This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Move more URLs from http:// to https://
[perl5.git] / pod / perlfilter.pod
CommitLineData
c7c04614
GS
1=head1 NAME
2
3perlfilter - Source Filters
c47ff5f1 4
c7c04614
GS
5=head1 DESCRIPTION
6
7This article is about a little-known feature of Perl called
8I<source filters>. Source filters alter the program text of a module
9before Perl sees it, much as a C preprocessor alters the source text of
10a C program before the compiler sees it. This article tells you more
11about what source filters are, how they work, and how to write your
12own.
13
14The original purpose of source filters was to let you encrypt your
15program source to prevent casual piracy. This isn't all they can do, as
16you'll soon learn. But first, the basics.
17
18=head1 CONCEPTS
19
20Before the Perl interpreter can execute a Perl script, it must first
4449c45d
GS
21read it from a file into memory for parsing and compilation. If that
22script itself includes other scripts with a C<use> or C<require>
23statement, then each of those scripts will have to be read from their
24respective files as well.
c7c04614
GS
25
26Now think of each logical connection between the Perl parser and an
27individual file as a I<source stream>. A source stream is created when
28the Perl parser opens a file, it continues to exist as the source code
29is read into memory, and it is destroyed when Perl is finished parsing
30the file. If the parser encounters a C<require> or C<use> statement in
31a source stream, a new and distinct stream is created just for that
32file.
33
34The diagram below represents a single source stream, with the flow of
35source from a Perl script file on the left into the Perl parser on the
36right. This is how Perl normally operates.
37
38 file -------> parser
39
40There are two important points to remember:
41
42=over 5
43
44=item 1.
45
46Although there can be any number of source streams in existence at any
47given time, only one will be active.
48
49=item 2.
50
51Every source stream is associated with only one file.
52
53=back
54
55A source filter is a special kind of Perl module that intercepts and
56modifies a source stream before it reaches the parser. A source filter
40b7eeef 57changes our diagram like this:
c7c04614
GS
58
59 file ----> filter ----> parser
60
61If that doesn't make much sense, consider the analogy of a command
62pipeline. Say you have a shell script stored in the compressed file
63I<trial.gz>. The simple pipeline command below runs the script without
64needing to create a temporary file to hold the uncompressed file.
65
66 gunzip -c trial.gz | sh
67
68In this case, the data flow from the pipeline can be represented as follows:
69
70 trial.gz ----> gunzip ----> sh
71
72With source filters, you can store the text of your script compressed and use a source filter to uncompress it for Perl's parser:
73
74 compressed gunzip
75 Perl program ---> source filter ---> parser
76
77=head1 USING FILTERS
78
79So how do you use a source filter in a Perl script? Above, I said that
80a source filter is just a special kind of module. Like all Perl
81modules, a source filter is invoked with a use statement.
82
83Say you want to pass your Perl source through the C preprocessor before
4c84d7f2
RGS
84execution. As it happens, the source filters distribution comes with a C
85preprocessor filter module called Filter::cpp.
c7c04614
GS
86
87Below is an example program, C<cpp_test>, which makes use of this filter.
88Line numbers have been added to allow specific lines to be referenced
89easily.
90
4358a253 91 1: use Filter::cpp;
c7c04614 92 2: #define TRUE 1
4358a253
SS
93 3: $a = TRUE;
94 4: print "a = $a\n";
c7c04614
GS
95
96When you execute this script, Perl creates a source stream for the
97file. Before the parser processes any of the lines from the file, the
98source stream looks like this:
99
100 cpp_test ---------> parser
101
102Line 1, C<use Filter::cpp>, includes and installs the C<cpp> filter
103module. All source filters work this way. The use statement is compiled
104and executed at compile time, before any more of the file is read, and
105it attaches the cpp filter to the source stream behind the scenes. Now
106the data flow looks like this:
107
108 cpp_test ----> cpp filter ----> parser
109
110As the parser reads the second and subsequent lines from the source
111stream, it feeds those lines through the C<cpp> source filter before
112processing them. The C<cpp> filter simply passes each line through the
113real C preprocessor. The output from the C preprocessor is then
114inserted back into the source stream by the filter.
115
116 .-> cpp --.
117 | |
118 | |
119 | <-'
120 cpp_test ----> cpp filter ----> parser
121
122The parser then sees the following code:
123
4358a253
SS
124 use Filter::cpp;
125 $a = 1;
126 print "a = $a\n";
c7c04614
GS
127
128Let's consider what happens when the filtered code includes another
129module with use:
130
4358a253 131 1: use Filter::cpp;
c7c04614 132 2: #define TRUE 1
4358a253
SS
133 3: use Fred;
134 4: $a = TRUE;
135 5: print "a = $a\n";
c7c04614
GS
136
137The C<cpp> filter does not apply to the text of the Fred module, only
138to the text of the file that used it (C<cpp_test>). Although the use
139statement on line 3 will pass through the cpp filter, the module that
140gets included (C<Fred>) will not. The source streams look like this
141after line 3 has been parsed and before line 4 is parsed:
142
143 cpp_test ---> cpp filter ---> parser (INACTIVE)
144
145 Fred.pm ----> parser
146
147As you can see, a new stream has been created for reading the source
148from C<Fred.pm>. This stream will remain active until all of C<Fred.pm>
149has been parsed. The source stream for C<cpp_test> will still exist,
150but is inactive. Once the parser has finished reading Fred.pm, the
151source stream associated with it will be destroyed. The source stream
152for C<cpp_test> then becomes active again and the parser reads line 4
153and subsequent lines from C<cpp_test>.
154
155You can use more than one source filter on a single file. Similarly,
156you can reuse the same filter in as many files as you like.
157
158For example, if you have a uuencoded and compressed source file, it is
159possible to stack a uudecode filter and an uncompression filter like
160this:
161
4358a253 162 use Filter::uudecode; use Filter::uncompress;
c7c04614
GS
163 M'XL(".H<US4''V9I;F%L')Q;>7/;1I;_>_I3=&E=%:F*I"T?22Q/
164 M6]9*<IQCO*XFT"0[PL%%'Y+IG?WN^ZYN-$'J.[.JE$,20/?K=_[>
165 ...
166
167Once the first line has been processed, the flow will look like this:
168
169 file ---> uudecode ---> uncompress ---> parser
170 filter filter
171
172Data flows through filters in the same order they appear in the source
173file. The uudecode filter appeared before the uncompress filter, so the
174source file will be uudecoded before it's uncompressed.
175
176=head1 WRITING A SOURCE FILTER
177
178There are three ways to write your own source filter. You can write it
179in C, use an external program as a filter, or write the filter in Perl.
180I won't cover the first two in any great detail, so I'll get them out
181of the way first. Writing the filter in Perl is most convenient, so
182I'll devote the most space to it.
183
184=head1 WRITING A SOURCE FILTER IN C
185
186The first of the three available techniques is to write the filter
187completely in C. The external module you create interfaces directly
188with the source filter hooks provided by Perl.
189
190The advantage of this technique is that you have complete control over
191the implementation of your filter. The big disadvantage is the
192increased complexity required to write the filter - not only do you
193need to understand the source filter hooks, but you also need a
194reasonable knowledge of Perl guts. One of the few times it is worth
195going to this trouble is when writing a source scrambler. The
196C<decrypt> filter (which unscrambles the source before Perl parses it)
197included with the source filter distribution is an example of a C
198source filter (see Decryption Filters, below).
c47ff5f1 199
c7c04614
GS
200
201=over 5
202
203=item B<Decryption Filters>
204
205All decryption filters work on the principle of "security through
206obscurity." Regardless of how well you write a decryption filter and
e8c6d8ab 207how strong your encryption algorithm is, anyone determined enough can
c7c04614
GS
208retrieve the original source code. The reason is quite simple - once
209the decryption filter has decrypted the source back to its original
210form, fragments of it will be stored in the computer's memory as Perl
211parses it. The source might only be in memory for a short period of
212time, but anyone possessing a debugger, skill, and lots of patience can
213eventually reconstruct your program.
214
215That said, there are a number of steps that can be taken to make life
216difficult for the potential cracker. The most important: Write your
217decryption filter in C and statically link the decryption module into
218the Perl binary. For further tips to make life difficult for the
219potential cracker, see the file I<decrypt.pm> in the source filters
e8c6d8ab 220distribution.
c7c04614
GS
221
222=back
223
224=head1 CREATING A SOURCE FILTER AS A SEPARATE EXECUTABLE
225
226An alternative to writing the filter in C is to create a separate
227executable in the language of your choice. The separate executable
228reads from standard input, does whatever processing is necessary, and
e8c6d8ab 229writes the filtered data to standard output. C<Filter::cpp> is an
c7c04614
GS
230example of a source filter implemented as a separate executable - the
231executable is the C preprocessor bundled with your C compiler.
232
233The source filter distribution includes two modules that simplify this
234task: C<Filter::exec> and C<Filter::sh>. Both allow you to run any
235external executable. Both use a coprocess to control the flow of data
236into and out of the external executable. (For details on coprocesses,
e8c6d8ab 237see Stephens, W.R., "Advanced Programming in the UNIX Environment."
c7c04614
GS
238Addison-Wesley, ISBN 0-210-56317-7, pages 441-445.) The difference
239between them is that C<Filter::exec> spawns the external command
240directly, while C<Filter::sh> spawns a shell to execute the external
241command. (Unix uses the Bourne shell; NT uses the cmd shell.) Spawning
242a shell allows you to make use of the shell metacharacters and
243redirection facilities.
244
245Here is an example script that uses C<Filter::sh>:
246
4358a253
SS
247 use Filter::sh 'tr XYZ PQR';
248 $a = 1;
249 print "XYZ a = $a\n";
c7c04614
GS
250
251The output you'll get when the script is executed:
252
253 PQR a = 1
254
255Writing a source filter as a separate executable works fine, but a
256small performance penalty is incurred. For example, if you execute the
257small example above, a separate subprocess will be created to run the
258Unix C<tr> command. Each use of the filter requires its own subprocess.
259If creating subprocesses is expensive on your system, you might want to
260consider one of the other options for creating source filters.
261
262=head1 WRITING A SOURCE FILTER IN PERL
263
264The easiest and most portable option available for creating your own
265source filter is to write it completely in Perl. To distinguish this
266from the previous two techniques, I'll call it a Perl source filter.
267
268To help understand how to write a Perl source filter we need an example
269to study. Here is a complete source filter that performs rot13
270decoding. (Rot13 is a very simple encryption scheme used in Usenet
271postings to hide the contents of offensive posts. It moves every letter
272forward thirteen places, so that A becomes N, B becomes O, and Z
273becomes M.)
274
275
4358a253 276 package Rot13;
c7c04614 277
4358a253 278 use Filter::Util::Call;
c7c04614
GS
279
280 sub import {
4358a253
SS
281 my ($type) = @_;
282 my ($ref) = [];
283 filter_add(bless $ref);
c7c04614
GS
284 }
285
286 sub filter {
4358a253
SS
287 my ($self) = @_;
288 my ($status);
c7c04614
GS
289
290 tr/n-za-mN-ZA-M/a-zA-Z/
4358a253
SS
291 if ($status = filter_read()) > 0;
292 $status;
c7c04614
GS
293 }
294
295 1;
296
0f292d69
KW
297=for apidoc filter_add
298=for apidoc filter_read
299
c7c04614
GS
300All Perl source filters are implemented as Perl classes and have the
301same basic structure as the example above.
302
303First, we include the C<Filter::Util::Call> module, which exports a
304number of functions into your filter's namespace. The filter shown
305above uses two of these functions, C<filter_add()> and
306C<filter_read()>.
307
308Next, we create the filter object and associate it with the source
309stream by defining the C<import> function. If you know Perl well
310enough, you know that C<import> is called automatically every time a
311module is included with a use statement. This makes C<import> the ideal
312place to both create and install a filter object.
313
314In the example filter, the object (C<$ref>) is blessed just like any
315other Perl object. Our example uses an anonymous array, but this isn't
316a requirement. Because this example doesn't need to store any context
317information, we could have used a scalar or hash reference just as
318well. The next section demonstrates context data.
319
320The association between the filter object and the source stream is made
321with the C<filter_add()> function. This takes a filter object as a
322parameter (C<$ref> in this case) and installs it in the source stream.
323
324Finally, there is the code that actually does the filtering. For this
325type of Perl source filter, all the filtering is done in a method
326called C<filter()>. (It is also possible to write a Perl source filter
327using a closure. See the C<Filter::Util::Call> manual page for more
328details.) It's called every time the Perl parser needs another line of
329source to process. The C<filter()> method, in turn, reads lines from
330the source stream using the C<filter_read()> function.
331
332If a line was available from the source stream, C<filter_read()>
333returns a status value greater than zero and appends the line to C<$_>.
334A status value of zero indicates end-of-file, less than zero means an
335error. The filter function itself is expected to return its status in
336the same way, and put the filtered line it wants written to the source
337stream in C<$_>. The use of C<$_> accounts for the brevity of most Perl
338source filters.
339
340In order to make use of the rot13 filter we need some way of encoding
341the source file in rot13 format. The script below, C<mkrot13>, does
342just that.
343
4358a253
SS
344 die "usage mkrot13 filename\n" unless @ARGV;
345 my $in = $ARGV[0];
346 my $out = "$in.tmp";
c7c04614
GS
347 open(IN, "<$in") or die "Cannot open file $in: $!\n";
348 open(OUT, ">$out") or die "Cannot open file $out: $!\n";
349
4358a253 350 print OUT "use Rot13;\n";
c7c04614 351 while (<IN>) {
4358a253
SS
352 tr/a-zA-Z/n-za-mN-ZA-M/;
353 print OUT;
c7c04614
GS
354 }
355
356 close IN;
357 close OUT;
358 unlink $in;
359 rename $out, $in;
360
361If we encrypt this with C<mkrot13>:
362
4358a253 363 print " hello fred \n";
c7c04614
GS
364
365the result will be this:
366
367 use Rot13;
4358a253 368 cevag "uryyb serq\a";
c7c04614
GS
369
370Running it produces this output:
371
372 hello fred
373
374=head1 USING CONTEXT: THE DEBUG FILTER
375
376The rot13 example was a trivial example. Here's another demonstration
377that shows off a few more features.
378
379Say you wanted to include a lot of debugging code in your Perl script
380during development, but you didn't want it available in the released
381product. Source filters offer a solution. In order to keep the example
382simple, let's say you wanted the debugging output to be controlled by
383an environment variable, C<DEBUG>. Debugging code is enabled if the
384variable exists, otherwise it is disabled.
385
386Two special marker lines will bracket debugging code, like this:
387
388 ## DEBUG_BEGIN
389 if ($year > 1999) {
4358a253 390 warn "Debug: millennium bug in year $year\n";
c7c04614
GS
391 }
392 ## DEBUG_END
393
e8c6d8ab
FC
394The filter ensures that Perl parses the code between the <DEBUG_BEGIN>
395and C<DEBUG_END> markers only when the C<DEBUG> environment variable
396exists. That means that when C<DEBUG> does exist, the code above
c7c04614
GS
397should be passed through the filter unchanged. The marker lines can
398also be passed through as-is, because the Perl parser will see them as
399comment lines. When C<DEBUG> isn't set, we need a way to disable the
400debug code. A simple way to achieve that is to convert the lines
401between the two markers into comments:
402
403 ## DEBUG_BEGIN
404 #if ($year > 1999) {
4358a253 405 # warn "Debug: millennium bug in year $year\n";
c7c04614
GS
406 #}
407 ## DEBUG_END
408
409Here is the complete Debug filter:
410
411 package Debug;
412
413 use strict;
9f1b1f2d 414 use warnings;
4358a253 415 use Filter::Util::Call;
c7c04614 416
4358a253
SS
417 use constant TRUE => 1;
418 use constant FALSE => 0;
c7c04614
GS
419
420 sub import {
4358a253 421 my ($type) = @_;
c7c04614
GS
422 my (%context) = (
423 Enabled => defined $ENV{DEBUG},
424 InTraceBlock => FALSE,
425 Filename => (caller)[1],
426 LineNo => 0,
427 LastBegin => 0,
4358a253
SS
428 );
429 filter_add(bless \%context);
c7c04614
GS
430 }
431
432 sub Die {
4358a253
SS
433 my ($self) = shift;
434 my ($message) = shift;
435 my ($line_no) = shift || $self->{LastBegin};
c7c04614
GS
436 die "$message at $self->{Filename} line $line_no.\n"
437 }
438
439 sub filter {
4358a253
SS
440 my ($self) = @_;
441 my ($status);
442 $status = filter_read();
443 ++ $self->{LineNo};
c7c04614
GS
444
445 # deal with EOF/error first
446 if ($status <= 0) {
447 $self->Die("DEBUG_BEGIN has no DEBUG_END")
4358a253
SS
448 if $self->{InTraceBlock};
449 return $status;
c7c04614
GS
450 }
451
452 if ($self->{InTraceBlock}) {
453 if (/^\s*##\s*DEBUG_BEGIN/ ) {
454 $self->Die("Nested DEBUG_BEGIN", $self->{LineNo})
455 } elsif (/^\s*##\s*DEBUG_END/) {
4358a253 456 $self->{InTraceBlock} = FALSE;
c7c04614
GS
457 }
458
459 # comment out the debug lines when the filter is disabled
4358a253 460 s/^/#/ if ! $self->{Enabled};
c7c04614 461 } elsif ( /^\s*##\s*DEBUG_BEGIN/ ) {
4358a253
SS
462 $self->{InTraceBlock} = TRUE;
463 $self->{LastBegin} = $self->{LineNo};
c7c04614
GS
464 } elsif ( /^\s*##\s*DEBUG_END/ ) {
465 $self->Die("DEBUG_END has no DEBUG_BEGIN", $self->{LineNo});
466 }
4358a253 467 return $status;
c7c04614
GS
468 }
469
4358a253 470 1;
c7c04614
GS
471
472The big difference between this filter and the previous example is the
473use of context data in the filter object. The filter object is based on
474a hash reference, and is used to keep various pieces of context
475information between calls to the filter function. All but two of the
476hash fields are used for error reporting. The first of those two,
477Enabled, is used by the filter to determine whether the debugging code
478should be given to the Perl parser. The second, InTraceBlock, is true
479when the filter has encountered a C<DEBUG_BEGIN> line, but has not yet
480encountered the following C<DEBUG_END> line.
481
482If you ignore all the error checking that most of the code does, the
483essence of the filter is as follows:
484
485 sub filter {
4358a253
SS
486 my ($self) = @_;
487 my ($status);
488 $status = filter_read();
c7c04614
GS
489
490 # deal with EOF/error first
4358a253 491 return $status if $status <= 0;
c7c04614
GS
492 if ($self->{InTraceBlock}) {
493 if (/^\s*##\s*DEBUG_END/) {
494 $self->{InTraceBlock} = FALSE
495 }
496
497 # comment out debug lines when the filter is disabled
4358a253 498 s/^/#/ if ! $self->{Enabled};
c7c04614 499 } elsif ( /^\s*##\s*DEBUG_BEGIN/ ) {
4358a253 500 $self->{InTraceBlock} = TRUE;
c7c04614 501 }
4358a253 502 return $status;
c7c04614
GS
503 }
504
505Be warned: just as the C-preprocessor doesn't know C, the Debug filter
506doesn't know Perl. It can be fooled quite easily:
507
508 print <<EOM;
509 ##DEBUG_BEGIN
510 EOM
511
512Such things aside, you can see that a lot can be achieved with a modest
40b7eeef 513amount of code.
c7c04614
GS
514
515=head1 CONCLUSION
516
517You now have better understanding of what a source filter is, and you
518might even have a possible use for them. If you feel like playing with
519source filters but need a bit of inspiration, here are some extra
520features you could add to the Debug filter.
521
522First, an easy one. Rather than having debugging code that is
523all-or-nothing, it would be much more useful to be able to control
524which specific blocks of debugging code get included. Try extending the
525syntax for debug blocks to allow each to be identified. The contents of
526the C<DEBUG> environment variable can then be used to control which
527blocks get included.
528
529Once you can identify individual blocks, try allowing them to be
530nested. That isn't difficult either.
531
d1be9408 532Here is an interesting idea that doesn't involve the Debug filter.
c7c04614
GS
533Currently Perl subroutines have fairly limited support for formal
534parameter lists. You can specify the number of parameters and their
535type, but you still have to manually take them out of the C<@_> array
536yourself. Write a source filter that allows you to have a named
537parameter list. Such a filter would turn this:
538
539 sub MySub ($first, $second, @rest) { ... }
540
541into this:
542
543 sub MySub($$@) {
4358a253
SS
544 my ($first) = shift;
545 my ($second) = shift;
546 my (@rest) = @_;
c7c04614
GS
547 ...
548 }
549
550Finally, if you feel like a real challenge, have a go at writing a
551full-blown Perl macro preprocessor as a source filter. Borrow the
552useful features from the C preprocessor and any other macro processors
553you know. The tricky bit will be choosing how much knowledge of Perl's
554syntax you want your filter to have.
555
f686c54e
CBW
556=head1 LIMITATIONS
557
558Source filters only work on the string level, thus are highly limited
559in its ability to change source code on the fly. It cannot detect
560comments, quoted strings, heredocs, it is no replacement for a real
561parser.
562The only stable usage for source filters are encryption, compression,
563or the byteloader, to translate binary code back to source code.
564
356231b0 565See for example the limitations in L<Switch>, which uses source filters,
f686c54e 566and thus is does not work inside a string eval, the presence of
356231b0 567regexes with embedded newlines that are specified with raw C</.../>
0168e427 568delimiters and don't have a modifier C<//x> are indistinguishable from
356231b0 569code chunks beginning with the division operator C</>. As a workaround
0168e427
CBW
570you must use C<m/.../> or C<m?...?> for such patterns. Also, the presence of
571regexes specified with raw C<?...?> delimiters may cause mysterious
572errors. The workaround is to use C<m?...?> instead. See
71c89d21 573L<https://search.cpan.org/perldoc?Switch#LIMITATIONS>
356231b0
SH
574
575Currently the content of the C<__DATA__> block is not filtered.
f686c54e
CBW
576
577Currently internal buffer lengths are limited to 32-bit only.
578
579
6c3397db
CW
580=head1 THINGS TO LOOK OUT FOR
581
582=over 5
583
584=item Some Filters Clobber the C<DATA> Handle
585
586Some source filters use the C<DATA> handle to read the calling program.
587When using these source filters you cannot rely on this handle, nor expect
588any particular kind of behavior when operating on it. Filters based on
589Filter::Util::Call (and therefore Filter::Simple) do not alter the C<DATA>
356231b0 590filehandle, but on the other hand totally ignore the text after C<__DATA__>.
6c3397db
CW
591
592=back
593
c7c04614
GS
594=head1 REQUIREMENTS
595
596The Source Filters distribution is available on CPAN, in
597
598 CPAN/modules/by-module/Filter
599
83df6a1d
JH
600Starting from Perl 5.8 Filter::Util::Call (the core part of the
601Source Filters distribution) is part of the standard Perl distribution.
602Also included is a friendlier interface called Filter::Simple, by
603Damian Conway.
604
c7c04614
GS
605=head1 AUTHOR
606
607Paul Marquess E<lt>Paul.Marquess@btinternet.comE<gt>
608
356231b0
SH
609Reini Urban E<lt>rurban@cpan.orgE<gt>
610
c7c04614
GS
611=head1 Copyrights
612
356231b0
SH
613The first version of this article originally appeared in The Perl
614Journal #11, and is copyright 1998 The Perl Journal. It appears
615courtesy of Jon Orwant and The Perl Journal. This document may be
616distributed under the same terms as Perl itself.