X-Git-Url: https://git.decadent.org.uk/gitweb/?a=blobdiff_plain;f=tea;h=9c5b00b786dc5633c1fa21d29fbf5851e6ef1ea3;hb=7b23ac278ff189b7b56cdd1c317716bc787bff33;hp=54137c5826630d2babd9b9530bad7787c7e640f5;hpb=bb8283ad6274e461fc78a00b2a52c6594c53e13d;p=dak.git

diff --git a/tea b/tea
index 54137c58..9c5b00b7 100755
--- a/tea
+++ b/tea
@@ -1,8 +1,8 @@
 #!/usr/bin/env python
 
-# 'Fix' stable to make debian-cd and dpkg -BORGiE users happy
-# Copyright (C) 2000  James Troup <james@nocrew.org>
-# $Id: tea,v 1.1 2000-12-05 04:27:48 troup Exp $
+# Sanity check the database
+# Copyright (C) 2000, 2001, 2002  James Troup <james@nocrew.org>
+# $Id: tea,v 1.20 2002-06-08 00:23:51 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
@@ -25,45 +25,256 @@
 
 ################################################################################
 
-import pg, sys, os
+import pg, sys, os, string, stat, time
 import utils, db_access
-import apt_pkg;
+import apt_pkg, apt_inst;
 
 ################################################################################
 
 Cnf = None;
 projectB = None;
+db_files = {};
+waste = 0.0;
+excluded = {};
+current_file = None;
+future_files = {};
+current_time = time.time();
 
 ################################################################################
 
-def main ():
-    global Cnf, projectB;
-
-    apt_pkg.init();
-    
-    Cnf = apt_pkg.newConfiguration();
-    apt_pkg.ReadConfigFileISC(Cnf,utils.which_conf_file());
+def process_dir (unused, dirname, filenames):
+    global waste, db_files, excluded;
 
-    Arguments = [('d',"debug","Claire::Options::Debug", "IntVal"),
-                 ('h',"help","Claire::Options::Help"),
-                 ('v',"version","Claire::Options::Version")];
-
-    apt_pkg.ParseCommandLine(Cnf,Arguments,sys.argv);
+    if string.find(dirname, '/disks-') != -1 or string.find(dirname, 'upgrade-') != -1:
+        return;
+    # hack; can't handle .changes files
+    if string.find(dirname, 'proposed-updates') != -1:
+        return;
+    for name in filenames:
+        filename = os.path.abspath(dirname+'/'+name);
+        filename = string.replace(filename, 'potato-proposed-updates', 'proposed-updates');
+        if os.path.isfile(filename) and not os.path.islink(filename) and not db_files.has_key(filename) and not excluded.has_key(filename):
+            waste = waste + os.stat(filename)[stat.ST_SIZE];
+            print filename
 
-    projectB = pg.connect('projectb', 'localhost');
+################################################################################
 
-    db_access.init(Cnf, projectB);
+def check_files():
+    global db_files;
 
+    print "Building list of database files...";
     q = projectB.query("SELECT l.path, f.filename FROM files f, location l WHERE f.location = l.id")
     ql = q.getresult();
 
+    db_files.clear();
     for i in ql:
-	filename = i[0] + i[1];
+	filename = os.path.abspath(i[0] + i[1]);
+        db_files[filename] = "";
         if os.access(filename, os.R_OK) == 0:
-            print filename
+            utils.warn("'%s' doesn't exist." % (filename));
+
+    file = utils.open_file(Cnf["Dir::Override"]+'override.unreferenced');
+    for filename in file.readlines():
+        filename = filename[:-1];
+        excluded[filename] = "";
+
+    print "Checking against existent files...";
+
+    os.path.walk(Cnf["Dir::Root"]+'dists/', process_dir, None);
+
+    print
+    print "%s wasted..." % (utils.size_type(waste));
+
+################################################################################
+
+def check_dscs():
+    count = 0;
+    suite = 'unstable';
+    for component in Cnf.SubTree("Component").List():
+        if component == "mixed":
+            continue;
+        component = string.lower(component);
+        list_filename = '%s%s_%s_source.list' % (Cnf["Dir::Lists"], suite, component);
+        list_file = utils.open_file(list_filename);
+        for line in list_file.readlines():
+            file = line[:-1];
+            try:
+                utils.parse_changes(file, dsc_whitespace_rules=1);
+            except utils.invalid_dsc_format_exc, line:
+                utils.warn("syntax error in .dsc file '%s', line %s." % (file, line));
+                count = count + 1;
+
+    if count:
+        utils.warn("Found %s invalid .dsc files." % (count));
+
+################################################################################
+
+def check_override():
+    for suite in [ "stable", "unstable" ]:
+        print suite
+        print "-------------"
+        print
+        suite_id = db_access.get_suite_id(suite);
+        q = projectB.query("""
+SELECT DISTINCT b.package FROM binaries b, bin_associations ba
+ WHERE b.id = ba.bin AND ba.suite = %s AND NOT EXISTS
+       (SELECT * FROM override o WHERE o.suite = %s AND o.package = b.package)"""
+                           % (suite_id, suite_id));
+        print q
+        q = projectB.query("""
+SELECT DISTINCT s.source FROM source s, src_associations sa
+  WHERE s.id = sa.source AND sa.suite = %s AND NOT EXISTS
+       (SELECT * FROM override o WHERE o.suite = %s and o.package = s.source)"""
+                           % (suite_id, suite_id));
+        print q
+
+################################################################################
+
+# Ensure that the source files for any given package is all in one
+# directory so that 'apt-get source' works...
+
+def check_source_in_one_dir():
+    # Not the most enterprising method, but hey...
+    broken_count = 0;
+    q = projectB.query("SELECT id FROM source;");
+    for i in q.getresult():
+        source_id = i[0];
+        q2 = projectB.query("""
+SELECT l.path, f.filename FROM files f, dsc_files df, location l WHERE df.source = %s AND f.id = df.file AND l.id = f.location"""
+                            % (source_id));
+        first_path = "";
+        first_filename = "";
+        broken = 0;
+        for j in q2.getresult():
+            filename = j[0]+j[1];
+            path = os.path.dirname(filename);
+            if first_path == "":
+                first_path = path;
+                first_filename = filename;
+            elif first_path != path:
+                symlink = path + '/' + os.path.basename(first_filename);
+                if not os.path.exists(symlink):
+                    broken = 1;
+                    print "WOAH, we got a live one here... %s [%s] {%s}" % (filename, source_id, symlink);
+        if broken:
+            broken_count = broken_count + 1;
+    print "Found %d source packages where the source is not all in one directory." % (broken_count);
+
+################################################################################
+
+def check_md5sums():
+    print "Getting file information from database...";
+    q = projectB.query("SELECT l.path, f.filename, f.md5sum, f.size FROM files f, location l WHERE f.location = l.id")
+    ql = q.getresult();
+
+    print "Checking file md5sums & sizes...";
+    for i in ql:
+	filename = os.path.abspath(i[0] + i[1]);
+        db_md5sum = i[2];
+        db_size = int(i[3]);
+        try:
+            file = utils.open_file(filename);
+        except:
+            utils.warn("can't open '%s'." % (filename));
+            continue;
+        md5sum = apt_pkg.md5sum(file);
+        size = os.stat(filename)[stat.ST_SIZE];
+        if md5sum != db_md5sum:
+            utils.warn("**WARNING** md5sum mismatch for '%s' ('%s' [current] vs. '%s' [db])." % (filename, md5sum, db_md5sum));
+        if size != db_size:
+            utils.warn("**WARNING** size mismatch for '%s' ('%s' [current] vs. '%s' [db])." % (filename, size, db_size));
+
+    print "Done."
+
+################################################################################
+#
+# Check all files for timestamps in the future; common from hardware
+# (e.g. alpha) which have far-future dates as their default dates.
+
+
+
+def Ent(Kind,Name,Link,Mode,UID,GID,Size,MTime,Major,Minor):
+    global future_files;
+
+    if MTime > current_time:
+        future_files[current_file] = MTime;
+        print "%s: %s '%s','%s',%u,%u,%u,%u,%u,%u,%u" % (current_file, Kind,Name,Link,Mode,UID,GID,Size, MTime, Major, Minor);
+
+def check_timestamps():
+    global current_file;
+
+    q = projectB.query("SELECT l.path, f.filename FROM files f, location l WHERE f.location = l.id AND f.filename ~ '.deb$'")
+    ql = q.getresult();
+    db_files.clear();
+    count = 0;
+    for i in ql:
+	filename = os.path.abspath(i[0] + i[1]);
+        if os.access(filename, os.R_OK):
+            file = utils.open_file(filename);
+            current_file = filename;
+            sys.stderr.write("Processing %s.\n" % (filename));
+            apt_inst.debExtract(file,Ent,"control.tar.gz");
+            file.seek(0);
+            apt_inst.debExtract(file,Ent,"data.tar.gz");
+            count = count + 1;
+    print "Checked %d files (out of %d)." % (count, len(db_files.keys()));
+
+################################################################################
+
+def check_missing_tar_gz_in_dsc():
+    count = 0;
+
+    print "Building list of database files...";
+    q = projectB.query("SELECT l.path, f.filename FROM files f, location l WHERE f.location = l.id AND f.filename ~ '.dsc$'");
+    ql = q.getresult();
+    if ql:
+        print "Checking %d files..." % len(ql);
+    else:
+        print "No files to check."
+    for i in ql:
+        filename = os.path.abspath(i[0] + i[1]);
+        try:
+            # NB: don't enforce .dsc syntax
+            dsc = utils.parse_changes(filename);
+        except:
+            utils.fubar("error parsing .dsc file '%s'." % (filename));
+        dsc_files = utils.build_file_list(dsc, is_a_dsc=1);
+        has_tar = 0;
+        for file in dsc_files.keys():
+            m = utils.re_issource.match(file);
+            if not m:
+                utils.fubar("%s not recognised as source." % (file));
+            type = m.group(3);
+            if type == "orig.tar.gz" or type == "tar.gz":
+                has_tar = 1;
+        if not has_tar:
+            utils.warn("%s has no .tar.gz in the .dsc file." % (file));
+            count = count + 1;
+
+    if count:
+        utils.warn("Found %s invalid .dsc files." % (count));
+
+################################################################################
+
+def main ():
+    global Cnf, projectB, db_files, waste, excluded;
+
+    Cnf = utils.get_conf()
+
+    apt_pkg.ParseCommandLine(Cnf,[],sys.argv);
+    projectB = pg.connect(Cnf["DB::Name"], Cnf["DB::Host"], int(Cnf["DB::Port"]));
+    db_access.init(Cnf, projectB);
+
+    #check_md5sums();
+    #check_source_in_one_dir();
+    #check_override();
+    #check_dscs();
+    #check_files();
+    #check_timestamps();
+    check_missing_tar_gz_in_dsc();
 
 #######################################################################################
 
 if __name__ == '__main__':
-    main()
+    main();