This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
avoid decoding on input in the new tests added in 613c63b4
[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';
dd4d388c 25my $pod_file = 'dist/Module-CoreList/lib/Module/CoreList.pod';
e1018a69 26
59189dd7 27my %lines;
fb237dfd
NC
28my %module_to_file;
29my %modlist;
e1018a69 30
dc2f75c0
JV
31die "usage: $0 [ cpan-mirror/ ] [ 5.x.y] \n" unless @ARGV <= 2;
32my $cpan = shift;
33my $raw_version = shift || $];
34my $perl_version = version->parse("$raw_version");
35my $perl_vnum = $perl_version->numify;
36my $perl_vstring = $perl_version->normal; # how do we get version.pm to not give us leading v?
37$perl_vstring =~ s/^v//;
fb237dfd 38
dc2f75c0
JV
39if ( !-f 'MANIFEST' ) {
40 die "Must be run from the root of a clean perl tree\n";
e1018a69
DM
41}
42
54233949 43open( my $corelist_fh, '<', $corelist_file );
dc2f75c0
JV
44my $corelist = join( '', <$corelist_fh> );
45
fb237dfd 46if ($cpan) {
dc2f75c0
JV
47 my $modlistfile = File::Spec->catfile( $cpan, 'modules', '02packages.details.txt' );
48 my $content;
49
50 my $fh;
51 if ( -e $modlistfile ) {
52 warn "Reading the module list from $modlistfile";
54233949 53 open $fh, '<', $modlistfile;
dc2f75c0 54 } elsif ( -e $modlistfile . ".gz" ) {
a762e054 55 my $zcat = can_run('gzcat') || can_run('zcat') or die "Can't find gzcat or zcat";
dc2f75c0 56 warn "Reading the module list from $modlistfile.gz";
54233949 57 open $fh, '-|', "$zcat $modlistfile.gz";
dc2f75c0
JV
58 } else {
59 warn "About to fetch 02packages from ftp.funet.fi. This may take a few minutes\n";
9168ed13
LB
60 my $gzipped_content = fetch_url('http://ftp.funet.fi/pub/CPAN/modules/02packages.details.txt.gz');
61 unless ($gzipped_content) {
dc2f75c0
JV
62 die "Unable to read 02packages.details.txt from either your CPAN mirror or ftp.funet.fi";
63 }
9168ed13
LB
64 IO::Uncompress::Gunzip::gunzip(\$gzipped_content, \$content, Transparent => 0)
65 or die "Can't gunzip content: $IO::Uncompress::Gunzip::GunzipError";
dc2f75c0
JV
66 }
67
68 if ( $fh and !$content ) {
69 local $/ = "\n";
70 $content = join( '', <$fh> );
fb237dfd
NC
71 }
72
dc2f75c0
JV
73 die "Incompatible modlist format"
74 unless $content =~ /^Columns: +package name, version, path/m;
75
fb237dfd
NC
76 # Converting the file to a hash is about 5 times faster than a regexp flat
77 # lookup.
dc2f75c0
JV
78 for ( split( qr/\n/, $content ) ) {
79 next unless /^([A-Za-z_:0-9]+) +[-0-9.undefHASHVERSIONvsetwhenloadingbogus]+ +(\S+)/;
80 $modlist{$1} = $2;
fb237dfd
NC
81 }
82}
83
dc2f75c0
JV
84find(
85 sub {
86 /(\.pm|_pm\.PL)$/ or return;
87 /PPPort\.pm$/ and return;
88 my $module = $File::Find::name;
89 $module =~ /\b(demo|t|private)\b/ and return; # demo or test modules
90 my $version = MM->parse_version($_);
91 defined $version or $version = 'undef';
92 $version =~ /\d/ and $version = "'$version'";
93
94 # some heuristics to figure out the module name from the file name
f4ccb67a 95 $module =~ s{^(lib|cpan|dist|(?:symbian/)?ext)/}{}
3eae08df 96 and $1 ne 'lib'
dc2f75c0
JV
97 and (
98 $module =~ s{\b(\w+)/\1\b}{$1},
99 $module =~ s{^B/O}{O},
100 $module =~ s{^Devel-PPPort}{Devel},
3eae08df 101 $module =~ s{^libnet/}{},
dc2f75c0
JV
102 $module =~ s{^Encode/encoding}{encoding},
103 $module =~ s{^IPC-SysV/}{IPC/},
104 $module =~ s{^MIME-Base64/QuotedPrint}{MIME/QuotedPrint},
d9c6e45d
CBW
105 $module =~ s{^(?:DynaLoader|Errno|Opcode|XSLoader)/}{},
106 $module =~ s{^Sys-Syslog/win32}{Sys-Syslog},
107 $module =~ s{^Time-Piece/Seconds}{Time/Seconds},
dc2f75c0 108 );
f4ccb67a 109 $module =~ s{^vms/ext}{VMS};
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
JV
118 'vms/ext',
119 'symbian/ext',
dc2f75c0
JV
120 'lib',
121 'ext',
3eae08df 122 'cpan',
71c80a8f 123 'dist'
dc2f75c0 124);
59189dd7 125
0fdd9e5c 126-e 'configpm' and $lines{Config} = 'undef';
cc8432b2 127
dc2f75c0
JV
128if ( open my $ucdv, "<", "lib/unicore/version" ) {
129 chomp( my $ucd = <$ucdv> );
0fdd9e5c 130 $lines{Unicode} = "'$ucd'";
59189dd7 131 close $ucdv;
dc2f75c0 132}
fb237dfd 133
dc2f75c0
JV
134my $versions_in_release = " " . $perl_vnum . " => {\n";
135foreach my $key ( sort keys %lines ) {
136 $versions_in_release .= sprintf "\t%-24s=> %s,\n", "'$key'", $lines{$key};
0fdd9e5c 137}
dc2f75c0 138$versions_in_release .= " },\n";
fb237dfd 139
dc2f75c0 140$corelist =~ s/^(%version\s*=\s*.*?)(^\);)$/$1$versions_in_release$2/xism;
fb237dfd
NC
141
142exit unless %modlist;
143
144# We have to go through this two stage lookup, given how Maintainers.pl keys its
145# data by "Module", which is really a dist.
dc2f75c0 146my $file_to_M = files_to_modules( values %module_to_file );
fb237dfd 147
febeafff
Z
148sub slurp_utf8($) {
149 open my $fh, "<:utf8", "$_[0]"
150 or die "can't open $_[0] for reading: $!";
151 return do { local $/; <$fh> };
152}
153
154sub parse_cpan_meta($) {
155 return Parse::CPAN::Meta->${
156 $_[0] =~ /\A\x7b/ ? \"load_json_string" : \"load_yaml_string"
157 }($_[0]);
158}
159
fb237dfd
NC
160my %module_to_upstream;
161my %module_to_dist;
162my %dist_to_meta_YAML;
a762e054 163my %module_to_deprecated;
dc2f75c0 164while ( my ( $module, $file ) = each %module_to_file ) {
fb237dfd
NC
165 my $M = $file_to_M->{$file};
166 next unless $M;
564ac620 167 next if $Modules{$M}{MAINTAINER} && $Modules{$M}{MAINTAINER} eq 'p5p';
fb237dfd 168 $module_to_upstream{$module} = $Modules{$M}{UPSTREAM};
a762e054 169 $module_to_deprecated{$module} = 1 if $Modules{$M}{DEPRECATED};
dc2f75c0
JV
170 next
171 if defined $module_to_upstream{$module}
172 && $module_to_upstream{$module} =~ /^(?:blead|first-come)$/;
fb237dfd
NC
173 my $dist = $modlist{$module};
174 unless ($dist) {
dc2f75c0
JV
175 warn "Can't find a distribution for $module\n";
176 next;
fb237dfd
NC
177 }
178 $module_to_dist{$module} = $dist;
179
180 next if exists $dist_to_meta_YAML{$dist};
181
182 $dist_to_meta_YAML{$dist} = undef;
183
184 # Like it or lump it, this has to be Unix format.
dc2f75c0 185 my $meta_YAML_path = "authors/id/$dist";
df6d85b8 186 $meta_YAML_path =~ s/(?:tar\.gz|tar\.bz2|zip|tgz)$/meta/ or die "$meta_YAML_path";
dc2f75c0
JV
187 my $meta_YAML_url = 'http://ftp.funet.fi/pub/CPAN/' . $meta_YAML_path;
188
189 if ( -e "$cpan/$meta_YAML_path" ) {
febeafff 190 $dist_to_meta_YAML{$dist} = parse_cpan_meta(slurp_utf8( $cpan . "/" . $meta_YAML_path ));
dc2f75c0
JV
191 } elsif ( my $content = fetch_url($meta_YAML_url) ) {
192 unless ($content) {
193 warn "Failed to fetch $meta_YAML_url\n";
194 next;
195 }
febeafff 196 eval { $dist_to_meta_YAML{$dist} = parse_cpan_meta($content); };
dc2f75c0
JV
197 if ( my $err = $@ ) {
198 warn "$meta_YAML_path: ".$err;
199 next;
200 }
201 } else {
202 warn "$meta_YAML_path does not exist for $module\n";
203
204 # I tried code to open the tarballs with Archive::Tar to find and
205 # extract META.yml, but only Text-Tabs+Wrap-2006.1117.tar.gz had one,
206 # so it's not worth including.
207 next;
fb237dfd 208 }
fb237dfd
NC
209}
210
dc2f75c0
JV
211my $upstream_stanza = "%upstream = (\n";
212foreach my $module ( sort keys %module_to_upstream ) {
213 my $upstream = defined $module_to_upstream{$module} ? "'$module_to_upstream{$module}'" : 'undef';
214 $upstream_stanza .= sprintf " %-24s=> %s,\n", "'$module'", $upstream;
fb237dfd 215}
dc2f75c0
JV
216$upstream_stanza .= ");";
217
218$corelist =~ s/^%upstream .*? ;$/$upstream_stanza/ismx;
fb237dfd 219
a762e054
DG
220# Deprecation generation
221my $deprecated_stanza = " " . $perl_vnum . " => {\n";
222foreach my $module ( sort keys %module_to_deprecated ) {
223 my $deprecated = defined $module_to_deprecated{$module} ? "'$module_to_deprecated{$module}'" : 'undef';
224 $deprecated_stanza .= sprintf "\t%-24s=> %s,\n", "'$module'", $deprecated;
225}
226$deprecated_stanza .= " },\n";
227$corelist =~ s/^(%deprecated\s*=\s*.*?)(^\);)$/$1$deprecated_stanza$2/xism;
228
dc2f75c0
JV
229my $tracker = "%bug_tracker = (\n";
230foreach my $module ( sort keys %module_to_upstream ) {
fb237dfd 231 my $upstream = defined $module_to_upstream{$module};
dc2f75c0
JV
232 next
233 if defined $upstream
234 and $upstream eq 'blead' || $upstream eq 'first-come';
fb237dfd
NC
235
236 my $bug_tracker;
237
238 my $dist = $module_to_dist{$module};
239 $bug_tracker = $dist_to_meta_YAML{$dist}->{resources}{bugtracker}
dc2f75c0 240 if $dist;
d141a985 241 $bug_tracker = $bug_tracker->{web} if ref($bug_tracker) eq "HASH";
fb237dfd
NC
242
243 $bug_tracker = defined $bug_tracker ? "'$bug_tracker'" : 'undef';
71c80a8f 244 next if $bug_tracker eq "'http://rt.perl.org/perlbug/'";
dc2f75c0
JV
245 $tracker .= sprintf " %-24s=> %s,\n", "'$module'", $bug_tracker;
246}
247$tracker .= ");";
248
249$corelist =~ s/^%bug_tracker .*? ;/$tracker/eismx;
250
dc2f75c0 251unless (
c99cfaa7
FR
252 $corelist =~ /^%released \s* = \s* \(
253 .*?
254 $perl_vnum => .*?
dc2f75c0
JV
255 \);/ismx
256 )
257{
258 warn "Adding $perl_vnum to the list of released perl versions. Please consider adding a release date.\n";
259 $corelist =~ s/^(%released \s* = \s* .*?) ( \) )
fd48f21c 260 /$1 $perl_vnum => '????-??-??',\n $2/ismx;
dc2f75c0
JV
261}
262
dd4d388c
CBW
263write_corelist($corelist,$corelist_file);
264
265open( my $pod_fh, '<', $pod_file );
266my $pod = join( '', <$pod_fh> );
267
268unless ( $pod =~ /and $perl_vstring releases of perl/ ) {
269 warn "Adding $perl_vstring to the list of perl versions covered by Module::CoreList\n";
1c518207 270 $pod =~ s/(currently\s+covers\s+(?:.*?))\s*and\s+(.*?)\s+releases\s+of\s+perl/$1, $2 and $perl_vstring releases of perl/ism;
dd4d388c
CBW
271}
272
273write_corelist($pod,$pod_file);
dc2f75c0 274
dd4d388c 275warn "All done. Please check over $corelist_file and $pod_file carefully before committing. Thanks!\n";
dc2f75c0
JV
276
277
278sub write_corelist {
279 my $content = shift;
dd4d388c
CBW
280 my $filename = shift;
281 open (my $clfh, ">", $filename);
54233949 282 print $clfh $content;
dc2f75c0
JV
283 close($clfh);
284}
285
286sub fetch_url {
287 my $url = shift;
b87bfaf3
LB
288 my $http = HTTP::Tiny->new;
289 my $response = $http->get($url);
290 if ($response->{success}) {
291 return $response->{content};
292 } else {
293 warn "Error fetching $url: $response->{status} $response->{reason}\n";
294 return;
dc2f75c0 295 }
fb237dfd 296}