This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
corelist.pl - Fix make_coreutils_delta()
[perl5.git] / Porting / corelist.pl
CommitLineData
4a656c5e
RGS
1#!perl
2# Generates info for Module::CoreList from this perl tree
dc2f75c0 3# run this from the root of a perl tree
e1018a69
DM
4#
5# Data is on STDOUT.
6#
7# With an optional arg specifying the root of a CPAN mirror, outputs the
8# %upstream and %bug_tracker hashes too.
9
54233949 10use autodie;
4a656c5e
RGS
11use strict;
12use warnings;
13use File::Find;
14use ExtUtils::MM_Unix;
dc2f75c0 15use version;
fb237dfd
NC
16use lib "Porting";
17use Maintainers qw(%Modules files_to_modules);
18use File::Spec;
dc2f75c0 19use Parse::CPAN::Meta;
a762e054 20use IPC::Cmd 'can_run';
b87bfaf3 21use HTTP::Tiny;
9168ed13 22use IO::Uncompress::Gunzip;
4a656c5e 23
1f809cd9 24my $corelist_file = 'dist/Module-CoreList/lib/Module/CoreList.pm';
d990e30d 25my $utils_file = 'dist/Module-CoreList/lib/Module/CoreList/Utils.pm';
dd4d388c 26my $pod_file = 'dist/Module-CoreList/lib/Module/CoreList.pod';
e1018a69 27
59189dd7 28my %lines;
fb237dfd
NC
29my %module_to_file;
30my %modlist;
e1018a69 31
dc2f75c0
JV
32die "usage: $0 [ cpan-mirror/ ] [ 5.x.y] \n" unless @ARGV <= 2;
33my $cpan = shift;
34my $raw_version = shift || $];
35my $perl_version = version->parse("$raw_version");
36my $perl_vnum = $perl_version->numify;
37my $perl_vstring = $perl_version->normal; # how do we get version.pm to not give us leading v?
38$perl_vstring =~ s/^v//;
fb237dfd 39
dc2f75c0
JV
40if ( !-f 'MANIFEST' ) {
41 die "Must be run from the root of a clean perl tree\n";
e1018a69
DM
42}
43
54233949 44open( my $corelist_fh, '<', $corelist_file );
dc2f75c0
JV
45my $corelist = join( '', <$corelist_fh> );
46
fb237dfd 47if ($cpan) {
dc2f75c0
JV
48 my $modlistfile = File::Spec->catfile( $cpan, 'modules', '02packages.details.txt' );
49 my $content;
50
51 my $fh;
52 if ( -e $modlistfile ) {
53 warn "Reading the module list from $modlistfile";
54233949 54 open $fh, '<', $modlistfile;
dc2f75c0 55 } elsif ( -e $modlistfile . ".gz" ) {
a762e054 56 my $zcat = can_run('gzcat') || can_run('zcat') or die "Can't find gzcat or zcat";
dc2f75c0 57 warn "Reading the module list from $modlistfile.gz";
54233949 58 open $fh, '-|', "$zcat $modlistfile.gz";
dc2f75c0
JV
59 } else {
60 warn "About to fetch 02packages from ftp.funet.fi. This may take a few minutes\n";
9168ed13
LB
61 my $gzipped_content = fetch_url('http://ftp.funet.fi/pub/CPAN/modules/02packages.details.txt.gz');
62 unless ($gzipped_content) {
dc2f75c0
JV
63 die "Unable to read 02packages.details.txt from either your CPAN mirror or ftp.funet.fi";
64 }
9168ed13
LB
65 IO::Uncompress::Gunzip::gunzip(\$gzipped_content, \$content, Transparent => 0)
66 or die "Can't gunzip content: $IO::Uncompress::Gunzip::GunzipError";
dc2f75c0
JV
67 }
68
69 if ( $fh and !$content ) {
70 local $/ = "\n";
71 $content = join( '', <$fh> );
fb237dfd
NC
72 }
73
dc2f75c0
JV
74 die "Incompatible modlist format"
75 unless $content =~ /^Columns: +package name, version, path/m;
76
fb237dfd
NC
77 # Converting the file to a hash is about 5 times faster than a regexp flat
78 # lookup.
dc2f75c0
JV
79 for ( split( qr/\n/, $content ) ) {
80 next unless /^([A-Za-z_:0-9]+) +[-0-9.undefHASHVERSIONvsetwhenloadingbogus]+ +(\S+)/;
81 $modlist{$1} = $2;
fb237dfd
NC
82 }
83}
84
dc2f75c0
JV
85find(
86 sub {
87 /(\.pm|_pm\.PL)$/ or return;
88 /PPPort\.pm$/ and return;
89 my $module = $File::Find::name;
90 $module =~ /\b(demo|t|private)\b/ and return; # demo or test modules
91 my $version = MM->parse_version($_);
92 defined $version or $version = 'undef';
93 $version =~ /\d/ and $version = "'$version'";
94
95 # some heuristics to figure out the module name from the file name
f4ccb67a 96 $module =~ s{^(lib|cpan|dist|(?:symbian/)?ext)/}{}
3eae08df 97 and $1 ne 'lib'
dc2f75c0
JV
98 and (
99 $module =~ s{\b(\w+)/\1\b}{$1},
100 $module =~ s{^B/O}{O},
101 $module =~ s{^Devel-PPPort}{Devel},
3eae08df 102 $module =~ s{^libnet/}{},
dc2f75c0
JV
103 $module =~ s{^Encode/encoding}{encoding},
104 $module =~ s{^IPC-SysV/}{IPC/},
105 $module =~ s{^MIME-Base64/QuotedPrint}{MIME/QuotedPrint},
d9c6e45d
CBW
106 $module =~ s{^(?:DynaLoader|Errno|Opcode|XSLoader)/}{},
107 $module =~ s{^Sys-Syslog/win32}{Sys-Syslog},
108 $module =~ s{^Time-Piece/Seconds}{Time/Seconds},
dc2f75c0 109 );
71c80a8f 110 $module =~ s{^lib/}{}g;
dc2f75c0
JV
111 $module =~ s{/}{::}g;
112 $module =~ s{-}{::}g;
71c80a8f 113 $module =~ s{^.*::lib::}{}; # turns Foo/lib/Foo.pm into Foo.pm
dc2f75c0
JV
114 $module =~ s/(\.pm|_pm\.PL)$//;
115 $lines{$module} = $version;
116 $module_to_file{$module} = $File::Find::name;
117 },
71c80a8f 118 'symbian/ext',
dc2f75c0
JV
119 'lib',
120 'ext',
3eae08df 121 'cpan',
71c80a8f 122 'dist'
dc2f75c0 123);
59189dd7 124
e1b61e88 125-e 'configpm' and $lines{Config} = "$]";
cc8432b2 126
dc2f75c0
JV
127if ( open my $ucdv, "<", "lib/unicore/version" ) {
128 chomp( my $ucd = <$ucdv> );
0fdd9e5c 129 $lines{Unicode} = "'$ucd'";
59189dd7 130 close $ucdv;
dc2f75c0 131}
fb237dfd 132
1e955fab
RS
133my $delta_data = make_corelist_delta(
134 $perl_vnum,
135 \%lines,
136 \%Module::CoreList::version
137);
138
dc2f75c0 139my $versions_in_release = " " . $perl_vnum . " => {\n";
a272bf38
DL
140$versions_in_release .= " delta_from => $delta_data->{delta_from},\n";
141$versions_in_release .= " changed => {\n";
142foreach my $key (sort keys $delta_data->{changed}) {
143 $versions_in_release .= sprintf " %-24s=> %s,\n", "'$key'",
144 defined $delta_data->{changed}{$key} ? "'"
145 . $delta_data->{changed}{$key} . "'" : "undef";
0fdd9e5c 146}
a272bf38
DL
147$versions_in_release .= " },\n";
148$versions_in_release .= " removed => {\n";
149for my $key (sort keys($delta_data->{removed} || {})) {
621de13f 150 $versions_in_release .= sprintf " %-24s=> %s,\n", "'$key'", 1;
a272bf38
DL
151}
152$versions_in_release .= " }\n";
dc2f75c0 153$versions_in_release .= " },\n";
fb237dfd 154
dc852d37 155$corelist =~ s/^(my %delta\s*=\s*.*?)^\s*$perl_vnum\s*=>\s*{.*?},\s*(^\);)$/$1$2/ism;
a272bf38 156$corelist =~ s/^(my %delta\s*=\s*.*?)(^\);)$/$1$versions_in_release$2/ism;
fb237dfd
NC
157
158exit unless %modlist;
159
160# We have to go through this two stage lookup, given how Maintainers.pl keys its
161# data by "Module", which is really a dist.
dc2f75c0 162my $file_to_M = files_to_modules( values %module_to_file );
fb237dfd 163
febeafff
Z
164sub slurp_utf8($) {
165 open my $fh, "<:utf8", "$_[0]"
166 or die "can't open $_[0] for reading: $!";
167 return do { local $/; <$fh> };
168}
169
170sub parse_cpan_meta($) {
171 return Parse::CPAN::Meta->${
172 $_[0] =~ /\A\x7b/ ? \"load_json_string" : \"load_yaml_string"
173 }($_[0]);
174}
175
fb237dfd
NC
176my %module_to_upstream;
177my %module_to_dist;
178my %dist_to_meta_YAML;
a762e054 179my %module_to_deprecated;
dc2f75c0 180while ( my ( $module, $file ) = each %module_to_file ) {
fb237dfd
NC
181 my $M = $file_to_M->{$file};
182 next unless $M;
564ac620 183 next if $Modules{$M}{MAINTAINER} && $Modules{$M}{MAINTAINER} eq 'p5p';
fb237dfd 184 $module_to_upstream{$module} = $Modules{$M}{UPSTREAM};
a762e054 185 $module_to_deprecated{$module} = 1 if $Modules{$M}{DEPRECATED};
dc2f75c0
JV
186 next
187 if defined $module_to_upstream{$module}
0582d0ad 188 && $module_to_upstream{$module} eq 'blead';
fb237dfd
NC
189 my $dist = $modlist{$module};
190 unless ($dist) {
dc2f75c0
JV
191 warn "Can't find a distribution for $module\n";
192 next;
fb237dfd
NC
193 }
194 $module_to_dist{$module} = $dist;
195
196 next if exists $dist_to_meta_YAML{$dist};
197
198 $dist_to_meta_YAML{$dist} = undef;
199
200 # Like it or lump it, this has to be Unix format.
dc2f75c0 201 my $meta_YAML_path = "authors/id/$dist";
d9450ee2
DM
202 $meta_YAML_path =~ s/(?:tar\.gz|tar\.bz2|zip|tgz)$/meta/
203 or die "ERROR: bad meta YAML path: '$meta_YAML_path'";
dc2f75c0
JV
204 my $meta_YAML_url = 'http://ftp.funet.fi/pub/CPAN/' . $meta_YAML_path;
205
206 if ( -e "$cpan/$meta_YAML_path" ) {
febeafff 207 $dist_to_meta_YAML{$dist} = parse_cpan_meta(slurp_utf8( $cpan . "/" . $meta_YAML_path ));
dc2f75c0
JV
208 } elsif ( my $content = fetch_url($meta_YAML_url) ) {
209 unless ($content) {
210 warn "Failed to fetch $meta_YAML_url\n";
211 next;
212 }
febeafff 213 eval { $dist_to_meta_YAML{$dist} = parse_cpan_meta($content); };
dc2f75c0
JV
214 if ( my $err = $@ ) {
215 warn "$meta_YAML_path: ".$err;
216 next;
217 }
218 } else {
219 warn "$meta_YAML_path does not exist for $module\n";
220
221 # I tried code to open the tarballs with Archive::Tar to find and
222 # extract META.yml, but only Text-Tabs+Wrap-2006.1117.tar.gz had one,
223 # so it's not worth including.
224 next;
fb237dfd 225 }
fb237dfd
NC
226}
227
dc2f75c0
JV
228my $upstream_stanza = "%upstream = (\n";
229foreach my $module ( sort keys %module_to_upstream ) {
230 my $upstream = defined $module_to_upstream{$module} ? "'$module_to_upstream{$module}'" : 'undef';
231 $upstream_stanza .= sprintf " %-24s=> %s,\n", "'$module'", $upstream;
fb237dfd 232}
dc2f75c0
JV
233$upstream_stanza .= ");";
234
235$corelist =~ s/^%upstream .*? ;$/$upstream_stanza/ismx;
fb237dfd 236
a762e054 237# Deprecation generation
1e955fab
RS
238{
239 my $delta_data = make_corelist_delta(
240 $perl_vnum,
241 \%module_to_deprecated,
242 do { no warnings 'once'; \%Module::CoreList::deprecated },
243 );
244
245 my $deprecated_stanza = " " . $perl_vnum . " => {\n";
246 $deprecated_stanza .= " delta_from => $delta_data->{delta_from},\n";
247 $deprecated_stanza .= " changed => {\n";
248 foreach my $key (sort keys $delta_data->{changed}) {
249 $deprecated_stanza .= sprintf " %-24s=> %s,\n", "'$key'",
250 defined $delta_data->{changed}{$key} ? "'"
251 . $delta_data->{changed}{$key} . "'" : "undef";
252 }
253 $deprecated_stanza .= " },\n";
254 $deprecated_stanza .= " removed => {\n";
255 for my $key (sort keys($delta_data->{removed} || {})) {
256 $deprecated_stanza .= sprintf " %-24s=> %s,\n", "'$key'", 1;
257 }
258 $deprecated_stanza .= " }\n";
259 $deprecated_stanza .= " },\n";
260
dc852d37 261 $corelist =~ s/^(%deprecated\s*=\s*.*?)^\s*$perl_vnum\s*=>\s*{.*?},\s*(^\);)$/$1$2/ism;
1e955fab 262 $corelist =~ s/^(%deprecated\s*=\s*.*?)(^\);)$/$1$deprecated_stanza$2/xism;
a762e054 263}
a762e054 264
dc2f75c0
JV
265my $tracker = "%bug_tracker = (\n";
266foreach my $module ( sort keys %module_to_upstream ) {
fb237dfd 267 my $upstream = defined $module_to_upstream{$module};
dc2f75c0 268 next
0582d0ad 269 if defined $upstream and $upstream eq 'blead';
fb237dfd
NC
270
271 my $bug_tracker;
272
273 my $dist = $module_to_dist{$module};
274 $bug_tracker = $dist_to_meta_YAML{$dist}->{resources}{bugtracker}
dc2f75c0 275 if $dist;
d141a985 276 $bug_tracker = $bug_tracker->{web} if ref($bug_tracker) eq "HASH";
fb237dfd 277
a681e3bc 278 $bug_tracker = defined $bug_tracker ? quote($bug_tracker) : 'undef';
71c80a8f 279 next if $bug_tracker eq "'http://rt.perl.org/perlbug/'";
dc2f75c0
JV
280 $tracker .= sprintf " %-24s=> %s,\n", "'$module'", $bug_tracker;
281}
282$tracker .= ");";
283
284$corelist =~ s/^%bug_tracker .*? ;/$tracker/eismx;
285
dc2f75c0 286unless (
c99cfaa7
FR
287 $corelist =~ /^%released \s* = \s* \(
288 .*?
9f92b9be 289 $perl_vnum \s* => \s* .*?
dc2f75c0
JV
290 \);/ismx
291 )
292{
293 warn "Adding $perl_vnum to the list of released perl versions. Please consider adding a release date.\n";
294 $corelist =~ s/^(%released \s* = \s* .*?) ( \) )
fd48f21c 295 /$1 $perl_vnum => '????-??-??',\n $2/ismx;
dc2f75c0
JV
296}
297
dd4d388c
CBW
298write_corelist($corelist,$corelist_file);
299
300open( my $pod_fh, '<', $pod_file );
301my $pod = join( '', <$pod_fh> );
302
303unless ( $pod =~ /and $perl_vstring releases of perl/ ) {
304 warn "Adding $perl_vstring to the list of perl versions covered by Module::CoreList\n";
1c518207 305 $pod =~ s/(currently\s+covers\s+(?:.*?))\s*and\s+(.*?)\s+releases\s+of\s+perl/$1, $2 and $perl_vstring releases of perl/ism;
dd4d388c
CBW
306}
307
308write_corelist($pod,$pod_file);
dc2f75c0 309
d990e30d
SH
310open( my $utils_fh, '<', $utils_file );
311my $utils = join( '', <$utils_fh> );
312close $utils_fh;
313
1b3a25b0 314my %utils = map { ( $_ => 1 ) } parse_utils_lst();
d990e30d 315
1b3a25b0 316my $delta_utils = make_coreutils_delta($perl_vnum, \%utils);
d990e30d
SH
317
318write_corelist($utils,$utils_file);
dc2f75c0 319
1b3a25b0
SH
320warn "All done. Please check over the following files carefully before committing.\nThanks!\n";
321warn "$corelist_file\n$pod_file\n$utils_file\n";
322
dc2f75c0
JV
323sub write_corelist {
324 my $content = shift;
dd4d388c
CBW
325 my $filename = shift;
326 open (my $clfh, ">", $filename);
c89c9a06 327 binmode $clfh;
54233949 328 print $clfh $content;
dc2f75c0
JV
329 close($clfh);
330}
331
332sub fetch_url {
333 my $url = shift;
b87bfaf3
LB
334 my $http = HTTP::Tiny->new;
335 my $response = $http->get($url);
336 if ($response->{success}) {
337 return $response->{content};
338 } else {
339 warn "Error fetching $url: $response->{status} $response->{reason}\n";
340 return;
dc2f75c0 341 }
fb237dfd 342}
a272bf38
DL
343
344sub make_corelist_delta {
1e955fab 345 my($version, $lines, $existing) = @_;
a272bf38
DL
346 # Trust core perl, if someone does use a weird version number the worst that
347 # can happen is an extra delta entry for a module.
348 my %versions = map { $_ => eval $lines->{$_} } keys %$lines;
349
350 # Ensure we have the corelist data loaded from this perl checkout, not the system one.
351 require $corelist_file;
352
353 my %deltas;
354 # Search for the release with the least amount of changes (this avoids having
355 # to ask for where this perl was branched from).
1b3a25b0 356 for my $previous (reverse sort keys %$existing) {
a272bf38
DL
357 # Shouldn't happen, but ensure we don't load weird data...
358 next if $previous > $version || $previous == $version && $previous eq $version;
359
360 my $delta = $deltas{$previous} = {};
361 ($delta->{changed}, $delta->{removed}) = calculate_delta(
1e955fab 362 $existing->{$previous}, \%versions);
a272bf38
DL
363 }
364
365 my $smallest = (sort {
366 (keys($deltas{$a}->{changed}) + keys($deltas{$a}->{removed})) <=>
1b3a25b0 367 (keys($deltas{$b}->{changed}) + keys($deltas{$b}->{removed}))
a272bf38
DL
368 } keys %deltas)[0];
369
370 return {
371 delta_from => $smallest,
372 changed => $deltas{$smallest}{changed},
373 removed => $deltas{$smallest}{removed},
374 }
375}
376
758842b8
SH
377sub make_coreutils_delta {
378 my($version, $lines) = @_;
379 # Trust core perl, if someone does use a weird version number the worst that
380 # can happen is an extra delta entry for a module.
ae864c3a 381 my %utilities = map { $_ => eval $lines->{$_} } keys %$lines;
758842b8
SH
382
383 # Ensure we have the corelist data loaded from this perl checkout, not the system one.
ae864c3a 384 require $utils_file;
758842b8
SH
385
386 my %deltas;
387 # Search for the release with the least amount of changes (this avoids having
388 # to ask for where this perl was branched from).
ae864c3a 389 for my $previous (reverse sort keys %Module::CoreList::Utils::utilities) {
758842b8
SH
390 # Shouldn't happen, but ensure we don't load weird data...
391 next if $previous > $version || $previous == $version && $previous eq $version;
392
393 my $delta = $deltas{$previous} = {};
394 ($delta->{changed}, $delta->{removed}) = calculate_delta(
ae864c3a 395 $Module::CoreList::Utils::utilities{$previous}, \%utilities);
758842b8
SH
396 }
397
398 my $smallest = (sort {
399 (keys($deltas{$a}->{changed}) + keys($deltas{$a}->{removed})) <=>
1b3a25b0 400 (keys($deltas{$b}->{changed}) + keys($deltas{$b}->{removed}))
758842b8
SH
401 } keys %deltas)[0];
402
403 return {
404 delta_from => $smallest,
405 changed => $deltas{$smallest}{changed},
406 removed => $deltas{$smallest}{removed},
407 }
408}
409
a272bf38
DL
410# Calculate (changed, removed) modules between two versions.
411sub calculate_delta {
412 my($from, $to) = @_;
413 my(%changed, %removed);
414
415 for my $package(keys $from) {
416 if(not exists $to->{$package}) {
417 $removed{$package} = 1;
418 }
419 }
420
421 for my $package(keys $to) {
422 if(!exists $from->{$package}
423 || (defined $from->{$package} && !defined $to->{$package})
424 || (!defined $from->{$package} && defined $to->{$package})
425 || (defined $from->{$package} && defined $to->{$package}
426 && $from->{$package} ne $to->{$package})) {
427 $changed{$package} = $to->{$package};
428 }
429 }
430
431 return \%changed, \%removed;
432}
a681e3bc
FR
433
434sub quote {
435 my ($str) = @_;
436 # There's gotta be something already doing this properly that we could just
437 # reuse, but I can't quite thing of where to look for it, so I'm gonna do
0af431d3 438 # the simplest possible thing that'll allow me to release 5.17.7. --rafl
a681e3bc
FR
439 $str =~ s/'/\\'/g;
440 "'${str}'";
441}
d990e30d
SH
442
443sub parse_utils_lst {
444 require File::Spec::Unix;
445 my @scripts;
446 open my $fh, '<', 'utils.lst' or die "$!\n";
447 while (<$fh>) {
448 chomp;
449 my ($file,$extra) = split m!#!;
450 $file =~ s!\s+!!g;
451 push @scripts, $file;
452 $extra =~ s!\s+!!g if $extra;
453 if ( $extra and my ($link) = $extra =~ m!^link=(.+?)$! ) {
454 push @scripts, $link;
455 }
456 }
457 return map { +( File::Spec::Unix->splitpath( $_ ) )[-1] } @scripts;
458}