stop using saved state

We can reconstruct the state based on log messages. Otherwise,
it's incorrect as the trimming function removes exported parts
anyway and then we have missing commits

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(args.project, 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/db_revision.py

@@ -204,6 +204,11 @@ class DBRevision:
                     and self.package == "_project"
                 ):
                     continue
+
+                # do not import _service:* files as those are created by OBS on source imports
+                if entry.get("name")[0:9] == "_service:":
+                    continue
+
                 cur.execute(
                     """INSERT INTO files (name, md5, size, mtime, revision_id)
                             VALUES (%s,%s,%s,%s,%s)""",

lib/flat_walker.py

@@ -20,7 +20,7 @@ class FlatTreeWalker(AbstractWalker):
 
     def __init__(self, rebase_devel=False) -> None:
         super().__init__()
-        self.flats = []
+        self.flats:list[FlatNode] = []
         # the rebase_devel won't work as such as rebasing the branch needs an explicit action
         self.rebase_devel = rebase_devel
         # remember the last merge point so we can know the parent of it for the root of the sources

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}"
@@ -159,6 +160,12 @@ class Git:
             .strip()
         )
 
+    def branch_commit(self, branch="HEAD"):
+      try:
+        return (self.git_run(["cat-file", "commit", branch], stdout=subprocess.PIPE).stdout.decode("utf-8").strip())
+      except:
+        return ''
+
     def set_branch_head(self, branch, commit):
         return self.git_run(["update-ref", f"refs/heads/{branch}", commit])
 
@@ -180,7 +187,13 @@ 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 +248,7 @@ 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 +257,7 @@ 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 +266,7 @@ 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 +284,11 @@ 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

@@ -1,8 +1,12 @@
 import logging
 import os
+from urllib.parse import parse_qs
 
+import psycopg
+from urllib3.util import url
 import yaml
 
+from lib import db
 from lib.binary import is_binary_or_large
 from lib.db import DB
 from lib.git import Git
@@ -12,6 +16,12 @@ from lib.proxy_sha256 import ProxySHA256
 from lib.tree_builder import TreeBuilder
 from lib.user import User
 
+def is_number(s):
+    try:
+        float(s)
+        return True
+    except ValueError:
+        return False
 
 class GitExporter:
     def __init__(self, api_url, project, package, repodir, cachedir):
@@ -30,49 +40,106 @@ class GitExporter:
         else:
             self.git.create()
             self.git.add_gitea_remote(package)
-        self.state_file = os.path.join(self.git.path, ".git", "_flat_state.yaml")
         self.gc_interval = 200
         self.cachedir = cachedir
 
     def set_gc_interval(self, gc):
         self.gc_interval = gc
 
-    def check_repo_state(self, flats, branch_state):
+    def reconstruct_state(self, flats):
         state_data = dict()
-        if os.path.exists(self.state_file):
-            with open(self.state_file) as f:
-                state_data = yaml.safe_load(f)
-                if not isinstance(state_data, dict):
-                    state_data = {}
+        prefix = "OBS-URL: "
+        for line in self.git.branch_commit("factory").splitlines():
+            if line.startswith(prefix):
+                u = url.parse_url(line.strip(prefix))
+                if u.path != f"/package/show/openSUSE:Factory/{self.package}" or "rev=" not in u.query:
+                    continue
+                v = parse_qs(u.query)
+                rev = v['rev'][0]
+                with self.db.cursor() as cur:
+                    try:
+                        if is_number(rev):
+                            cur.execute("SELECT id FROM revisions WHERE project=%s AND package=%s AND rev=%s", ('openSUSE:Factory', self.package, rev,))
+                        else:
+                            cur.execute("SELECT id FROM revisions WHERE project=%s AND package=%s AND expanded_srcmd5=%s", ('openSUSE:Factory', self.package, rev, rev))
+                    except psycopg.Error as e:
+                        logging.error(e)
+                        self.db.conn.rollback()
+                    row = cur.fetchone()
+                    if not row:
+                        return state_data
+                    state_data['factory'] = row[0]
+        try:
+            print("devel reconstruct")
+            d = self.devel_rev(flats)
+            if d is not None:
+                prj = d.commit.project
+                for line in self.git.branch_commit("devel").splitlines():
+                    if line.startswith(prefix):
+                        u = url.parse_url(line.strip(prefix))
+                        if u.path != f"/package/show/{prj}/{self.package}" or u.query is None or "rev=" not in u.query:
+                            continue
+                        v = parse_qs(u.query)
+                        rev = v['rev'][0]
+                        try:
+                            with self.db.cursor() as cur:
+                                logging.debug(f"finding id for ({prj, self.package, rev}")
+                                if is_number(rev):
+                                    cur.execute("SELECT id FROM revisions WHERE project=%s AND package=%s AND rev=%s", (prj, self.package, rev,))
+                                else:
+                                    cur.execute("SELECT id FROM revisions WHERE project=%s AND package=%s AND expanded_srcmd5=%s", (prj, self.package, rev,))
+                                row = cur.fetchone()
+                                if not row:
+                                    logging.info(" ** cannot find revision for devel branch:", rev)
+                                    return state_data
+                                state_data['devel'] = row[0]
+                        except psycopg.Error as e:
+                            logging.error(e)
+                            self.db.conn.rollback()
+                            if state_data['factory'] is not None:
+                                state_data['devel'] = state_data['factory']
+        except:
+            if state_data['factory'] is not None:
+                state_data['devel'] = state_data['factory']
+        return state_data
+
+    def check_repo_state(self, flats, branch_state, branch):
+        state_data = self.reconstruct_state(flats)
+
+        logging.debug(f"state data: {state_data}")
         left_to_commit = []
         for flat in reversed(flats):
             found_state = False
-            for branch in ["factory", "devel"]:
-                if flat.commit.dbid == state_data.get(branch):
-                    branch_state[branch] = flat.commit
-                    flat.commit.git_commit = self.git.branch_head(branch)
-                    logging.debug(
-                        f"Found {self.git.path}'s {branch} branch in state {flat}"
-                    )
-                    left_to_commit = []
-                    found_state = True
+            if flat.commit.dbid == state_data.get(branch):
+                branch_state[branch] = flat.commit
+                flat.commit.git_commit = self.git.branch_head(branch)
+                logging.debug(
+                    f"Found {self.git.path}'s {branch} branch in state {flat}"
+                )
+                left_to_commit = []
+                found_state = True
             if not found_state:
                 left_to_commit.append(flat)
         return left_to_commit
 
+    def devel_rev(self, tree):
+        for flat in tree:
+            if flat.branch == "devel":
+                return flat
+        return None
+
     def export_as_git(self):
         if os.getenv("CHECK_ALL_LFS"):
             LFSOid.check_all(self.db, self.package)
         tree = TreeBuilder(self.db).build(self.project, self.package)
-        flats = tree.as_flat_list()
+        added_commits = False
 
-        branch_state = {"factory": None, "devel": None}
-        left_to_commit = self.check_repo_state(flats, branch_state)
-
-        if not left_to_commit:
+        if tree == None:  # eg. python-M2Crypto errors
             return
-
-        logging.info(f"Commiting into {self.git.path}")
+        flats = tree.as_flat_list()
+        branch_state = {"factory": None, "devel": None}
+        left_to_commit = self.check_repo_state(flats, branch_state, "factory")
+        logging.info(f"Commiting into {self.git.path} {len(left_to_commit)} into factory")
         self.run_gc()
         users = dict()
 
@@ -85,8 +152,40 @@ class GitExporter:
                 self.run_gc()
             logging.debug(f"Committing {flat}")
             self.commit_flat(flat, branch_state)
+            added_commits = True
 
-        self.git.push(force=True)
+        # export the devel_tree head commits based on the devel branch
+        if self.project == "openSUSE:Factory":
+            devel_head = self.devel_rev(flats)
+            flat_devel = None
+            if devel_head is not None:
+                logging.debug(f"building devel revisions chain for {devel_head.commit.project} / {self.package}")
+                flat_devel = TreeBuilder(self.db).revisions_chain(devel_head.commit.project, self.package).as_flat_list()
+                for f in flat_devel:
+                    f.branch = "devel"
+
+            if flat_devel is not None:
+                left_to_commit = self.check_repo_state(flat_devel, branch_state, "devel")
+                logging.debug(branch_state)
+                logging.debug(f"appending {len(left_to_commit)} items on top of devel")
+                for flat in left_to_commit:
+                    if flat.commit.userid not in users:
+                        users[flat.commit.userid] = User.find(self.db, flat.commit.userid)
+                    flat.user = users[flat.commit.userid]
+                    self.gc_cnt -= 1
+                    if self.gc_cnt <= 0 and self.gc_interval:
+                        self.run_gc()
+                    logging.debug(f"Committing {flat}")
+                    self.commit_flat(flat, branch_state)
+                    added_commits = True
+
+        # 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"))
+
+        #if added_commits:
+        #    self.git.push(force=True)
 
     def run_gc(self):
         self.gc_cnt = self.gc_interval
@@ -98,6 +197,10 @@ class GitExporter:
         return not self.proxy_sha256.is_text(package, filename)
 
     def commit_file(self, flat, file, size, md5):
+        # don't export imported _service: files, if any
+        if file.name[0:9] == '_service:':
+            return
+
         # have such files been detected as text mimetype before?
         if self.is_lfs_file(flat.commit.package, file.name, size):
             file_sha256 = self.proxy_sha256.get_or_put(
@@ -150,6 +253,7 @@ 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:
@@ -166,10 +270,3 @@ class GitExporter:
         )
         flat.commit.git_commit = commit
         branch_state[flat.branch] = flat.commit
-        with open(self.state_file, "w") as f:
-            data = {}
-            for branch in ["factory", "devel"]:
-                commit = branch_state[branch]
-                if commit:
-                    data[branch] = commit.dbid
-            yaml.dump(data, f)

lib/importer.py

@@ -26,19 +26,24 @@ 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)
 
     def update_db_package(self, project, package):
         root = self.obs._history(project, package)
         if root is None:
+            if self.project == "openSUSE:Factory" and project == self.project:
+                exit(10)
             return
         latest = DBRevision.max_rev(self.db, project, package)
         for r in root.findall("revision"):
@@ -213,6 +218,10 @@ 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((project, 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 +264,18 @@ 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_exists = False
+        if root is not None:
+            scmsync_exists = root.find('scmsync') is not None
+        self.scmsync_cache[key] = scmsync_exists
+        return scmsync_exists
+
+    def package_with_scmsync(self, project, package):
+        return (project, 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
@@ -148,12 +148,21 @@ class OBS:
         ]
 
     def _download(self, project, package, name, revision):
-        url = osc.core.makeurl(
-            self.url,
-            ["source", project, package, name],
-            {"rev": revision, "expand": 1},
-        )
-        return osc.core.http_GET(url)
+        try:
+            url = osc.core.makeurl(
+                self.url,
+                ["source", project, package, name],
+                {"rev": revision, "expand": 1},
+            )
+            return osc.core.http_GET(url)
+        except HTTPError as e:
+            if e.status == 404:
+                url = osc.core.makeurl(
+                    self.url,
+                    ["source", project, package, name],
+                    {"rev": revision, "expand": 1, "deleted": 1},
+                )
+                return osc.core.http_GET(url)
 
     def download(
         self,
@@ -189,7 +198,7 @@ class OBS:
         try:
             root = self._xml(f"source/{project}/{package}", **params)
         except HTTPError as e:
-            if e.code == 400:
+            if e.code == 400 or e.code == 404:
                 logging.error(
                     f"Package [{project}/{package} {params}] can't be expanded: {e}"
                 )

lib/proxy_sha256.py

@@ -50,12 +50,12 @@ class ProxySHA256:
         sha = hashlib.sha256()
         while True:
             buffer = fin.read(10000)
-            if not buffer:
-                break
-            sha.update(buffer)
             # only guess from the first 10K
             if not mimetype:
                 mimetype = self.mime.from_buffer(buffer)
+            if not buffer:
+                break
+            sha.update(buffer)
         fin.close()
         LFSOid(self.db).add(
             project, package, name, revision, sha.hexdigest(), size, mimetype, file_md5

lib/tree_builder.py

@@ -216,6 +216,8 @@ class TreeBuilder:
     def build(self, project, package):
         """Create a Factory tree (returning the top)"""
         factory_revisions = self.revisions_chain(project, package)
+        if factory_revisions == None:
+            return None
         self.add_merge_points(factory_revisions)
         # factory_revisions.print()
         self.prune_loose_end(factory_revisions)

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