File Coverage

File:lib/CheckSpelling/Apply.pm
Coverage:72.8%

linestmtbrancondsubtimecode
1
1
1
1
1
255302
1
20
package CheckSpelling::Apply; use CheckSpelling::Util; #!/usr/bin/env perl
2":" || q@<<"=END_OF_PERL"@;
3
4
1
1
1
1
2
1
22
use Symbol 'gensym';
5
1
1
1
1
189
1230
24
use IPC::Open3;
6
1
1
1
1
2
0
19
use File::Basename qw(dirname);
7
1
1
1
1
2
0
14
use File::Path qw(make_path);
8
1
1
1
1
155
286
33
use File::Spec::Functions qw(catfile path);
9
1
1
1
1
2
0
15
use File::Temp qw/ tempfile tempdir /;
10
1
1
1
1
2
0
24
use JSON::PP;
11
1
1
1
1
4
0
2455
use warnings;
12
13my @safe_path = qw(
14    /opt/homebrew/bin
15    /opt/homebrew/sbin
16    /usr/local/bin
17    /usr/bin
18    /bin
19    /usr/sbin
20    /sbin
21);
22
23my $bin = glob("~/bin");
24push @safe_path, $bin if -d $bin;
25
26my $ua = 'check-spelling-agent/0.0.4';
27
28$ENV{'PATH'} = join ':', @safe_path unless defined $ENV{SYSTEMROOT};
29
30sub check_exists_command {
31
12
12
9
    my ($command) = @_;
32
33
12
15
    my @path = path;
34
12
85
    my @pathext = ('');
35
36
12
100
20
    if ($^O eq 'MSWin32') {
37
1
4
3
4
        push @pathext, map { lc } split /;/, $ENV{PATHEXT};
38    }
39
40
12
4
    for my $dir (@path) {
41
51
29
        for my $suffix (@pathext) {
42
63
79
            my $f = catfile $dir, "$command$suffix";
43
63
100
297
            return $f if -x $f;
44        }
45    }
46}
47
48sub needs_command_because {
49
11
11
1944
    my ($command, $reason) = @_;
50
11
100
11
    return if check_exists_command($command);
51
1
7
    CheckSpelling::Util::die_custom $program, 51, 'Please install `'.$command.'` - it is needed to '.$reason;
52}
53
54sub check_basic_tools {
55
3
3
1427
    needs_command_because('git', 'interact with git repositories');
56
3
3
    needs_command_because('curl', 'download other tools');
57
3
3
    needs_command_because('gh', 'interact with github');
58    #needs_command_because('magic-magic', 'debugging');
59}
60
61sub get_token {
62
7
7
1386
    our $token;
63
7
100
67
35
    return $token if defined $token && $token ne '';
64
4
100
9
    $token = $ENV{'GH_TOKEN'} || $ENV{'GITHUB_TOKEN'};
65
4
100
67
10
    return $token if defined $token && $token ne '';
66
1
0
    my ($err, $exit);
67
1
2
    ($token, $err, $exit) = capture_system('gh', 'auth', 'token');
68
1
5
    chomp $token;
69
1
1
    chomp $err;
70
1
5
    return ($token, $err, $exit);
71};
72
73sub download_with_curl {
74
3
3
3
    my ($url, $dest, $flags) = @_;
75
3
50
10
    $flags = '-fsL' unless defined $flags;
76
3
136247
    system('curl',
77        '--connect-timeout', 3,
78        '-A', $ua,
79        $flags,
80        '-o', $dest,
81        $url
82    );
83}
84
85sub tempfile_name {
86
8
8
26
    my ($fh, $filename) = tempfile();
87
8
1323
    close $fh;
88
8
14
    return $filename;
89}
90
91sub strip_comments {
92
8
8
7
    my ($file) = @_;
93
8
32
    my ($fh, $filename) = tempfile();
94
8
1481
    open INPUT, '<', $file;
95
8
106
    while (<INPUT>) {
96
4728
100
2513
        next if /^\s*(?:#.*)/;
97
4655
3038
        print $fh $_;
98    }
99
8
26
    close INPUT;
100
8
96
    close $fh;
101
8
15
    return $filename;
102}
103
104sub capture_system {
105
28
28
84
    my @args = @_;
106
28
125
    my $pid = open3(my $child_in, my $child_out, my $child_err = gensym, @args);
107
28
62667
    my (@err, @out);
108
28
1223363
    while (my $output = <$child_out>) {
109
22
2586
        push @out, $output;
110    }
111
28
675
    while (my $error = <$child_err>) {
112
21
112
        push @err, $error;
113    }
114
28
371
    waitpid( $pid, 0 );
115
28
131
    my $child_exit_status = $?;
116
28
79
    my $output_joined = join '', @out;
117
28
52
    my $error_joined = join '', @err;
118
28
798
    return ($output_joined, $error_joined, $child_exit_status);
119}
120
121sub capture_merged_system {
122
10
10
26
    my ($output_joined, $error_joined, $child_exit_status) = capture_system(@_);
123
10
50
42
    my $joiner = ($output_joined ne '') ? "\n" : '';
124
10
38
    return ($output_joined.$joiner.$error_joined, $child_exit_status);
125}
126
127sub compare_files {
128
4
4
1370
    my ($one, $two) = @_;
129
4
5
    my $one_stripped = strip_comments($one);
130
4
4
    my $two_stripped = strip_comments($two);
131
4
6
    my $exit_code;
132
4
8
    (undef, undef, $exit_code) = capture_system(
133            'diff',
134            '-qwB',
135            $one_stripped, $two_stripped
136        );
137
4
50
9
    if ($? == -1) {
138
0
0
        print "could not compare '$one' and '$two': $!\n";
139
0
0
        return 0;
140    }
141
4
50
4
    if ($? & 127) {
142
0
0
0
        printf "child died with signal %d, %s core dump\n",
143        ($? & 127),  ($? & 128) ? 'with' : 'without';
144
0
0
        return 0;
145    }
146
4
50
5
    return 0 if $? == 0;
147
4
14
    return 1;
148}
149
150my $bash_script=q{
151=END_OF_PERL@
152# bash
153set -e
154if [ "$OUTPUT" = "$ERROR" ]; then
155    ("$@" 2>&1) > "$OUTPUT"
156else
157    "$@" > "$OUTPUT" 2> "$ERROR"
158fi
159exit
160};
161
162sub check_current_script {
163
3
50
3
12777
    if ("$0" eq '-') {
164
0
0
        my ($bash_script) = @_;
165
0
0
        my $fh;
166
0
0
        ($fh, $0) = tempfile();
167
0
0
        $bash_script =~ s/^=.*\@$//m;
168
0
0
        print $fh $bash_script;
169
0
0
        close $fh;
170
0
0
        return;
171    }
172
3
5
    my $filename = tempfile_name();
173
3
5
    my $source = 'https://raw.githubusercontent.com/check-spelling/check-spelling/prerelease/apply.pl';
174
3
5
    download_with_curl($source, $filename);
175
3
50
53
    if ($? == 0) {
176
3
50
18
        if (compare_files($filename, $0)) {
177
3
47
            print "Current apply script differs from '$source' (locally downloaded to `$filename`). You may wish to upgrade.\n";
178        }
179    }
180}
181
182sub die_with_message {
183
5
5
5
    our $program;
184
5
7
    my ($gh_err_text) = @_;
185
5
50
33
29
    if ($gh_err_text =~ /error connecting to / && $gh_err_text =~ /check your internet connection/) {
186
0
0
        print "$program: Internet access may be limited. Check your connection (this often happens with lousy cable internet service providers where their CG-NAT or whatever strands the modem).\n\n$gh_err_text";
187
0
0
0
0
0
0
        { CheckSpelling::Util::tear_here(5); die "exiting"; }
188    }
189
5
100
17
    if ($gh_err_text =~ /proxyconnect tcp:.*connect: connection refused/) {
190
1
27
        print "$program: Proxy is not accepting connections.\n";
191
1
6
        for my $proxy (qw(http_proxy HTTP_PROXY https_proxy HTTPS_PROXY)) {
192
4
100
9
            if (defined $ENV{$proxy}) {
193
1
5
                print "  $proxy: '$ENV{$proxy}'\n";
194            }
195        }
196
1
5
        print "\n$gh_err_text";
197
1
1
1
0
5
36
        { CheckSpelling::Util::tear_here(6); die "exiting"; }
198    }
199
4
100
12
    if ($gh_err_text =~ /dial unix .*: connect: .*/) {
200
1
26
        print "$program: Unix http socket is not working.\n";
201
1
27972
        my $gh_http_unix_socket = `gh config get http_unix_socket`;
202
1
20
        print "  http_unix_socket: $gh_http_unix_socket\n";
203
1
6
        print "\n$gh_err_text";
204
1
1
1
8
13
21
        { CheckSpelling::Util::tear_here(7); die "exiting"; }
205    }
206}
207
208sub gh_is_happy_internal {
209
5
5
17
    my ($output, $exit_code) = capture_merged_system(qw(gh api /installation/repositories));
210
5
50
14
    return ($exit_code, $output) if $exit_code == 0;
211
5
43
    ($output, $exit_code) = capture_merged_system(qw(gh api /user));
212
5
14
    return ($exit_code, $output);
213}
214
215sub gh_is_happy {
216
3
3
6
    my ($program) = @_;
217
3
7
    my ($gh_auth_status, $gh_status_lines) = gh_is_happy_internal();
218
3
50
11
    return 1 if $gh_auth_status == 0;
219
3
11
    die_with_message($gh_status_lines);
220
221
1
2
    my @problematic_env_variables;
222
1
3
    for my $variable (qw(GH_TOKEN GITHUB_TOKEN GITHUB_ACTIONS CI)) {
223
4
100
9
        if (defined $ENV{$variable}) {
224
2
23
            delete $ENV{$variable};
225
2
4
            push @problematic_env_variables, $variable;
226
2
2
            ($gh_auth_status, $gh_status_lines) = gh_is_happy_internal();
227
2
50
12
            if ($gh_auth_status == 0) {
228
0
0
                print STDERR "$0: gh program did not like these environment variables: ".join(', ', @problematic_env_variables)." -- consider unsetting them.\n";
229
0
0
                return 1;
230            }
231        }
232    }
233
234
1
35
    print $gh_status_lines;
235
1
9
    return 0;
236}
237
238sub tools_are_ready {
239
3
3
100903
    my ($program) = @_;
240
3
50
7
    unless (gh_is_happy($program)) {
241
1
4
        $! = 1;
242
1
50
33
9
        my $or_gh_token = (defined $ENV{CI} && $ENV{CI}) ? ' or set the GH_TOKEN environment variable' : '';
243
1
7
        CheckSpelling::Util::die_custom $program, 243, "$program requires a happy gh, please try 'gh auth login'$or_gh_token\n";
244    }
245}
246
247sub run_pipe {
248
8
8
26
    my @args = @_;
249
8
15
    my ($out, undef, $exit_code) = capture_system(@args);
250
8
45
    return $out;
251}
252
253sub unzip_pipe {
254
6
6
19
    my ($artifact, $file) = @_;
255
6
14
    return run_pipe(
256        'unzip',
257        '-p', $artifact,
258        $file
259    );
260}
261
262sub retrieve_spell_check_this {
263
1
1
3
    my ($artifact, $config_ref) = @_;
264
1
1
    my $spell_check_this_config = unzip_pipe($artifact, 'spell_check_this.json');
265
1
50
9
    return unless $spell_check_this_config =~ /\{.*\}/s;
266
1
2
    my %config;
267
1
1
1
50
2
2
6
    eval { %config = %{decode_json $spell_check_this_config}; } || CheckSpelling::Util::die_custom $program, 267, "decode_json failed in retrieve_spell_check_this with '$spell_check_this_config'";
268
1
388
    my ($repo, $branch, $destination, $path) = ($config{url}, $config{branch}, $config{config}, $config{path});
269
1
4
    my $spell_check_this_dir = tempdir();
270
1
189
    my $exit_code;
271
1
2
    (undef, undef, $exit_code) = capture_system(
272            'git', 'clone',
273            '--depth', '1',
274            '--no-tags',
275            $repo,
276            '--branch', $branch,
277            $spell_check_this_dir
278        );
279
1
50
4
    if ($?) {
280
0
0
        CheckSpelling::Util::die_custom $program, 280, "git clone $repo#$branch failed";
281    }
282
283
1
168
    make_path($destination);
284
1
2276
    system('cp', '-i', '-R', glob("$spell_check_this_dir/$path/*"), $destination);
285
1
2715
    system('git', 'add', '-f', $destination);
286}
287
288sub case_biased {
289
1
1
5
    lc($a)."-".$a cmp lc($b)."-".$b;
290}
291
292sub add_to_excludes {
293
1
1
3
    my ($artifact, $config_ref) = @_;
294
1
2
    my $excludes = $config_ref->{"excludes_file"};
295
1
2
    my $should_exclude_patterns = unzip_pipe($artifact, 'should_exclude.patterns');
296
1
50
6
    unless ($should_exclude_patterns =~ /\w/) {
297
1
2
        $should_exclude_patterns = unzip_pipe($artifact, 'should_exclude.txt');
298
1
50
10
        return unless $should_exclude_patterns =~ /\w/;
299
1
10
        $should_exclude_patterns =~ s{^(.*)}{^\\Q$1\\E\$}gm;
300    }
301
1
1
    my $need_to_add_excludes;
302    my %excludes;
303
1
50
8
    if (-f $excludes) {
304
1
10
        open EXCLUDES, '<', $excludes;
305
1
9
        while (<EXCLUDES>) {
306
1
1
            chomp;
307
1
50
3
            next unless /./;
308
1
4
            $excludes{$_."\n"} = 1;
309        }
310
1
3
        close EXCLUDES;
311    } else {
312
0
0
        $need_to_add_excludes = 1;
313    }
314
1
4
    for $pattern (split /\n/, $should_exclude_patterns) {
315
1
50
2
        next unless $pattern =~ /./;
316
1
2
        $excludes{$pattern."\n"} = 1;
317    }
318
1
25
    open EXCLUDES, '>', $excludes;
319
1
10
    print EXCLUDES join "", sort case_biased keys %excludes;
320
1
38
    close EXCLUDES;
321
1
50
4
    system('git', 'add', '--', $excludes) if $need_to_add_excludes;
322}
323
324sub remove_stale {
325
1
1
2
    my ($artifact, $config_ref) = @_;
326
1
3
    my @stale = split /\s+/s, unzip_pipe($artifact, 'remove_words.txt');
327
1
50
4
    return unless @stale;
328
1
1
1
4
    my @expect_files = @{$config_ref->{"expect_files"}};
329    @expect_files = grep {
330
1
1
50
4
7
        print STDERR "Could not find $_\n" unless -f $_;
331
1
5
        -f $_;
332    } @expect_files;
333
1
50
1
    unless (@expect_files) {
334
0
0
        CheckSpelling::Util::die_custom $program, 334, "Could not find any of the processed expect files, are you on the wrong branch?";
335    }
336
337
1
2
    my $re = join "|", @stale;
338
1
1
    for my $file (@expect_files) {
339
1
9
        open INPUT, '<', $file;
340
1
2
        my @keep;
341
1
12
        while (<INPUT>) {
342
2
100
44
            next if /^(?:$re)(?:(?:\r|\n)*$|[# ].*)/;
343
1
2
            push @keep, $_;
344        }
345
1
4
        close INPUT;
346
347
1
27
        open OUTPUT, '>', $file;
348
1
5
        print OUTPUT join '', @keep;
349
1
40
        close OUTPUT;
350    };
351}
352
353sub add_expect {
354
1
1
1
    my ($artifact, $config_ref) = @_;
355
1
1
    my @add = split /\s+/s, (unzip_pipe($artifact, 'tokens.txt'));
356
1
50
5
    return unless @add;
357
1
1
    my $new_expect_file = $config_ref->{"new_expect_file"};
358
1
1
    my @words;
359
1
119
    make_path (dirname($new_expect_file));
360
1
50
9
    if (-s $new_expect_file) {
361
0
0
        open FILE, q{<}, $new_expect_file;
362
0
0
        local $/ = undef;
363
0
0
        @words = split /\s+/, <FILE>;
364
0
0
        close FILE;
365    }
366
1
1
    my %items;
367
1
1
    @items{@words} = @words x (1);
368
1
7
    @items{@add} = @add x (1);
369
1
5
    @words = sort case_biased keys %items;
370
1
24
    open FILE, q{>}, $new_expect_file;
371
1
2
    for my $word (@words) {
372
1
50
6
        print FILE "$word\n" if $word =~ /\S/;
373    };
374
1
18
    close FILE;
375
1
2820
    system("git", "add", $new_expect_file);
376}
377
378sub get_artifact_metadata {
379
3
3
3
    my ($url) = @_;
380
3
4
    my $json_file = tempfile_name();
381
3
3
    my ($curl_stdout, $curl_stderr, $curl_result);
382
3
14
    my @curl_args = (
383        'curl',
384        '-L',
385        $url,
386        '-A',
387        $ua,
388        '-s',
389        '--fail-with-body',
390    );
391
3
7
    my ($gh_token) = get_token();
392
3
50
6
    push @curl_args, '-u', "token:$gh_token" if defined $gh_token;
393
3
5
    push @curl_args, (
394        '-o',
395        $json_file
396    );
397
3
3
    ($curl_stdout, $curl_stderr, $curl_result) = capture_system(
398        @curl_args
399    );
400
3
100
8
    unless ($curl_result == 0) {
401
1
50
6
        if ($curl_stdout eq '') {
402
1
7
            local $/;
403
1
18
            open my $error_fh, '<', $json_file;
404
1
12
            $curl_stdout = <$error_fh>;
405
1
5
            close $error_fh;
406        }
407        return (
408
1
12
            out    => $curl_stdout,
409            err    => $curl_stderr,
410            result => $curl_result,
411        );
412    }
413
2
4
    my $link;
414
2
32
    open my $json_file_fh, '<', $json_file;
415
2
29
    my ($id, $download_url, $count);
416    {
417
2
2
2
9
        local $/;
418
2
21
        my $content = <$json_file_fh>;
419
2
14
        my $json = decode_json $content;
420
2
5109
        my $artifact = $json->{'artifacts'}->[0];
421
2
3
        $id = $artifact->{'id'};
422
2
1
        $download_url = $artifact->{'archive_download_url'};
423
2
8
        $count = $json->{'total_count'};
424    }
425
2
7
    close $json_file_fh;
426
2
50
3
    if ($count == 0) {
427        return (
428
0
0
            out => '',
429            err => 'no artifact matches any of the names or patterns provided',
430            result => (3 << 8),
431        );
432    }
433    return (
434
2
19
        id       => $id,
435        download => $download_url,
436        count    => $count,
437    );
438}
439
440sub get_latest_artifact_metadata {
441
2
2
4
    my ($artifact_dir, $repo, $run, $artifact_name) = @_;
442
2
9
    my $page = 1;
443
2
5
    my $url = "$ENV{GITHUB_API_URL}/repos/$repo/actions/runs/$run/artifacts?name=$artifact_name&per_page=1&page=";
444
2
6
    my %first = get_artifact_metadata($url.$page);
445
2
4
    $page = $first{'count'};
446
2
100
7
    if (defined $page) {
447
1
3
        my %second = get_artifact_metadata($url.$page);
448
1
5
        my ($id_1, $id_2) = ($first{'id'}, $second{'id'});
449
1
50
33
7
        if (defined $id_1 && defined $id_2) {
450
1
50
3
            if ($id_2 > $id_1) {
451                return (
452
0
0
                    download => $second{'download'},
453                );
454            }
455        }
456    }
457
2
4
    my $download = $first{'download'};
458
2
100
6
    if (defined $download) {
459        return (
460
1
5
            download => $download,
461        );
462    }
463
1
6
    return %first;
464}
465
466sub download_latest_artifact {
467
2
2
4
    my %maybe_download = get_latest_artifact_metadata(@_);
468
2
2
    my $download = $maybe_download{'download'};
469
2
7
    my $zip_file = tempfile_name();
470
2
100
2
    if (defined $download) {
471
1
7
        my @curl_args = (
472            'curl',
473            $download,
474            '-L',
475            '-A',
476            $ua,
477            '-s',
478            '--fail-with-body',
479        );
480
1
2
        my ($gh_token) = get_token();
481
1
50
2
        push @curl_args, '-u', "token:$gh_token" if defined $gh_token;
482
1
2
        push @curl_args, (
483            '-o',
484            $zip_file
485        );
486
1
1
        ($curl_stdout, $curl_stderr, $curl_result) = capture_system(
487            @curl_args
488        );
489
1
50
7
        if ($curl_result != 0) {
490
1
50
3
            if ($curl_stdout eq '') {
491
1
4
                local $/;
492
1
17
                open my $error_fh, '<', $zip_file;
493
1
11
                $curl_stdout = <$error_fh>;
494
1
5
                close $error_fh;
495            }
496
1
10
            return ("$curl_stdout\n$curl_stderr", $curl_result);
497        }
498
0
0
        my ($artifact_dir, $repo, $run, $artifact_name) = @_;
499
0
0
        ($out, $err, $result) = capture_system(
500            'unzip',
501            '-q',
502            $zip_file,
503            '-d',
504            $artifact_dir,
505            );
506
0
0
        return ("$out\n$err", $result);
507    }
508
1
4
    my ($out, $err, $result) = ($maybe_download{'out'}, $maybe_download{'err'}, $maybe_download{'result'});
509
1
5
    return ("$out\n$err", $result);
510}
511
512sub get_artifacts {
513
2
2
3631
    my ($repo, $run, $suffix) = @_;
514
2
2
    our $program;
515
2
4
    my $artifact_dir = tempdir(CLEANUP => 1);
516
2
357
    my $gh_err_text;
517
2
2
    my $artifact_name = 'check-spelling-comment';
518
2
50
3
    if ($suffix) {
519
0
0
        $artifact_name .= "-$suffix";
520    }
521
2
2
    my $retries_remaining = 3;
522
2
11
    while ($retries_remaining-- > 0) {
523
2
6
        ($gh_err_text, $ret) = download_latest_artifact(
524            $artifact_dir,
525            $repo,
526            $run,
527            $artifact_name
528        );
529
2
50
4
        return glob("$artifact_dir/artifact*.zip") unless ($ret >> 8);
530
531
2
9
        die_with_message($gh_err_text);
532
2
100
17
        if ($gh_err_text =~ /no valid artifacts found to download|"Artifact has expired"/) {
533
1
8
            my $expired_json = run_pipe(
534                'gh', 'api',
535                "/repos/$repo/actions/runs/$run/artifacts",
536                '-q',
537                '[.artifacts.[]|select(.name=="'.$artifact_name.'")][-1]|.expired'
538            );
539
1
50
8
            if ($expired_json ne '') {
540
1
4
                chomp $expired_json;
541
1
1
                my $expired;
542
1
1
50
3
11
                eval { $expired = decode_json $expired_json } || CheckSpelling::Util::die_custom $program, 542, "decode_json failed in update_repository with '$expired_json'";
543
1
50
146
                if ($expired) {
544
1
29
                    print "$program: GitHub Run Artifact expired. You will need to trigger a new run.\n";
545
1
1
1
1
5
17
                    { CheckSpelling::Util::tear_here(1); die "exiting"; }
546                }
547            }
548
0
0
            print "$program: GitHub Run may not have completed. If so, please wait for it to finish and try again.\n";
549
0
0
0
0
0
0
            { CheckSpelling::Util::tear_here(2); die "exiting"; }
550        }
551
1
50
2
        if ($gh_err_text =~ /no artifact matches any of the names or patterns provided/) {
552
0
0
0
            $github_server_url = $ENV{GITHUB_SERVER_URL} || '';
553
0
0
            my $run_link;
554
0
0
0
            if ($github_server_url) {
555
0
0
                $run_link = "[$run]($github_server_url/$repo/actions/runs/$run)";
556            } else {
557
0
0
                $run_link = "$run";
558            }
559
0
0
            print "$program: The referenced repository ($repo) run ($run_link) does not have a corresponding artifact ($artifact_name). If it was deleted, that's unfortunate. Consider pushing a change to the branch to trigger a new run?\n";
560
0
0
            print "If you don't think anyone deleted the artifact, please file a bug to https://github.com/check-spelling/check-spelling/issues/new including as much information about how you triggered this error as possible.\n";
561
0
0
0
0
0
0
            { CheckSpelling::Util::tear_here(3); die "exiting"; }
562        }
563
1
50
5
        if ($gh_err_text =~ /HTTP 404: Not Found|"status":"404"/) {
564
1
15
            print "$program: The referenced repository ($repo) may not exist, perhaps you do not have permission to see it. If the repository is hosted by GitHub Enterprise, check-spelling does not know how to integrate with it.\n";
565
1
1
1
1
3
9
            { CheckSpelling::Util::tear_here(8); die "exiting"; }
566        }
567
0
0
0
0
        if ($gh_err_text =~ /HTTP 403: API rate limit exceeded for .*?./) {
568        } elsif ($gh_err_text =~ m{dial tcp \S+:\d+: i/o timeout$}) {
569
0
0
0
            if ($retries_remaining <= 0) {
570
0
0
                print "$program: Timeout connecting to GitHub. This is probably caused by an outage of sorts.\nCheck https://www.githubstatus.com/history\nTry again later.";
571
0
0
0
0
0
0
                { CheckSpelling::Util::tear_here(9); die "exiting"; }
572            }
573        } else {
574
0
0
            print "$program: Unknown error, please check the list of known issues https://github.com/check-spelling/check-spelling/issues?q=is%3Aissue%20apply.pl and file a bug to https://github.com/check-spelling/check-spelling/issues/new?title=%60apply.pl%60%20scenario&body=Please%20provide%20details+preferably%20including%20a%20link%20to%20a%20workflow%20run,%20the%20configuration%20of%20the%20repository,%20and%20anything%20else%20you%20may%20know%20about%20the%20problem%2e\n";
575
0
0
            print $gh_err_text;
576
0
0
0
0
0
0
            { CheckSpelling::Util::tear_here(4); die "exiting"; }
577        }
578
0
0
0
        my $request_id = $1 if ($gh_err_text =~ /\brequest ID\s+(\S+)/);
579
0
0
0
        my $timestamp = $1 if ($gh_err_text =~ /\btimestamp\s+(.*? UTC)/);
580
0
0
0
        my $has_gh_token = defined $ENV{GH_TOKEN} || defined $ENV{GITHUB_TOKEN};
581
0
0
        my $meta_url = 'https://api.github.com/meta';
582
0
0
        while (1) {
583
0
0
            my @curl_args = qw(curl);
584
0
0
0
            unless ($has_gh_token) {
585
0
0
                my ($gh_token) = get_token();
586
0
0
0
                push @curl_args, '-u', "token:$gh_token" if defined $gh_token;
587            }
588
0
0
            push @curl_args, '-I', $meta_url;
589
0
0
            my ($curl_stdout, $curl_stderr, $curl_result);
590
0
0
            ($curl_stdout, $curl_stderr, $curl_result) = capture_system(@curl_args);
591
0
0
            my $delay = 1;
592
0
0
0
0
            if ($curl_stdout =~ m{^HTTP/\S+\s+200}) {
593
0
0
0
                if ($curl_stdout =~ m{^x-ratelimit-remaining:\s+(\d+)$}m) {
594
0
0
                    my $ratelimit_remaining = $1;
595
0
0
0
                    last if ($ratelimit_remaining > 10);
596
597
0
0
                    $delay = 5;
598
0
0
                    print STDERR "Sleeping for $delay seconds because $ratelimit_remaining is close to 0\n";
599                } else {
600
0
0
                    print STDERR "Couldn't find x-ratelimit-remaining, will sleep for $delay\n";
601                }
602            } elsif ($curl_stdout =~ m{^HTTP/\S+\s+403}) {
603
0
0
0
                if ($curl_stdout =~ /^retry-after:\s+(\d+)/m) {
604
0
0
                    $delay = $1;
605
0
0
                    print STDERR "Sleeping for $delay seconds (presumably due to API rate limit)\n";
606                } else {
607
0
0
                    print STDERR "Couldn't find retry-after, will sleep for $delay\n";
608                }
609            } else {
610
0
0
0
                my $response = $1 if $curl_stdout =~ m{^(HTTP/\S+)};
611
0
0
                print STDERR "Unexpected response ($response) from $meta_url; sleeping for $delay\n";
612            }
613
0
0
            sleep $delay;
614        }
615    }
616}
617
618sub update_repository {
619
1
1
2
    my ($artifact) = @_;
620
1
1
    our $program;
621
1
50
4
    CheckSpelling::Util::die_custom $program, 621, "$program: artifact argument contains quote characters" if $artifact =~ /'/;
622
1
5
    my $apply = unzip_pipe($artifact, 'apply.json');
623
1
50
8
    unless ($apply =~ /\{.*\}/s) {
624
0
0
        print STDERR "$program: Could not retrieve valid apply.json from artifact\n";
625
0
0
        $apply = '{
626            "expect_files": [".github/actions/spelling/expect.txt"],
627            "new_expect_file": ".github/actions/spelling/expect.txt",
628            "excludes_file": ".github/actions/spelling/excludes.txt",
629            "spelling_config": ".github/actions/spelling"
630        }';
631    }
632
1
14
    my $config_ref;
633
1
1
50
0
8
    eval { $config_ref = decode_json($apply); } ||
634        CheckSpelling::Util::die_custom $program, 634, "$program: decode_json failed in update_repository with '$apply'";
635
636
1
473
    my $git_repo_root = run_pipe('git', 'rev-parse', '--show-toplevel');
637
1
4
    chomp $git_repo_root;
638
1
50
7
    CheckSpelling::Util::die_custom $program, 638, "$program: Could not find git repo root..." unless $git_repo_root =~ /\w/;
639
1
5
    chdir $git_repo_root;
640
641
1
10
    retrieve_spell_check_this($artifact, $config_ref);
642
1
12
    remove_stale($artifact, $config_ref);
643
1
4
    add_expect($artifact, $config_ref);
644
1
12
    add_to_excludes($artifact, $config_ref);
645
1
2618
    system('git', 'add', '-u', '--', $config_ref->{'spelling_config'});
646}
647
648sub extract_artifacts_from_file {
649
1
1
2
    my ($artifact) = @_;
650
1
1276
    open my $artifact_reader, '-|', 'unzip', '-l', $artifact;
651
1
10
    my ($has_artifact, $only_file) = (0, 0);
652
1
866
    while (my $line = <$artifact_reader>) {
653
6
5
        chomp $line;
654
6
100
15
        if ($line =~ /\s+artifact\.zip$/) {
655
1
2
            $has_artifact = 1;
656
1
4
            next;
657        }
658
5
100
7
        if ($line =~ /\s+1 file$/) {
659
1
3
            $only_file = 1;
660
1
144
            next;
661        }
662
4
50
10
        $only_file = 0 if $only_file;
663    }
664
1
10
    close $artifact_reader;
665
1
1
    my @artifacts;
666
1
50
33
12
    if ($has_artifact && $only_file) {
667
1
8
        my $artifact_dir = tempdir(CLEANUP => 1);
668
1
255
        my ($fh, $gh_err) = tempfile();
669
1
132
        close $fh;
670
1
2433
        system('unzip', '-q', '-d', $artifact_dir, $artifact, 'artifact.zip');
671
1
26
        @artifacts = ("$artifact_dir/artifact.zip");
672    } else {
673
0
0
        @artifacts = ($artifact);
674    }
675
1
16
    return @artifacts;
676}
677
678sub main {
679
2
2
196053
    our $program;
680
2
3
    my ($bash_script, $first, $run);
681
2
4
    ($program, $bash_script, $first, $run) = @_;
682
2
2
    my $syntax = "$program <RUN_URL | OWNER/REPO RUN | ARTIFACT.zip>";
683    # Stages
684    # - 1 check for tools basic
685
2
4
    check_basic_tools();
686    # - 2 check for current
687    # -> 1. download the latest version to a temp file
688    # -> 2. parse current and latest (stripping comments) and compare (whitespace insensitively)
689    # -> 3. offer to update if the latest version is different
690
2
11
    check_current_script($bash_script);
691    # - 4 parse arguments
692
2
50
3
    CheckSpelling::Util::die_custom $program, 692, $syntax unless defined $first;
693
2
50
4
    $ENV{'GITHUB_API_URL'} ||= 'https://api.github.com';
694
2
3
    my $repo;
695    my @artifacts;
696
2
100
15
    if (-s $first) {
697
1
8
        @artifacts = extract_artifacts_from_file($first);
698    } else {
699
1
1
        my $suffix;
700
1
50
4
        if ($first =~ m{^\s*https://.*/([^/]+/[^/]+)/actions/runs/(\d+)(?:/attempts/\d+|)(?:#(\S+)|)\s*$}) {
701
0
0
            ($repo, $run, $suffix) = ($1, $2, $3);
702        } else {
703
1
2
            $repo = $first;
704        }
705
1
50
33
8
        CheckSpelling::Util::die_custom $program, 705, $syntax unless defined $repo && defined $run;
706        # - 3 check for tool readiness (is `gh` working)
707
0
0
        tools_are_ready($program);
708
0
0
        @artifacts = get_artifacts($repo, $run, $suffix);
709    }
710
711    # - 5 do work
712
1
2
    for my $artifact (@artifacts) {
713
1
5
        update_repository($artifact);
714    }
715}
716
717# main($0 ne '-' ? $0 : 'apply.pl', $bash_script, @ARGV);