#!/usr/bin/python -tt

#I need a script which will go through Rawhide and make a list of
#packages where:

#* The source package generates a subpackage that does NOT depend (either
#implicitly or explicitly) on another binary package/subpackage generated
#from that same source package.

#E.g.

#libfoo-1.2.3.src.rpm generates libfoo, libfoo-devel, and libfoo-utils.
#libfoo-devel depends on libfoo, but libfoo-utils doesn't depend on
#either libfoo or libfoo-devel.

#So, the results would flag libfoo-utils as an independent subpackage
#(source: libfoo).

#I don't really need the script as much as I need the results, but the
#script might be useful in the future.


# written by seth vidal
# copyright 2010 Red Hat, Inc
# GPLv2+




import yum
import re
import os


license_match = re.compile('(copying)|(license).*$').match

def has_likely_license_file(pkg):
    for fn in pkg.filelist:
        test = os.path.basename(fn).lower()
        if license_match(test):
            return True
    return False

       


my = yum.YumBase()
my.preconf.debuglevel = 1
my.preconf.disabled_plugins = ['auto-update-debuginfo', 'addl-history-info']
#my.arch.archlist = ['x86_64', 'noarch']
my.setCacheDir()
my.repos.disableRepo('*')
my.repos.enableRepo('rawhide')

pkgs_by_sourcerpm = {}

for pkg in my.pkgSack.returnNewestByNameArch():
    if not pkg.base_package_name in pkgs_by_sourcerpm:
        pkgs_by_sourcerpm[pkg.base_package_name] = set()
    pkgs_by_sourcerpm[pkg.base_package_name].add(pkg)
    
bases = pkgs_by_sourcerpm.keys()
for basepkg in bases:
    if len(pkgs_by_sourcerpm[basepkg]) < 2:
        del pkgs_by_sourcerpm[basepkg]


indy_rockers = {}
for basepkg in pkgs_by_sourcerpm:
    #print basepkg
    siblings = pkgs_by_sourcerpm[basepkg]
    for pkg in siblings:
        if pkg.name == basepkg: # if our name is the same as the basepkg then ignore us
            continue
        #print '  %s' % pkg
        still_indy = True # assume it is indy until proven otherwise
        for brother in siblings: # look at each sibling pkg
            if pkg == brother:
                continue
            #print brother
            for req in pkg.requires: 
                if brother.provides_for(req):
                    still_indy = False # if the sibling provides something the pkg requires 
                    break              # then we're not indy
            #print 'end'
        #print 'no more sibs'
        if still_indy: # if we're here and still indy then add us to the indy_rockers.
            if not basepkg in indy_rockers:
                indy_rockers[basepkg] = set()
            indy_rockers[basepkg].add(pkg)

# go through all the indyrockers and if their filelists includes a file named COPYING or LICENSE then toss them out
# of the list

for basepkg in sorted(indy_rockers.keys()):
    print '%s:' % basepkg
    for pkg in indy_rockers[basepkg]:
        if has_likely_license_file(pkg):
            print '  %s - has license or copying file' % pkg
        else:
            print '  %s' % pkg