Only stop importing when it isn't a jengelh repository

Reviewed-on: importers/git-importer#32
Reviewed-by: Dirk Mueller <dirkmueller@noreply@src.opensuse.org>

git-importer.py

@@ -42,8 +42,8 @@ PROJECTS = [
 ]
 
 
-def export_package(package, repodir, cachedir, gc):
-    exporter = GitExporter(URL_OBS, "openSUSE:Factory", package, repodir, cachedir)
+def export_package(project, package, repodir, cachedir, gc):
+    exporter = GitExporter(URL_OBS, project, package, repodir, cachedir)
     exporter.set_gc_interval(gc)
     exporter.export_as_git()
 
@@ -51,6 +51,12 @@ def export_package(package, repodir, cachedir, gc):
 def main():
     parser = argparse.ArgumentParser(description="OBS history importer into git")
     parser.add_argument("packages", help="OBS package names", nargs="*")
+    parser.add_argument(
+        "-p",
+        "--project",
+        default="openSUSE:Factory",
+        help="Project to import/export, default is openSUSE:Factory",
+    )
     parser.add_argument(
         "-r",
         "--repodir",
@@ -110,10 +116,13 @@ def main():
     if not args.cachedir:
         args.cachedir = pathlib.Path("~/.cache/git-import/").expanduser()
 
-    importer = Importer(URL_OBS, "openSUSE:Factory", args.packages)
+    importer = Importer(URL_OBS, args.project, args.packages)
     importer.import_into_db()
     for package in args.packages:
-        export_package(package, args.repodir, args.cachedir, args.gc)
+        if not importer.package_with_scmsync(package):
+            export_package(args.project, package, args.repodir, args.cachedir, args.gc)
+        else:
+            logging.debug(f"{args.project}/{package} has scmsync links - skipping export")
 
 
 if __name__ == "__main__":

lib/git.py

@@ -48,6 +48,7 @@ class Git:
     def open(self):
         if not self.exists():
             self.git_run(["init", "--object-format=sha256", "-b", "factory"])
+        self.git_run(["config", "lfs.allowincompletepush", "true"])
 
     def is_dirty(self):
         """Check if there is something to commit"""
@@ -85,7 +86,7 @@ class Git:
         """Checkout into the branch HEAD"""
         new_branch = False
         if branch not in self.branches():
-            self.git_run(["branch", "-q", branch, "HEAD"])
+            self.git_run(["switch", "-q", "--orphan", branch])
             new_branch = True
         else:
             ref = f"refs/heads/{branch}"
@@ -180,7 +181,13 @@ class Git:
     #                logging.warning(f"Error removing file {path}: {e}")
 
     def add(self, filename):
-        self.git_run(["add", filename])
+        self.git_run(["add", ":(literal)" + str(filename)])
+
+    def add_default_gitignore(self):
+        if not (self.path / ".gitignore").exists():
+            with (self.path / ".gitignore").open("w") as f:
+                f.write(".osc\n")
+            self.add(".gitignore")
 
     def add_default_lfs_gitattributes(self, force=False):
         if not (self.path / ".gitattributes").exists() or force:
@@ -235,7 +242,7 @@ class Git:
 
     def remove(self, file: pathlib.Path):
         self.git_run(
-            ["rm", "-q", "-f", "--ignore-unmatch", file.name],
+            ["rm", "-q", "-f", "--ignore-unmatch", ":(literal)" + file.name],
         )
         patterns = self.get_specific_lfs_gitattributes()
         if file.name in patterns:
@@ -244,7 +251,7 @@ class Git:
 
     def add_gitea_remote(self, package):
         repo_name = package.replace("+", "_")
-        org_name = "rpm"
+        org_name = "pool"
 
         if not os.getenv("GITEA_TOKEN"):
             logging.warning("Not adding a remote due to missing $GITEA_TOKEN")
@@ -253,7 +260,7 @@ class Git:
         url = f"https://src.opensuse.org/api/v1/org/{org_name}/repos"
         response = requests.post(
             url,
-            data={"name": repo_name},
+            data={"name": repo_name, "object_format_name": "sha256"},
             headers={"Authorization": f"token {os.getenv('GITEA_TOKEN')}"},
             timeout=10,
         )
@@ -271,13 +278,11 @@ class Git:
             ["remote"],
             stdout=subprocess.PIPE,
         ).stdout.decode("utf-8"):
-            logger.warning("Not pushing to remote because no 'origin' configured")
+            logging.warning("Not pushing to remote because no 'origin' configured")
             return
 
         cmd = ["push"]
         if force:
             cmd.append("-f")
-        cmd.append("origin")
-        cmd.append("refs/heads/factory")
-        cmd.append("refs/heads/devel")
+        cmd += ["origin", "--all"]
         self.git_run(cmd)

lib/git_exporter.py

@@ -47,7 +47,7 @@ class GitExporter:
         left_to_commit = []
         for flat in reversed(flats):
             found_state = False
-            for branch in ["factory", "devel"]:
+            for branch in ["factory"]:
                 if flat.commit.dbid == state_data.get(branch):
                     branch_state[branch] = flat.commit
                     flat.commit.git_commit = self.git.branch_head(branch)
@@ -86,6 +86,11 @@ class GitExporter:
             logging.debug(f"Committing {flat}")
             self.commit_flat(flat, branch_state)
 
+        # make sure that we create devel branch
+        if not branch_state["devel"]:
+            logging.debug("force creating devel")
+            self.git.set_branch_head("devel", self.git.branch_head("factory"))
+
         self.git.push(force=True)
 
     def run_gc(self):
@@ -150,6 +155,7 @@ class GitExporter:
 
         # create file if not existant
         self.git.add_default_lfs_gitattributes(force=False)
+        self.git.add_default_gitignore()
 
         to_download, to_delete = flat.commit.calc_delta(branch_state[flat.branch])
         for file in to_delete:

lib/importer.py

@@ -26,13 +26,16 @@ class Importer:
         # Import multiple Factory packages into the database
         self.packages = packages
         self.project = project
+        self.scmsync_cache = dict()
+        self.packages_with_scmsync = set()
 
         self.db = DB()
         self.obs = OBS(api_url)
-        assert project == "openSUSE:Factory"
+        assert not self.has_scmsync(project)
         self.refreshed_packages = set()
         self.gone_packages_set = None
 
+
     def import_request(self, number):
         self.obs.request(number).import_into_db(self.db)
 
@@ -213,6 +216,10 @@ class Importer:
             return
         logging.debug(f"Refresh {project}/{package}")
         self.refreshed_packages.add(key)
+        if self.has_scmsync(project) or self.has_scmsync(key):
+            self.packages_with_scmsync.add(package)
+            logging.debug(f"{project}/{package} already in Git - skipping")
+            return
         self.update_db_package(project, package)
         self.fetch_all_linked_packages(project, package)
 
@@ -255,3 +262,21 @@ class Importer:
                 for line in f.readlines():
                     self.gone_packages_set.add(line.strip())
         return key in self.gone_packages_set
+
+    def has_scmsync(self, key):
+        if key in self.scmsync_cache:
+            return self.scmsync_cache[key]
+
+        root = self.obs._meta(key)
+        scmsync = None
+        scmsync_exists = False
+        if root and root.find('scmsync') is not None:
+            scmsync = root.find('scmsync').text
+        if scmsync:
+            scmsync_exists = scmsync.startswith('https://src.opensuse.org/pool/')
+        self.scmsync_cache[key] = scmsync_exists
+        return scmsync_exists
+
+    def package_with_scmsync(self, package):
+        return package in self.packages_with_scmsync
+

lib/obs.py

@@ -73,11 +73,11 @@ class OBS:
         logging.debug(f"GET {url}")
         return ET.parse(osc.core.http_GET(url)).getroot()
 
-    def _meta(self, project, package, **params):
+    def _meta(self, key, **params):
         try:
-            root = self._xml(f"source/{project}/{package}/_meta", **params)
+            root = self._xml(f"source/{key}/_meta", **params)
         except HTTPError:
-            logging.error(f"Package [{project}/{package} {params}] has no meta")
+            logging.error(f"Project/Package [{key} {params}] has no meta")
             return None
         return root
 
@@ -118,13 +118,13 @@ class OBS:
         return root
 
     def exists(self, project, package):
-        root = self._meta(project, package)
+        root = self._meta(f"{project}/{package}")
         if root is None:
             return False
         return root.get("project") == project
 
     def devel_project(self, project, package):
-        root = self._meta(project, package)
+        root = self._meta(f"{project}/{package}")
         devel = root.find("devel")
         if devel is None:
             return None

opensuse-monitor.py

@@ -2,36 +2,60 @@
 import json
 from pathlib import Path
 import pika
-import sys
+import random
+import time
 
 MY_TASKS_DIR = Path(__file__).parent / "tasks"
 
-connection = pika.BlockingConnection(pika.URLParameters("amqps://opensuse:opensuse@rabbit.opensuse.org"))
-channel = connection.channel()
 
-channel.exchange_declare(exchange='pubsub', exchange_type='topic', passive=True, durable=True)
+def listen_events():
+    connection = pika.BlockingConnection(
+        pika.URLParameters("amqps://opensuse:opensuse@rabbit.opensuse.org")
+    )
+    channel = connection.channel()
 
-result = channel.queue_declare("", exclusive=True)
-queue_name = result.method.queue
+    channel.exchange_declare(
+        exchange="pubsub", exchange_type="topic", passive=True, durable=False
+    )
 
-channel.queue_bind(exchange='pubsub',
-                   queue=queue_name,routing_key='#')
+    result = channel.queue_declare("", exclusive=True)
+    queue_name = result.method.queue
 
-print(' [*] Waiting for logs. To exit press CTRL+C')
+    channel.queue_bind(
+        exchange="pubsub", queue=queue_name, routing_key="opensuse.obs.package.commit"
+    )
 
-def callback(ch, method, properties, body):
-    if method.routing_key not in ("opensuse.obs.package.commit",):
-        return
-    body = json.loads(body)
-    if 'project' in body and 'package' in body and body['project'] == 'openSUSE:Factory':
-        if '/' in body['package']:
+    print(" [*] Waiting for logs. To exit press CTRL+C")
+
+    def callback(ch, method, properties, body):
+        if method.routing_key not in ("opensuse.obs.package.commit",):
             return
+        body = json.loads(body)
+        if (
+            "project" in body
+            and "package" in body
+            and body["project"] == "openSUSE:Factory"
+        ):
+            # Strip multibuild flavors
+            package = body["package"].partition(':')[0]
+            if "/" in package:
+                return
 
-        (MY_TASKS_DIR / body['package']).touch()
-        print(" [x] %r:%r" % (method.routing_key, body['package']))
+            (MY_TASKS_DIR / package).touch()
+            print(" [x] %r:%r" % (method.routing_key, body["package"]))
 
-channel.basic_consume(queue_name,
-                      callback,
-                      auto_ack=True)
+    channel.basic_consume(queue_name, callback, auto_ack=True)
 
-channel.start_consuming()
+    channel.start_consuming()
+
+
+def main():
+    while True:
+        try:
+            listen_events()
+        except (pika.exceptions.ConnectionClosed, pika.exceptions.AMQPHeartbeatTimeout):
+            time.sleep(random.randint(10, 100))
+
+
+if __name__ == "__main__":
+    main()

update-tasks.sh

@@ -0,0 +1,19 @@
+#!/bin/bash
+#
+cd /space/dmueller/git-importer
+
+source credentials.sh
+
+while true; do
+   for i in $PWD/tasks/*; do
+       if test -f "$i"; then
+            echo "$(date): Importing $(basename $i)"
+            if ! python3 ./git-importer.py -c repos/.cache $(basename $i); then
+                mkdir -p $PWD/failed-tasks
+                mv -f $i $PWD/failed-tasks
+            fi
+            rm -f $i
+        fi
+   done
+   inotifywait -q -e create $PWD/tasks
+done