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 => 0 };
135 my $dbh = DBI->connect("dbi:SQLite:dbname=$output", q{}, q{}, $options);
136 $dbh->do($SQLITE_CREATE);
138 # Prepare the insert statement for each word.
139 my $sth = $dbh->prepare($SQLITE_INSERT);
141 # Walk through the input word list and add each word that passes the
142 # filter to the database, both as-is and reversed.
143 while (defined(my $word = <$in_fh>)) {
145 next if !$filter->($word);
146 my $reversed = reverse($word);
147 $sth->execute($word, $reversed);
150 # Commit and close the database.
156 # Filter the given input file and write the results to a new wordlist.
158 # $in_fh - Input file handle for the source wordlist
159 # $output - Output file name to which to write the resulting wordlist
160 # $filter - Reference to sub that returns true to keep a word, false otherwise
163 # Throws: Text exception on output failure
165 my ($in_fh, $output, $filter) = @_;
166 open(my $out_fh, '>', $output)
167 or die "$0: cannot create output file $output: $!\n";
169 # Walk through the input word list and write each word that passes the
170 # filter to the output file handle.
171 while (defined(my $word = <$in_fh>)) {
173 next if !$filter->($word);
174 print_fh($out_fh, "$word\n");
178 close($out_fh) or die "$0: cannot write to output file $output: $!\n";
182 ##############################################################################
184 ##############################################################################
186 # Given the parsed command-line options as a hash, construct a filter for the
187 # word list and return it. The filter will, given a word, return true if the
188 # word should be included in the dictionary and false otherwise.
190 # $config_ref - Hash of configuration options
191 # ascii - Strip non-printable or non-ASCII words
192 # exclude - Reference to array of regex patterns to exclude
193 # min_length - Minimum word length
194 # max_length - Maximum word length
196 # Returns: Filter function to check a word.
198 my ($config_ref) = @_;
200 # Build a filter from our command-line parameters. This is an anonymous
201 # sub that returns true to keep a word and false otherwise.
204 my $length = length($word);
205 my $min_length = $config_ref->{'min-length'};
206 my $max_length = $config_ref->{'max-length'};
209 return if (defined($min_length) && $length < $min_length);
210 return if (defined($max_length) && $length > $max_length);
212 # Check character classes.
213 if ($config_ref->{ascii}) {
214 return if $word =~ m{ [^[:ascii:]] }xms;
215 return if $word =~ m{ [[:cntrl:]] }xms;
218 # Check regex exclusions.
219 if ($config_ref->{exclude}) {
220 for my $pattern (@{ $config_ref->{exclude} }) {
221 return if $word =~ m{ $pattern }xms;
225 # Word passes. Return success.
231 ##############################################################################
233 ##############################################################################
235 # Always flush output.
238 # Clean up the script name for error reporting.
240 local $0 = basename($0);
242 # Parse the argument list.
245 'ascii|a', 'cdb|c=s', 'max-length|L=i', 'min-length|l=i',
246 'manual|man|m', 'output|o=s', 'sqlite|s=s', 'exclude|x=s@',
248 Getopt::Long::config('bundling', 'no_ignore_case');
249 GetOptions(\%config, @options);
250 if ($config{manual}) {
251 print_fh(\*STDOUT, "Feeding myself to perldoc, please wait...\n");
252 exec('perldoc', '-t', $fullpath);
255 die "Usage: krb5-strength-wordlist <wordlist>\n";
257 if ($config{cdb} && ($config{output} || $config{sqlite})) {
258 die "$0: -c cannot be used with -o or -s\n";
259 } elsif ($config{output} && $config{sqlite}) {
260 die "$0: -o cannot be used with -c or -s\n";
262 my $input = $ARGV[0];
264 # Build the filter closure.
265 my $filter = build_filter(\%config);
267 # Process the input file into either wordlist output or a CDB file.
268 open(my $in_fh, '<', $input)
269 or die "$0: cannot open input file $input: $!\n";
270 if ($config{output}) {
271 write_wordlist($in_fh, $config{output}, $filter);
272 } elsif ($config{cdb}) {
273 write_cdb($in_fh, $config{cdb}, $filter);
274 } elsif ($config{sqlite}) {
275 write_sqlite($in_fh, $config{sqlite}, $filter);
277 close($in_fh) or die "$0: cannot read all of input file $input: $!\n";
283 ##############################################################################
285 ##############################################################################
288 krb5-strength-wordlist krb5-strength cdb whitespace lookups lookup
289 sublicense MERCHANTABILITY NONINFRINGEMENT krb5-strength --ascii Allbery
290 regexes output-wordlist heimdal-strength SQLite output-wordlist
291 output-sqlite DBI wordlist
295 krb5-strength-wordlist - Create a krb5-strength database from a word list
299 B<krb5-strength-wordlist> [B<-am>] [B<-c> I<output-cdb>] [B<-l> I<min-length>]
300 [B<-L> I<max-length>] [B<-o> I<output-wordlist>] [B<-s> I<output-sqlite>]
301 [B<-x> I<exclude> ...] I<wordlist>
305 B<krb5-strength-wordlist> converts a word list (a file containing one word
306 per line) into a database that can be used by the krb5-strength plugin or
307 B<heimdal-strength> command for checking passwords. Two database formats
308 are supported, with different features. CDB is more space-efficient and
309 possibly faster, but supports checking passwords only against exact
310 matches or simple transformations (removing small numbers of leading and
311 trailing characters). SQLite creates a much larger database, but supports
312 rejecting any password within edit distance one of a word in the word
315 CDB is a format invented by Dan Bernstein for fast, constant databases.
316 The database is fixed during creation and cannot be changed without
317 rebuilding it, and is optimized for very fast access. For cdb, the
318 database generated by this program will have keys for each word in the
319 word list and the constant C<1> as the value.
321 SQLite stores the word list in a single table containing both each word
322 and each word reversed. This allows the krb5-strength plugin or
323 B<heimdal-strength> command to reject passwords within edit distance one
324 of any word in the word list. (Edit distance one means that the word list
325 entry can be formed by changing a single character of the password, either
326 by adding one character, removing one character, or changing one character
327 to a different character.) However, the SQLite database will be much
328 larger and lookups may be somewhat slower.
330 B<krb5-strength-wordlist> takes one argument, the input word list file.
331 Use the B<-c> option to specify an output CDB file, B<-s> to specify an
332 output SQLite file, or B<-o> to just filter the word list against the
333 criteria given on the command line and generate a new word list.
334 The input word list file does not have to be sorted. See the individual
335 option descriptions for more information.
341 =item B<-a>, B<--ascii>
343 Filter all words that contain non-ASCII characters or control characters
344 from the resulting cdb file, leaving only words that consist solely of
345 ASCII non-control characters.
347 =item B<-c> I<output-cdb>, B<--cdb>=I<output-cdb>
349 Create a CDB database in I<output-cdb>. A temporary file named after
350 I<output-cdb> with C<.data> appended will be created in the same directory
351 and used to stage the database contents. The actual CDB file will be
352 built using the B<cdb> command, which must be on the user's path. If
353 either file already exists, B<krb5-strength-wordlist> will abort with an
356 This option cannot be used with B<-o> or B<-s>.
358 =item B<-L> I<maximum>, B<--max-length>=I<maximum>
360 Filter all words of length greater than I<maximum> from the resulting cdb
361 database. The length of each line (minus the separating newline) in the
362 input word list will be checked against I<minimum> and will be filtered
363 out of the resulting database if it is shorter. Useful for generating
364 password dictionaries from word lists that contain random noise that's
365 highly unlikely to be used as a password.
367 The default is to not filter out any words for maximum length.
369 =item B<-l> I<minimum>, B<--min-length>=I<minimum>
371 Filter all words of length less than I<minimum> from the resulting cdb
372 database. The length of each line (minus the separating newline) in the
373 input word list will be checked against I<minimum> and will be filtered
374 out of the resulting database if it is shorter. Useful for generating
375 password dictionaries where shorter passwords will be rejected by a
376 generic length check and no dictionary lookup will be done for a transform
377 of the password shorter than the specified minimum.
379 The default is not to filter out any words for minimum length.
381 =item B<-m>, B<--man>, B<--manual>
383 Print out this documentation (which is done simply by feeding the script to
386 =item B<-o> I<wordlist>, B<--output>=I<wordlist>
388 Rather than creating a database, apply the filter rules given by the other
389 command-line arguments and generate a new word list in the file name given
390 by the I<wordlist> option. This can be used to reduce the size of a raw
391 word list file (such as one taken from Internet sources) by removing the
392 words that will be filtered out of the dictionary anyway, thus reducing
393 the size of the source required to regenerate the dictionary.
395 This option cannot be used with B<-c> or B<-s>.
397 =item B<-s> I<output-sqlite>, B<--sqlite>=I<output-sqlite>
399 Create a SQLite database in I<output-sqlite>. If this file already
400 exists, B<krb5-strength-wordlist> will abort with an error. The resulting
401 SQLite database will have one table, C<passwords>, with two columns,
402 C<password> and C<drowssap>. The first holds a word from the word list,
403 and the second holds the same word reversed.
405 Using this option requires the DBI and DBD::SQLite Perl modules be
408 This option cannot be used with B<-c> or B<-o>.
410 =item B<-x> I<exclude>, B<--exclude>=I<exclude>
412 Filter all words matching the regular expression I<exclude> from the
413 resulting cdb database. This regular expression will be matched against
414 each line of the source word list after the trailing newline is removed.
415 This option may be given repeatedly to add multiple exclusion regexes.
421 Russ Allbery <eagle@eyrie.org>
423 =head1 COPYRIGHT AND LICENSE
425 Copyright 2013, 2014 The Board of Trustees of the Leland Stanford Junior
428 Permission is hereby granted, free of charge, to any person obtaining a
429 copy of this software and associated documentation files (the "Software"),
430 to deal in the Software without restriction, including without limitation
431 the rights to use, copy, modify, merge, publish, distribute, sublicense,
432 and/or sell copies of the Software, and to permit persons to whom the
433 Software is furnished to do so, subject to the following conditions:
435 The above copyright notice and this permission notice shall be included in
436 all copies or substantial portions of the Software.
438 THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
439 IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
440 FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
441 THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
442 LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
443 FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
444 DEALINGS IN THE SOFTWARE.
448 cdb(1), L<DBI>, L<DBD::SQLite>
450 The cdb file format is defined at L<http://cr.yp.to/cdb.html>.
452 The current version of this program is available from its web page at
453 L<http://www.eyrie.org/~eagle/software/krb5-strength/> as part of the
454 krb5-strength package.