3 # Turn a wordlist into a CDB or SQLite database.
5 # This program takes as input a word list (a file of words separated by
6 # newlines) and turns it into either a CDB or a SQLite database that can be
7 # used by the krb5-strength plugin or heimdal-strength program to check
8 # passwords against a password dictionary. It can also filter a word list in
9 # various ways to create a new word list.
11 ##############################################################################
12 # Declarations and configuration
13 ##############################################################################
19 use File::Basename qw(basename);
20 use Getopt::Long qw(GetOptions);
22 # The path to the cdb utility, used to create the final database. By default,
23 # the user's PATH is searched for cdb.
26 # The SQL used to create the SQLite database.
27 ## no critic (ValuesAndExpressions::ProhibitImplicitNewlines)
28 my $SQLITE_CREATE = q{
29 CREATE TABLE passwords (
30 password TEXT UNIQUE NOT NULL,
31 drowssap TEXT UNIQUE NOT NULL
35 # The SQL used to insert passwords into the database.
36 my $SQLITE_INSERT = q{
37 INSERT OR IGNORE INTO passwords (password, drowssap) values (?, ?)
41 ##############################################################################
43 ##############################################################################
45 # print with error checking and an explicit file handle.
47 # $fh - Output file handle
48 # @args - Remaining arguments to print
51 # Throws: Text exception on output failure
54 print {$fh} @args or croak('print failed');
58 ##############################################################################
60 ##############################################################################
62 # Filter the given input file and write it to a CDB data file, and then use
63 # cdb to turn that into a database.
65 # $in_fh - Input file handle for the source wordlist
66 # $output - Name of the output CDB file
67 # $filter - Reference to sub that returns true to keep a word, false otherwise
70 # Throws: Text exception on output failure or pre-existing temporary file
72 my ($in_fh, $output, $filter) = @_;
74 # Check that the output CDB file doesn't exist.
76 die "$0: output file $output already exists\n";
79 # Create a temporary file to write the CDB input into.
80 my $tmp = $output . '.data';
82 die "$0: temporary output file $tmp already exists\n";
84 open(my $tmp_fh, '>', $tmp)
85 or die "$0: cannot create output file $tmp: $!\n";
87 # Walk through the input word list and write each word that passes the
88 # filter to the output file handle as CDB data.
89 while (defined(my $word = <$in_fh>)) {
91 next if !$filter->($word);
92 my $length = length($word);
93 print_fh($tmp_fh, "+$length,1:$word->1\n");
96 # Add a trailing newline, required by the CDB data format, and close.
97 print_fh($tmp_fh, "\n");
98 close($tmp_fh) or die "$0: cannot write to temporary file $tmp: $!\n";
100 # Run cdb to turn the result into a CDB database. Ignore duplicate keys.
101 system($CDB, '-c', '-u', $output, $tmp) == 0
102 or die "$0: cdb -c failed\n";
104 # Remove the temporary file and return.
105 unlink($tmp) or die "$0: cannot remove temporary file $tmp: $!\n";
109 # Filter the given input file and write it to a newly-created SQLite database.
110 # Requires the DBI and DBD::SQLite modules be installed. The database will
111 # contain one table, passwords, with two columns, password and drowssap, which
112 # store the word and the word reversed for each word that passes the filter.
114 # $in_fh - Input file handle for the source wordlist
115 # $output - Name of the output SQLite database
116 # $filter - Reference to sub that returns true to keep a word, false otherwise
119 # Throws: Text exception on output failure, pre-existing output file, or
120 # missing Perl modules
122 my ($in_fh, $output, $filter) = @_;
124 # Check that the output SQLite file doesn't exist.
126 die "$0: output file $output already exists\n";
129 # Load the required modules.
133 # Open and create the database.
134 my $options = { PrintError => 0, RaiseError => 1, AutoCommit => 1 };
135 my $dbh = DBI->connect("dbi:SQLite:dbname=$output", q{}, q{}, $options);
136 $dbh->do($SQLITE_CREATE);
138 # Tune SQLite to improve the speed of bulk inserts. Use unsafe insert
139 # processing and increase the index cache to 500MB.
140 $dbh->do('PRAGMA synchronous = 0');
141 $dbh->do('PRAGMA cache_size = 500000');
143 # Start a transaction and prepare the insert statement for each word.
145 my $sth = $dbh->prepare($SQLITE_INSERT);
147 # Walk through the input word list and add each word that passes the
148 # filter to the database, both as-is and reversed.
149 while (defined(my $word = <$in_fh>)) {
151 next if !$filter->($word);
152 my $reversed = reverse($word);
153 $sth->execute($word, $reversed);
156 # Commit and close the database.
162 # Filter the given input file and write the results to a new wordlist.
164 # $in_fh - Input file handle for the source wordlist
165 # $output - Output file name to which to write the resulting wordlist
166 # $filter - Reference to sub that returns true to keep a word, false otherwise
169 # Throws: Text exception on output failure
171 my ($in_fh, $output, $filter) = @_;
172 open(my $out_fh, '>', $output)
173 or die "$0: cannot create output file $output: $!\n";
175 # Walk through the input word list and write each word that passes the
176 # filter to the output file handle.
177 while (defined(my $word = <$in_fh>)) {
179 next if !$filter->($word);
180 print_fh($out_fh, "$word\n");
184 close($out_fh) or die "$0: cannot write to output file $output: $!\n";
188 ##############################################################################
190 ##############################################################################
192 # Given the parsed command-line options as a hash, construct a filter for the
193 # word list and return it. The filter will, given a word, return true if the
194 # word should be included in the dictionary and false otherwise.
196 # $config_ref - Hash of configuration options
197 # ascii - Strip non-printable or non-ASCII words
198 # exclude - Reference to array of regex patterns to exclude
199 # min_length - Minimum word length
200 # max_length - Maximum word length
202 # Returns: Filter function to check a word.
204 my ($config_ref) = @_;
206 # Build a filter from our command-line parameters. This is an anonymous
207 # sub that returns true to keep a word and false otherwise.
210 my $length = length($word);
211 my $min_length = $config_ref->{'min-length'};
212 my $max_length = $config_ref->{'max-length'};
215 return if (defined($min_length) && $length < $min_length);
216 return if (defined($max_length) && $length > $max_length);
218 # Check character classes.
219 if ($config_ref->{ascii}) {
220 return if $word =~ m{ [^[:ascii:]] }xms;
221 return if $word =~ m{ [[:cntrl:]] }xms;
224 # Check regex exclusions.
225 if ($config_ref->{exclude}) {
226 for my $pattern (@{ $config_ref->{exclude} }) {
227 return if $word =~ m{ $pattern }xms;
231 # Word passes. Return success.
237 ##############################################################################
239 ##############################################################################
241 # Always flush output.
244 # Clean up the script name for error reporting.
246 local $0 = basename($0);
248 # Parse the argument list.
251 'ascii|a', 'cdb|c=s', 'max-length|L=i', 'min-length|l=i',
252 'manual|man|m', 'output|o=s', 'sqlite|s=s', 'exclude|x=s@',
254 Getopt::Long::config('bundling', 'no_ignore_case');
255 GetOptions(\%config, @options);
256 if ($config{manual}) {
257 print_fh(\*STDOUT, "Feeding myself to perldoc, please wait...\n");
258 exec('perldoc', '-t', $fullpath);
261 die "Usage: krb5-strength-wordlist <wordlist>\n";
263 if ($config{cdb} && ($config{output} || $config{sqlite})) {
264 die "$0: -c cannot be used with -o or -s\n";
265 } elsif ($config{output} && $config{sqlite}) {
266 die "$0: -o cannot be used with -c or -s\n";
268 my $input = $ARGV[0];
270 # Build the filter closure.
271 my $filter = build_filter(\%config);
273 # Process the input file into either wordlist output or a CDB file.
274 open(my $in_fh, '<', $input)
275 or die "$0: cannot open input file $input: $!\n";
276 if ($config{output}) {
277 write_wordlist($in_fh, $config{output}, $filter);
278 } elsif ($config{cdb}) {
279 write_cdb($in_fh, $config{cdb}, $filter);
280 } elsif ($config{sqlite}) {
281 write_sqlite($in_fh, $config{sqlite}, $filter);
283 close($in_fh) or die "$0: cannot read all of input file $input: $!\n";
289 ##############################################################################
291 ##############################################################################
294 krb5-strength-wordlist krb5-strength cdb whitespace lookups lookup
295 sublicense MERCHANTABILITY NONINFRINGEMENT krb5-strength --ascii Allbery
296 regexes output-wordlist heimdal-strength SQLite output-wordlist
297 output-sqlite DBI wordlist
301 krb5-strength-wordlist - Create a krb5-strength database from a word list
305 B<krb5-strength-wordlist> [B<-am>] [B<-c> I<output-cdb>] [B<-l> I<min-length>]
306 [B<-L> I<max-length>] [B<-o> I<output-wordlist>] [B<-s> I<output-sqlite>]
307 [B<-x> I<exclude> ...] I<wordlist>
311 B<krb5-strength-wordlist> converts a word list (a file containing one word
312 per line) into a database that can be used by the krb5-strength plugin or
313 B<heimdal-strength> command for checking passwords. Two database formats
314 are supported, with different features. CDB is more space-efficient and
315 possibly faster, but supports checking passwords only against exact
316 matches or simple transformations (removing small numbers of leading and
317 trailing characters). SQLite creates a much larger database, but supports
318 rejecting any password within edit distance one of a word in the word
321 CDB is a format invented by Dan Bernstein for fast, constant databases.
322 The database is fixed during creation and cannot be changed without
323 rebuilding it, and is optimized for very fast access. For cdb, the
324 database generated by this program will have keys for each word in the
325 word list and the constant C<1> as the value.
327 SQLite stores the word list in a single table containing both each word
328 and each word reversed. This allows the krb5-strength plugin or
329 B<heimdal-strength> command to reject passwords within edit distance one
330 of any word in the word list. (Edit distance one means that the word list
331 entry can be formed by changing a single character of the password, either
332 by adding one character, removing one character, or changing one character
333 to a different character.) However, the SQLite database will be much
334 larger and lookups may be somewhat slower.
336 B<krb5-strength-wordlist> takes one argument, the input word list file.
337 Use the B<-c> option to specify an output CDB file, B<-s> to specify an
338 output SQLite file, or B<-o> to just filter the word list against the
339 criteria given on the command line and generate a new word list.
340 The input word list file does not have to be sorted. See the individual
341 option descriptions for more information.
347 =item B<-a>, B<--ascii>
349 Filter all words that contain non-ASCII characters or control characters
350 from the resulting cdb file, leaving only words that consist solely of
351 ASCII non-control characters.
353 =item B<-c> I<output-cdb>, B<--cdb>=I<output-cdb>
355 Create a CDB database in I<output-cdb>. A temporary file named after
356 I<output-cdb> with C<.data> appended will be created in the same directory
357 and used to stage the database contents. The actual CDB file will be
358 built using the B<cdb> command, which must be on the user's path. If
359 either file already exists, B<krb5-strength-wordlist> will abort with an
362 This option cannot be used with B<-o> or B<-s>.
364 =item B<-L> I<maximum>, B<--max-length>=I<maximum>
366 Filter all words of length greater than I<maximum> from the resulting cdb
367 database. The length of each line (minus the separating newline) in the
368 input word list will be checked against I<minimum> and will be filtered
369 out of the resulting database if it is shorter. Useful for generating
370 password dictionaries from word lists that contain random noise that's
371 highly unlikely to be used as a password.
373 The default is to not filter out any words for maximum length.
375 =item B<-l> I<minimum>, B<--min-length>=I<minimum>
377 Filter all words of length less than I<minimum> from the resulting cdb
378 database. The length of each line (minus the separating newline) in the
379 input word list will be checked against I<minimum> and will be filtered
380 out of the resulting database if it is shorter. Useful for generating
381 password dictionaries where shorter passwords will be rejected by a
382 generic length check and no dictionary lookup will be done for a transform
383 of the password shorter than the specified minimum.
385 The default is not to filter out any words for minimum length.
387 =item B<-m>, B<--man>, B<--manual>
389 Print out this documentation (which is done simply by feeding the script to
392 =item B<-o> I<wordlist>, B<--output>=I<wordlist>
394 Rather than creating a database, apply the filter rules given by the other
395 command-line arguments and generate a new word list in the file name given
396 by the I<wordlist> option. This can be used to reduce the size of a raw
397 word list file (such as one taken from Internet sources) by removing the
398 words that will be filtered out of the dictionary anyway, thus reducing
399 the size of the source required to regenerate the dictionary.
401 This option cannot be used with B<-c> or B<-s>.
403 =item B<-s> I<output-sqlite>, B<--sqlite>=I<output-sqlite>
405 Create a SQLite database in I<output-sqlite>. If this file already
406 exists, B<krb5-strength-wordlist> will abort with an error. The resulting
407 SQLite database will have one table, C<passwords>, with two columns,
408 C<password> and C<drowssap>. The first holds a word from the word list,
409 and the second holds the same word reversed.
411 Using this option requires the DBI and DBD::SQLite Perl modules be
414 This option cannot be used with B<-c> or B<-o>.
416 =item B<-x> I<exclude>, B<--exclude>=I<exclude>
418 Filter all words matching the regular expression I<exclude> from the
419 resulting cdb database. This regular expression will be matched against
420 each line of the source word list after the trailing newline is removed.
421 This option may be given repeatedly to add multiple exclusion regexes.
427 Russ Allbery <eagle@eyrie.org>
429 =head1 COPYRIGHT AND LICENSE
431 Copyright 2016 Russ Allbery <eagle@eyrie.org>
433 Copyright 2013, 2014 The Board of Trustees of the Leland Stanford Junior
436 Permission is hereby granted, free of charge, to any person obtaining a
437 copy of this software and associated documentation files (the "Software"),
438 to deal in the Software without restriction, including without limitation
439 the rights to use, copy, modify, merge, publish, distribute, sublicense,
440 and/or sell copies of the Software, and to permit persons to whom the
441 Software is furnished to do so, subject to the following conditions:
443 The above copyright notice and this permission notice shall be included in
444 all copies or substantial portions of the Software.
446 THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
447 IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
448 FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
449 THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
450 LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
451 FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
452 DEALINGS IN THE SOFTWARE.
456 cdb(1), L<DBI>, L<DBD::SQLite>
458 The cdb file format is defined at L<http://cr.yp.to/cdb.html>.
460 The current version of this program is available from its web page at
461 L<https://www.eyrie.org/~eagle/software/krb5-strength/> as part of the
462 krb5-strength package.