lib/_emerge/MetadataRegen.py

# Copyright 1999-2020 Gentoo Authors
# Distributed under the terms of the GNU General Public License v2

from _emerge.EbuildMetadataPhase import EbuildMetadataPhase

import portage
from portage import os
from portage.cache.cache_errors import CacheError
from portage.dep import _repo_separator
from portage.util._async.AsyncScheduler import AsyncScheduler


class MetadataRegen(AsyncScheduler):
    def __init__(self, portdb, cp_iter=None, consumer=None, write_auxdb=True, **kwargs):
        AsyncScheduler.__init__(self, **kwargs)
        self._portdb = portdb
        self._write_auxdb = write_auxdb
        self._global_cleanse = False
        if cp_iter is None:
            cp_iter = self._iter_every_cp()
            # We can globally cleanse stale cache only if we
            # iterate over every single cp.
            self._global_cleanse = True
        self._cp_iter = cp_iter
        self._consumer = consumer

        self._valid_pkgs = set()
        self._cp_set = set()
        self._process_iter = self._iter_metadata_processes()
        self._running_tasks = set()

    def _next_task(self):
        return next(self._process_iter)

    def _iter_every_cp(self):
        # List categories individually, in order to start yielding quicker,
        # and in order to reduce latency in case of a signal interrupt.
        cp_all = self._portdb.cp_all
        for category in sorted(self._portdb.categories):
            for cp in cp_all(categories=(category,)):
                yield cp

    def _iter_metadata_processes(self):
        portdb = self._portdb
        valid_pkgs = self._valid_pkgs
        cp_set = self._cp_set
        consumer = self._consumer

        portage.writemsg_stdout("Regenerating cache entries...\n")
        for cp in self._cp_iter:
            if self._terminated.is_set():
                break
            cp_set.add(cp)
            portage.writemsg_stdout("Processing %s\n" % cp)
            # We iterate over portdb.porttrees, since it's common to
            # tweak this attribute in order to adjust repo selection.
            for mytree in portdb.porttrees:
                repo = portdb.repositories.get_repo_for_location(mytree)
                cpv_list = portdb.cp_list(cp, mytree=[repo.location])
                for cpv in cpv_list:
                    if self._terminated.is_set():
                        break
                    valid_pkgs.add(cpv)
                    ebuild_path, repo_path = portdb.findname2(cpv, myrepo=repo.name)
                    if ebuild_path is None:
                        raise AssertionError(
                            "ebuild not found for '%s%s%s'"
                            % (cpv, _repo_separator, repo.name)
                        )
                    metadata, ebuild_hash = portdb._pull_valid_cache(
                        cpv, ebuild_path, repo_path
                    )
                    if metadata is not None:
                        if consumer is not None:
                            consumer(cpv, repo_path, metadata, ebuild_hash, True)
                        continue

                    yield EbuildMetadataPhase(
                        cpv=cpv,
                        ebuild_hash=ebuild_hash,
                        portdb=portdb,
                        repo_path=repo_path,
                        settings=portdb.doebuild_settings,
                        write_auxdb=self._write_auxdb,
                    )

    def _cleanup(self):
        super(MetadataRegen, self)._cleanup()

        portdb = self._portdb
        dead_nodes = {}

        if self._terminated.is_set():
            portdb.flush_cache()
            return

        if self._global_cleanse:
            for mytree in portdb.porttrees:
                try:
                    dead_nodes[mytree] = set(portdb.auxdb[mytree])
                except CacheError as e:
                    portage.writemsg(
                        "Error listing cache entries for "
                        + "'%s': %s, continuing...\n" % (mytree, e),
                        noiselevel=-1,
                    )
                    del e
                    dead_nodes = None
                    break
        else:
            cp_set = self._cp_set
            cpv_getkey = portage.cpv_getkey
            for mytree in portdb.porttrees:
                try:
                    dead_nodes[mytree] = set(
                        cpv for cpv in portdb.auxdb[mytree] if cpv_getkey(cpv) in cp_set
                    )
                except CacheError as e:
                    portage.writemsg(
                        "Error listing cache entries for "
                        + "'%s': %s, continuing...\n" % (mytree, e),
                        noiselevel=-1,
                    )
                    del e
                    dead_nodes = None
                    break

        if dead_nodes:
            for y in self._valid_pkgs:
                for mytree in portdb.porttrees:
                    if portdb.findname2(y, mytree=mytree)[0]:
                        dead_nodes[mytree].discard(y)

            for mytree, nodes in dead_nodes.items():
                auxdb = portdb.auxdb[mytree]
                for y in nodes:
                    try:
                        del auxdb[y]
                    except (KeyError, CacheError):
                        pass

        portdb.flush_cache()

    def _task_exit(self, metadata_process):

        if metadata_process.returncode != os.EX_OK:
            self._valid_pkgs.discard(metadata_process.cpv)
            if not self._terminated_tasks:
                portage.writemsg(
                    "Error processing %s, continuing...\n" % (metadata_process.cpv,),
                    noiselevel=-1,
                )

        if self._consumer is not None:
            # On failure, still notify the consumer (in this case the metadata
            # argument is None).
            self._consumer(
                metadata_process.cpv,
                metadata_process.repo_path,
                metadata_process.metadata,
                metadata_process.ebuild_hash,
                metadata_process.eapi_supported,
            )

        AsyncScheduler._task_exit(self, metadata_process)