[entropy.server] add drained and merged repositories dependencies testing
It happens that during our workflow, we bump packages into overlay repositories like sabayon-limbo and sabayon-hell. However, potential dependencies issues may pop up only when repositories are merged (their content drained) into others. This features assumes that repositories are listed in order of priority, where the first one is the base repo and others are layers on top of the first one.
This commit is contained in:
Fabio Erculiani
@@ -3942,6 +3942,133 @@ class Server(Client):
|
||||
|
||||
return deps_not_satisfied
|
||||
|
||||
def drained_dependencies_test(self, repository_ids):
|
||||
"""
|
||||
Test repositories against missing dependencies taking into
|
||||
consideration the possibility of repositories being drained
|
||||
and their content merged into the repositories before them.
|
||||
So, if you have 3 repositories, say: main, limbo, hell, this
|
||||
method will test:
|
||||
- hell drained and merged into main and limbo
|
||||
- limbo and hell drained and merged into main
|
||||
|
||||
@param repository_ids: ordered list of repository identifiers
|
||||
to test, like the output of Server.repositories()
|
||||
@type repository_ids: list
|
||||
@return: list (set) of unsatisfied dependencies
|
||||
@rtype: set
|
||||
"""
|
||||
repos = reversed(repository_ids)
|
||||
missing_dependencies = set()
|
||||
cases = []
|
||||
for case in reversed(range(len(repos) - 1)):
|
||||
cases.append((repos[0:case+1], repos[case+1:]))
|
||||
|
||||
spm = self.Spm()
|
||||
for merged, drained in cases:
|
||||
|
||||
mytxt = "%s..." % (
|
||||
teal(_("Calculating dependencies")),)
|
||||
self.output(
|
||||
mytxt,
|
||||
importance = 2,
|
||||
level = "info",
|
||||
header = brown(" @@ ")
|
||||
)
|
||||
mytxt = "%s: %s" % (
|
||||
blue(_("merged repositories")),
|
||||
", ".join([darkgreen(x) for x in merged]))
|
||||
self.output(
|
||||
mytxt,
|
||||
header = " "
|
||||
)
|
||||
mytxt = "%s: %s" % (
|
||||
blue(_("drained repositories")),
|
||||
", ".join([purple(x) for x in merged]))
|
||||
self.output(
|
||||
mytxt,
|
||||
header = " "
|
||||
)
|
||||
|
||||
deps_cache = set() # used for memoization
|
||||
for repository_id in merged:
|
||||
|
||||
repo = self.open_server_repository(
|
||||
repository_id, read_only = True,
|
||||
no_upload = True, do_treeupdates = False)
|
||||
|
||||
package_ids = repo.listAllPackageIds()
|
||||
total = len(package_ids)
|
||||
for count, package_id in enumerate(package_ids, 1):
|
||||
|
||||
mytxt = "%s: %s" % (
|
||||
darkgreen(_("checking repository")),
|
||||
purple(repository_id))
|
||||
self.output(mytxt, header = blue(" :: "),
|
||||
count = (count, total),
|
||||
back = True)
|
||||
|
||||
xdeps = repo.retrieveDependencies(package_id)
|
||||
xdeps = [x for x in xdeps if x not in deps_cache]
|
||||
deps_cache.update(xdeps)
|
||||
|
||||
for dependency in xdeps:
|
||||
|
||||
# need to check inside merged.
|
||||
pkg_id, pkg_repo = self.atom_match(
|
||||
dependency, match_repo=merged)
|
||||
if pkg_id == -1:
|
||||
# potentially broken candidate, but this is
|
||||
# detected by typical dep testing.
|
||||
continue
|
||||
|
||||
pkg_keyslot = self.open_repository(
|
||||
pkg_repo).retrieveKeySlotAggregated(pkg_id)
|
||||
# match keyslot inside drained, if there is something
|
||||
# we check with dependency.
|
||||
drained_pkg_id, drained_repo = self.atom_match(
|
||||
pkg_keyslot, match_repo=drained)
|
||||
if drained_pkg_id == -1:
|
||||
# nothing, all good
|
||||
continue
|
||||
|
||||
# then match with dependency.
|
||||
drained_pkg_id, drained_repo = self.atom_match(
|
||||
dependency, match_repo=drained)
|
||||
if drained_pkg_id != -1:
|
||||
# all good then, we are still able to match
|
||||
# a dependency there.
|
||||
continue
|
||||
|
||||
# in this case, we need to check if the top level
|
||||
# package match (package_id, repository_id) is going
|
||||
# away. If it does, then there is no need to worry.
|
||||
package_keyslot = repo.retrieveKeySlotAggregated(
|
||||
package_id)
|
||||
pkg_id, repo_id = self.atom_match(
|
||||
package_keyslot, match_repo=drained)
|
||||
if pkg_id != -1:
|
||||
continue
|
||||
|
||||
# check if the package is still installed on the system
|
||||
package_atom = spm.convert_from_entropy_package_name(
|
||||
repo.retrieveAtom(package_id))
|
||||
inst_matches = spm.match_installed_package(
|
||||
package_atom)
|
||||
if not inst_matches:
|
||||
continue
|
||||
|
||||
atom = repo.retrieveAtom(package_id)
|
||||
mytxt = "[%s] %s, %s: %s" % (
|
||||
brown(repository_id),
|
||||
teal(atom),
|
||||
darkgreen(_("missing dependency")),
|
||||
blue(dependency))
|
||||
self.output(mytxt, header = "- ")
|
||||
missing_dependencies.add(dependency)
|
||||
|
||||
return missing_dependencies
|
||||
|
||||
def extended_dependencies_test(self, repository_ids):
|
||||
"""
|
||||
Test repository against missing dependencies.
|
||||
@@ -3965,6 +4092,12 @@ class Server(Client):
|
||||
# self-contained
|
||||
unsatisfied_deps |= self.dependencies_test(base_repository_id,
|
||||
match_repo = [base_repository_id])
|
||||
|
||||
# test draining and merging as well, since we don't want
|
||||
# to get surprises when stuff is moved.
|
||||
unsatisfied_deps |= self.drained_dependencies_test(
|
||||
self.repositories())
|
||||
|
||||
return unsatisfied_deps
|
||||
|
||||
def dependencies_test(self, repository_id, match_repo = None):
|
||||
|
||||
Reference in New Issue
Block a user