forked from adamm/git-importer
196 lines
7.3 KiB
Python
196 lines
7.3 KiB
Python
import logging
|
|
import xml.etree.ElementTree as ET
|
|
|
|
from lib.db import DB
|
|
from lib.db_revision import DBRevision
|
|
from lib.obs import OBS
|
|
from lib.obs_revision import OBSRevision
|
|
|
|
from lib.user import User
|
|
|
|
class Importer:
|
|
def __init__(self, api_url, project, package):
|
|
# Import a Factory package into the database
|
|
self.package = package
|
|
self.project = project
|
|
|
|
self.obs = OBS()
|
|
assert project == "openSUSE:Factory"
|
|
self.obs.change_url(api_url)
|
|
|
|
def update_db_package(self, db, project, package):
|
|
root = self.obs._history(project, package)
|
|
if root is None:
|
|
return
|
|
latest = DBRevision.latest_revision(db, project, package)
|
|
for r in root.findall("revision"):
|
|
rev = OBSRevision(self.obs, project, package).parse(r)
|
|
if not latest or rev.rev > latest.rev:
|
|
dbrev = DBRevision.import_obs_rev(db, rev)
|
|
try:
|
|
root = rev.read_link()
|
|
except ET.ParseError:
|
|
dbrev.set_broken(db)
|
|
continue
|
|
if root is not None:
|
|
tprj = root.get("project") or project
|
|
tpkg = root.get("package") or package
|
|
dbrev.links_to(db, tprj, tpkg)
|
|
|
|
def find_linked_revs(self, db):
|
|
with db.cursor() as cur:
|
|
cur.execute(
|
|
"""SELECT * from revisions WHERE id in (SELECT l.revision_id FROM links l
|
|
LEFT JOIN linked_revs lrevs ON lrevs.revision_id=l.revision_id
|
|
WHERE lrevs.id IS NULL) and broken is FALSE;"""
|
|
)
|
|
for row in cur.fetchall():
|
|
rev = DBRevision(row)
|
|
linked_rev = rev.linked_rev(db)
|
|
if not linked_rev:
|
|
logging.debug(f"No link {rev}")
|
|
continue
|
|
cur.execute(
|
|
"""INSERT INTO linked_revs (revision_id, linked_id)
|
|
VALUES (%s,%s)""",
|
|
(rev.dbid, linked_rev.dbid),
|
|
)
|
|
|
|
def fetch_all_linked_packages(self, db, project, package):
|
|
with db.cursor() as cur:
|
|
cur.execute(
|
|
"""SELECT DISTINCT l.project, l.package from links l JOIN revisions r
|
|
on r.id=l.revision_id WHERE r.project=%s AND r.package=%s""",
|
|
(project, package),
|
|
)
|
|
for row in cur.fetchall():
|
|
(lproject, lpackage) = row
|
|
self.update_db_package(db, lproject, lpackage)
|
|
|
|
def find_fake_revisions(self, db):
|
|
with db.cursor() as cur:
|
|
cur.execute(
|
|
"SELECT * from revisions WHERE id in (SELECT linked_id from linked_revs WHERE considered=FALSE)"
|
|
)
|
|
for row in cur.fetchall():
|
|
self._find_fake_revision(db, DBRevision(row))
|
|
|
|
def _find_fake_revision(self, db, rev):
|
|
prev = rev.previous_commit(db)
|
|
if not prev:
|
|
with db.cursor() as cur:
|
|
cur.execute(
|
|
"UPDATE linked_revs SET considered=TRUE where linked_id=%s",
|
|
(rev.dbid,),
|
|
)
|
|
return
|
|
with db.cursor() as cur:
|
|
cur.execute(
|
|
"""SELECT * FROM revisions WHERE id IN
|
|
(SELECT revision_id from linked_revs WHERE linked_id=%s)
|
|
AND commit_time <= %s ORDER BY commit_time""",
|
|
(prev.dbid, rev.commit_time),
|
|
)
|
|
last_linked = None
|
|
for linked in cur.fetchall():
|
|
linked = DBRevision(linked)
|
|
nextrev = linked.next_commit(db)
|
|
if nextrev and nextrev.commit_time < rev.commit_time:
|
|
continue
|
|
last_linked = linked
|
|
cur.execute(
|
|
"UPDATE linked_revs SET considered=TRUE where linked_id=%s",
|
|
(rev.dbid,),
|
|
)
|
|
if not last_linked:
|
|
return
|
|
|
|
with db.cursor() as cur:
|
|
linked = last_linked
|
|
cur.execute(
|
|
"SELECT 1 FROM fake_revs where revision_id=%s AND linked_id=%s",
|
|
(rev.dbid, linked.dbid),
|
|
)
|
|
if cur.fetchone():
|
|
cur.execute(
|
|
"UPDATE linked_revs SET considered=TRUE where linked_id=%s",
|
|
(rev.dbid,),
|
|
)
|
|
return
|
|
fake_rev = linked.rev + rev.rev / 1000.0
|
|
comment = f"Updating link to change in {rev.project}/{rev.package} revision {rev.rev}"
|
|
cur.execute(
|
|
"""INSERT INTO revisions (project,package,rev,unexpanded_srcmd5,
|
|
commit_time, userid, comment) VALUES(%s,%s,%s,%s,%s,%s,%s) RETURNING id""",
|
|
(
|
|
linked.project,
|
|
linked.package,
|
|
fake_rev,
|
|
linked.unexpanded_srcmd5,
|
|
rev.commit_time,
|
|
"buildservice-autocommit",
|
|
comment,
|
|
),
|
|
)
|
|
new_id = cur.fetchone()[0]
|
|
cur.execute(
|
|
"""INSERT INTO linked_revs (revision_id, linked_id) VALUES (%s,%s)""",
|
|
(new_id, rev.dbid),
|
|
)
|
|
cur.execute(
|
|
"""INSERT INTO fake_revs (revision_id, linked_id) VALUES (%s,%s)""",
|
|
(rev.dbid, linked.dbid),
|
|
)
|
|
|
|
def revisions_without_files(self, db):
|
|
with db.cursor() as cur:
|
|
cur.execute(
|
|
"SELECT * FROM revisions WHERE broken=FALSE AND expanded_srcmd5 IS NULL"
|
|
)
|
|
return [DBRevision(row) for row in cur.fetchall()]
|
|
|
|
|
|
def import_into_db(self):
|
|
db = DB()
|
|
self.update_db_package(db, self.project, self.package)
|
|
self.fetch_all_linked_packages(db, self.project, self.package)
|
|
|
|
# all remaining, no filtering here
|
|
self.find_linked_revs(db)
|
|
|
|
missing_users = User.missing_users(db)
|
|
for userid in missing_users:
|
|
missing_user = self.obs.user(userid)
|
|
if missing_user:
|
|
missing_user.import_into_db(db)
|
|
|
|
self.find_fake_revisions(db)
|
|
for rev in self.revisions_without_files(db):
|
|
with db.cursor() as cur:
|
|
cur.execute(
|
|
"""SELECT unexpanded_srcmd5 from revisions WHERE
|
|
id=(SELECT linked_id FROM linked_revs WHERE revision_id=%s)""",
|
|
(rev.dbid,),
|
|
)
|
|
linked_rev = cur.fetchone()
|
|
if linked_rev:
|
|
linked_rev = linked_rev[0]
|
|
list = self.obs.list(
|
|
rev.project, rev.package, rev.unexpanded_srcmd5, linked_rev
|
|
)
|
|
if list:
|
|
rev.import_dir_list(db, list)
|
|
md5 = rev.calculate_files_hash(db)
|
|
with db.cursor() as cur:
|
|
cur.execute(
|
|
"UPDATE revisions SET files_hash=%s WHERE id=%s",
|
|
(md5, rev.dbid),
|
|
)
|
|
else:
|
|
rev.set_broken(db)
|
|
|
|
for number in DBRevision.requests_to_fetch(db):
|
|
self.obs.request(number).import_into_db(db)
|
|
|
|
db.conn.commit()
|