# rhona, cleans up unassociated binary and source packages
# Copyright (C) 2000, 2001 James Troup <james@nocrew.org>
-# $Id: rhona,v 1.9 2001-03-14 20:31:56 troup Exp $
+# $Id: rhona,v 1.14 2001-05-31 02:19:30 troup Exp $
# This program is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
projectB = None
Cnf = None
-delete_date = None;
+now_date = None; # mark newly "deleted" things as deleted "now"
+delete_date = None; # delete things marked "deleted" earler than this
overrides = {};
+tried_too_hard_exc = "Tried too hard to find a free filename for %s; something's gone Pete Tong";
+
###################################################################################################
def usage (exit_code):
###################################################################################################
+def find_next_free (dest):
+ extra = 0;
+ orig_dest = dest;
+ too_much = 100;
+ while os.path.exists(dest) and extra < too_much:
+ dest = orig_dest + '.' + repr(extra);
+ extra = extra + 1;
+ if extra >= too_much:
+ raise tried_too_hard_exc;
+ return dest;
+
# FIXME: why can't we make (sane speed) UPDATEs out of these SELECTs?
def check_binaries():
- global delete_date;
+ global delete_date, now_date;
print "Checking for orphaned binary packages..."
projectB.query("BEGIN WORK");
for i in ql:
file_id = i[0];
- projectB.query("UPDATE files SET last_used = '%s' WHERE id = %s" % (delete_date, file_id))
+ projectB.query("UPDATE files SET last_used = '%s' WHERE id = %s AND last_used IS NULL" % (now_date, file_id))
projectB.query("COMMIT WORK");
# Check for any binaries which are marked for eventual deletion
projectB.query("COMMIT WORK");
def check_sources():
- global delete_date;
+ global delete_date, now_date;
print "Checking for orphaned source packages..."
WHERE NOT EXISTS
(SELECT sa.suite FROM src_associations sa WHERE sa.source = s.id)
AND NOT EXISTS (SELECT b.id FROM binaries b WHERE b.source = s.id)""");
- ql = q.getresult();
+ #### XXX: this should ignore cases where the files for the binary b
+ #### have been marked for deletion (so the delay between bins go
+ #### byebye and sources go byebye is 0 instead of StayOfExecution)
+
+ ql = q.getresult();
+
projectB.query("BEGIN WORK");
for i in ql:
source_id = i[0];
dsc_file_id = i[1];
# Mark the .dsc file for deletion
- projectB.query("UPDATE files SET last_used = '%s' WHERE id = %s" % (delete_date, dsc_file_id))
+ projectB.query("UPDATE files SET last_used = '%s' WHERE id = %s AND last_used IS NULL" % (now_date, dsc_file_id))
# Mark all other files references by .dsc too if they're not used by anyone else
x = projectB.query("SELECT f.id FROM files f, dsc_files d WHERE d.source = %s AND d.file = f.id" % (source_id));
for j in x.getresult():
- file_id = i[0];
- y = projectB.query("SELECT id FROM dsc_files d WHERE file = %s" % (file_id));
+ file_id = j[0];
+ y = projectB.query("SELECT id FROM dsc_files d WHERE d.file = %s" % (file_id));
if len(y.getresult()) == 1:
- projectB.query("UPDATE files SET last_used = '%s' WHERE id = %s" % (delete_date, file_id));
+ projectB.query("UPDATE files SET last_used = '%s' WHERE id = %s AND last_used IS NULL" % (now_date, file_id));
projectB.query("COMMIT WORK");
# Check for any sources which are marked for deletion but which
WHERE f.last_used IS NOT NULL AND s.id = df.source AND df.file = f.id
AND ((EXISTS (SELECT sa.suite FROM src_associations sa WHERE sa.source = s.id))
OR (EXISTS (SELECT b.id FROM binaries b WHERE b.source = s.id)))""");
+
+ #### XXX: this should also handle deleted binaries specially (ie, not
+ #### reinstate sources because of them
+
ql = q.getresult();
# Could be done in SQL; but left this way for hysterical raisins
# [and freedom to innovate don'cha know?]
projectB.query("COMMIT WORK");
def check_files():
- global delete_date;
+ global delete_date, now_date;
# FIXME: this is evil; nothing should ever be in this state. if
# they are, it's a bug and the files should not be auto-deleted.
print "Checking for unused files..."
- # Check for files not references in either binaries or dsc_files
- used = {};
- q = projectB.query("SELECT file FROM binaries");
- for i in q.getresult():
- used[i[0]] = "";
- q = projectB.query("SELECT file FROM dsc_files");
- for i in q.getresult():
- used[i[0]] = "";
-
- all = {};
- q = projectB.query("SELECT f.id, l.path, f.filename FROM files f, location l WHERE f.location = l.id;");
- for i in q.getresult():
- all[i[0]] = i[1] + i[2];
+ q = projectB.query("""
+SELECT id FROM files f
+ WHERE NOT EXISTS (SELECT id FROM binaries b WHERE b.file = f.id)
+ AND NOT EXISTS (SELECT id FROM dsc_files df WHERE df.file = f.id)""");
projectB.query("BEGIN WORK");
- for id in all.keys():
- if not used.has_key(id):
- projectB.query("UPDATE files SET last_used = '%s' WHERE id = %s" % (delete_date, id));
+ for i in q.getresult():
+ file_id = i[0];
+ projectB.query("UPDATE files SET last_used = '%s' WHERE id = %s" % (now_date, file_id));
projectB.query("COMMIT WORK");
def clean_binaries():
- global delete_date;
+ global delete_date, now_date;
# We do this here so that the binaries we remove will have their
# source also removed (if possible).
+ # XXX: why doesn't this remove the files here as well? I don't think it
+ # buys anything keeping this separate
print "Cleaning binaries from the DB..."
if not Cnf["Rhona::Options::No-Action"]:
before = time.time();
sys.stdout.write("[Deleting from binaries table... ");
projectB.query("DELETE FROM binaries WHERE EXISTS (SELECT id FROM files WHERE binaries.file = files.id AND files.last_used <= '%s')" % (delete_date));
- sys.stdout.write("done. (%d)]\n" % (int(time.time()-before)));
+ sys.stdout.write("done. (%d seconds)]\n" % (int(time.time()-before)));
def clean():
- global delete_date;
+ global delete_date, now_date;
count = 0;
size = 0;
print "Cleaning out packages..."
- # Ensure destination directory exists
- dest = Cnf["Dir::Morgue"] + '/' + Cnf["Rhona::MorgueSubDir"];
+ date = time.strftime("%Y-%m-%d", time.localtime(time.time()));
+ dest = Cnf["Dir::Morgue"] + '/' + Cnf["Rhona::MorgueSubDir"] + '/' + date;
if not os.path.exists(dest):
os.mkdir(dest);
sys.stdout.write("[Deleting from source table... ");
projectB.query("DELETE FROM dsc_files WHERE EXISTS (SELECT df.id FROM source s, files f, dsc_files df WHERE f.last_used <= '%s' AND s.file = f.id AND s.id = df.source AND df.id = dsc_files.id)" % (delete_date));
projectB.query("DELETE FROM source WHERE EXISTS (SELECT id FROM files WHERE source.file = files.id AND files.last_used <= '%s')" % (delete_date));
- sys.stdout.write("done. (%d)]\n" % (int(time.time()-before)));
+ sys.stdout.write("done. (%d seconds)]\n" % (int(time.time()-before)));
# Delete files from the pool
q = projectB.query("SELECT l.path, f.filename FROM location l, files f WHERE f.last_used <= '%s' AND l.id = f.location" % (delete_date));
else:
size = size + os.stat(filename)[stat.ST_SIZE];
count = count + 1;
+
+ dest_filename = dest + '/' + os.path.basename(filename);
+ # If the destination file exists; try to find another filename to use
+ if os.path.exists(dest_filename):
+ dest_filename = find_next_free(dest_filename);
+
if Cnf["Rhona::Options::No-Action"]:
- print "Cleaning %s to %s..." % (filename, dest);
+ print "Cleaning %s -> %s ..." % (filename, dest_filename);
else:
- utils.move(filename, dest);
+ utils.move(filename, dest_filename);
else:
sys.stderr.write("%s is neither symlink nor file?!\n" % (filename));
sys.exit(1);
- # delete from files
+
+ # Delete from the 'files' table
if not Cnf["Rhona::Options::No-Action"]:
before = time.time();
sys.stdout.write("[Deleting from files table... ");
projectB.query("DELETE FROM files WHERE last_used <= '%s'" % (delete_date));
- sys.stdout.write("done. (%d)]\n" % (int(time.time()-before)));
+ sys.stdout.write("done. (%d seconds)]\n" % (int(time.time()-before)));
if count > 0:
sys.stderr.write("Cleaned %d files, %s.\n" % (count, utils.size_type(size)));
def clean_maintainers():
print "Cleaning out unused Maintainer entries..."
- used = {};
- q = projectB.query("SELECT maintainer FROM binaries WHERE maintainer IS NOT NULL");
- for i in q.getresult():
- used[i[0]] = "";
- q = projectB.query("SELECT maintainer FROM source WHERE maintainer IS NOT NULL");
- for i in q.getresult():
- used[i[0]] = "";
-
- all = {};
- q = projectB.query("SELECT id, name FROM maintainer");
- for i in q.getresult():
- all[i[0]] = i[1];
+ q = projectB.query("""
+SELECT m.id FROM maintainer m
+ WHERE NOT EXISTS (SELECT id FROM binaries b WHERE b.maintainer = m.id)
+ AND NOT EXISTS (SELECT id FROM source s WHERE s.maintainer = m.id)""");
+ ql = q.getresult();
count = 0;
projectB.query("BEGIN WORK");
- for id in all.keys():
- if not used.has_key(id):
- if not Cnf["Rhona::Options::No-Action"]:
- projectB.query("DELETE FROM maintainer WHERE id = %s" % (id));
+ for i in ql:
+ maintainer_id = i[0];
+ if not Cnf["Rhona::Options::No-Action"]:
+ projectB.query("DELETE FROM maintainer WHERE id = %s" % (maintainer_id));
count = count + 1;
projectB.query("COMMIT WORK");
sys.stderr.write("Cleared out %d maintainer entries.\n" % (count));
def main():
- global Cnf, projectB, delete_date;
+ global Cnf, projectB, delete_date, now_date;
- projectB = pg.connect('projectb', 'localhost');
-
apt_pkg.init();
Cnf = apt_pkg.newConfiguration();
apt_pkg.ReadConfigFileISC(Cnf,utils.which_conf_file());
+ projectB = pg.connect(Cnf["DB::Name"], Cnf["DB::Host"], int(Cnf["DB::Port"]));
+
Arguments = [('D',"debug","Rhona::Options::Debug", "IntVal"),
('h',"help","Rhona::Options::Help"),
('n',"no-action","Rhona::Options::No-Action"),
if not os.access(override_filename, os.R_OK):
sys.stderr.write("W: Could not find source-only override file '%s'.\n" % (override_filename));
+ now_date = time.strftime("%Y-%m-%d %H:%M", time.localtime(time.time()));
delete_date = time.strftime("%Y-%m-%d %H:%M", time.localtime(time.time()-int(Cnf["Rhona::StayOfExecution"])));
-
+
check_binaries();
clean_binaries();
check_sources();
- #check_files();
+ check_files();
clean();
clean_maintainers();