Replace md5 with haslib, md5 is deprecated since 2.5 and gone in python3

[projects/pld-ftp-admin.git] / modules / ftptree.py

# vi: encoding=utf-8 ts=8 sts=4 sw=4 et

from __future__ import print_function

import os, config, string, urllib, re, rpm
from common import fileexists, noarchcachedir
from baseftptree import BasePkg, BaseFtpTree
from sign import is_signed

errnum = 0
quietmode = False

class SomeError(Exception):
    def __init__(self):
        return

    def __str__(self):
        return "An Error occured!"

def bailoutonerror():
    if not errnum == 0:
        print("%d error(s) encountered... aborting" % errnum)
        raise SomeError()

def pinfo(msg):
    print('INFO: ' + msg)

def perror(msg):
    global errnum
    errnum = errnum + 1
    print('ERR: ' + msg)

def pwarning(msg):
    print('WARN: ' + msg)

def rm(file, test = False):
    if test:
        if not os.path.exists(file):
            pinfo("TEST os.remove(%s): file doesn't exists" % file)
    else:
        try:
            os.remove(file)
        except OSError as e:
            pinfo("os.remove(%s): %s" % (file, e))
            #raise

def mv(src, dst, test = False):
    fsrc = src
    fdst = dst + '/' + src.split('/')[-1]
    if test:
        if not os.path.exists(fsrc):
            pinfo("TEST os.rename(%s, %s): source doesn't exists" % (fsrc, fdst))
        if not os.path.exists(dst):
            pinfo("TEST destination doesn't exist: %s" % dst)
    else:
        try:
            os.rename(fsrc, fdst)
        except OSError as e:
            pinfo("os.rename(%s, %s): %s" % (fsrc, fdst, e))
            raise

class Pkg(BasePkg):
    def __init__(self, nvr, tree):
        BasePkg.__init__(self, nvr, tree)
        self.name = string.join(nvr.split('-')[:-2], '-')
        self.version = nvr.split('-')[-2]
        self.release = nvr.split('-')[-1]
        self.marked4removal = False
        self.marked4moving = False
        self.marked4movingpool = []
        self.errors = []
        self.warnings = []

    def __cmp__(self, pkg):
        if self.name > pkg.name:
            return 1
        elif self.name < pkg.name:
            return -1
        else:
            return rpm.labelCompare(('0', self.version, self.release),
                                    ('0', pkg.version, pkg.release))


    # unfortunately can't do new Pkg(NVR), and have no "tree" in this pkg context
    # so this static function
    def is_debuginfo(self, nvr):
        """
        returns true if NVR is debuginfo package and separate debuginfo is enabled
        """
        if not config.separate_debuginfo:
            return False
        pkg = nvr.split('-')[:-2]
        return pkg[-1] == 'debuginfo'

    def is_sourcefile(self, file):
        """
        returns true if file is source package
        """
        return file[-8:] == '.src.rpm'

    # returns true if package build is integer
    def is_release(self):
        """
        To account Release tags with subver macros, we consider integer release
        if it contains odd number of dots:

        1 -> True
        0.1 -> False
        0.%{subver}.%{rel}, %{rel} = 1 -> 0.20010.1 -> True
        0.%{subver}.%{rel}, %{rel} = 0.1 -> 0.20010.0.1 -> False
        """
        return self.release.count('.') % 2 == 0

    def mark4moving(self):
        if not self.marked4moving:
            # Only one pkg in this pool can be marked for moving
            for pkg in self.marked4movingpool:
                pkg.unmark4moving()
            self.tree.marked4moving.append(self)
            self.marked4moving=True

    def unmark4moving(self):
        if self.marked4moving:
            self.tree.marked4moving.remove(self)
            self.marked4moving=False

    def mark4removal(self):
        if not self.marked4removal:
            self.tree.marked4removal.append(self)
            self.marked4removal=True

    def error(self, msg):
        self.errors.append(msg)
        if not quietmode:
            perror('%s %s' % (self.nvr, msg))

    def warning(self, msg):
        self.warnings.append(msg)
        if not quietmode:
            pwarning('%s %s' % (self.nvr, msg))

    def load(self, content=None):
        BasePkg.load(self, content)
        if 'move' in self.info:
            self.mark4moving()

    def writeinfo(self):
        f = open(self.tree.basedir+'/SRPMS/.metadata/'+self.nvr+'.src.rpm.info', 'w')
        for bid in self.build.keys():
            f.write("info:build:%s:requester:%s\ninfo:build:%s:requester_email:%s\n" % (bid, self.build[bid].requester, bid, self.build[bid].requester_email))
        for key in self.info.keys():
            f.write("info:%s:%s\n" % (key, string.join(self.info[key], ':')))
        for arch in self.files.keys():
            for rpm in self.files[arch]:
                f.write("file:%s:%s\n" % (arch, rpm))

    def remove(self, test = False):
        """
        Remove package from ftp
        """
        for arch in self.files.keys():
            for rpm in self.files[arch]:
                if self.is_debuginfo(rpm):
                    rm(self.tree.basedir + '/' + arch + '/debuginfo/' + rpm, test)
                else:
                    rm(self.tree.basedir + '/' + arch + '/RPMS/' + rpm, test)
                if arch == 'noarch':
                    if fileexists(noarchcachedir + rpm + '.filelist'):
                        rm(noarchcachedir + rpm + '.filelist', test)
                    if fileexists(noarchcachedir + rpm + '.reqlist'):
                        rm(noarchcachedir + rpm + '.reqlist', test)
        rm(self.tree.basedir + '/SRPMS/.metadata/' + self.nvr + '.src.rpm.info', test)

    def rpmfiles(self, debugfiles = True, sourcefiles  = True):
        """
        Return rpm files related to this package
        """
        files = []
        for arch, rpms in self.files.items():
            for nvr in rpms:
                if self.is_debuginfo(nvr):
                    if debugfiles:
                        files.append(self.tree.basedir + '/' + arch + '/debuginfo/' + nvr)
                else:
                    if self.is_sourcefile(nvr):
                        if sourcefiles:
                            files.append(self.tree.basedir + '/' + arch + '/RPMS/' + nvr)
                    else:
                        files.append(self.tree.basedir + '/' + arch + '/RPMS/' + nvr)
        return files

    def obsoletes(self):
        """
        Return obsoletes for all packages in Pkg:

        {'php-geshi': set(['geshi'])}

        """
        def rpmhdr(pkg):
            ts = rpm.ts()
            ts.setVSFlags(rpm.RPMVSF_NODSAHEADER)
            fdno = os.open(pkg, os.O_RDONLY)
            hdr = ts.hdrFromFdno(fdno)
            os.close(fdno)
            return hdr

        obsoletes = {}
        for rpmfile in self.rpmfiles():
            if not os.path.exists(rpmfile):
                continue
            hdr = rpmhdr(rpmfile)
            if not hdr[rpm.RPMTAG_OBSOLETES]:
                continue

            name = hdr[rpm.RPMTAG_NAME]
            if not name in obsoletes:
                obsoletes[name] = set()

            for tag in hdr[rpm.RPMTAG_OBSOLETES]:
                obsoletes[name].add(tag)

        return obsoletes

    def move(self, dsttree, test = False):
        if dsttree.has_key(self.nvr):
            movedany = False
            for arch in self.files.keys():
                if arch in dsttree[self.nvr].files.keys():
                    msg = ""
                    if test:
                        msg = "TEST "
                    pinfo("%sArch %s for %s is already present in dest tree; removing from srctree" % (msg, arch, self.nvr))
                    for rpm in self.files[arch]:
                        if self.is_debuginfo(rpm):
                            rm(self.tree.basedir + '/' + arch + '/debuginfo/' + rpm, test)
                        else:
                            rm(self.tree.basedir + '/' + arch + '/RPMS/' + rpm, test)
                else:
                    movedany = True
                    dsttree[self.nvr].files[arch] = self.files[arch]
                    for rpm in self.files[arch]:
                        if self.is_debuginfo(rpm):
                            mv(self.tree.basedir + '/' + arch + '/debuginfo/' + rpm, dsttree.basedir + '/' + arch + '/debuginfo/', test)
                        else:
                            mv(self.tree.basedir + '/' + arch + '/RPMS/' + rpm, dsttree.basedir + '/' + arch + '/RPMS/', test)
            if not test and movedany:
                for bid in self.build.keys():
                    dsttree[self.nvr].build[bid] = self.build[bid]
                dsttree[self.nvr].writeinfo()
            rm(self.tree.basedir + '/SRPMS/.metadata/' + self.nvr + '.src.rpm.info', test)
        else:
            # move files
            for arch in self.files.keys():
                for rpm in self.files[arch]:
                    if self.is_debuginfo(rpm):
                        mv(self.tree.basedir + '/' + arch + '/debuginfo/' + rpm, dsttree.basedir + '/' + arch + '/debuginfo/', test)
                    else:
                        mv(self.tree.basedir + '/' + arch + '/RPMS/' + rpm, dsttree.basedir + '/' + arch + '/RPMS/', test)

            # move metadata
            mv(self.tree.basedir + '/SRPMS/.metadata/' + self.nvr + '.src.rpm.info', dsttree.basedir + '/SRPMS/.metadata/', test)

class FtpTree(BaseFtpTree):
    def __init__(self, tree, loadall=False):
        BaseFtpTree.__init__(self, tree)
        self.loadedpkgs = {}
        self.marked4removal = []
        self.marked4moving = []
        self.pkgnames = []
        self.__loadpkgnames()
        if loadall:
            for pkgname in self.pkgnames:
                self.loadedpkgs[pkgname] = Pkg(pkgname, self)
        # Tests:
        self.do_checkbuild = True

    def __getitem__(self, key):
        if key in self.loadedpkgs:
            return self.loadedpkgs[key]
        elif key in self.pkgnames:
            pkg=Pkg(key, self)
            self.loadedpkgs[key]=pkg
            return pkg
        else:
            raise KeyError(key)

    def has_key(self, key):
        if key in self.pkgnames:
            return True
        else:
            return False

    def keys(self):
        return self.pkgnames

    def values(self):
        return self.loadedpkgs.values()

    def checktree(self, dsttree):
        self.__checkbuild(self.loadedpkgs.values())
        self.__checkarchs(dsttree, self.loadedpkgs.values())

    def testmove(self, dsttree, archivetree = None):
        self.__checkbuild(self.marked4moving)
        self.__checkarchs(dsttree, self.marked4moving)
        if not dsttree.treename.count("archive"):
            self.__checkduplicates(self.marked4moving)

        self.__checksigns(dsttree, self.marked4moving, test = True)
        self.__checkforobsoletes(dsttree, self.marked4moving, test = True)
        self.__checkforrelease(dsttree, self.marked4moving, test = True)

        if not self.treename.count("archive"):
            self.__rmolderfromsrc(test = True)
        if not dsttree.treename.count("archive"):
            self.__rmotherfromdst(dsttree, test = True, archivetree = archivetree)

        for pkg in self.marked4moving:
            pkg.move(dsttree, test = True)

    def movepkgs(self, dsttree, archivetree = None):
        if self.do_checkbuild:
            self.__checkbuild(self.marked4moving)
        bailoutonerror()

        self.__checkarchs(dsttree, self.marked4moving)
        bailoutonerror()

        self.__checksigns(dsttree, self.marked4moving)
        bailoutonerror()

        if not self.treename.count("archive"):
            self.__rmolderfromsrc()
        if not dsttree.treename.count("archive"):
            self.__rmotherfromdst(dsttree, archivetree = archivetree)

        for pkg in self.marked4moving:
            pkg.move(dsttree)

    def rpmfiles(self, debugfiles = True, sourcefiles = True):
        if self.do_checkbuild:
            self.__checkbuild(self.marked4moving)

        files = []
        for pkg in self.marked4moving:
            files += pkg.rpmfiles(debugfiles = debugfiles, sourcefiles = sourcefiles)
        return files

    def removepkgs(self):
        if self.do_checkbuild:
            self.__checkbuild(self.marked4removal)
        bailoutonerror()
        for pkg in self.marked4removal:
            pkg.remove()

    def mark4removal(self, wannabepkgs):
        self.__mark4something(wannabepkgs, Pkg.mark4removal)

    def mark4moving(self, wannabepkgs):
        self.__mark4something(wannabepkgs, Pkg.mark4moving)

    # Internal functions below
    def __arch_stringify(self, list):
        ret = []
        dist = config.ftp_dist;
        for arch in list:
            ret.append(dist + '-' + arch)
        return ' '.join(ret)

    def __loadpkgnames(self):
        def checkfiletype(name):
            if name[-13:]=='.src.rpm.info':
                return True
            else:
                return False
        list = filter(checkfiletype, os.listdir(self.basedir+'/SRPMS/.metadata'))
        self.pkgnames = map((lambda x: x[:-13]), list)

    def __mark4something(self, wannabepkgs, markfunction):
        def chopoffextension(pkg):
            found = pkg.find('.src.rpm')
            if found == -1:
                return pkg
            else:
                return pkg[:found]

        for wannabepkg in wannabepkgs:
            pkgname = chopoffextension(wannabepkg)
            if pkgname in self.pkgnames:
                if not pkgname in self.loadedpkgs.keys():
                    self.loadedpkgs[pkgname]=Pkg(pkgname, self)
                markfunction(self.loadedpkgs[pkgname])
            else:
                perror('%s not found in source tree' % pkgname)
        bailoutonerror()

    def __checkbuild(self, marked):
        """
        Checks queue file if all arches are built

        Reads config.builderqueue to grab the info
        """
        f = urllib.urlopen(config.builderqueue)
        requests = {}
        reid = re.compile(r'^.*id=(.*) pri.*$')
        regb = re.compile(r'^group:.*$|builders:.*$', re.M)
        for i in re.findall(regb, f.read()):
            if i[0] == 'g':
                id = reid.sub(r'\1', i)
                requests[id] = ""
            elif i[0]=='b':
                requests[id] = requests[id] + i
        f.close()

        for pkg in marked:
            for bid in pkg.build.keys():
                if bid in requests and not requests[bid].find('?') == -1:
                    pkg.error("(buildid %s) building not finished" % bid)

    def __checkarchs(self, dsttree, marked):
        """
        Checks marked pkgs it is built on all archs.
        """
        for pkg in marked:
            if len(pkg.files.keys()) <= 1:
                pkg.error('has only src.rpm built')
                continue
            otherpkgnames = self.__find_other_pkgs(pkg, dsttree)

            # check if we're not removing some archs
            if otherpkgnames:
                curarchs = []
                missingarchs = []
                for somepkg in otherpkgnames:
                    curarchs.extend(Pkg(somepkg, dsttree).files.keys())
                for arch in curarchs:
                    if arch not in pkg.files.keys():
                        missingarchs.append(arch)
                if missingarchs:
                    pkg.error('moving would remove archs: %s' % self.__arch_stringify(missingarchs))
            else:
                # warn if a package isn't built for all archs
                # ftp_archs + SRPMS
                ftp_archs_num = len(config.ftp_archs) + 1
                if (config.separate_noarch and 'noarch' in pkg.files.keys()):
                    # ftp_archs + SRPMS + noarch subpackages
                    ftp_archs_num += 1
                    # plain simple noarch package
                    if (len(pkg.files.keys()) == 2):
                        continue

                if len(pkg.files.keys()) != ftp_archs_num:
                    missingarchs = []
                    for arch in config.ftp_archs:
                        if arch not in pkg.files.keys():
                            missingarchs.append(arch)
                    pkg.warning('not built for archs: %s' % self.__arch_stringify(missingarchs))

    def __checkduplicates(self, marked):
        """
        Checks if marked packages contain duplicate packages (with different versions)
        """
        for pkg in marked:
            olderpkgnames = self.__find_older_pkgs(pkg)
            for i in olderpkgnames:
                markednames = [str(x) for x in marked]
                if i in markednames:
                    pkg.error('duplicate package: %s' % i)

    def __rmolderfromsrc(self, test = False):
        for pkg in self.marked4moving:
            olderpkgnames = self.__find_older_pkgs(pkg)
            for i in olderpkgnames:
                Pkg(i, self).remove(test)

    def __rmotherfromdst(self, dsttree, test = False, archivetree = None):
        for pkg in self.marked4moving:
            pkgnames = self.__find_other_pkgs(pkg, dsttree)
            for i in pkgnames:
                if archivetree == None:
                    Pkg(i, dsttree).remove(test)
                else:
                    Pkg(i, dsttree).move(archivetree, test = test)

    # Used more than once filter functions
    def __find_other_pkgs(self, pkg, tree):
        escapedpkgname = pkg.name.replace('.', '\.').replace('+', '\+')
        ziewre = re.compile(escapedpkgname + '-[^-]*-[^-]*$')
        def filter_other_pkgs(x):
            if ziewre.match(x) and not x == pkg.nvr:
                return True
            else:
                return False
        return filter(filter_other_pkgs, tree.pkgnames)

    def __find_older_pkgs(self, pkg):
        def filter_older_pkgs(x):
            c = x.split('-')
            rc = rpm.labelCompare(('0', pkg.version, pkg.release),
                                                        ('0', c[-2], c[-1]))
            if rc == 1: # pkg > x
                return True
            else:
                return False
        return filter(filter_older_pkgs, self.__find_other_pkgs(pkg, self))

    def __checksigns(self, tree, pkgs, test = False):
        """
        Checks if pkgs in tree are all signed.

        in case of test = true, error flag is set for unsigned packages
        """
        if not tree.treename in config.signed_trees:
            return

        for pkg in pkgs:
            unsigned = 0
            for file in pkg.rpmfiles():
                if not is_signed(file):
                    unsigned += 1

            if unsigned != 0:
                if test == True:
                    if not quietmode:
                        pkg.warning('%d files not signed' % unsigned)
                else:
                    pkg.error('%d files not signed' % unsigned)

    def __checkforobsoletes(self, tree, pkgs, test = False):
        """
        Checks queue file if package obsoletes something in destination tree and suggest for removal.

        Only NAME tag is compared, i.e virtual packages do not get reported.

        """
        if test != True:
            return

        def findbyname(name):
            def x(nvr):
                return '-'.join(nvr.split('-')[:-2]) == name
            return filter(x, tree.pkgnames)

        for pkg in pkgs:
            obsoletes = pkg.obsoletes()
            if not obsoletes:
                continue

            for pn, setlist in obsoletes.items():
                for item in setlist:
                    p = findbyname(item)
                    if p:
                        pkg.warning('obsoletes %s (via %s) in dest tree, perhaps you want rmpkg' % (p,pn))

    def __checkforrelease(self, tree, pkgs, test = False):
        """
        Checks queue file if package release is non integer.

        """
        if test != True:
            return

        for pkg in pkgs:
            if not pkg.is_release():
                pkg.warning('non-integer release: %s' % pkg.release)