does the cleanup

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

diff --git a/scripts/clean-backupdb.py b/scripts/clean-backupdb.py
index 106ba94..b689943 100755 (executable)
--- a/scripts/clean-backupdb.py
+++ b/scripts/clean-backupdb.py
@@ -18,6 +18,9 @@ import traceback
 
 from optparse import OptionParser
 
+now=datetime.now()
+counter=0
+
 class FileIgnored (Exception): pass
 
 # in days
@@ -58,8 +61,6 @@ def parse_filename (filename):
             return parsing_failed
 
 # one entry like this per file, managed in the Kind class
-now=datetime.now()
-
 class File:
 
     def __init__ (self, dir, filename, datetime, options):
@@ -75,7 +76,10 @@ class File:
         self.group = self._group_string()
 
     def __repr__ (self):
-        return "%s (%s) -- weekday %s"%(self.filename,self.datetime,self.datetime.weekday())
+        return "%s (%s) -- weekday %s"%(self.path(),self.datetime,self.datetime.weekday())
+
+    def path (self):
+        return os.path.normpath(os.path.join(self.dir,self.filename))
 
     def age_days (self):
         return self.age.days
@@ -144,6 +148,16 @@ class File:
             weeks += 1
             return "week%02d"%weeks
 
+    def cleanup (self, preserved):
+        global counter
+        counter+=1
+        if self.options.dry_run:
+            print "Would cleanup %s"%(self.path())
+            print "    (keeping %s)"%preserved.path()
+        else:
+            if self.options.verbose: print "unlink",self.path()
+            os.unlink (self.path())
+
 # all files in a given timeslot (either month or week)
 class Group:
     def __init__ (self, groupname):
@@ -153,9 +167,9 @@ class Group:
         self.files.append(file)
     def epilogue (self):
         self.files.sort (File.sort_relevance)
-#        print 20*'*','after sort'
-#        for file in self.files:
-#            print "%s"%file
+    def keep_one (self):
+        for file in self.files[1:]:
+            file.cleanup(self.files[0])
 
 # all files with the same (prefix, suffix)
 class Kind:
@@ -220,6 +234,13 @@ class Kind:
     def sort_size (k1, k2):
         return len(k1.list)-len(k2.list)
 
+    def cleanup (self):
+        groupnames=self.groups.keys()
+        groupnames.sort()
+        for groupname in groupnames:
+            if self.options.extra_verbose: print 'GROUP',groupname
+            self.groups[groupname].keep_one()
+
 # keeps an index of all files found, index by (prefix, suffix), then sorted by time
 class Index:
     def __init__ (self,options):
@@ -252,7 +273,10 @@ class Index:
                 continue
             self.insert (dir, filename, p, s, d)
 
-def handle_arg (index, dir, pattern):
+    def cleanup (self):
+        for kind in self.index.values(): kind.cleanup()
+
+def handle_dir_pattern (index, dir, pattern):
     try:
         os.chdir(dir)
     except:
@@ -262,7 +286,8 @@ def handle_arg (index, dir, pattern):
     index.insert_many (dir, filenames)
 
 def main ():
-    parser=OptionParser()
+    usage="Usage: %prog [options] dir_or_files"
+    parser=OptionParser(usage=usage)
     parser.add_option ("-v","--verbose",dest='verbose',action='store_true',default=False,
                        help="run in verbose mode")
     parser.add_option ("-x","--extra-verbose",dest='extra_verbose',action='store_true',default=False,
@@ -274,8 +299,6 @@ def main ():
     (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)
@@ -284,19 +307,29 @@ def main ():
         print "Offset not understood %s - expect an int. number of days"%options.offset
         sys.exit(1)
     
+    if len(args) ==0:
+        parser.print_help()
+        sys.exit(1)
+
     # args can be directories, or patterns, like 
     # main /db-backup /db-backup-f8/*bz2
     # in any case we handle each arg completely separately
-    index = Index (options)
+    dir_patterns=[]
     for arg in args:
         if os.path.isdir (arg):
-            handle_arg (index, arg, "*")
+            if arg=='.': arg=os.getcwd()
+            dir_patterns.append ( (arg, '*',) )
         else:
             (dir,pattern)=os.path.split(arg)
-            if not dir: dir='.'
-            handle_arg (index, dir, pattern)
+            if not dir: dir=os.getcwd()
+            dir_patterns.append ( (dir, pattern,) )
+
+    index = Index (options)
+    for (dir, pattern) in dir_patterns: handle_dir_pattern (index, dir, pattern)
     index.epilogue()
     index.show()
+    index.cleanup()
+    print 'Found %d entries to unlink'%counter
             
 if __name__ == '__main__':
     main()