reconstruct state data

If the state file is missing, we can reconstruct which parts were
exported based on revision ids

Also, packages could have branches in Git, but not be in Git. We
need to check (project,package) tuple for this and not just abort
based on the package name alone.

git-importer.py

@@ -119,7 +119,7 @@ def main():
     importer = Importer(URL_OBS, args.project, args.packages)
     importer.import_into_db()
     for package in args.packages:
-        if not importer.package_with_scmsync(package):
+        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")

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

@@ -160,6 +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])
 
@@ -251,7 +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")

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):
@@ -37,17 +47,77 @@ class GitExporter:
     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()
+        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 = 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 = {}
+        else:
+            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"]:
             if flat.commit.dbid == state_data.get(branch):
                 branch_state[branch] = flat.commit
                 flat.commit.git_commit = self.git.branch_head(branch)
@@ -60,19 +130,24 @@ class GitExporter:
                 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()
 
@@ -85,13 +160,40 @@ class GitExporter:
                 self.run_gc()
             logging.debug(f"Committing {flat}")
             self.commit_flat(flat, branch_state)
+            added_commits = 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"))
 
-        self.git.push(force=True)
+        #if added_commits:
+        #    self.git.push(force=True)
 
     def run_gc(self):
         self.gc_cnt = self.gc_interval
@@ -103,6 +205,10 @@ 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(

lib/importer.py

@@ -42,6 +42,8 @@ class Importer:
     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"):
@@ -217,7 +219,7 @@ class Importer:
         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)
+            self.packages_with_scmsync.add((project, package))
             logging.debug(f"{project}/{package} already in Git - skipping")
             return
         self.update_db_package(project, package)
@@ -274,6 +276,6 @@ class Importer:
         self.scmsync_cache[key] = scmsync_exists
         return scmsync_exists
 
-    def package_with_scmsync(self, package):
-        return package in self.packages_with_scmsync
+    def package_with_scmsync(self, project, package):
+        return (project, package) in self.packages_with_scmsync
 

lib/obs.py

@@ -148,12 +148,21 @@ class OBS:
         ]
 
     def _download(self, project, package, name, revision):
+        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

@@ -36,10 +36,12 @@ def listen_events():
             and "package" in body
             and body["project"] == "openSUSE:Factory"
         ):
-            if "/" in body["package"]:
+            # Strip multibuild flavors
+            package = body["package"].partition(':')[0]
+            if "/" in package:
                 return
 
-            (MY_TASKS_DIR / body["package"]).touch()
+            (MY_TASKS_DIR / package).touch()
             print(" [x] %r:%r" % (method.routing_key, body["package"]))
 
     channel.basic_consume(queue_name, callback, auto_ack=True)