more accurate grouping

[infrastructure.git] / scripts / clean-backupdb.py

diff --git a/scripts/clean-backupdb.py b/scripts/clean-backupdb.py
index 45e5026..106ba94 100755 (executable)
--- a/scripts/clean-backupdb.py
+++ b/scripts/clean-backupdb.py
@@ -9,8 +9,9 @@
 # (*) recent stuff: unchanged
 # 
 
+import sys
 import os, os.path
-from datetime import datetime
+from datetime import datetime, timedelta
 from glob import glob
 import re
 import traceback
@@ -79,9 +80,6 @@ class File:
     def age_days (self):
         return self.age.days
 
-    def age_weeks (self):
-        return self.age.days/7
-
     # oldest first
     @staticmethod
     def sort_age (file1, file2):
@@ -105,23 +103,31 @@ class File:
         elif w1!=PREFERRED_WEEKDAY and w2==PREFERRED_WEEKDAY:
             return 1
     
-    month_marrier=None
+    month_barrier=None
+    week_barrier=None
     @staticmethod
-    def compute_month_barrier ():
-        if not File.month_marrier: 
-            # find the exact datetime for a month change 
-            # KEEP_FILES_IN_MONTH_AFTER months ago +
-            year=now.year
-            month=now.month
-            day=1
-            if now.month>=KEEP_FILES_IN_MONTH_AFTER+1:
-                month -= KEEP_FILES_IN_MONTH_AFTER
-            else:
-                year -= 1
-                month += (12-KEEP_FILES_IN_MONTH_AFTER)
-            File.month_marrier = datetime (year=year, month=month, day=day)
-                
-        return File.month_marrier
+    def _compute_barriers ():
+        if File.month_barrier:
+            return
+        # find the exact datetime for a month change 
+        # KEEP_FILES_IN_MONTH_AFTER months ago +
+        year=now.year
+        month=now.month
+        day=1
+        if now.month>=KEEP_FILES_IN_MONTH_AFTER+1:
+            month -= KEEP_FILES_IN_MONTH_AFTER
+        else:
+            year -= 1
+            month += (12-KEEP_FILES_IN_MONTH_AFTER)
+        File.month_barrier = datetime (year=year, month=month, day=day)
+        # find the next monday morning
+        remaining_days=(7-File.month_barrier.weekday())%7
+        File.week_barrier=File.month_barrier+timedelta(days=remaining_days)
+
+    @staticmethod
+    def compute_month_barrier(): File._compute_barriers(); return File.month_barrier
+    @staticmethod
+    def compute_week_barrier(): File._compute_barriers(); return File.week_barrier
 
     # returns a key for grouping files, the cleanup then
     # preserving one entry in the set of files with same group
@@ -134,7 +140,9 @@ class File:
         if self.datetime <= File.compute_month_barrier():
             return self.datetime.strftime("%Y%m")
         else:
-            return "week%d"%self.age_weeks()
+            weeks=(self.datetime-File.compute_week_barrier()).days/7
+            weeks += 1
+            return "week%02d"%weeks
 
 # all files in a given timeslot (either month or week)
 class Group:
@@ -191,7 +199,7 @@ class Kind:
             print " << %s - %s d old"%(f.filename, f.age_days()),
         if entries >=2:
             f=self.list[-1]
-            print ">> %s - %s d old"%(f.filename, f.age_days())
+            print "|| %s - %s d old >>"%(f.filename, f.age_days())
         groupnames=self.groups.keys()
         groupnames.sort()
         groupnames.reverse()
@@ -204,7 +212,7 @@ class Kind:
                     print "    %s"%file
         elif self.options.verbose:
             print " Found %d groups"%len(groupnames),
-            for g in groupnames: print "%s->%d"%(k,len(self.groups[g].files)),
+            for g in groupnames: print "%s->%d"%(g,len(self.groups[g].files)),
             print ''
 
     # sort on number of entries
@@ -261,8 +269,20 @@ def main ():
                        help="run in extra verbose mode")
     parser.add_option ("-n","--dry-run",dest='dry_run',action='store_true',default=False,
                        help="dry run")
+    parser.add_option ("-o","--offset",dest='offset',action='store',type='int',default=0,
+                       help="pretend we run <offset> days in the future")
     (options, args) = parser.parse_args()
     if options.extra_verbose: options.verbose=True
+    try:
+        #options.offset=int(options.offset)
+        print 'offset=%d'%options.offset
+        if options.offset !=0:
+            global now
+            now += timedelta(days=options.offset)
+    except:
+        traceback.print_exc()
+        print "Offset not understood %s - expect an int. number of days"%options.offset
+        sys.exit(1)
     
     # args can be directories, or patterns, like 
     # main /db-backup /db-backup-f8/*bz2