Skip to content
Snippets Groups Projects
Verified Commit fbfdf88e authored by Antoine R. Dumont's avatar Antoine R. Dumont
Browse files

nixguix: Add lister

Related to T3781
parent fa1205c4
No related branches found
No related tags found
No related merge requests found
swh.core[db,github] >= 2.8
swh.core[db,github] >= 2.15
swh.scheduler >= 0.8
......@@ -29,6 +29,28 @@ LISTERS = {
SUPPORTED_LISTERS = list(LISTERS)
TARBALL_EXTENSIONS = [
"crate",
"gem",
"jar",
"zip",
"tar",
"gz",
"tgz",
"tbz",
"bz2",
"bzip2",
"lzma",
"lz",
"txz",
"xz",
"z",
"Z",
"7z",
"zst",
]
"""Tarball recognition pattern"""
def get_lister(lister_name, db_url=None, **conf):
"""Instantiate a lister given its name.
......
# Copyright (C) 2019-2021 The Software Heritage developers
# Copyright (C) 2019-2022 The Software Heritage developers
# See the AUTHORS file at the top-level directory of this distribution
# License: GNU General Public License version 3, or any later version
# See top-level LICENSE file for more information
......@@ -15,6 +15,8 @@ from urllib.parse import urlparse
import requests
from swh.lister import TARBALL_EXTENSIONS
logger = logging.getLogger(__name__)
......@@ -186,21 +188,6 @@ def check_filename_is_archive(filename: str) -> bool:
return False
# to recognize existing naming pattern
EXTENSIONS = [
"zip",
"tar",
"gz",
"tgz",
"bz2",
"bzip2",
"lzma",
"lz",
"xz",
"Z",
"7z",
]
VERSION_KEYWORDS = [
"cygwin_me",
"w32",
......@@ -269,7 +256,7 @@ PATTERN = r"""
(?P<extension>(?:\.(?:{extensions}))+)
$
""".format(
extensions="|".join(EXTENSIONS),
extensions="|".join(TARBALL_EXTENSIONS),
vkeywords="|".join("%s[-]?" % k for k in VERSION_KEYWORDS),
)
......
# Copyright (C) 2022 the Software Heritage developers
# License: GNU General Public License version 3, or any later version
# See top-level LICENSE file for more information
def register():
from .lister import NixGuixLister
return {
"lister": NixGuixLister,
"task_modules": ["%s.tasks" % __name__],
}
# Copyright (C) 2020-2022 The Software Heritage developers
# See the AUTHORS file at the top-level directory of this distribution
# License: GNU General Public License version 3, or any later version
# See top-level LICENSE file for more information
"""NixGuix lister definition.
This lists artifacts out of manifest for Guix or Nixpkgs manifests.
Artifacts can be of types:
- upstream git repository (NixOS/nixpkgs, Guix)
- VCS repositories (svn, git, hg, ...)
- unique file
- unique tarball
"""
import base64
from dataclasses import dataclass
from enum import Enum
import logging
from pathlib import Path
import random
from typing import Any, Dict, Iterator, List, Optional, Tuple, Union
from urllib.parse import urlparse
import requests
from swh.core.github.utils import GitHubSession
from swh.core.tarball import MIMETYPE_TO_ARCHIVE_FORMAT
from swh.lister import TARBALL_EXTENSIONS
from swh.lister.pattern import CredentialsType, StatelessLister
from swh.scheduler.model import ListedOrigin
logger = logging.getLogger(__name__)
class ArtifactNatureUndetected(ValueError):
"""Raised when a remote artifact's nature (tarball, file) cannot be detected."""
pass
@dataclass
class OriginUpstream:
"""Upstream origin (e.g. NixOS/nixpkgs, Guix/Guix)."""
origin: str
"""Canonical url of the repository"""
version: int
"""Version of the repository (dismissed?)"""
revision: str
"""Revision of the repository (dismissed?)"""
@dataclass
class Artifact:
"""Metadata information on Remote Artifact with url (tarball or file)."""
origin: str
"""Canonical url retrieve the tarball artifact."""
visit_type: str
"""Either 'tar' or 'file' """
fallback_urls: List[str]
"""List of urls to retrieve tarball artifact if canonical url no longer works."""
checksums: Dict[str, str]
"""Integrity hash converted into a checksum dict."""
@dataclass
class VCS:
"""Metadata information on VCS."""
origin: str
"""Origin url of the vcs"""
ref: Optional[str]
"""Reference either a svn commit id, a git commit, ..."""
type: str
"""Type of (d)vcs, e.g. svn, git, hg, ..."""
class ArtifactType(Enum):
"""The possible artifact types listed out of the manifest."""
ARTIFACT = "artifact"
ORIGIN = "origin"
VCS = "vcs"
PageResult = Tuple[ArtifactType, Union[Artifact, VCS, OriginUpstream]]
VCS_SUPPORTED = ("git", "svn", "hg")
# Rough approximation of what we can find of mimetypes for tarballs "out there"
POSSIBLE_TARBALL_MIMETYPES = set(MIMETYPE_TO_ARCHIVE_FORMAT.keys())
def is_tarball(urls: List[str], request: Optional[Any] = None) -> Tuple[bool, str]:
"""Determine whether a list of files actually are tarballs or simple files.
When this cannot be answered simply out of the url, when request is provided, this
executes a HTTP `HEAD` query on the url to determine the information. If request is
not provided, this raises an ArtifactNatureUndetected exception.
Args:
urls: name of the remote files for which the extension needs to be checked.
Raises:
ArtifactNatureUndetected when the artifact's nature cannot be detected out
of its url
Returns: A tuple (bool, url). The boolean represents whether the url is an archive
or not. The second parameter is the actual url once the head request is issued
as a fallback of not finding out whether the urls are tarballs or not.
"""
def _is_tarball(url):
"""Determine out of an extension whether url is a tarball.
Raises:
IndexError in case no extension is available
"""
return Path(urlparse(url).path).suffixes[-1].lstrip(".") in TARBALL_EXTENSIONS
index = random.randrange(len(urls))
url = urls[index]
try:
is_tar = _is_tarball(url)
return is_tar, urls[0]
except IndexError:
if request is None:
raise ArtifactNatureUndetected(
"Cannot determine artifact type from url %s", url
)
logger.warning(
"Cannot detect extension for '%s'. Fallback to http head query",
url,
)
response = request.head(url)
if not response.ok or response.status_code == 404:
raise ArtifactNatureUndetected(
"Cannot determine artifact type from url %s", url
)
location = response.headers.get("Location")
if location: # It's not always present
logger.debug("Location: %s", location)
try:
# FIXME: location is also returned as it's considered the true origin,
# true enough?
return _is_tarball(location), location
except IndexError:
logger.warning(
"Still cannot detect extension through location '%s'...",
url,
)
content_type = response.headers.get("Content-Type")
if content_type:
logger.debug("Content-Type: %s", content_type)
if content_type == "application/json":
return False, urls[0]
return content_type in POSSIBLE_TARBALL_MIMETYPES, urls[0]
raise ArtifactNatureUndetected(
"Cannot determine artifact type from url %s", url
)
VCS_KEYS_MAPPING = {
"git": {
"ref": "git_ref",
"url": "git_url",
},
"svn": {
"ref": "svn_revision",
"url": "svn_url",
},
"hg": {
"ref": "hg_changeset",
"url": "hg_url",
},
}
class NixGuixLister(StatelessLister[PageResult]):
"""List Guix or Nix sources out of a public json manifest.
This lister can output:
- unique tarball (.tar.gz, .tbz2, ...)
- vcs repositories (e.g. git, hg, svn)
- unique file (.lisp, .py, ...)
Note that no `last_update` is available in either manifest.
For `url` types artifacts, this tries to determine the artifact's nature, tarball or
file. It first tries to compute out of the "url" extension. In case of no extension,
it fallbacks to query (HEAD) the url to retrieve the origin out of the `Location`
response header, and then checks the extension again.
"""
LISTER_NAME = "nixguix"
def __init__(
self,
scheduler,
url: str,
origin_upstream: str,
instance: Optional[str] = None,
credentials: Optional[CredentialsType] = None,
# canonicalize urls, can be turned off during docker runs
canonicalize: bool = True,
**kwargs: Any,
):
super().__init__(
scheduler=scheduler,
url=url.rstrip("/"),
instance=instance,
credentials=credentials,
)
# either full fqdn NixOS/nixpkgs or guix repository urls
# maybe add an assert on those specific urls?
self.origin_upstream = origin_upstream
self.session = requests.Session()
# for testing purposes, we may want to skip this step (e.g. docker run and rate
# limit)
self.github_session = (
GitHubSession(
credentials=self.credentials,
user_agent=str(self.session.headers["User-Agent"]),
)
if canonicalize
else None
)
def get_pages(self) -> Iterator[PageResult]:
"""Yield one page per "typed" origin referenced in manifest."""
# fetch and parse the manifest...
response = self.http_request(self.url)
# ... if any
raw_data = response.json()
version = raw_data["version"]
revision = raw_data["revision"]
yield ArtifactType.ORIGIN, OriginUpstream(
self.origin_upstream,
version,
revision,
)
# grep '"type"' guix-sources.json | sort | uniq
# "type": false <<<<<<<<< noise
# "type": "git",
# "type": "hg",
# "type": "no-origin", <<<<<<<<< noise
# "type": "svn",
# "type": "url",
# grep '"type"' nixpkgs-sources-unstable.json | sort | uniq
# "type": "url",
for artifact in raw_data["sources"]:
artifact_type = artifact["type"]
if artifact_type in VCS_SUPPORTED:
plain_url = artifact[VCS_KEYS_MAPPING[artifact_type]["url"]]
plain_ref = artifact[VCS_KEYS_MAPPING[artifact_type]["ref"]]
artifact_url = (
self.github_session.get_canonical_url(plain_url)
if self.github_session
else plain_url
)
if not artifact_url:
continue
yield ArtifactType.VCS, VCS(
origin=artifact_url, type=artifact_type, ref=plain_ref
)
elif artifact_type == "url":
# It's either a tarball or a file
urls = artifact.get("urls")
if not urls:
# Nothing to fetch
logger.warning("Skipping url '%s': empty artifact", artifact)
continue
assert urls is not None
# FIXME: T3294: Fix missing scheme in urls
origin, *fallback_urls = urls
integrity = artifact.get("integrity")
if integrity is None:
logger.warning("Skipping url '%s': missing integrity field", origin)
continue
try:
is_tar, origin = is_tarball(urls, self.session)
except ArtifactNatureUndetected:
logger.warning(
"Skipping url '%s': undetected remote artifact type", origin
)
continue
# Determine the content checksum stored in the integrity field and
# convert into a dict of checksums. This only parses the
# `hash-expression` (hash-<b64-encoded-checksum>) as defined in
# https://w3c.github.io/webappsec-subresource-integrity/#the-integrity-attribute
chksum_algo, chksum_b64 = integrity.split("-")
checksums: Dict[str, str] = {
chksum_algo: base64.decodebytes(chksum_b64.encode()).hex()
}
logger.debug("%s: %s", "dir" if is_tar else "cnt", origin)
yield ArtifactType.ARTIFACT, Artifact(
origin=origin,
fallback_urls=fallback_urls,
checksums=checksums,
visit_type="directory" if is_tar else "content",
)
else:
logger.warning(
"Skipping artifact '%s': unsupported type %s",
artifact,
artifact_type,
)
def vcs_to_listed_origin(self, artifact: VCS) -> Iterator[ListedOrigin]:
"""Given a vcs repository, yield a ListedOrigin."""
assert self.lister_obj.id is not None
# FIXME: What to do with the "ref" (e.g. git/hg/svn commit, ...)
yield ListedOrigin(
lister_id=self.lister_obj.id,
url=artifact.origin,
visit_type=artifact.type,
)
def origin_to_listed_origin(
self, origin_upstream: OriginUpstream
) -> Iterator[ListedOrigin]:
"""Given an upstream origin, yield a ListedOrigin."""
assert self.lister_obj.id is not None
yield ListedOrigin(
lister_id=self.lister_obj.id,
url=origin_upstream.origin,
visit_type="git", # both nixpkgs and guix are git origins so far
)
def artifact_to_listed_origin(self, artifact: Artifact) -> Iterator[ListedOrigin]:
"""Given an artifact (tarball, file), yield one ListedOrigin."""
assert self.lister_obj.id is not None
yield ListedOrigin(
lister_id=self.lister_obj.id,
url=artifact.origin,
visit_type=artifact.visit_type,
extra_loader_arguments={
"checksums": artifact.checksums,
"fallback_urls": artifact.fallback_urls,
},
)
def get_origins_from_page(
self, artifact_tuple: PageResult
) -> Iterator[ListedOrigin]:
"""Given an artifact tuple (type, artifact), yield a ListedOrigin."""
artifact_type, artifact = artifact_tuple
mapping_type_fn = getattr(self, f"{artifact_type.value}_to_listed_origin")
yield from mapping_type_fn(artifact)
{
"sources": [
{"type": "git", "git_url": "", "git_ref": ""},
{"type": false},
{"type": "no-origin"},
{"type": "url", "urls": []},
{
"type": "url",
"urls": ["https://crates.io/api/v1/0.1.5/no-extension-and-head-404-so-skipped"],
"integrity": "sha256-HW6jxFlbljY8E5Q0l9s0r0Rg+0dKlcQ/REatNBuMl4U="
},
{
"type": "url",
"urls": [ "https://example.org/another-file-no-integrity-so-skipped.txt" ]
}
],
"version":"1",
"revision":"ab59155c5a38dda7efaceb47c7528578fcf0def4"
}
{
"sources": [
{
"type": "url",
"urls": [ "https://github.com/owner-1/repository-1/revision-1.tgz" ],
"integrity": "sha256-3vm2Nt+O4zHf3Ovd/qsv1gKTEUwodX9FLxlrQdry0zs="
},
{
"type": "url",
"urls": [ "https://github.com/owner-3/repository-1/revision-1.tgz" ],
"integrity": "sha256-3vm2Nt+O4zHf3Ovd/qsv1gKTEUwodX9FLxlrQdry0zs="
},
{
"type": "url",
"urls": [ "https://example.com/file.txt" ],
"integrity": "sha256-Q0copBCnj1b8G1iZw1k0NuYasMcx6QctleltspAgXlM="
},
{
"type": "url",
"urls": [
"https://releases.wildfiregames.com/0ad-0.0.25b-alpha-unix-build.tar.xz"
],
"integrity": "sha256-1w3NdfRzp9XIFDLD2SYJJr+Nnf9c1UF5YWlJfRxSLt0="
},
{
"type": "url",
"urls": [
"http://downloads.sourceforge.net/project/nmon/lmon16n.c",
"http://ufpr.dl.sourceforge.net/project/nmon/lmon16n.c",
"http://netassist.dl.sourceforge.net/project/nmon/lmon16n.c"
],
"integrity": "sha256-wAEswtkl3ulAw3zq4perrGS6Wlww5XXnQYsEAoYT9fI="
},
{
"type": "git",
"git_url": "https://example.org/pali/0xffff",
"git_ref": "0.9"
},
{
"type": "hg",
"hg_url": "https://example.org/vityok/cl-string-match",
"hg_changeset": "5048480a61243e6f1b02884012c8f25cdbee6d97"
},
{
"type": "svn",
"svn_url": "https://code.call-cc.org/svn/chicken-eggs/release/5/iset/tags/2.2",
"svn_revision": 39057
}
],
"version": "1",
"revision": "cc4e04c26672dd74e5fd0fecb78b435fb55368f7"
}
# Copyright (C) 2022 The Software Heritage developers
# See the AUTHORS file at the top-level directory of this distribution
# License: GNU General Public License version 3, or any later version
# See top-level LICENSE file for more information
from collections import defaultdict
import json
import logging
from pathlib import Path
from typing import Dict, List
import pytest
import requests
from swh.lister import TARBALL_EXTENSIONS
from swh.lister.nixguix.lister import (
POSSIBLE_TARBALL_MIMETYPES,
ArtifactNatureUndetected,
NixGuixLister,
is_tarball,
)
from swh.lister.pattern import ListerStats
logger = logging.getLogger(__name__)
def page_response(datadir, instance: str) -> List[Dict]:
"""Return list of repositories (out of test dataset)"""
datapath = Path(datadir, f"{instance}-swh_sources.json")
return json.loads(datapath.read_text()) if datapath.exists else []
@pytest.mark.parametrize(
"urls",
[[f"one.{ext}", f"two.{ext}"] for ext in TARBALL_EXTENSIONS]
+ [[f"one.{ext}?foo=bar"] for ext in TARBALL_EXTENSIONS],
)
def test_is_tarball_simple(urls):
"""Simple check on tarball should discriminate betwenn tarball and file"""
is_tar, origin = is_tarball(urls)
assert is_tar is True
assert origin == urls[0]
@pytest.mark.parametrize(
"urls",
[
["abc.lisp"],
["one.abc", "two.bcd"],
["abc.c", "other.c"],
["one.scm?foo=bar", "two.scm?foo=bar"],
["config.nix", "flakes.nix"],
],
)
def test_is_tarball_simple_not_tarball(urls):
"""Simple check on tarball should discriminate betwenn tarball and file"""
is_tar, origin = is_tarball(urls)
assert is_tar is False
assert origin == urls[0]
def test_is_tarball_complex_with_no_result(requests_mock):
"""Complex tarball detection without proper information should fail."""
# No extension, this won't detect immediately the nature of the url
url = "https://example.org/crates/package/download"
urls = [url]
with pytest.raises(ArtifactNatureUndetected):
is_tarball(url) # no request parameter, this cannot fallback, raises
with pytest.raises(ArtifactNatureUndetected):
requests_mock.head(
url,
status_code=404, # not found so cannot detect anything
)
is_tarball(urls, requests)
with pytest.raises(ArtifactNatureUndetected):
requests_mock.head(
url, headers={}
) # response ok without headers, cannot detect anything
is_tarball(urls, requests)
with pytest.raises(ArtifactNatureUndetected):
fallback_url = "https://example.org/mirror/crates/package/download"
requests_mock.head(
url, headers={"location": fallback_url} # still no extension, cannot detect
)
is_tarball(urls, requests)
@pytest.mark.parametrize(
"fallback_url, expected_result",
[
("https://example.org/mirror/crates/package/download.tar.gz", True),
("https://example.org/mirror/package/download.lisp", False),
],
)
def test_is_tarball_complex_with_location_result(
requests_mock, fallback_url, expected_result
):
"""Complex tarball detection with information should detect artifact nature"""
# No extension, this won't detect immediately the nature of the url
url = "https://example.org/crates/package/download"
urls = [url]
# One scenario where the url renders a location with a proper extension
requests_mock.head(url, headers={"location": fallback_url})
is_tar, origin = is_tarball(urls, requests)
assert is_tar == expected_result
if is_tar:
assert origin == fallback_url
@pytest.mark.parametrize(
"content_type, expected_result",
[("application/json", False), ("application/something", False)]
+ [(ext, True) for ext in POSSIBLE_TARBALL_MIMETYPES],
)
def test_is_tarball_complex_with_content_type_result(
requests_mock, content_type, expected_result
):
"""Complex tarball detection with information should detect artifact nature"""
# No extension, this won't detect immediately the nature of the url
url = "https://example.org/crates/package/download"
urls = [url]
# One scenario where the url renders a location with a proper extension
requests_mock.head(url, headers={"Content-Type": content_type})
is_tar, origin = is_tarball(urls, requests)
assert is_tar == expected_result
if is_tar:
assert origin == url
def test_lister_nixguix(datadir, swh_scheduler, requests_mock):
"""NixGuixLister should list all origins per visit type"""
url = "https://nix-community.github.io/nixpkgs-swh/sources-unstable.json"
origin_upstream = "https://github.com/NixOS/nixpkgs"
lister = NixGuixLister(swh_scheduler, url=url, origin_upstream=origin_upstream)
response = page_response(datadir, "nixpkgs")
requests_mock.get(
url,
[{"json": response}],
)
expected_visit_types = defaultdict(int)
# origin upstream is added as origin
expected_nb_origins = 1
expected_visit_types["git"] += 1
for artifact in response["sources"]:
# Each artifact is considered an origin (even "url" artifacts with mirror urls)
expected_nb_origins += 1
artifact_type = artifact["type"]
if artifact_type in [
"git",
"svn",
"hg",
]:
expected_visit_types[artifact_type] += 1
elif artifact_type == "url":
url = artifact["urls"][0]
if url.endswith(".c") or url.endswith(".txt"):
expected_visit_types["content"] += 1
else:
expected_visit_types["directory"] += 1
assert set(expected_visit_types.keys()) == {
"content",
"git",
"svn",
"hg",
"directory",
}
listed_result = lister.run()
# 1 page read is 1 origin
nb_pages = expected_nb_origins
assert listed_result == ListerStats(pages=nb_pages, origins=expected_nb_origins)
scheduler_origins = lister.scheduler.get_listed_origins(
lister.lister_obj.id
).results
assert len(scheduler_origins) == expected_nb_origins
mapping_visit_types = defaultdict(int)
for listed_origin in scheduler_origins:
assert listed_origin.visit_type in expected_visit_types
# no last update is listed on those manifests
assert listed_origin.last_update is None
mapping_visit_types[listed_origin.visit_type] += 1
assert dict(mapping_visit_types) == expected_visit_types
def test_lister_nixguix_mostly_noop(datadir, swh_scheduler, requests_mock):
"""NixGuixLister should ignore unsupported or incomplete origins"""
url = "https://guix.gnu.org/sources.json"
origin_upstream = "https://git.savannah.gnu.org/git/guix.git"
lister = NixGuixLister(swh_scheduler, url=url, origin_upstream=origin_upstream)
response = page_response(datadir, "guix")
requests_mock.get(
url,
[{"json": response}],
)
# Amongst artifacts, this url does not allow to determine its nature (tarball, file)
# It's ending up doing a http head query which ends up being 404, so it's skipped.
requests_mock.head(
"https://crates.io/api/v1/0.1.5/no-extension-and-head-404-so-skipped",
status_code=404,
)
listed_result = lister.run()
# only the origin upstream is listed, every other entries are unsupported or incomplete
assert listed_result == ListerStats(pages=1, origins=1)
scheduler_origins = lister.scheduler.get_listed_origins(
lister.lister_obj.id
).results
assert len(scheduler_origins) == 1
assert scheduler_origins[0].visit_type == "git"
def test_lister_nixguix_fail(datadir, swh_scheduler, requests_mock):
url = "https://nix-community.github.io/nixpkgs-swh/sources-unstable.json"
origin_upstream = "https://github.com/NixOS/nixpkgs"
lister = NixGuixLister(swh_scheduler, url=url, origin_upstream=origin_upstream)
requests_mock.get(
url,
status_code=404,
)
with pytest.raises(requests.HTTPError): # listing cannot continues so stop
lister.run()
scheduler_origins = swh_scheduler.get_listed_origins(lister.lister_obj.id).results
assert len(scheduler_origins) == 0
0% Loading or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment