Commit | Line | Data |
---|---|---|
920dae64 AT |
1 | package AutoSplit; |
2 | ||
3 | use 5.006_001; | |
4 | use Exporter (); | |
5 | use Config qw(%Config); | |
6 | use Carp qw(carp); | |
7 | use File::Basename (); | |
8 | use File::Path qw(mkpath); | |
9 | use File::Spec::Functions qw(curdir catfile catdir); | |
10 | use strict; | |
11 | our($VERSION, @ISA, @EXPORT, @EXPORT_OK, $Verbose, $Keep, $Maxlen, | |
12 | $CheckForAutoloader, $CheckModTime); | |
13 | ||
14 | $VERSION = "1.04"; | |
15 | @ISA = qw(Exporter); | |
16 | @EXPORT = qw(&autosplit &autosplit_lib_modules); | |
17 | @EXPORT_OK = qw($Verbose $Keep $Maxlen $CheckForAutoloader $CheckModTime); | |
18 | ||
19 | =head1 NAME | |
20 | ||
21 | AutoSplit - split a package for autoloading | |
22 | ||
23 | =head1 SYNOPSIS | |
24 | ||
25 | autosplit($file, $dir, $keep, $check, $modtime); | |
26 | ||
27 | autosplit_lib_modules(@modules); | |
28 | ||
29 | =head1 DESCRIPTION | |
30 | ||
31 | This function will split up your program into files that the AutoLoader | |
32 | module can handle. It is used by both the standard perl libraries and by | |
33 | the MakeMaker utility, to automatically configure libraries for autoloading. | |
34 | ||
35 | The C<autosplit> interface splits the specified file into a hierarchy | |
36 | rooted at the directory C<$dir>. It creates directories as needed to reflect | |
37 | class hierarchy, and creates the file F<autosplit.ix>. This file acts as | |
38 | both forward declaration of all package routines, and as timestamp for the | |
39 | last update of the hierarchy. | |
40 | ||
41 | The remaining three arguments to C<autosplit> govern other options to | |
42 | the autosplitter. | |
43 | ||
44 | =over 2 | |
45 | ||
46 | =item $keep | |
47 | ||
48 | If the third argument, I<$keep>, is false, then any | |
49 | pre-existing C<*.al> files in the autoload directory are removed if | |
50 | they are no longer part of the module (obsoleted functions). | |
51 | $keep defaults to 0. | |
52 | ||
53 | =item $check | |
54 | ||
55 | The | |
56 | fourth argument, I<$check>, instructs C<autosplit> to check the module | |
57 | currently being split to ensure that it includes a C<use> | |
58 | specification for the AutoLoader module, and skips the module if | |
59 | AutoLoader is not detected. | |
60 | $check defaults to 1. | |
61 | ||
62 | =item $modtime | |
63 | ||
64 | Lastly, the I<$modtime> argument specifies | |
65 | that C<autosplit> is to check the modification time of the module | |
66 | against that of the C<autosplit.ix> file, and only split the module if | |
67 | it is newer. | |
68 | $modtime defaults to 1. | |
69 | ||
70 | =back | |
71 | ||
72 | Typical use of AutoSplit in the perl MakeMaker utility is via the command-line | |
73 | with: | |
74 | ||
75 | perl -e 'use AutoSplit; autosplit($ARGV[0], $ARGV[1], 0, 1, 1)' | |
76 | ||
77 | Defined as a Make macro, it is invoked with file and directory arguments; | |
78 | C<autosplit> will split the specified file into the specified directory and | |
79 | delete obsolete C<.al> files, after checking first that the module does use | |
80 | the AutoLoader, and ensuring that the module is not already currently split | |
81 | in its current form (the modtime test). | |
82 | ||
83 | The C<autosplit_lib_modules> form is used in the building of perl. It takes | |
84 | as input a list of files (modules) that are assumed to reside in a directory | |
85 | B<lib> relative to the current directory. Each file is sent to the | |
86 | autosplitter one at a time, to be split into the directory B<lib/auto>. | |
87 | ||
88 | In both usages of the autosplitter, only subroutines defined following the | |
89 | perl I<__END__> token are split out into separate files. Some | |
90 | routines may be placed prior to this marker to force their immediate loading | |
91 | and parsing. | |
92 | ||
93 | =head2 Multiple packages | |
94 | ||
95 | As of version 1.01 of the AutoSplit module it is possible to have | |
96 | multiple packages within a single file. Both of the following cases | |
97 | are supported: | |
98 | ||
99 | package NAME; | |
100 | __END__ | |
101 | sub AAA { ... } | |
102 | package NAME::option1; | |
103 | sub BBB { ... } | |
104 | package NAME::option2; | |
105 | sub BBB { ... } | |
106 | ||
107 | package NAME; | |
108 | __END__ | |
109 | sub AAA { ... } | |
110 | sub NAME::option1::BBB { ... } | |
111 | sub NAME::option2::BBB { ... } | |
112 | ||
113 | =head1 DIAGNOSTICS | |
114 | ||
115 | C<AutoSplit> will inform the user if it is necessary to create the | |
116 | top-level directory specified in the invocation. It is preferred that | |
117 | the script or installation process that invokes C<AutoSplit> have | |
118 | created the full directory path ahead of time. This warning may | |
119 | indicate that the module is being split into an incorrect path. | |
120 | ||
121 | C<AutoSplit> will warn the user of all subroutines whose name causes | |
122 | potential file naming conflicts on machines with drastically limited | |
123 | (8 characters or less) file name length. Since the subroutine name is | |
124 | used as the file name, these warnings can aid in portability to such | |
125 | systems. | |
126 | ||
127 | Warnings are issued and the file skipped if C<AutoSplit> cannot locate | |
128 | either the I<__END__> marker or a "package Name;"-style specification. | |
129 | ||
130 | C<AutoSplit> will also emit general diagnostics for inability to | |
131 | create directories or files. | |
132 | ||
133 | =cut | |
134 | ||
135 | # for portability warn about names longer than $maxlen | |
136 | $Maxlen = 8; # 8 for dos, 11 (14-".al") for SYSVR3 | |
137 | $Verbose = 1; # 0=none, 1=minimal, 2=list .al files | |
138 | $Keep = 0; | |
139 | $CheckForAutoloader = 1; | |
140 | $CheckModTime = 1; | |
141 | ||
142 | my $IndexFile = "autosplit.ix"; # file also serves as timestamp | |
143 | my $maxflen = 255; | |
144 | $maxflen = 14 if $Config{'d_flexfnam'} ne 'define'; | |
145 | if (defined (&Dos::UseLFN)) { | |
146 | $maxflen = Dos::UseLFN() ? 255 : 11; | |
147 | } | |
148 | my $Is_VMS = ($^O eq 'VMS'); | |
149 | ||
150 | # allow checking for valid ': attrlist' attachments | |
151 | # (we use 'our' rather than 'my' here, due to the rather complex and buggy | |
152 | # behaviour of lexicals with qr// and (??{$lex}) ) | |
153 | our $nested; | |
154 | $nested = qr{ \( (?: (?> [^()]+ ) | (??{ $nested }) )* \) }x; | |
155 | our $one_attr = qr{ (?> (?! \d) \w+ (?:$nested)? ) (?:\s*\:\s*|\s+(?!\:)) }x; | |
156 | our $attr_list = qr{ \s* : \s* (?: $one_attr )* }x; | |
157 | ||
158 | ||
159 | ||
160 | sub autosplit{ | |
161 | my($file, $autodir, $keep, $ckal, $ckmt) = @_; | |
162 | # $file - the perl source file to be split (after __END__) | |
163 | # $autodir - the ".../auto" dir below which to write split subs | |
164 | # Handle optional flags: | |
165 | $keep = $Keep unless defined $keep; | |
166 | $ckal = $CheckForAutoloader unless defined $ckal; | |
167 | $ckmt = $CheckModTime unless defined $ckmt; | |
168 | autosplit_file($file, $autodir, $keep, $ckal, $ckmt); | |
169 | } | |
170 | ||
171 | ||
172 | # This function is used during perl building/installation | |
173 | # ./miniperl -e 'use AutoSplit; autosplit_lib_modules(@ARGV)' ... | |
174 | ||
175 | sub autosplit_lib_modules{ | |
176 | my(@modules) = @_; # list of Module names | |
177 | ||
178 | while(defined($_ = shift @modules)){ | |
179 | while (m#(.*?[^:])::([^:].*)#) { # in case specified as ABC::XYZ | |
180 | $_ = catfile($1, $2); | |
181 | } | |
182 | s|\\|/|g; # bug in ksh OS/2 | |
183 | s#^lib/##s; # incase specified as lib/*.pm | |
184 | my($lib) = catfile(curdir(), "lib"); | |
185 | if ($Is_VMS) { # may need to convert VMS-style filespecs | |
186 | $lib =~ s#^\[\]#.\/#; | |
187 | } | |
188 | s#^$lib\W+##s; # incase specified as ./lib/*.pm | |
189 | if ($Is_VMS && /[:>\]]/) { # may need to convert VMS-style filespecs | |
190 | my ($dir,$name) = (/(.*])(.*)/s); | |
191 | $dir =~ s/.*lib[\.\]]//s; | |
192 | $dir =~ s#[\.\]]#/#g; | |
193 | $_ = $dir . $name; | |
194 | } | |
195 | autosplit_file(catfile($lib, $_), catfile($lib, "auto"), | |
196 | $Keep, $CheckForAutoloader, $CheckModTime); | |
197 | } | |
198 | 0; | |
199 | } | |
200 | ||
201 | ||
202 | # private functions | |
203 | ||
204 | my $self_mod_time = (stat __FILE__)[9]; | |
205 | ||
206 | sub autosplit_file { | |
207 | my($filename, $autodir, $keep, $check_for_autoloader, $check_mod_time) | |
208 | = @_; | |
209 | my(@outfiles); | |
210 | local($_); | |
211 | local($/) = "\n"; | |
212 | ||
213 | # where to write output files | |
214 | $autodir ||= catfile(curdir(), "lib", "auto"); | |
215 | if ($Is_VMS) { | |
216 | ($autodir = VMS::Filespec::unixpath($autodir)) =~ s|/\z||; | |
217 | $filename = VMS::Filespec::unixify($filename); # may have dirs | |
218 | } | |
219 | unless (-d $autodir){ | |
220 | mkpath($autodir,0,0755); | |
221 | # We should never need to create the auto dir | |
222 | # here. installperl (or similar) should have done | |
223 | # it. Expecting it to exist is a valuable sanity check against | |
224 | # autosplitting into some random directory by mistake. | |
225 | print "Warning: AutoSplit had to create top-level " . | |
226 | "$autodir unexpectedly.\n"; | |
227 | } | |
228 | ||
229 | # allow just a package name to be used | |
230 | $filename .= ".pm" unless ($filename =~ m/\.pm\z/); | |
231 | ||
232 | open(my $in, "<$filename") or die "AutoSplit: Can't open $filename: $!\n"; | |
233 | my($pm_mod_time) = (stat($filename))[9]; | |
234 | my($autoloader_seen) = 0; | |
235 | my($in_pod) = 0; | |
236 | my($def_package,$last_package,$this_package,$fnr); | |
237 | while (<$in>) { | |
238 | # Skip pod text. | |
239 | $fnr++; | |
240 | $in_pod = 1 if /^=\w/; | |
241 | $in_pod = 0 if /^=cut/; | |
242 | next if ($in_pod || /^=cut/); | |
243 | next if /^\s*#/; | |
244 | ||
245 | # record last package name seen | |
246 | $def_package = $1 if (m/^\s*package\s+([\w:]+)\s*;/); | |
247 | ++$autoloader_seen if m/^\s*(use|require)\s+AutoLoader\b/; | |
248 | ++$autoloader_seen if m/\bISA\s*=.*\bAutoLoader\b/; | |
249 | last if /^__END__/; | |
250 | } | |
251 | if ($check_for_autoloader && !$autoloader_seen){ | |
252 | print "AutoSplit skipped $filename: no AutoLoader used\n" | |
253 | if ($Verbose>=2); | |
254 | return 0; | |
255 | } | |
256 | $_ or die "Can't find __END__ in $filename\n"; | |
257 | ||
258 | $def_package or die "Can't find 'package Name;' in $filename\n"; | |
259 | ||
260 | my($modpname) = _modpname($def_package); | |
261 | ||
262 | # this _has_ to match so we have a reasonable timestamp file | |
263 | die "Package $def_package ($modpname.pm) does not ". | |
264 | "match filename $filename" | |
265 | unless ($filename =~ m/\Q$modpname.pm\E$/ or | |
266 | ($^O eq 'dos') or ($^O eq 'MSWin32') or ($^O eq 'NetWare') or | |
267 | $Is_VMS && $filename =~ m/$modpname.pm/i); | |
268 | ||
269 | my($al_idx_file) = catfile($autodir, $modpname, $IndexFile); | |
270 | ||
271 | if ($check_mod_time){ | |
272 | my($al_ts_time) = (stat("$al_idx_file"))[9] || 1; | |
273 | if ($al_ts_time >= $pm_mod_time and | |
274 | $al_ts_time >= $self_mod_time){ | |
275 | print "AutoSplit skipped ($al_idx_file newer than $filename)\n" | |
276 | if ($Verbose >= 2); | |
277 | return undef; # one undef, not a list | |
278 | } | |
279 | } | |
280 | ||
281 | my($modnamedir) = catdir($autodir, $modpname); | |
282 | print "AutoSplitting $filename ($modnamedir)\n" | |
283 | if $Verbose; | |
284 | ||
285 | unless (-d $modnamedir){ | |
286 | mkpath($modnamedir,0,0777); | |
287 | } | |
288 | ||
289 | # We must try to deal with some SVR3 systems with a limit of 14 | |
290 | # characters for file names. Sadly we *cannot* simply truncate all | |
291 | # file names to 14 characters on these systems because we *must* | |
292 | # create filenames which exactly match the names used by AutoLoader.pm. | |
293 | # This is a problem because some systems silently truncate the file | |
294 | # names while others treat long file names as an error. | |
295 | ||
296 | my $Is83 = $maxflen==11; # plain, case INSENSITIVE dos filenames | |
297 | ||
298 | my(@subnames, $subname, %proto, %package); | |
299 | my @cache = (); | |
300 | my $caching = 1; | |
301 | $last_package = ''; | |
302 | my $out; | |
303 | while (<$in>) { | |
304 | $fnr++; | |
305 | $in_pod = 1 if /^=\w/; | |
306 | $in_pod = 0 if /^=cut/; | |
307 | next if ($in_pod || /^=cut/); | |
308 | # the following (tempting) old coding gives big troubles if a | |
309 | # cut is forgotten at EOF: | |
310 | # next if /^=\w/ .. /^=cut/; | |
311 | if (/^package\s+([\w:]+)\s*;/) { | |
312 | $this_package = $def_package = $1; | |
313 | } | |
314 | ||
315 | if (/^sub\s+([\w:]+)(\s*(?:\(.*?\))?(?:$attr_list)?)/) { | |
316 | print $out "# end of $last_package\::$subname\n1;\n" | |
317 | if $last_package; | |
318 | $subname = $1; | |
319 | my $proto = $2 || ''; | |
320 | if ($subname =~ s/(.*):://){ | |
321 | $this_package = $1; | |
322 | } else { | |
323 | $this_package = $def_package; | |
324 | } | |
325 | my $fq_subname = "$this_package\::$subname"; | |
326 | $package{$fq_subname} = $this_package; | |
327 | $proto{$fq_subname} = $proto; | |
328 | push(@subnames, $fq_subname); | |
329 | my($lname, $sname) = ($subname, substr($subname,0,$maxflen-3)); | |
330 | $modpname = _modpname($this_package); | |
331 | my($modnamedir) = catdir($autodir, $modpname); | |
332 | mkpath($modnamedir,0,0777); | |
333 | my($lpath) = catfile($modnamedir, "$lname.al"); | |
334 | my($spath) = catfile($modnamedir, "$sname.al"); | |
335 | my $path; | |
336 | ||
337 | if (!$Is83 and open($out, ">$lpath")){ | |
338 | $path=$lpath; | |
339 | print " writing $lpath\n" if ($Verbose>=2); | |
340 | } else { | |
341 | open($out, ">$spath") or die "Can't create $spath: $!\n"; | |
342 | $path=$spath; | |
343 | print " writing $spath (with truncated name)\n" | |
344 | if ($Verbose>=1); | |
345 | } | |
346 | push(@outfiles, $path); | |
347 | my $lineno = $fnr - @cache; | |
348 | print $out <<EOT; | |
349 | # NOTE: Derived from $filename. | |
350 | # Changes made here will be lost when autosplit is run again. | |
351 | # See AutoSplit.pm. | |
352 | package $this_package; | |
353 | ||
354 | #line $lineno "$filename (autosplit into $path)" | |
355 | EOT | |
356 | print $out @cache; | |
357 | @cache = (); | |
358 | $caching = 0; | |
359 | } | |
360 | if($caching) { | |
361 | push(@cache, $_) if @cache || /\S/; | |
362 | } else { | |
363 | print $out $_; | |
364 | } | |
365 | if(/^\}/) { | |
366 | if($caching) { | |
367 | print $out @cache; | |
368 | @cache = (); | |
369 | } | |
370 | print $out "\n"; | |
371 | $caching = 1; | |
372 | } | |
373 | $last_package = $this_package if defined $this_package; | |
374 | } | |
375 | if ($subname) { | |
376 | print $out @cache,"1;\n# end of $last_package\::$subname\n"; | |
377 | close($out); | |
378 | } | |
379 | close($in); | |
380 | ||
381 | if (!$keep){ # don't keep any obsolete *.al files in the directory | |
382 | my(%outfiles); | |
383 | # @outfiles{@outfiles} = @outfiles; | |
384 | # perl downcases all filenames on VMS (which upcases all filenames) so | |
385 | # we'd better downcase the sub name list too, or subs with upper case | |
386 | # letters in them will get their .al files deleted right after they're | |
387 | # created. (The mixed case sub name won't match the all-lowercase | |
388 | # filename, and so be cleaned up as a scrap file) | |
389 | if ($Is_VMS or $Is83) { | |
390 | %outfiles = map {lc($_) => lc($_) } @outfiles; | |
391 | } else { | |
392 | @outfiles{@outfiles} = @outfiles; | |
393 | } | |
394 | my(%outdirs,@outdirs); | |
395 | for (@outfiles) { | |
396 | $outdirs{File::Basename::dirname($_)}||=1; | |
397 | } | |
398 | for my $dir (keys %outdirs) { | |
399 | opendir(my $outdir,$dir); | |
400 | foreach (sort readdir($outdir)){ | |
401 | next unless /\.al\z/; | |
402 | my($file) = catfile($dir, $_); | |
403 | $file = lc $file if $Is83 or $Is_VMS; | |
404 | next if $outfiles{$file}; | |
405 | print " deleting $file\n" if ($Verbose>=2); | |
406 | my($deleted,$thistime); # catch all versions on VMS | |
407 | do { $deleted += ($thistime = unlink $file) } while ($thistime); | |
408 | carp "Unable to delete $file: $!" unless $deleted; | |
409 | } | |
410 | closedir($outdir); | |
411 | } | |
412 | } | |
413 | ||
414 | open(my $ts,">$al_idx_file") or | |
415 | carp "AutoSplit: unable to create timestamp file ($al_idx_file): $!"; | |
416 | print $ts "# Index created by AutoSplit for $filename\n"; | |
417 | print $ts "# (file acts as timestamp)\n"; | |
418 | $last_package = ''; | |
419 | for my $fqs (@subnames) { | |
420 | my($subname) = $fqs; | |
421 | $subname =~ s/.*:://; | |
422 | print $ts "package $package{$fqs};\n" | |
423 | unless $last_package eq $package{$fqs}; | |
424 | print $ts "sub $subname $proto{$fqs};\n"; | |
425 | $last_package = $package{$fqs}; | |
426 | } | |
427 | print $ts "1;\n"; | |
428 | close($ts); | |
429 | ||
430 | _check_unique($filename, $Maxlen, 1, @outfiles); | |
431 | ||
432 | @outfiles; | |
433 | } | |
434 | ||
435 | sub _modpname ($) { | |
436 | my($package) = @_; | |
437 | my $modpname = $package; | |
438 | if ($^O eq 'MSWin32') { | |
439 | $modpname =~ s#::#\\#g; | |
440 | } else { | |
441 | my @modpnames = (); | |
442 | while ($modpname =~ m#(.*?[^:])::([^:].*)#) { | |
443 | push @modpnames, $1; | |
444 | $modpname = $2; | |
445 | } | |
446 | $modpname = catfile(@modpnames, $modpname); | |
447 | } | |
448 | if ($Is_VMS) { | |
449 | $modpname = VMS::Filespec::unixify($modpname); # may have dirs | |
450 | } | |
451 | $modpname; | |
452 | } | |
453 | ||
454 | sub _check_unique { | |
455 | my($filename, $maxlen, $warn, @outfiles) = @_; | |
456 | my(%notuniq) = (); | |
457 | my(%shorts) = (); | |
458 | my(@toolong) = grep( | |
459 | length(File::Basename::basename($_)) | |
460 | > $maxlen, | |
461 | @outfiles | |
462 | ); | |
463 | ||
464 | foreach (@toolong){ | |
465 | my($dir) = File::Basename::dirname($_); | |
466 | my($file) = File::Basename::basename($_); | |
467 | my($trunc) = substr($file,0,$maxlen); | |
468 | $notuniq{$dir}{$trunc} = 1 if $shorts{$dir}{$trunc}; | |
469 | $shorts{$dir}{$trunc} = $shorts{$dir}{$trunc} ? | |
470 | "$shorts{$dir}{$trunc}, $file" : $file; | |
471 | } | |
472 | if (%notuniq && $warn){ | |
473 | print "$filename: some names are not unique when " . | |
474 | "truncated to $maxlen characters:\n"; | |
475 | foreach my $dir (sort keys %notuniq){ | |
476 | print " directory $dir:\n"; | |
477 | foreach my $trunc (sort keys %{$notuniq{$dir}}) { | |
478 | print " $shorts{$dir}{$trunc} truncate to $trunc\n"; | |
479 | } | |
480 | } | |
481 | } | |
482 | } | |
483 | ||
484 | 1; | |
485 | __END__ | |
486 | ||
487 | # test functions so AutoSplit.pm can be applied to itself: | |
488 | sub test1 ($) { "test 1\n"; } | |
489 | sub test2 ($$) { "test 2\n"; } | |
490 | sub test3 ($$$) { "test 3\n"; } | |
491 | sub testtesttesttest4_1 { "test 4\n"; } | |
492 | sub testtesttesttest4_2 { "duplicate test 4\n"; } | |
493 | sub Just::Another::test5 { "another test 5\n"; } | |
494 | sub test6 { return join ":", __FILE__,__LINE__; } | |
495 | package Yet::Another::AutoSplit; | |
496 | sub testtesttesttest4_1 ($) { "another test 4\n"; } | |
497 | sub testtesttesttest4_2 ($$) { "another duplicate test 4\n"; } | |
498 | package Yet::More::Attributes; | |
499 | sub test_a1 ($) : locked :locked { 1; } | |
500 | sub test_a2 : locked { 1; } |