3 perlfilter - Source Filters
7 This article is about a little-known feature of Perl called
8 I<source filters>. Source filters alter the program text of a module
9 before Perl sees it, much as a C preprocessor alters the source text of
10 a C program before the compiler sees it. This article tells you more
11 about what source filters are, how they work, and how to write your
14 The original purpose of source filters was to let you encrypt your
15 program source to prevent casual piracy. This isn't all they can do, as
16 you'll soon learn. But first, the basics.
20 Before the Perl interpreter can execute a Perl script, it must first
21 read it from a file into memory for parsing and compilation. If that
22 script itself includes other scripts with a C<use> or C<require>
23 statement, then each of those scripts will have to be read from their
24 respective files as well.
26 Now think of each logical connection between the Perl parser and an
27 individual file as a I<source stream>. A source stream is created when
28 the Perl parser opens a file, it continues to exist as the source code
29 is read into memory, and it is destroyed when Perl is finished parsing
30 the file. If the parser encounters a C<require> or C<use> statement in
31 a source stream, a new and distinct stream is created just for that
34 The diagram below represents a single source stream, with the flow of
35 source from a Perl script file on the left into the Perl parser on the
36 right. This is how Perl normally operates.
40 There are two important points to remember:
46 Although there can be any number of source streams in existence at any
47 given time, only one will be active.
51 Every source stream is associated with only one file.
55 A source filter is a special kind of Perl module that intercepts and
56 modifies a source stream before it reaches the parser. A source filter
57 changes our diagram like this:
59 file ----> filter ----> parser
61 If that doesn't make much sense, consider the analogy of a command
62 pipeline. Say you have a shell script stored in the compressed file
63 I<trial.gz>. The simple pipeline command below runs the script without
64 needing to create a temporary file to hold the uncompressed file.
66 gunzip -c trial.gz | sh
68 In this case, the data flow from the pipeline can be represented as follows:
70 trial.gz ----> gunzip ----> sh
72 With source filters, you can store the text of your script compressed and use a source filter to uncompress it for Perl's parser:
75 Perl program ---> source filter ---> parser
79 So how do you use a source filter in a Perl script? Above, I said that
80 a source filter is just a special kind of module. Like all Perl
81 modules, a source filter is invoked with a use statement.
83 Say you want to pass your Perl source through the C preprocessor before
84 execution. As it happens, the source filters distribution comes with a C
85 preprocessor filter module called Filter::cpp.
87 Below is an example program, C<cpp_test>, which makes use of this filter.
88 Line numbers have been added to allow specific lines to be referenced
96 When you execute this script, Perl creates a source stream for the
97 file. Before the parser processes any of the lines from the file, the
98 source stream looks like this:
100 cpp_test ---------> parser
102 Line 1, C<use Filter::cpp>, includes and installs the C<cpp> filter
103 module. All source filters work this way. The use statement is compiled
104 and executed at compile time, before any more of the file is read, and
105 it attaches the cpp filter to the source stream behind the scenes. Now
106 the data flow looks like this:
108 cpp_test ----> cpp filter ----> parser
110 As the parser reads the second and subsequent lines from the source
111 stream, it feeds those lines through the C<cpp> source filter before
112 processing them. The C<cpp> filter simply passes each line through the
113 real C preprocessor. The output from the C preprocessor is then
114 inserted back into the source stream by the filter.
120 cpp_test ----> cpp filter ----> parser
122 The parser then sees the following code:
128 Let's consider what happens when the filtered code includes another
137 The C<cpp> filter does not apply to the text of the Fred module, only
138 to the text of the file that used it (C<cpp_test>). Although the use
139 statement on line 3 will pass through the cpp filter, the module that
140 gets included (C<Fred>) will not. The source streams look like this
141 after line 3 has been parsed and before line 4 is parsed:
143 cpp_test ---> cpp filter ---> parser (INACTIVE)
147 As you can see, a new stream has been created for reading the source
148 from C<Fred.pm>. This stream will remain active until all of C<Fred.pm>
149 has been parsed. The source stream for C<cpp_test> will still exist,
150 but is inactive. Once the parser has finished reading Fred.pm, the
151 source stream associated with it will be destroyed. The source stream
152 for C<cpp_test> then becomes active again and the parser reads line 4
153 and subsequent lines from C<cpp_test>.
155 You can use more than one source filter on a single file. Similarly,
156 you can reuse the same filter in as many files as you like.
158 For example, if you have a uuencoded and compressed source file, it is
159 possible to stack a uudecode filter and an uncompression filter like
162 use Filter::uudecode; use Filter::uncompress;
163 M'XL(".H<US4''V9I;F%L')Q;>7/;1I;_>_I3=&E=%:F*I"T?22Q/
164 M6]9*<IQCO*XFT"0[PL%%'Y+IG?WN^ZYN-$'J.[.JE$,20/?K=_[>
167 Once the first line has been processed, the flow will look like this:
169 file ---> uudecode ---> uncompress ---> parser
172 Data flows through filters in the same order they appear in the source
173 file. The uudecode filter appeared before the uncompress filter, so the
174 source file will be uudecoded before it's uncompressed.
176 =head1 WRITING A SOURCE FILTER
178 There are three ways to write your own source filter. You can write it
179 in C, use an external program as a filter, or write the filter in Perl.
180 I won't cover the first two in any great detail, so I'll get them out
181 of the way first. Writing the filter in Perl is most convenient, so
182 I'll devote the most space to it.
184 =head1 WRITING A SOURCE FILTER IN C
186 The first of the three available techniques is to write the filter
187 completely in C. The external module you create interfaces directly
188 with the source filter hooks provided by Perl.
190 The advantage of this technique is that you have complete control over
191 the implementation of your filter. The big disadvantage is the
192 increased complexity required to write the filter - not only do you
193 need to understand the source filter hooks, but you also need a
194 reasonable knowledge of Perl guts. One of the few times it is worth
195 going to this trouble is when writing a source scrambler. The
196 C<decrypt> filter (which unscrambles the source before Perl parses it)
197 included with the source filter distribution is an example of a C
198 source filter (see Decryption Filters, below).
203 =item B<Decryption Filters>
205 All decryption filters work on the principle of "security through
206 obscurity." Regardless of how well you write a decryption filter and
207 how strong your encryption algorithm is, anyone determined enough can
208 retrieve the original source code. The reason is quite simple - once
209 the decryption filter has decrypted the source back to its original
210 form, fragments of it will be stored in the computer's memory as Perl
211 parses it. The source might only be in memory for a short period of
212 time, but anyone possessing a debugger, skill, and lots of patience can
213 eventually reconstruct your program.
215 That said, there are a number of steps that can be taken to make life
216 difficult for the potential cracker. The most important: Write your
217 decryption filter in C and statically link the decryption module into
218 the Perl binary. For further tips to make life difficult for the
219 potential cracker, see the file I<decrypt.pm> in the source filters
224 =head1 CREATING A SOURCE FILTER AS A SEPARATE EXECUTABLE
226 An alternative to writing the filter in C is to create a separate
227 executable in the language of your choice. The separate executable
228 reads from standard input, does whatever processing is necessary, and
229 writes the filtered data to standard output. C<Filter::cpp> is an
230 example of a source filter implemented as a separate executable - the
231 executable is the C preprocessor bundled with your C compiler.
233 The source filter distribution includes two modules that simplify this
234 task: C<Filter::exec> and C<Filter::sh>. Both allow you to run any
235 external executable. Both use a coprocess to control the flow of data
236 into and out of the external executable. (For details on coprocesses,
237 see Stephens, W.R., "Advanced Programming in the UNIX Environment."
238 Addison-Wesley, ISBN 0-210-56317-7, pages 441-445.) The difference
239 between them is that C<Filter::exec> spawns the external command
240 directly, while C<Filter::sh> spawns a shell to execute the external
241 command. (Unix uses the Bourne shell; NT uses the cmd shell.) Spawning
242 a shell allows you to make use of the shell metacharacters and
243 redirection facilities.
245 Here is an example script that uses C<Filter::sh>:
247 use Filter::sh 'tr XYZ PQR';
249 print "XYZ a = $a\n";
251 The output you'll get when the script is executed:
255 Writing a source filter as a separate executable works fine, but a
256 small performance penalty is incurred. For example, if you execute the
257 small example above, a separate subprocess will be created to run the
258 Unix C<tr> command. Each use of the filter requires its own subprocess.
259 If creating subprocesses is expensive on your system, you might want to
260 consider one of the other options for creating source filters.
262 =head1 WRITING A SOURCE FILTER IN PERL
264 The easiest and most portable option available for creating your own
265 source filter is to write it completely in Perl. To distinguish this
266 from the previous two techniques, I'll call it a Perl source filter.
268 To help understand how to write a Perl source filter we need an example
269 to study. Here is a complete source filter that performs rot13
270 decoding. (Rot13 is a very simple encryption scheme used in Usenet
271 postings to hide the contents of offensive posts. It moves every letter
272 forward thirteen places, so that A becomes N, B becomes O, and Z
278 use Filter::Util::Call;
283 filter_add(bless $ref);
290 tr/n-za-mN-ZA-M/a-zA-Z/
291 if ($status = filter_read()) > 0;
297 =for apidoc filter_add
298 =for apidoc filter_read
300 All Perl source filters are implemented as Perl classes and have the
301 same basic structure as the example above.
303 First, we include the C<Filter::Util::Call> module, which exports a
304 number of functions into your filter's namespace. The filter shown
305 above uses two of these functions, C<filter_add()> and
308 Next, we create the filter object and associate it with the source
309 stream by defining the C<import> function. If you know Perl well
310 enough, you know that C<import> is called automatically every time a
311 module is included with a use statement. This makes C<import> the ideal
312 place to both create and install a filter object.
314 In the example filter, the object (C<$ref>) is blessed just like any
315 other Perl object. Our example uses an anonymous array, but this isn't
316 a requirement. Because this example doesn't need to store any context
317 information, we could have used a scalar or hash reference just as
318 well. The next section demonstrates context data.
320 The association between the filter object and the source stream is made
321 with the C<filter_add()> function. This takes a filter object as a
322 parameter (C<$ref> in this case) and installs it in the source stream.
324 Finally, there is the code that actually does the filtering. For this
325 type of Perl source filter, all the filtering is done in a method
326 called C<filter()>. (It is also possible to write a Perl source filter
327 using a closure. See the C<Filter::Util::Call> manual page for more
328 details.) It's called every time the Perl parser needs another line of
329 source to process. The C<filter()> method, in turn, reads lines from
330 the source stream using the C<filter_read()> function.
332 If a line was available from the source stream, C<filter_read()>
333 returns a status value greater than zero and appends the line to C<$_>.
334 A status value of zero indicates end-of-file, less than zero means an
335 error. The filter function itself is expected to return its status in
336 the same way, and put the filtered line it wants written to the source
337 stream in C<$_>. The use of C<$_> accounts for the brevity of most Perl
340 In order to make use of the rot13 filter we need some way of encoding
341 the source file in rot13 format. The script below, C<mkrot13>, does
344 die "usage mkrot13 filename\n" unless @ARGV;
347 open(IN, "<$in") or die "Cannot open file $in: $!\n";
348 open(OUT, ">$out") or die "Cannot open file $out: $!\n";
350 print OUT "use Rot13;\n";
352 tr/a-zA-Z/n-za-mN-ZA-M/;
361 If we encrypt this with C<mkrot13>:
363 print " hello fred \n";
365 the result will be this:
368 cevag "uryyb serq\a";
370 Running it produces this output:
374 =head1 USING CONTEXT: THE DEBUG FILTER
376 The rot13 example was a trivial example. Here's another demonstration
377 that shows off a few more features.
379 Say you wanted to include a lot of debugging code in your Perl script
380 during development, but you didn't want it available in the released
381 product. Source filters offer a solution. In order to keep the example
382 simple, let's say you wanted the debugging output to be controlled by
383 an environment variable, C<DEBUG>. Debugging code is enabled if the
384 variable exists, otherwise it is disabled.
386 Two special marker lines will bracket debugging code, like this:
390 warn "Debug: millennium bug in year $year\n";
394 The filter ensures that Perl parses the code between the <DEBUG_BEGIN>
395 and C<DEBUG_END> markers only when the C<DEBUG> environment variable
396 exists. That means that when C<DEBUG> does exist, the code above
397 should be passed through the filter unchanged. The marker lines can
398 also be passed through as-is, because the Perl parser will see them as
399 comment lines. When C<DEBUG> isn't set, we need a way to disable the
400 debug code. A simple way to achieve that is to convert the lines
401 between the two markers into comments:
405 # warn "Debug: millennium bug in year $year\n";
409 Here is the complete Debug filter:
415 use Filter::Util::Call;
417 use constant TRUE => 1;
418 use constant FALSE => 0;
423 Enabled => defined $ENV{DEBUG},
424 InTraceBlock => FALSE,
425 Filename => (caller)[1],
429 filter_add(bless \%context);
434 my ($message) = shift;
435 my ($line_no) = shift || $self->{LastBegin};
436 die "$message at $self->{Filename} line $line_no.\n"
442 $status = filter_read();
445 # deal with EOF/error first
447 $self->Die("DEBUG_BEGIN has no DEBUG_END")
448 if $self->{InTraceBlock};
452 if ($self->{InTraceBlock}) {
453 if (/^\s*##\s*DEBUG_BEGIN/ ) {
454 $self->Die("Nested DEBUG_BEGIN", $self->{LineNo})
455 } elsif (/^\s*##\s*DEBUG_END/) {
456 $self->{InTraceBlock} = FALSE;
459 # comment out the debug lines when the filter is disabled
460 s/^/#/ if ! $self->{Enabled};
461 } elsif ( /^\s*##\s*DEBUG_BEGIN/ ) {
462 $self->{InTraceBlock} = TRUE;
463 $self->{LastBegin} = $self->{LineNo};
464 } elsif ( /^\s*##\s*DEBUG_END/ ) {
465 $self->Die("DEBUG_END has no DEBUG_BEGIN", $self->{LineNo});
472 The big difference between this filter and the previous example is the
473 use of context data in the filter object. The filter object is based on
474 a hash reference, and is used to keep various pieces of context
475 information between calls to the filter function. All but two of the
476 hash fields are used for error reporting. The first of those two,
477 Enabled, is used by the filter to determine whether the debugging code
478 should be given to the Perl parser. The second, InTraceBlock, is true
479 when the filter has encountered a C<DEBUG_BEGIN> line, but has not yet
480 encountered the following C<DEBUG_END> line.
482 If you ignore all the error checking that most of the code does, the
483 essence of the filter is as follows:
488 $status = filter_read();
490 # deal with EOF/error first
491 return $status if $status <= 0;
492 if ($self->{InTraceBlock}) {
493 if (/^\s*##\s*DEBUG_END/) {
494 $self->{InTraceBlock} = FALSE
497 # comment out debug lines when the filter is disabled
498 s/^/#/ if ! $self->{Enabled};
499 } elsif ( /^\s*##\s*DEBUG_BEGIN/ ) {
500 $self->{InTraceBlock} = TRUE;
505 Be warned: just as the C-preprocessor doesn't know C, the Debug filter
506 doesn't know Perl. It can be fooled quite easily:
512 Such things aside, you can see that a lot can be achieved with a modest
517 You now have better understanding of what a source filter is, and you
518 might even have a possible use for them. If you feel like playing with
519 source filters but need a bit of inspiration, here are some extra
520 features you could add to the Debug filter.
522 First, an easy one. Rather than having debugging code that is
523 all-or-nothing, it would be much more useful to be able to control
524 which specific blocks of debugging code get included. Try extending the
525 syntax for debug blocks to allow each to be identified. The contents of
526 the C<DEBUG> environment variable can then be used to control which
529 Once you can identify individual blocks, try allowing them to be
530 nested. That isn't difficult either.
532 Here is an interesting idea that doesn't involve the Debug filter.
533 Currently Perl subroutines have fairly limited support for formal
534 parameter lists. You can specify the number of parameters and their
535 type, but you still have to manually take them out of the C<@_> array
536 yourself. Write a source filter that allows you to have a named
537 parameter list. Such a filter would turn this:
539 sub MySub ($first, $second, @rest) { ... }
545 my ($second) = shift;
550 Finally, if you feel like a real challenge, have a go at writing a
551 full-blown Perl macro preprocessor as a source filter. Borrow the
552 useful features from the C preprocessor and any other macro processors
553 you know. The tricky bit will be choosing how much knowledge of Perl's
554 syntax you want your filter to have.
558 Source filters only work on the string level, thus are highly limited
559 in its ability to change source code on the fly. It cannot detect
560 comments, quoted strings, heredocs, it is no replacement for a real
562 The only stable usage for source filters are encryption, compression,
563 or the byteloader, to translate binary code back to source code.
565 See for example the limitations in L<Switch>, which uses source filters,
566 and thus is does not work inside a string eval, the presence of
567 regexes with embedded newlines that are specified with raw C</.../>
568 delimiters and don't have a modifier C<//x> are indistinguishable from
569 code chunks beginning with the division operator C</>. As a workaround
570 you must use C<m/.../> or C<m?...?> for such patterns. Also, the presence of
571 regexes specified with raw C<?...?> delimiters may cause mysterious
572 errors. The workaround is to use C<m?...?> instead. See
573 L<https://search.cpan.org/perldoc?Switch#LIMITATIONS>
575 Currently the content of the C<__DATA__> block is not filtered.
577 Currently internal buffer lengths are limited to 32-bit only.
580 =head1 THINGS TO LOOK OUT FOR
584 =item Some Filters Clobber the C<DATA> Handle
586 Some source filters use the C<DATA> handle to read the calling program.
587 When using these source filters you cannot rely on this handle, nor expect
588 any particular kind of behavior when operating on it. Filters based on
589 Filter::Util::Call (and therefore Filter::Simple) do not alter the C<DATA>
590 filehandle, but on the other hand totally ignore the text after C<__DATA__>.
596 The Source Filters distribution is available on CPAN, in
598 CPAN/modules/by-module/Filter
600 Starting from Perl 5.8 Filter::Util::Call (the core part of the
601 Source Filters distribution) is part of the standard Perl distribution.
602 Also included is a friendlier interface called Filter::Simple, by
607 Paul Marquess E<lt>Paul.Marquess@btinternet.comE<gt>
609 Reini Urban E<lt>rurban@cpan.orgE<gt>
613 The first version of this article originally appeared in The Perl
614 Journal #11, and is copyright 1998 The Perl Journal. It appears
615 courtesy of Jon Orwant and The Perl Journal. This document may be
616 distributed under the same terms as Perl itself.