pungi/bin/comps_filter

#!/usr/bin/env python
# -*- coding: utf-8 -*-


import sys
import fnmatch
import argparse
import lxml.etree
import re
from io import StringIO


class CompsFilter(object):
    def __init__(self, file_obj, reindent=False):
        self.reindent = reindent
        parser = None
        if self.reindent:
            parser = lxml.etree.XMLParser(remove_blank_text=True)
        self.tree = lxml.etree.parse(file_obj, parser=parser)
        self.encoding = "utf-8"

    def _filter_elements_by_arch(self, xpath, arch, only_arch=False):
        if only_arch:
            # remove all elements without the 'arch' attribute
            for i in self.tree.xpath(xpath + "[not(@arch)]"):
                i.getparent().remove(i)

        for i in self.tree.xpath(xpath + "[@arch]"):
            arches = i.attrib.get("arch")
            arches = re.split(r"[, ]+", arches)
            arches = [j for j in arches if j]
            if arch not in arches:
                # remove elements not matching the arch
                i.getparent().remove(i)
            else:
                # remove the 'arch' attribute
                del i.attrib["arch"]

    def filter_packages(self, arch, only_arch=False):
        """
        Filter packages according to arch.
        If only_arch is set, then only packages for the specified arch are preserved.
        Multiple arches separated by comma can be specified in the XML.
        """
        self._filter_elements_by_arch("/comps/group/packagelist/packagereq", arch, only_arch)

    def filter_groups(self, arch, only_arch=False):
        """
        Filter groups according to arch.
        If only_arch is set, then only groups for the specified arch are preserved.
        Multiple arches separated by comma can be specified in the XML.
        """
        self._filter_elements_by_arch("/comps/group", arch, only_arch)

    def filter_environments(self, arch, only_arch=False):
        """
        Filter environments according to arch.
        If only_arch is set, then only environments for the specified arch are preserved.
        Multiple arches separated by comma can be specified in the XML.
        """
        self._filter_elements_by_arch("/comps/environment", arch, only_arch)

    def filter_category_groups(self):
        """
        Remove undefined groups from categories.
        """
        all_groups = self.tree.xpath("/comps/group/id/text()")
        for category in self.tree.xpath("/comps/category"):
            for group in category.xpath("grouplist/groupid"):
                if group.text not in all_groups:
                    group.getparent().remove(group)

    def remove_empty_groups(self, keep_empty=None):
        """
        Remove all groups without packages.
        """
        keep_empty = keep_empty or []
        for group in self.tree.xpath("/comps/group"):
            if not group.xpath("packagelist/packagereq"):
                group_id = group.xpath("id/text()")[0]
                found = False
                for pattern in keep_empty:
                    if fnmatch.fnmatch(group_id, pattern):
                        found = True
                        break
                if found:
                    continue
                group.getparent().remove(group)

    def remove_empty_categories(self):
        """
        Remove all categories without groups.
        """
        for category in self.tree.xpath("/comps/category"):
            if not category.xpath("grouplist/groupid"):
                category.getparent().remove(category)

    def remove_categories(self):
        """
        Remove all categories.
        """
        categories = self.tree.xpath("/comps/category")
        for i in categories:
            i.getparent().remove(i)

    def remove_langpacks(self):
        """
        Remove all langpacks.
        """
        langpacks = self.tree.xpath("/comps/langpacks")
        for i in langpacks:
            i.getparent().remove(i)

    def remove_translations(self):
        """
        Remove all translations.
        """
        for i in self.tree.xpath("//*[@xml:lang]"):
            i.getparent().remove(i)

    def filter_environment_groups(self):
        """
        Remove undefined groups from environments.
        """
        all_groups = self.tree.xpath("/comps/group/id/text()")
        for environment in self.tree.xpath("/comps/environment"):
            for group in environment.xpath("grouplist/groupid"):
                if group.text not in all_groups:
                    group.getparent().remove(group)

    def remove_empty_environments(self):
        """
        Remove all environments without groups.
        """
        for environment in self.tree.xpath("/comps/environment"):
            if not environment.xpath("grouplist/groupid"):
                environment.getparent().remove(environment)

    def remove_environments(self):
        """
        Remove all langpacks.
        """
        environments = self.tree.xpath("/comps/environment")
        for i in environments:
            i.getparent().remove(i)

    def write(self, file_obj):
        self.tree.write(file_obj, pretty_print=self.reindent, xml_declaration=True, encoding=self.encoding)
        file_obj.write("\n")

    def pprint(self):
        self.write(sys.stdout)

    def xml(self):
        io = StringIO()
        self.write(io)
        io.seek(0)
        return io.read()


def main():
    parser = argparse.ArgumentParser()
    parser.add_argument("--output", help="redirect output to a file")
    parser.add_argument("--arch", help="filter groups and packagews according to an arch")
    parser.add_argument("--arch-only-groups", default=False, action="store_true",
                        help="keep only arch groups, remove the rest")
    parser.add_argument("--arch-only-packages", default=False, action="store_true",
                        help="keep only arch packages, remove the rest")
    parser.add_argument("--arch-only-environments", default=False, action="store_true",
                        help="keep only arch environments, remove the rest")
    parser.add_argument("--remove-categories", default=False, action="store_true",
                        help="remove all categories")
    parser.add_argument("--remove-langpacks", default=False, action="store_true",
                        help="remove the langpacks section")
    parser.add_argument("--remove-translations", default=False, action="store_true",
                        help="remove all translations")
    parser.add_argument("--remove-environments", default=False, action="store_true",
                        help="remove all environment sections")
    parser.add_argument("--keep-empty-group", default=[], action="append", metavar="GROUPID",
                        help="keep groups even if they are empty")
    parser.add_argument("--no-cleanup", default=False, action="store_true",
                        help="don't remove empty groups and categories")
    parser.add_argument("--no-reindent", default=False, action="store_true",
                        help="don't re-indent the output")
    parser.add_argument("comps_file", metavar='COMPS_FILE')

    opts = parser.parse_args()

    if opts.arch is None:
        parser.error("please specify arch")

    file_obj = open(opts.comps_file, "r")
    f = CompsFilter(file_obj, reindent=not opts.no_reindent)
    f.filter_packages(opts.arch, opts.arch_only_packages)
    f.filter_groups(opts.arch, opts.arch_only_groups)
    f.filter_environments(opts.arch, opts.arch_only_environments)

    if not opts.no_cleanup:
        f.remove_empty_groups(keep_empty=opts.keep_empty_group)
        f.filter_category_groups()
        f.remove_empty_categories()
        f.filter_environment_groups()
        f.remove_empty_environments()

    if opts.remove_categories:
        f.remove_categories()

    if opts.remove_langpacks:
        f.remove_langpacks()

    if opts.remove_translations:
        f.remove_translations()

    if opts.remove_environments:
        f.remove_environments()

    if opts.output:
        out = open(opts.output, "w")
        f.write(out)
    else:
        f.pprint()


if __name__ == "__main__":
    main()