Add and remove literal files

pathspec in git has special characters that we should not trigger.
Assume every filespec as literal

README.md

@@ -1,4 +1,4 @@
-sudo zypper in python3-psycopg2
+sudo zypper in python3-psycopg
 sudo su - postgres
 # `createdb -O <LOCAL_USER> imported_git`
 

lib/config.py

@@ -14,8 +14,6 @@ def config(filename="database.ini", section="production"):
         for param in params:
             db[param[0]] = param[1]
     else:
-        raise Exception(
-            "Section {0} not found in the {1} file".format(section, filename)
-        )
+        raise Exception(f"Section {section} not found in the {filename} file")
 
     return db

lib/db.py

@@ -1,7 +1,6 @@
 import logging
 
-import psycopg2
-from psycopg2.extras import LoggingConnection
+import psycopg
 
 from lib.config import config
 
@@ -17,22 +16,20 @@ class DB:
             # read the connection parameters
             params = config(section=self.config_section)
             # connect to the PostgreSQL server
-            self.conn = psycopg2.connect(connection_factory=LoggingConnection, **params)
-            logger = logging.getLogger(__name__)
-            self.conn.initialize(logger)
+            self.conn = psycopg.connect(conninfo=f"dbname={params['database']}")
+            logging.getLogger("psycopg.pool").setLevel(logging.INFO)
 
-        except (Exception, psycopg2.DatabaseError) as error:
+        except (Exception, psycopg.DatabaseError) as error:
             print(error)
             raise error
 
     def schema_version(self):
         # create a cursor
         with self.conn.cursor() as cur:
-
             # execute a statement
             try:
                 cur.execute("SELECT MAX(version) from scheme")
-            except psycopg2.errors.UndefinedTable as error:
+            except psycopg.errors.UndefinedTable:
                 cur.close()
                 self.close()
                 self.connect()
@@ -146,9 +143,9 @@ class DB:
         )
         schemes[10] = (
             "ALTER TABLE revisions ADD COLUMN request_id INTEGER",
-            """ALTER TABLE revisions 
+            """ALTER TABLE revisions
                ADD CONSTRAINT request_id_foreign_key
-               FOREIGN KEY (request_id) 
+               FOREIGN KEY (request_id)
                REFERENCES requests (id)""",
             "UPDATE scheme SET version=10",
         )
@@ -273,7 +270,7 @@ class DB:
                         cur.execute(command)
             # commit the changes
             self.conn.commit()
-        except (Exception, psycopg2.DatabaseError) as error:
+        except (Exception, psycopg.DatabaseError) as error:
             print(error)
             self.close()
             raise error

lib/db_revision.py

@@ -2,7 +2,6 @@ from __future__ import annotations
 
 from hashlib import md5
 from pathlib import Path
-from typing import Optional
 
 from lib.db import DB
 from lib.obs_revision import OBSRevision
@@ -206,7 +205,7 @@ class DBRevision:
                 ):
                     continue
                 cur.execute(
-                    """INSERT INTO files (name, md5, size, mtime, revision_id) 
+                    """INSERT INTO files (name, md5, size, mtime, revision_id)
                             VALUES (%s,%s,%s,%s,%s)""",
                     (
                         entry.get("name"),
@@ -255,7 +254,7 @@ class DBRevision:
         self._files.sort(key=lambda x: x["name"])
         return self._files
 
-    def calc_delta(self, current_rev: Optional[DBRevision]):
+    def calc_delta(self, current_rev: DBRevision | None):
         """Calculate the list of files to download and to delete.
         Param current_rev is the revision that's currently checked out.
         If it's None, the repository is empty.

lib/git.py

@@ -4,7 +4,6 @@ import os
 import pathlib
 import subprocess
 
-import pygit2
 import requests
 
 from lib.binary import BINARY
@@ -20,11 +19,6 @@ class Git:
         self.committer = committer
         self.committer_email = committer_email
 
-        self.repo = None
-
-    def is_open(self):
-        return self.repo is not None
-
     def exists(self):
         """Check if the path is a valid git repository"""
         return (self.path / ".git").exists()
@@ -34,36 +28,69 @@ class Git:
         self.path.mkdir(parents=True, exist_ok=True)
         self.open()
 
+    def git_run(self, args, **kwargs):
+        """Run a git command"""
+        if "env" in kwargs:
+            envs = kwargs["env"].copy()
+            del kwargs["env"]
+        else:
+            envs = os.environ.copy()
+        envs["GIT_LFS_SKIP_SMUDGE"] = "1"
+        envs["GIT_CONFIG_GLOBAL"] = "/dev/null"
+        return subprocess.run(
+            ["git"] + args,
+            cwd=self.path,
+            check=True,
+            env=envs,
+            **kwargs,
+        )
+
     def open(self):
-        # Convert the path to string, to avoid some limitations in
-        # older pygit2
-        self.repo = pygit2.init_repository(str(self.path))
+        if not self.exists():
+            self.git_run(["init", "--object-format=sha256", "-b", "factory"])
 
     def is_dirty(self):
         """Check if there is something to commit"""
-        assert self.is_open()
-
-        return self.repo.status()
+        status_str = self.git_run(
+            ["status", "--porcelain=2"],
+            stdout=subprocess.PIPE,
+        ).stdout.decode("utf-8")
+        return len(list(filter(None, status_str.split("\n")))) > 0
 
     def branches(self):
-        return list(self.repo.branches)
+        br = (
+            self.git_run(
+                ["for-each-ref", "--format=%(refname:short)", "refs/heads/"],
+                stdout=subprocess.PIPE,
+            )
+            .stdout.decode("utf-8")
+            .split()
+        )
+        if len(br) == 0:
+            br.append("factory")  # unborn branch?
+        return br
 
-    def branch(self, branch, commit=None):
-        if not commit:
-            commit = self.repo.head
-        else:
-            commit = self.repo.get(commit)
-        self.repo.branches.local.create(branch, commit)
+    def branch(self, branch, commit="HEAD"):
+        commit = (
+            self.git_run(
+                ["rev-parse", "--verify", "--end-of-options", commit + "^{commit}"],
+                stdout=subprocess.PIPE,
+            )
+            .stdout.decode("utf-8")
+            .strip()
+        )
+        return self.git_run(["branch", branch, commit])
 
     def checkout(self, branch):
         """Checkout into the branch HEAD"""
         new_branch = False
-        ref = f"refs/heads/{branch}"
         if branch not in self.branches():
-            self.repo.references["HEAD"].set_target(ref)
+            self.git_run(["switch", "-q", "--orphan", branch])
             new_branch = True
         else:
-            self.repo.checkout(ref)
+            ref = f"refs/heads/{branch}"
+            if (self.path / ".git" / ref).exists():
+                self.git_run(["switch", "--no-guess", "-q", branch])
         return new_branch
 
     def commit(
@@ -87,51 +114,79 @@ class Git:
             committer_time = committer_time if committer_time else user_time
 
         if self.is_dirty():
-            self.repo.index.add_all()
+            self.git_run(["add", "--all", "."])
 
-        self.repo.index.write()
-        author = pygit2.Signature(user, user_email, int(user_time.timestamp()))
-        committer = pygit2.Signature(
-            committer, committer_email, int(committer_time.timestamp())
+        tree_id = (
+            self.git_run(["write-tree"], stdout=subprocess.PIPE)
+            .stdout.decode("utf-8")
+            .strip()
         )
 
-        tree = self.repo.index.write_tree()
-        return self.repo.create_commit(
-            "HEAD", author, committer, message, tree, parents
+        parent_array = []
+        if isinstance(parents, list):
+            for parent in filter(None, parents):
+                parent_array = parent_array + ["-p", parent]
+        elif isinstance(parents, str):
+            parent_array = ["-p", parents]
+
+        commit_id = (
+            self.git_run(
+                ["commit-tree"] + parent_array + [tree_id],
+                env={
+                    "GIT_AUTHOR_NAME": user,
+                    "GIT_AUTHOR_EMAIL": user_email,
+                    "GIT_AUTHOR_DATE": f"{int(user_time.timestamp())} +0000",
+                    "GIT_COMMITTER_NAME": committer,
+                    "GIT_COMMITTER_EMAIL": committer_email,
+                    "GIT_COMMITTER_DATE": f"{int(committer_time.timestamp())} +0000",
+                },
+                input=message.encode("utf-8"),
+                stdout=subprocess.PIPE,
+            )
+            .stdout.decode("utf-8")
+            .rstrip()
         )
+        self.git_run(["reset", "--soft", commit_id])
+        return commit_id
 
-    def last_commit(self):
-        try:
-            return self.repo.head.target
-        except:
-            return None
-
-    def branch_head(self, branch):
-        return self.repo.references["refs/heads/" + branch].target
+    def branch_head(self, branch="HEAD"):
+        return (
+            self.git_run(
+                ["rev-parse", "--verify", "--end-of-options", branch],
+                stdout=subprocess.PIPE,
+            )
+            .stdout.decode("utf-8")
+            .strip()
+        )
 
     def set_branch_head(self, branch, commit):
-        self.repo.references["refs/heads/" + branch].set_target(commit)
+        return self.git_run(["update-ref", f"refs/heads/{branch}", commit])
 
     def gc(self):
         logging.debug(f"Garbage recollect and repackage {self.path}")
-        subprocess.run(
-            ["git", "gc", "--auto"],
-            cwd=self.path,
+        self.git_run(
+            ["gc", "--auto"],
             stdout=subprocess.PIPE,
             stderr=subprocess.STDOUT,
         )
 
-    def clean(self):
-        for path, _ in self.repo.status().items():
-            logging.debug(f"Cleaning {path}")
-            try:
-                (self.path / path).unlink()
-                self.repo.index.remove(path)
-            except Exception as e:
-                logging.warning(f"Error removing file {path}: {e}")
+    #    def clean(self):
+    #        for path, _ in self.repo.status().items():
+    #            logging.debug(f"Cleaning {path}")
+    #            try:
+    #                (self.path / path).unlink()
+    #                self.repo.index.remove(path)
+    #            except Exception as e:
+    #                logging.warning(f"Error removing file {path}: {e}")
 
     def add(self, filename):
-        self.repo.index.add(filename)
+        self.git_run(["add", ":(literal)" + 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:
@@ -185,9 +240,9 @@ class Git:
             return any(fnmatch.fnmatch(filename, line) for line in patterns)
 
     def remove(self, file: pathlib.Path):
-        self.repo.index.remove(file.name)
-        (self.path / file).unlink()
-
+        self.git_run(
+            ["rm", "-q", "-f", "--ignore-unmatch", ":(literal)" + file.name],
+        )
         patterns = self.get_specific_lfs_gitattributes()
         if file.name in patterns:
             patterns.remove(file.name)
@@ -201,7 +256,7 @@ class Git:
             logging.warning("Not adding a remote due to missing $GITEA_TOKEN")
             return
 
-        url = f"https://gitea.opensuse.org/api/v1/org/{org_name}/repos"
+        url = f"https://src.opensuse.org/api/v1/org/{org_name}/repos"
         response = requests.post(
             url,
             data={"name": repo_name},
@@ -212,20 +267,23 @@ class Git:
         # 201 Created
         if response.status_code not in (201, 409):
             print(response.data)
-        url = f"gitea@gitea.opensuse.org:{org_name}/{repo_name}.git"
-        self.repo.remotes.create("origin", url)
+        url = f"gitea@src.opensuse.org:{org_name}/{repo_name}.git"
+        self.git_run(
+            ["remote", "add", "origin", url],
+        )
 
     def push(self, force=False):
-        remo = self.repo.remotes["origin"]
+        if "origin" not in self.git_run(
+            ["remote"],
+            stdout=subprocess.PIPE,
+        ).stdout.decode("utf-8"):
+            logging.warning("Not pushing to remote because no 'origin' configured")
+            return
 
-        keypair = pygit2.KeypairFromAgent("gitea")
-        callbacks = pygit2.RemoteCallbacks(credentials=keypair)
-
-        refspecs = ["refs/heads/factory"]
-        develspec = "refs/heads/devel"
-        if develspec in self.repo.references:
-            if force:
-                refspecs.append(f"+{develspec}:{develspec}")
-            else:
-                refspecs.append("{develspec}:{develspec}")
-        remo.push(refspecs, callbacks=callbacks)
+        cmd = ["push"]
+        if force:
+            cmd.append("-f")
+        cmd.append("origin")
+        cmd.append("refs/heads/factory")
+        cmd.append("refs/heads/devel")
+        self.git_run(cmd)

lib/git_exporter.py

@@ -40,9 +40,9 @@ class GitExporter:
     def check_repo_state(self, flats, branch_state):
         state_data = dict()
         if os.path.exists(self.state_file):
-            with open(self.state_file, "r") as f:
+            with open(self.state_file) as f:
                 state_data = yaml.safe_load(f)
-                if type(state_data) != dict:
+                if not isinstance(state_data, dict):
                     state_data = {}
         left_to_commit = []
         for flat in reversed(flats):
@@ -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

@@ -1,5 +1,5 @@
-import concurrent.futures
 import logging
+import pathlib
 import xml.etree.ElementTree as ET
 
 from lib.db import DB
@@ -31,6 +31,7 @@ class Importer:
         self.obs = OBS(api_url)
         assert project == "openSUSE:Factory"
         self.refreshed_packages = set()
+        self.gone_packages_set = None
 
     def import_request(self, number):
         self.obs.request(number).import_into_db(self.db)
@@ -105,7 +106,7 @@ class Importer:
         with self.db.cursor() as cur:
             cur.execute(
                 """SELECT * FROM revisions WHERE id IN
-                            (SELECT revision_id from linked_revs WHERE linked_id=%s) 
+                            (SELECT revision_id from linked_revs WHERE linked_id=%s)
                             AND commit_time <= %s ORDER BY commit_time""",
                 (prev.dbid, rev.commit_time),
             )
@@ -138,7 +139,7 @@ class Importer:
             fake_rev = linked.rev + rev.rev / 1000.0
             comment = f"Updating link to change in {rev.project}/{rev.package} revision {int(rev.rev)}"
             cur.execute(
-                """INSERT INTO revisions (project,package,rev,unexpanded_srcmd5, 
+                """INSERT INTO revisions (project,package,rev,unexpanded_srcmd5,
                             commit_time, userid, comment, api_url) VALUES(%s,%s,%s,%s,%s,%s,%s,%s) RETURNING id""",
                 (
                     linked.project,
@@ -161,10 +162,12 @@ class Importer:
                 (rev.dbid, linked.dbid),
             )
 
-    def revisions_without_files(self):
+    def revisions_without_files(self, package):
+        logging.debug(f"revisions_without_files({package})")
         with self.db.cursor() as cur:
             cur.execute(
-                "SELECT * FROM revisions WHERE broken=FALSE AND expanded_srcmd5 IS NULL"
+                "SELECT * FROM revisions WHERE package=%s AND broken=FALSE AND expanded_srcmd5 IS NULL",
+                (package,),
             )
             return [DBRevision(self.db, row) for row in cur.fetchall()]
 
@@ -178,11 +181,11 @@ class Importer:
             linked_rev = cur.fetchone()
         if linked_rev:
             linked_rev = linked_rev[0]
-        list = self.obs.list(
+        obs_dir_list = self.obs.list(
             rev.project, rev.package, rev.unexpanded_srcmd5, linked_rev
         )
-        if list:
-            rev.import_dir_list(list)
+        if obs_dir_list:
+            rev.import_dir_list(obs_dir_list)
             md5 = rev.calculate_files_hash()
             with self.db.cursor() as cur:
                 cur.execute(
@@ -196,53 +199,43 @@ class Importer:
         self.find_linked_revs()
 
         self.find_fake_revisions()
-        with concurrent.futures.ThreadPoolExecutor(max_workers=8) as executor:
-            fs = [
-                executor.submit(import_rev, self, rev)
-                for rev in self.revisions_without_files()
-            ]
-        concurrent.futures.wait(fs)
+        for package in self.packages:
+            for rev in self.revisions_without_files(package):
+                print(f"rev {rev} is without files")
+                self.import_rev(rev)
 
     def refresh_package(self, project, package):
         key = f"{project}/{package}"
         if key in self.refreshed_packages:
             # refreshing once is good enough
             return
+        if self.package_gone(key):
+            return
         logging.debug(f"Refresh {project}/{package}")
         self.refreshed_packages.add(key)
         self.update_db_package(project, package)
         self.fetch_all_linked_packages(project, package)
 
     def import_into_db(self):
+        for package in self.packages:
+            refresh_package(self, self.project, package)
 
-        with concurrent.futures.ThreadPoolExecutor(max_workers=8) as executor:
-            fs = [
-                executor.submit(refresh_package, self, self.project, package)
-                for package in self.packages
-            ]
-            concurrent.futures.wait(fs)
+        self.db.conn.commit()
 
-            self.db.conn.commit()
+        for number in DBRevision.requests_to_fetch(self.db):
+            self.import_request(number)
 
-            fs = [
-                executor.submit(import_request, self, number)
-                for number in DBRevision.requests_to_fetch(self.db)
-            ]
-            concurrent.futures.wait(fs)
+        self.db.conn.commit()
 
-            self.db.conn.commit()
+        with self.db.cursor() as cur:
+            cur.execute(
+                """SELECT DISTINCT source_project,source_package FROM requests
+                        WHERE id IN (SELECT request_id FROM revisions WHERE project=%s and package = ANY(%s));""",
+                (self.project, self.packages),
+            )
+            for project, package in cur.fetchall():
+                self.refresh_package(project, package)
 
-            with self.db.cursor() as cur:
-                cur.execute(
-                    """SELECT DISTINCT source_project,source_package FROM requests 
-                            WHERE id IN (SELECT request_id FROM revisions WHERE project=%s and package = ANY(%s));""",
-                    (self.project, self.packages),
-                )
-                fs = [
-                    executor.submit(refresh_package, self, project, package)
-                    for project, package in cur.fetchall()
-                ]
-                concurrent.futures.wait(fs)
         self.db.conn.commit()
 
         missing_users = User.missing_users(self.db)
@@ -254,3 +247,11 @@ class Importer:
 
         self.fill_file_lists()
         self.db.conn.commit()
+
+    def package_gone(self, key):
+        if not self.gone_packages_set:
+            self.gone_packages_set = set()
+            with open(pathlib.Path(__file__).parent.parent / "gone-packages.txt") as f:
+                for line in f.readlines():
+                    self.gone_packages_set.add(line.strip())
+        return key in self.gone_packages_set

lib/lfs_oid.py

@@ -68,7 +68,7 @@ class LFSOid:
             row = cur.fetchone()
             lfs_oid_id = row[0]
             cur.execute(
-                """INSERT INTO lfs_oid_in_package (package,filename,lfs_oid_id) 
+                """INSERT INTO lfs_oid_in_package (package,filename,lfs_oid_id)
                         VALUES (%s,%s,%s)""",
                 (package, filename, lfs_oid_id),
             )
@@ -83,7 +83,7 @@ class LFSOid:
             self.register()
 
     def check(self):
-        url = f"http://gitea.opensuse.org:9999/check/{self.sha256}/{self.size}"
+        url = f"http://localhost:9999/check/{self.sha256}/{self.size}"
         response = requests.get(
             url,
             timeout=10,
@@ -127,12 +127,13 @@ class LFSOid:
             "size": self.size,
         }
 
-        url = "http://gitea.opensuse.org:9999/register"
+        url = "http://localhost:9999/register"
         response = requests.post(
             url,
             json=data,
             timeout=10,
         )
+        response.raise_for_status()
         logging.info(f"Register LFS returned {response.status_code}")
 
 
@@ -167,7 +168,7 @@ if __name__ == "__main__":
         cur.execute(
             """
             CREATE TEMPORARY TABLE lfs_oid_in_revision (
-                revision_id INTEGER, 
+                revision_id INTEGER,
                 lfs_oid_id  INTEGER NOT NULL,
                 name        VARCHAR(255) NOT NULL
             )

lib/obs.py

@@ -150,7 +150,7 @@ class OBS:
     def _download(self, project, package, name, revision):
         url = osc.core.makeurl(
             self.url,
-            ["source", project, package, urllib.parse.quote(name)],
+            ["source", project, package, name],
             {"rev": revision, "expand": 1},
         )
         return osc.core.http_GET(url)
@@ -165,7 +165,6 @@ class OBS:
         cachedir: str,
         file_md5: str,
     ) -> None:
-
         cached_file = self._path_from_md5(name, cachedir, file_md5)
         if not self.in_cache(name, cachedir, file_md5):
             with (dirpath / name).open("wb") as f:

lib/proxy_sha256.py

@@ -7,8 +7,6 @@ except:
     print("Install python3-python-magic, not python3-magic")
     raise
 
-import requests
-
 from lib.db import DB
 from lib.lfs_oid import LFSOid
 from lib.obs import OBS
@@ -43,7 +41,6 @@ class ProxySHA256:
             }
 
     def put(self, project, package, name, revision, file_md5, size):
-
         if not self.mime:
             self.mime = magic.Magic(mime=True)
 

lib/tree_builder.py

@@ -1,4 +1,3 @@
-from typing import Dict
 from xmlrpc.client import Boolean
 
 from lib.db_revision import DBRevision
@@ -114,7 +113,7 @@ class TreeBuilder:
                 candidates.append(node)
             if node.merged_into:
                 # we can't have candidates that are crossing previous merges
-                # see https://gitea.opensuse.org/importers/git-importer/issues/14
+                # see https://src.opensuse.org/importers/git-importer/issues/14
                 candidates = []
             node = node.parent
         if candidates:
@@ -138,7 +137,7 @@ class TreeBuilder:
                 self.requests.add(node.revision.request_id)
 
         class FindMergeWalker(AbstractWalker):
-            def __init__(self, builder: TreeBuilder, requests: Dict) -> None:
+            def __init__(self, builder: TreeBuilder, requests: dict) -> None:
                 super().__init__()
                 self.source_revisions = dict()
                 self.builder = builder

opensuse-monitor.py

@@ -0,0 +1,37 @@
+#!/usr/bin/python3
+import json
+from pathlib import Path
+import pika
+import sys
+
+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)
+
+result = channel.queue_declare("", exclusive=True)
+queue_name = result.method.queue
+
+channel.queue_bind(exchange='pubsub',
+                   queue=queue_name,routing_key='#')
+
+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':
+        if '/' in body['package']:
+            return
+
+        (MY_TASKS_DIR / body['package']).touch()
+        print(" [x] %r:%r" % (method.routing_key, body['package']))
+
+channel.basic_consume(queue_name,
+                      callback,
+                      auto_ack=True)
+
+channel.start_consuming()

tasks/.gitignore

@@ -0,0 +1 @@
+*