editor', 'vi', 'pico', 'nano', 'joe'))

    for editor in editors:
        try:
            ret = subprocess.call([editor, filename])
        except OSError, e:
            if e.errno == 2:
                continue
            raise

        if ret == 127:
            continue

        return True

class urllister(SGMLParser):
    def reset(self):
        SGMLParser.reset(self)
        self.urls = []

    def start_a(self, attrs):
        href = [v for k, v in attrs if k=='href']
        if href:
            self.urls.extend(href)

class XzTarFile(tarfile.TarFile):

    OPEN_METH = tarfile.TarFile.OPEN_METH.copy()
    OPEN_METH["xz"] = "xzopen"

    @classmethod
    def xzopen(cls, name, mode="r", fileobj=None, **kwargs):
        """Open gzip compressed tar archive name for reading or writing.
           Appending is not allowed.
        """
        if len(mode) > 1 or mode not in "rw":
            raise ValueError("mode must be 'r' or 'w'")

        if fileobj is not None:
            fileobj = _LMZAProxy(fileobj, mode)
        else:
            fileobj = lzma.LZMAFile(name, mode)

        try:
            # lzma doesn't immediately return an error
            # try and read a bit of data to determine if it is a valid xz file
            fileobj.read(_LZMAProxy.blocksize)
            fileobj.seek(0)
            t = cls.taropen(name, mode, fileobj, **kwargs)
        except IOError:
            raise tarfile.ReadError("not a xz file")
        except lzma.error:
            raise tarfile.ReadError("not a xz file")
        t._extfileobj = False
        return t

if not hasattr(tarfile.TarFile, 'xzopen'):
    tarfile.open = XzTarFile.open

def is_valid_hash(path, algo, hexdigest):
    if algo not in hashlib.algorithms:
        raise ValueError("Unknown hash algorithm: %s" % algo)

    local_hash = getattr(hashlib, algo)()

    with open(path, 'rb') as fp:
        data = fp.read(32768)
        while data:
            local_hash.update(data)
            data = fp.read(32768)

    return local_hash.hexdigest() == hexdigest

class SpecFile(object):
    re_update_version = re.compile(r'^(?P<pre>Version:\s*)(?P<version>.+)(?P<post>\s*)$', re.MULTILINE + re.IGNORECASE)
    re_update_release = re.compile(r'^(?P<pre>Release:\s*)(?P<release>%mkrel \d+)(?P<post>\s*)$', re.MULTILINE + re.IGNORECASE)

    def __init__(self, path):
        self.path = path
        self.cwd = os.path.dirname(path)

    @property
    def version(self):
        return subprocess.check_output(["rpm", "--specfile", self.path, "--queryformat", "%{VERSION}\n"]).splitlines()[0]
    @property
    def sources(self):
        ts = rpm.ts()
        spec = ts.parseSpec(self.path)
        srclist = spec.sources if isinstance(spec.sources, (list, tuple)) \
                        else spec.sources()
        return dict((os.path.basename(name), name) for name, no, flags in srclist)

    def update(self, version):
        """Update specfile (increase version)"""
        cur_version = self.version

        (judgement, msg) = judge_version_increase(cur_version, version)

        if judgement < 0:
            print >>sys.stderr, "ERROR: %s!" % (msg)
            return False

        if judgement < 5:
            print "WARNING: %s!" % (msg)
            return False

        # XXX - os.path.join is hackish
        if subprocess.check_output(["svn", "diff", os.path.join(self.path, '..')]) != '':
            print >>sys.stderr, "ERROR: Package has uncommitted changes!"
            return False

        with open(self.path, "rw") as f:
            data = f.read()

            if data.count("%mkrel") != 1:
                print >>sys.stderr, "ERROR: Multiple %mkrel found; don't know what to do!"
                return False

            data, nr = self.re_update_version.subn(r'\g<pre>%s\g<post>' % version, data, 1)
            if nr != 1:
                print >>sys.stderr, "ERROR: Could not increase version!"
                return False

            data, nr = self.re_update_release.subn(r'\g<pre>%mkrel 1\g<post>', data, 1)
            if nr != 1:
                print >>sys.stderr, "ERROR: Could not reset release!"
                return False

            # Overwrite file with new version number
            write_file(self.path, data)


        # Verify that RPM also agrees that version number has changed
        if self.version != version:
            print "ERROR: Increased version to %s, but RPM doesn't agree!?!" % version
            return False

        try:
            # Download new tarball
            subprocess.check_call(['mgarepo', 'sync', '-d'], cwd=self.cwd)
            # Check patches still apply
            subprocess.check_call(['bm', '-p', '--nodeps'], cwd=self.cwd)
        except subprocess.CalledProcessError:
            return False

        return True

class Patch(object):
    """Do things with patches"""

    re_dep3 = re.compile(r'^(?:#\s*)?(?P<header>[-A-Za-z0-9]+?):\s*(?P<data>.*)$')
    re_dep3_cont = re.compile(r'^#?\s+(?P<data>.*)$')

    def __init__(self, path, show_path=False):
        """Path: path to patch (might not exist)"""
        self.path = path
        self.show_path = show_path

    def __str__(self):
        return self.path if self.show_path else os.path.basename(self.path)

    def add_dep3(self):
        """Add DEP-3 headers to a patch file"""
        if self.dep3['valid']:
            return False

        new_headers = (
            ('Author', self.svn_author),
            ('Subject', ''),
            ('Applied-Upstream', ''),
            ('Forwarded', ''),
            ('Bug', ''),
        )

        with tempfile.NamedTemporaryFile(dir=os.path.dirname(self.path), delete=False) as fdst:
            with open(self.path, "r") as fsrc:
                # Start with any existing DEP3 headers
                for i in range(self.dep3['last_nr']):
                    fdst.write(fsrc.read())

                # After that add the DEP3 headers
                add_line = False
                for header, data in new_headers:
                    if header in self.dep3['headers']:
                        continue

                    # XXX - wrap this at 80 chars
                    add_line = True
                    print >>fdst, "%s: %s" % (header, "" if data is None else data)

                if add_line: print >>fdst, ""
                # Now copy any other data and the patch
                shutil.copyfileobj(fsrc, fdst)

            fdst.flush()
            os.rename(fdst.name, self.path)

        call_editor(self.path)

    #Author: fwang
    #Subject: Build fix: Fix glib header inclusion
    #Applied-Upstream: commit:30602
    #Forwarded: yes
    #Bug: http://bugzilla.abisource.com/show_bug.cgi?id=13247

    def _read_dep3(self):
        """Read DEP-3 headers from an existing patch file

        This will also parse git headers"""
        dep3 = {}
        headers = {}

        last_header = None
        last_nr = 0
        nr = 0
        try:
            with open(self.path, "r") as f:
                for line in line_input(f):
                    nr += 1
                    # stop trying to parse when real patch begins
                    if line == '---':
                        break

                    r = self.re_dep3.match(line)
                    if r:
                        info = r.groupdict()

                        # Avoid matching URLS
                        if info['data'].startswith('//') and info['header'].lower () == info['header']:
                            continue

                        headers[info['header']] = info['data']
                        last_header = info['header']
                        last_nr = nr
                        continue

                    r = self.re_dep3_cont.match(line)
                    if r:
                        info = r.groupdict()
                        if last_header:
                            headers[last_header] = " ".join((headers[last_header], info['data']))
                            last_nr = nr
                        continue

                    last_header = None
        except IOError:
            pass

        dep3['valid'] = \
            (('Description' in headers and headers['Description'].strip() != '')
                or ('Subject' in headers and headers['Subject'].strip() != '')) \
            and (('Origin' in headers and headers['Origin'].strip() != '') \
                or ('Author' in headers and headers['Author'].strip() != '') \
                or ('From' in headers and headers['From'].strip() != ''))
        dep3['last_nr'] = last_nr
        dep3['headers'] = headers

        self._dep3 = dep3

    @property
    def dep3(self):
        if not hasattr(self, '_dep3'):
            self._read_dep3()

        return self._dep3

    @property
    def svn_author(self):
        if not hasattr(self, '_svn_author'):
            try:
                contents = subprocess.check_output(['svn', 'log', '-q', "--", self.path], close_fds=True).strip("\n").splitlines()

                for line in contents:
                    if ' | ' not in line:
                        continue

                    fields = line.split(' | ')
                    if len(fields) >= 3:
                        self._svn_author = fields[1]
            except subprocess.CalledProcessError:
                pass

        if not hasattr(self, '_svn_author'):
            return None

        return self._svn_author

def get_upstream_names():
    urlopen = urllib2.build_opener()

    good_dir = re.compile('^[-A-Za-z0-9_+.]+/$')

    # Get the files
    usock = urlopen.open(URL)
    parser = urllister()
    parser.feed(usock.read())
    usock.close()
    parser.close()
    files = parser.urls

    tarballs = set([filename.replace('/', '') for filename in files if good_dir.search(filename)])

    return tarballs

def get_downstream_names():
    re_file = re.compile(r'^(?P<module>.*?)[_-](?:(?P<oldversion>([0-9]+[\.])*[0-9]+)-)?(?P<version>([0-9]+[\.\-])*[0-9]+)\.(?P<format>(?:tar\.|diff\.)?[a-z][a-z0-9]*)$')

    contents = subprocess.check_output(['urpmf', '--files', '.', "--media", MEDIA], close_fds=True).strip("\n").splitlines()

    FILES = {}
    TARBALLS = {}

    for line in  contents:
        try:
            srpm, filename = line.split(":")
        except ValueError:
            print >>sys.stderr, line
            continue

        if '.tar' in filename:
            r = re_file.match(filename)
            if r:
                fileinfo = r.groupdict()
                module = fileinfo['module']

                if module not in TARBALLS:
                    TARBALLS[module] = set()
                TARBALLS[module].add(srpm)

        if srpm not in FILES:
            FILES[srpm] = set()
        FILES[srpm].add(filename)

    return TARBALLS, FILES

def get_downstream_from_upstream(upstream, version):
    # Determine the package name
    downstream, downstream_files = get_downstream_names()

    if upstream not in downstream:
        raise ValueError("No packages for upstream name: %s" % upstream)

    if len(downstream[upstream]) != 1:
        # XXX - Make it more intelligent
        raise ValueError("Multiple packages found for %s: %s" % (upstream, ", ".join(downstream[upstream])))

    return list(downstream[upstream])

def write_file(path, data):
    with tempfile.NamedTemporaryFile(dir=os.path.dirname(path), delete=False) as fdst:
        fdst.write(data)
        fdst.flush()
        os.rename(fdst.name, path)

def cmd_co(options, parser):
    upstream = get_upstream_names()
    downstream, downstream_files = get_downstream_names()

    cwd = os.path.expanduser(PKGROOT)

    matches = upstream & set(downstream.keys())
    for module in matches:
        print module, "\t".join(downstream[module])
        for package in downstream[module]:
            subprocess.call(['mgarepo', 'co', package], cwd=cwd)

def join_streams():
    upstream = get_upstream_names()
    downstream, downstream_files = get_downstream_names()

    matches = upstream & set(downstream.keys())
    for module in matches:
        for package in downstream[module]:
            yield (package, module)

def cmd_ls(options, parser):
    for package, module in sorted(join_streams()):
        print "\t".join((package, module)) if options.upstream else package

def cmd_patches(options, parser):
    upstream = get_upstream_names()
    downstream, downstream_files = get_downstream_names()

    path = os.path.expanduser(PKGROOT)

    import pprint

    matches = upstream & set(downstream.keys())
    for module in sorted(matches):
        for srpm in downstream[module]:
            for filename in downstream_files[srpm]:
                if '.patch' in filename or '.diff' in filename:

                    p = Patch(os.path.join(path, srpm, "SOURCES", filename), show_path=options.path)
                    valid = ""
                    forwarded = ""
                    if p.dep3['headers']:
                        forwarded = p.dep3['headers'].get('Forwarded', "no")
                        if p.dep3['valid']:
                            valid="VALID"
                    print "\t".join((module, srpm, str(p), forwarded, valid))

def cmd_dep3(options, parser):
    p = Patch(options.patch)
    p.add_dep3()

def cmd_package_new_version(options, parser):
    # Determine the package name
    if options.upstream:
        try:
            package = get_downstream_from_upstream(options.package, options.version)[0]
        except ValueError, e:
            print >>sys.stderr, "ERROR: %s" % e
            sys.exit(1)
    else:
        package = options.package

    # Directories packages are located in
    root = os.path.expanduser(PKGROOT)
    cwd = os.path.join(root, package)

    # Checkout package to ensure the checkout reflects the latest changes
    try:
        subprocess.check_call(['mgarepo', 'co', package], cwd=root)
    except subprocess.CalledProcessError:
        sys.exit(1)

    # SpecFile class handles the actual version+release change
    s = SpecFile(os.path.join(cwd, "SPECS", "%s.spec" % package))
    print "%s => %s" % (s.version, options.version)
    if not s.update(options.version):
        sys.exit(1)

    # Check hash, if given
    if options.hexdigest is not None:
        sources = [name for name, origname in s.sources.iteritems() if '://' in origname]
        if not len(sources):
            print >>sys.stderr, "ERROR: Cannot determine source file (for hash check)!"
            sys.stderr(1)

        for filename in sources:
            if not is_valid_hash(os.path.join(cwd, "SOURCES", filename), options.algo, options.hexdigest):
                print >>sys.stderr, "ERROR: Hash file failed check for %s!" % path
                print >>sys.stderr, "ERROR: Reverting changes!"
                subprocess.call(['svn', 'revert', '-R', cwd], cwd=cwd)
                sys.exit(1)

    # We can even checkin and submit :-)
    if options.submit:
        try:
            # checkin changes
            subprocess.check_call(['mgarepo', 'ci', '-m', 'new version %s' % options.version], cwd=cwd)
            # and submit
            subprocess.check_call(['mgarepo', 'submit'], cwd=cwd)
        except subprocess.CalledProcessError:
            sys.exit(1)

def cmd_parse_ftp_release_list(options, parser):
    # XXX - not working yet
    def _send_reply_mail(contents, orig_msg, to, error=False):
        """Send an reply email"""
        contents.seek(0)
        msg = MIMEText(contents.read(), _charset='utf-8')
        if error:
            # XXX - ugly
            contents.seek(0)
            lastline = contents.read().splitlines()[-1]
            # Remove things like "ERROR: " and so on from the last line
            lastline = re.sub(r'^(?:[^ :]+:\s+)+', '', lastline)
            subjecterror = " (ERROR: %s)" % lastline if lastline else " (ERROR)"
        else:
            subjecterror = ""
        msg['Subject'] = "Re: %s%s" % (orig_msg['Subject'], subjecterror)
        msg['To'] = to
        msg["In-Reply-To"] = orig_msg["Message-ID"]
        msg["References"] = orig_msg["Message-ID"]

        # Call sendmail program directly so it doesn't matter if the service is running
        cmd = ['/usr/sbin/sendmail', '-oi', '--']
        cmd.extend([to])
        p = subprocess.Popen(cmd, stdin=subprocess.PIPE)
        p.stdin.write(msg.as_string())
        p.stdin.flush()
        p.stdin.close()
        p.wait()


    msg = email.email.message_from_file(sys.stdin)

    if options.mail:
        stdout = tempfile.TemporaryFile()
        stderr = stdout
    else:
        stdout = sys.stdout
        stderr = sys.stderr

    version_freeze = datetime.datetime(2012,3,7,23,59,0)
    now = datetime.datetime.utcnow()
    if now > version_freeze:
        print >>stderr, "ERROR: Past version freeze"
        if options.mail: _send_reply_mail(stdout, msg, options.mail, error=True)
        sys.exit(1)

    try:
        module = msg['X-Module-Name']
        version = msg['X-Module-Version']
        hexdigest = msg['X-Module-SHA256-tar.xz']
    except KeyError, e:
        print >>stderr, "ERROR: %s" % e
        if options.mail: _send_reply_mail(stdout, msg, options.mail, error=True)
        sys.exit(1)

    try:
        packages = get_downstream_from_upstream(module, version)
    except ValueError, e:
        print >>stderr, "ERROR: %s" % e
        if options.mail: _send_reply_mail(stdout, msg, options.mail, error=True)
        sys.exit(1)

    if options.wait:
        # maildrop aborts and will try to deliver after 5min
        # fork to avoid this
        if os.fork() != 0: sys.exit(0)
        time.sleep(SLEEP_INITIAL)

    error = False
    for package in packages:
        if subprocess.call(['mga-gnome', 'increase', '--submit', '--hash', hexdigest, package, version], stdout=stdout, stderr=stderr):
            error = True

    if options.mail: _send_reply_mail(stdout, msg, options.mail, error=error)

def main():
    description = """Mageia GNOME commands."""
    epilog="""Report bugs to Olav Vitters"""
    parser = argparse.ArgumentParser(description=description,epilog=epilog)

    # SUBPARSERS
    subparsers = parser.add_subparsers(title='subcommands')
    #   install
    subparser = subparsers.add_parser('co', help='checkout all GNOME modules')
    subparser.set_defaults(
        func=cmd_co
    )

    subparser = subparsers.add_parser('packages', help='list all GNOME packages')
    subparser.add_argument("-m", "--m", action="store_true", dest="upstream",
                                       help="Show upstream module")
    subparser.set_defaults(
        func=cmd_ls, upstream=False
    )

    subparser = subparsers.add_parser('patches', help='list all GNOME patches')
    subparser.add_argument("-p", "--path", action="store_true", dest="path",
                                       help="Show full path to patch")
    subparser.set_defaults(
        func=cmd_patches, path=False
    )

    subparser = subparsers.add_parser('dep3', help='Add dep3 headers')
    subparser.add_argument("patch", help="Patch")
    subparser.set_defaults(
        func=cmd_dep3, path=False
    )

    subparser = subparsers.add_parser('increase', help='Increase version number')
    subparser.add_argument("package", help="Package name")
    subparser.add_argument("version", help="Version number")
    subparser.add_argument("-u", "--upstream", action="store_true", dest="upstream",
                                       help="Package name reflects the upstream name")
    subparser.add_argument("-s", "--submit", action="store_true", dest="submit",
                                       help="Commit changes and submit")
    subparser.add_argument("-a", "--algorithm", choices=hashlib.algorithms, dest="algo",
                                       help="Hash algorithm")
    subparser.add_argument("--hash", dest="hexdigest",
                                       help="Hexdigest of the hash")
    subparser.set_defaults(
        func=cmd_package_new_version, submit=False, upstream=False, hexdigest=None, algo="sha256"
    )

    subparser = subparsers.add_parser('gnome-release-email', help='Submit packages based on GNOME ftp-release-list email')
    subparser.add_argument("-m", "--mail", help="Email address to send the progress to")
    subparser.add_argument("-w", "--wait", action="store_true",
                                 help="Wait before trying to retrieve the new version")
    subparser.set_defaults(
        func=cmd_parse_ftp_release_list
    )

    if len(sys.argv) == 1:
        parser.print_help()
        sys.exit(2)

    options = parser.parse_args()

    try:
        options.func(options, parser)
    except KeyboardInterrupt:
        print('Interrupted')
        sys.exit(1)
    except EOFError:
        print('EOF')
        sys.exit(1)
    except IOError, e:
        if e.errno != errno.EPIPE:
            raise
        sys.exit(0)

if __name__ == "__main__":
    main()