Merge pull request #1760 from dmach/osc-build-project-from-_ObsPrj

Extend 'osc build' and 'osc fork' to use project.build file from Gitea, 'osc fork' also follows the devel project now

osc/commands/fork.py

@@ -1,3 +1,4 @@
+import re
 import sys
 import urllib.parse
 
@@ -44,27 +45,58 @@ class ForkCommand(osc.commandline.OscCommand):
 
         self.add_argument_new_repo_name()
 
+        self.add_argument(
+            "--no-devel-project",
+            action="store_true",
+            help="Fork the specified package instead the package from the devel project (which is the place where the package is developed)",
+        )
+
     def run(self, args):
         from osc import conf as osc_conf
         from osc import gitea_api
         from osc import obs_api
+        from osc.git_scm import GitStore
         from osc.output import tty
 
-        is_package = args.package is not None
+        # make a copy of project, package; if we change them, the original values remain in args
+        project = args.project
+        package = args.package
+
+        is_package = package is not None
+        use_devel_project = False
 
         if not is_package and args.target_package:
             self.parser.error("The '--target-package' option requires the 'package' argument to be set")
 
+        if not is_package and args.no_devel_project:
+            self.parser.error("The '--no-devel-project' option can be used only when forking a package")
+
         if is_package:
             # get the package meta from the OBS API first
-            package = obs_api.Package.from_api(args.apiurl, args.project, args.package)
+            pkg = obs_api.Package.from_api(args.apiurl, project, package)
-            if not package.scmsync:
+
-                raise RuntimeError(
+            if not args.no_devel_project:
-                    "Forking is possible only with packages managed in Git (the <scmsync> element must be set in the package meta)"
+                # devel project is not set in package meta as usual but we parse it from "OBS:RejectBranch" attribute
-                )
+                attributes = obs_api.Attributes.from_api(args.apiurl, project, package, attr="OBS:RejectBranch").attribute_list
+                if attributes:
+                    attribute = attributes[0].value
+                    # the pattern starts with a non-greedy match so we capture the first url
+                    match = re.match(r".*?(https://[^ ]+).*", attribute)
+                    if match:
+                        devel_project_url = match.group(1)
+                        build_project = GitStore.get_build_project(devel_project_url)
+                        # override the package we're cloning with the one from the devel project
+                        use_devel_project = True
+                        project = build_project
+                        pkg = obs_api.Package.from_api(args.apiurl, project, package)
+
+            if not pkg.scmsync:
+                print(f"{tty.colorize('ERROR', 'red,bold')}: Forking is possible only with packages managed in Git (the <scmsync> element must be set in the package meta)")
+                sys.exit(1)
+
         else:
             # get the project meta from the OBS API first
-            project = obs_api.Project.from_api(args.apiurl, args.project)
+            project = obs_api.Project.from_api(args.apiurl, project)
             if not project.scmsync:
                 raise RuntimeError(
                     "Forking is possible only with projects managed in Git (the <scmsync> element must be set in the project meta)"
@@ -72,11 +104,16 @@ class ForkCommand(osc.commandline.OscCommand):
 
         # parse gitea url, owner, repo and branch from the scmsync url
         if is_package:
-            parsed_scmsync_url = urllib.parse.urlparse(package.scmsync, scheme="https")
+            parsed_scmsync_url = urllib.parse.urlparse(pkg.scmsync, scheme="https")
         else:
             parsed_scmsync_url = urllib.parse.urlparse(project.scmsync, scheme="https")
         url = urllib.parse.urlunparse((parsed_scmsync_url.scheme, parsed_scmsync_url.netloc, "", "", "", ""))
         owner, repo = parsed_scmsync_url.path.strip("/").split("/")
+
+        # remove trailing ".git" from repo
+        if repo.endswith(".git"):
+            repo = repo[:-4]
+
         # temporary hack to allow people using fork atm at all, when packages
         # are managed via git project.
         # fallback always to default branch for now, but we actually need to 
@@ -125,15 +162,17 @@ class ForkCommand(osc.commandline.OscCommand):
 
         print()
         if is_package:
-            print(f"Forking OBS package {args.project}/{args.package} ...")
+            print(f"Forking OBS package {project}/{package} ...")
+            if use_devel_project:
+                print(f" * {tty.colorize('NOTE', 'bold')}: Forking from the devel project instead of the specified {args.project}/{args.package}")
         else:
-            print(f"Forking OBS project {args.project} ...")
+            print(f"Forking OBS project {project} ...")
         print(f" * OBS apiurl: {args.apiurl}")
         # we use a single API endpoint for forking both projects and packages (project requires setting package to "_project")
         status = obs_api.Package.cmd_fork(
             args.apiurl,
-            args.project,
+            project,
-            args.package if is_package else "_project",
+            package if is_package else "_project",
             scmsync=fork_scmsync,
             target_project=args.target_project,
             target_package=args.target_package if is_package else None,

osc/git_scm/store.py

@@ -28,6 +28,66 @@ class GitStore:
             return False
         return store.is_package
 
+    @staticmethod
+    def get_build_project(git_repo_url: str):
+        """
+        Get the project we use for building from _ObsPrj git repo.
+        The _ObsPrj is located under the same owner as the repo with the package.
+        They share the same branch.
+        """
+        import tempfile
+
+        from osc import gitea_api
+
+        # parse the git_repo_url (which usually corresponds with the url of the 'origin' remote of the local git repo)
+        scheme, netloc, path, params, query, fragment = gitea_api.Git.urlparse(git_repo_url)
+
+        # scheme + host
+        gitea_host = urllib.parse.urlunparse((scheme, netloc, "", None, None, None))
+
+        # OBS and Gitea usernames are identical
+        # XXX: we're using the configured apiurl; it would be great to have a mapping from -G/--gitea-login to -A/--apiurl so we don't have to provide -A on the command-line
+        apiurl = osc_conf.config["apiurl"]
+        gitea_user = osc_conf.get_apiurl_usr(apiurl)
+
+        # remove trailing ".git" from path
+        if path.endswith(".git"):
+            path = path[:-4]
+
+        gitea_owner, gitea_repo = path.strip("/").split("/")[-2:]
+
+        # replace gitea_repo with _ObsPrj
+        gitea_repo = "_ObsPrj"
+
+        # XXX: we assume that the _ObsPrj project has the same branch as the package
+        gitea_branch = fragment
+
+        gitea_conf = gitea_api.Config()
+        try:
+            gitea_login = gitea_conf.get_login_by_url_user(url=gitea_host, user=gitea_user)
+        except gitea_api.Login.DoesNotExist:
+            # matching login entry doesn't exist in git-obs config
+            return None
+
+        gitea_conn = gitea_api.Connection(gitea_login)
+
+        with tempfile.TemporaryDirectory(prefix="osc_devel_project_git") as tmp_dir:
+            try:
+                gitea_api.Repo.clone(gitea_conn, gitea_owner, gitea_repo, branch=gitea_branch, quiet=True, directory=tmp_dir)
+                project_build_path = os.path.join(tmp_dir, "project.build")
+                with open(project_build_path, "r", encoding="utf-8") as f:
+                    project_build = f.readline().strip()
+                    return project_build
+            except gitea_api.GiteaException:
+                # "_ObsPrj" repo doesn't exist
+                return None
+            except subprocess.CalledProcessError:
+                # branch doesn't exist
+                return None
+            except FileNotFoundError:
+                # "project.build" file doesn't exist
+                return None
+
     @property
     def git_project_dir(self):
         if not hasattr(self, "_git_project_dir"):
@@ -135,6 +195,10 @@ class GitStore:
             if self.is_package and self.project_obs_scm_store:
                 # read project from parent directory that contains a project with .osc metadata
                 self._project = self.project_obs_scm_store.project
+            if not self._project:
+                # read project from Gitea (identical owner, repo: _ObsPrj, file: project.build)
+                origin = self._run_git(["remote", "get-url", "origin"])
+                self._project = self.get_build_project(origin)
             if not self._project:
                 # HACK: assume openSUSE:Factory project if project metadata is missing
                 self._project = "openSUSE:Factory"

osc/gitea_api/conf.py

@@ -121,9 +121,31 @@ class Config:
         """
         Return ``Login`` object for the given ``url`` and ``user``.
         """
+        from .git import Git
+
+        # exact match
         for login in self.list_logins():
             if (login.url, login.user) == (url, user):
                 return login
+
+        def url_to_hostname(value):
+            netloc = Git.urlparse(value).netloc
+
+            # remove user from hostname
+            if "@" in netloc:
+                netloc = netloc.split("@")[-1]
+
+            # remove port from hostname
+            if ":" in netloc:
+                netloc = netloc.split(":")[0]
+
+            return netloc
+
+        # match only hostname (netloc without 'user@' and ':port') + user
+        for login in self.list_logins():
+            if (url_to_hostname(login.url), login.user) == (url_to_hostname(url), user):
+                return login
+
         raise Login.DoesNotExist(url=url, user=user)
 
     def add_login(self, login: Login):

osc/gitea_api/git.py

@@ -9,6 +9,38 @@ from typing import Tuple
 
 
 class Git:
+    @staticmethod
+    def urlparse(url) -> urllib.parse.ParseResult:
+        """
+        Parse git url.
+
+        Supported formats:
+            - https://example.com/owner/repo.git
+            - https://example.com:1234/owner/repo.git
+            - example.com/owner/repo.git
+            - user@example.com:owner/repo.git
+            - user@example.com:1234:owner/repo.git"
+        """
+        # try ssh clone url first
+        pattern = r"(?P<netloc>[^@:]+@[^@:]+(:[0-9]+)?):(?P<path>.+)"
+        match = re.match(pattern, url)
+        if match:
+            scheme = ""
+            netloc = match.groupdict()["netloc"]
+            path = match.groupdict()["path"]
+            params = ''
+            query = ''
+            fragment = ''
+            result = urllib.parse.ParseResult(scheme, netloc, path, params, query, fragment)
+            return result
+
+        result = urllib.parse.urlparse(url)
+        if not result.netloc:
+            # empty netloc is most likely an error, prepend and then discard scheme to trick urlparse()
+            result = urllib.parse.urlparse("https://" + url)
+            result = urllib.parse.ParseResult("", *list(result)[1:])
+        return result
+
     def __init__(self, workdir):
         self.abspath = os.path.abspath(workdir)
 

osc/gitea_api/repo.py

@@ -44,6 +44,8 @@ class Repo:
         owner: str,
         repo: str,
         *,
+        branch: Optional[str] = None,
+        quiet: bool = False,
         directory: Optional[str] = None,
         cwd: Optional[str] = None,
         anonymous: bool = False,
@@ -112,6 +114,13 @@ class Repo:
 
         # clone
         cmd = ["git", "clone", clone_url, directory]
+
+        if branch:
+            cmd += ["--branch", branch]
+
+        if quiet:
+            cmd += ["--quiet"]
+
         subprocess.run(cmd, cwd=cwd, env=env, check=True)
 
         # setup remotes

osc/obs_api/__init__.py

@@ -1,3 +1,4 @@
+from .attributes import Attributes
 from .keyinfo import Keyinfo
 from .package import Package
 from .package_sources import PackageSources

osc/obs_api/attributes.py

@@ -0,0 +1,46 @@
+from typing import Any
+
+from ..util.models import *  # pylint: disable=wildcard-import,unused-wildcard-import
+
+
+class Attribute(XmlModel):
+    name: str = Field(
+        xml_attribute=True,
+    )
+    namespace: str = Field(
+        xml_attribute=True,
+    )
+    value: str = Field(
+        xml_set_text=True,
+        xml_wrapped=True,
+    )
+
+
+class Attributes(XmlModel):
+    XML_TAG = "attributes"
+
+    attribute_list: List[Attribute] = Field(
+        xml_name="attribute",
+    )
+
+    @classmethod
+    def from_api(
+        cls, apiurl: str, project: str, package: Optional[str] = None, *, attr: Optional[str] = None
+    ) -> "Attributes":
+        import urllib.error
+        from .. import oscerr
+        from ..connection import http_request
+        from ..core import makeurl
+
+        if package:
+            url_path = ["source", project, package, "_attribute"]
+        else:
+            url_path = ["source", project, "_attribute"]
+
+        if attr:
+            url_path.append(attr)
+
+        url_query: Dict[str, Any] = {}
+        url = makeurl(apiurl, url_path, url_query)
+        response = http_request("GET", url)
+        return cls.from_file(response)

tests/test_gitea_api_git.py

@@ -0,0 +1,36 @@
+import unittest
+
+from osc.gitea_api import Git
+
+
+class TestGiteaApiGit(unittest.TestCase):
+    def test_urlparse(self):
+        # https url without port
+        url = "https://example.com/owner/repo.git"
+        result = Git.urlparse(url)
+        self.assertEqual(list(result), ['https', 'example.com', '/owner/repo.git', '', '', ''])
+
+        # https url with port
+        url = "https://example.com:1234/owner/repo.git"
+        result = Git.urlparse(url)
+        self.assertEqual(list(result), ['https', 'example.com:1234', '/owner/repo.git', '', '', ''])
+
+        # url without scheme
+        # urllib.parse.urlparse() would normally return ['', '', 'example.com/owner/repo.git', '', '', '']
+        url = "example.com/owner/repo.git"
+        result = Git.urlparse(url)
+        self.assertEqual(list(result), ['', 'example.com', '/owner/repo.git', '', '', ''])
+
+        # ssh url
+        url = "user@example.com:owner/repo.git"
+        result = Git.urlparse(url)
+        self.assertEqual(list(result), ['', 'user@example.com', 'owner/repo.git', '', '', ''])
+
+        # ssh url with port
+        url = "user@example.com:1234:owner/repo.git"
+        result = Git.urlparse(url)
+        self.assertEqual(list(result), ['', 'user@example.com:1234', 'owner/repo.git', '', '', ''])
+
+
+if __name__ == "__main__":
+    unittest.main()