#!/usr/local/bin/perl -w #---------#---------#---------#---------#---------#---------#---------#--------- # # title # read LDIF files, make LDIF delete records for all except ones in a list # author # idc@planetlarg.net 28 nov 03 # description # a variation on ldap-ldif2ldif-hash.pl # This one checks each record, selects a few and prints delete statements # # # bugs # does not allow for multivalued attributes # does not deal with multiline entries # # modifications # #---------#---------#---------#---------#---------#---------#---------#--------- # setup use 5.005; use strict; use vars qw( $DEBUG $VERBOSE); $DEBUG = 0; # 1 to print internal information, 0 for normal $VERBOSE = 1; # 1 to print internal information, 0 for normal # logs to examine my @logs = @ARGV if (@ARGV) ; usage() unless (@logs); die "no log to process! \n" unless @logs; # vars # store the results in these vars my %tally; # count of various totals $tally{del_records} = 0; # tally of records to delete $tally{match_records} = 0; # tally of target records in all log files my %records; # ldif entries # oooh, let's get our first statistic $tally{logs} = @logs; # number of logs # my $log; # current logfile (with path) being read # #---------#---------#---------#---------#---------#---------#---------#--------- # main # let's rock. open the logfile # examine # add up the stats # display the results # # get stuff $/ = "\n\n"; # change from line reading mode to paragraph reading mode oreach $log ( @logs ) { print STDERR "\$log $log \n" if $DEBUG; my $handle = &open_log ("$log"); my %conns; while (<$handle>) { $tally{total_records}++; &ldif2hash ( $_, \%tally, \%records); } close $handle or die "can't close file $log: $!"; } # foreach $/ = "\n"; # change from paragraph reading mode to line reading mode # # put stuff my $dels = &hash2txt (\%tally, \%records); print STDOUT $dels; # # report my $rep = &report (\%tally, @logs); print STDERR $rep; #---------#---------#---------#---------#---------#---------#---------#--------- # subs # all subroutines go below here. # #---------#---------#---------#---------#---------#---------#---------#--------- # subroutine # hash2txt # description # sub hash2txt { my $tally_ptr = shift; # counts for lines read, files examined, etc. my $rec_ptr = shift; my $key; my $txt; # list of delete LDIFs my $dn; # LDAP's idea of a key my $new_attribute; # what the old attributes get translated to my $value; # value to copy # # agent labels. Each one must be checked. my $val_to_check; # current value being checked my $val_in_key; # value in record key my @vals_to_keep = qw( 0000 0212 0402 0558 0573 0723 0727 K481 K484 K486 K498 K508 K519 K541 K547 K635 K654 K674 M035 M051 M059 M070 M071 M073 M088 NQ029 WT001 ); my $val_found; # one of the vals_to_keep # work through the whole hash oreach $key (sort keys %$rec_ptr) { print STDERR "checking \$key $key \n" if $DEBUG; # # skip certain records # # does the key contain the text # "ou=XXX,ou=agent,o=isp01.co.uk"? unless ($key =~ m/ ou=(.....),\s* # Note the ou label. ou=[Ii]ntermediary,\s* # Match the fields, commas, o=pp # and any spaces seperating them. /x) { print STDERR "keep $key\n" if $VERBOSE; next; }; # unless $val_in_key = uc ($1); # # does a certain field contain one value? if ($rec_ptr->{$key}{destinationindicator}) { if ($rec_ptr->{$key}{destinationindicator} =~ m/International|Both/i) { print STDERR "uid $val_in_key has value 'International' or 'Both'\n" if $VERBOSE; next; }; }; # # does the key contain one of a range of values? $val_found = ''; oreach $val_to_check (@vals_to_keep) { #print STDERR "\$val_to_check $val_to_check\n" if $DEBUG; if ( $val_to_check eq $val_in_key ) { print STDERR "keep $key\n" if $VERBOSE; $val_found = $key; last; } }; # foreach next if ($val_found); # found one of our keepers # print STDERR "delete $key\n" if $VERBOSE; $tally_ptr->{del_records}++; $txt .= "dn: $key changetype: delete "; } # foreach # return the new hash return $txt; } # #---------#---------#---------#---------#---------#---------#---------#--------- # subroutine # open_log # description # sub open_log { my $file = shift; die "empty file: $file" if -z $file; die "not a text file: $file" unless -T $file; open( IN, "<$file") or die "can't open $file: $!"; return *IN; } # #---------#---------#---------#---------#---------#---------#---------#--------- # subroutine # ldif2hash # description # Take a multiline ldap record. # Copy attributes and values from this record using regular expressions. # Regex flags are # c (continue past end of line characters), # m (treat it as a multiline record by matching start and end of each line) # and # x (extended format, so we can use whitespace and comments in the regex). # # people records start with lines like this # # entry-id: 25 # dn: uid=john smith,ou=people,o=isp01.co.uk # ... # and have entries like this # sn: SMITH # ... sub ldif2hash { my $record = shift; # multiline record my $tally_ptr = shift; # counts for lines read, files examined, etc. my $records_ptr = shift; # processed records # # extract the record key $record =~ m/ dn:\s* # Match "dn:", then any spaces, (.*) # then everything to the end of the line /x; my $key = $1; # # copy fields from the LDIF to the hash # # BUGS # # PROBLEMS WITH # - \s*$ IF NO VALUE, MISSES THE END OF LINE AND MATCHES THE NEXT LINE # - \s*(.*)$ NOT REMOVING LEADING SPACES FROM VALUE # # DOES NOT MATCH MULTILINE VALUE CORRECTLY # eg. these two fields won't be split properly # teamdescription: The e-commerce team look after the web application # hosting for the UK # mail: john.smith@isp01-tech.com # # would Grahame Barr's Net::LDAP::LDIF be better? # $tally_ptr->{match_records}++; # how many records parsed so far my $field; my $attr; # lots of "attribute: value" in LDAP land. my $value; while ($record =~ m/ ^ # from the start of line, find ([^:]*) # attribute (anything that isn't a colon), : # first colon (the inter-field seperator), (.*) # and the value is everything else $ # to the end of the line /gmox) { $attr = $1; $value = &trim ($2); $records_ptr->{$key}{$attr} = $value; # add an entry to the hash print STDERR "\$records_ptr->{\$key}{\$attr} = \$value ", "$records_ptr ->{ $key }{ $attr } = $value \n" if $DEBUG; } # while } # #---------#---------#---------#---------#---------#---------#---------#--------- # subroutine # trim # description # see perl cookbook p.30 sub trim { my @out = @_; or (@out) { # print STDERR "trim before >>>$_<<< \n"; s/^\s+//; s/\s+$//; # print STDERR "trim after >>>$_<<< \n"; } return wantarray ? @out : $out[0]; } #---------#---------#---------#---------#---------#---------#---------#--------- # subroutine # report # description # Create a report and shove it in a scalar. # note the presence of one of those weird # numeric value sort criteria operators "<=>" # and a format string "%this %that %the-other" # sub report { my $tally_ptr = shift; # counts of everything else eg. lines read my @logs = @_; my $today = scalar localtime; # date eg.Tue Apr 17 11:29:19 2001 # host use Sys::Hostname; my $host = hostname(); # the machine this script runs on my $report = "\n"; # the scalar to hold the final report my $threshold = 20; # number of top records to report # add some pretty commas # this should be done inline, but I have some format problems creeping in my $tally_logs = &commas($tally_ptr->{logs}); my $tally_del_records = &commas($tally_ptr->{del_records}); my $tally_match_records = &commas($tally_ptr->{match_records}); # report header $report = <