File Coverage

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

linestmtbrancondsubtimecode
1
1
1
1
1
264287
1
21
package CheckSpelling::Apply; use CheckSpelling::Util; #!/usr/bin/env perl
2":" || q@<<"=END_OF_PERL"@;
3
4
1
1
1
1
1
1
25
use Symbol 'gensym';
5
1
1
1
1
171
1195
23
use IPC::Open3;
6
1
1
1
1
2
0
21
use File::Basename qw(dirname);
7
1
1
1
1
2
0
15
use File::Path qw(make_path);
8
1
1
1
1
158
329
34
use File::Spec::Functions qw(catfile path);
9
1
1
1
1
2
2
19
use File::Temp qw/ tempfile tempdir /;
10
1
1
1
1
1
1
18
use JSON::PP;
11
1
1
1
1
1
5
2471
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 $ENV{SYSTEMROOT};
29
30sub check_exists_command {
31
12
12
10
    my ($command) = @_;
32
33
12
21
    my @path = path;
34
12
81
    my @pathext = ('');
35
36
12
100
15
    if ($^O eq 'MSWin32') {
37
1
4
4
6
        push @pathext, map { lc } split /;/, $ENV{PATHEXT};
38    }
39
40
12
14
    for my $dir (@path) {
41
51
22
        for my $suffix (@pathext) {
42
63
72
            my $f = catfile $dir, "$command$suffix";
43
63
100
278
            return $f if -x $f;
44        }
45    }
46}
47
48sub needs_command_because {
49
11
11
2258
    my ($command, $reason) = @_;
50
11
100
8
    return if check_exists_command($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
1469
    needs_command_because('git', 'interact with git repositories');
56
3
3
    needs_command_because('curl', 'download other tools');
57
3
7
    needs_command_because('gh', 'interact with github');
58    #needs_command_because('magic-magic', 'debugging');
59}
60
61sub get_token {
62
7
7
1471
    our $token;
63
7
100
67
58
    return $token if defined $token && $token ne '';
64
4
100
11
    $token = $ENV{'GH_TOKEN'} || $ENV{'GITHUB_TOKEN'};
65
4
100
67
17
    return $token if defined $token && $token ne '';
66
1
2
    my ($err, $exit);
67
1
5
    ($token, $err, $exit) = capture_system('gh', 'auth', 'token');
68
1
3
    chomp $token;
69
1
3
    chomp $err;
70
1
7
    return ($token, $err, $exit);
71};
72
73sub download_with_curl {
74
3
3
4
    my ($url, $dest, $flags) = @_;
75
3
50
9
    $flags = '-fsL' unless defined $flags;
76
3
184407
    system('curl',
77        '--connect-timeout', 3,
78        '-A', $ua,
79        $flags,
80        '-o', $dest,
81        $url
82    );
83}
84
85sub tempfile_name {
86
8
8
36
    my ($fh, $filename) = tempfile();
87
8
1500
    close $fh;
88
8
17
    return $filename;
89}
90
91sub strip_comments {
92
8
8
10
    my ($file) = @_;
93
8
41
    my ($fh, $filename) = tempfile();
94
8
1747
    open INPUT, '<', $file;
95
8
119
    while (<INPUT>) {
96
4719
100
2595
        next if /^\s*(?:#.*)/;
97
4646
3261
        print $fh $_;
98    }
99
8
26
    close INPUT;
100
8
82
    close $fh;
101
8
21
    return $filename;
102}
103
104sub capture_system {
105
28
28
88
    my @args = @_;
106
28
146
    my $pid = open3(my $child_in, my $child_out, my $child_err = gensym, @args);
107
28
73045
    my (@err, @out);
108
28
1651433
    while (my $output = <$child_out>) {
109
22
2683
        push @out, $output;
110    }
111
28
702
    while (my $error = <$child_err>) {
112
21
100
        push @err, $error;
113    }
114
28
444
    waitpid( $pid, 0 );
115
28
153
    my $child_exit_status = $?;
116
28
88
    my $output_joined = join '', @out;
117
28
53
    my $error_joined = join '', @err;
118
28
909
    return ($output_joined, $error_joined, $child_exit_status);
119}
120
121sub capture_merged_system {
122
10
10
49
    my ($output_joined, $error_joined, $child_exit_status) = capture_system(@_);
123
10
50
82
    my $joiner = ($output_joined ne '') ? "\n" : '';
124
10
66
    return ($output_joined.$joiner.$error_joined, $child_exit_status);
125}
126
127sub compare_files {
128
4
4
1716
    my ($one, $two) = @_;
129
4
9
    my $one_stripped = strip_comments($one);
130
4
6
    my $two_stripped = strip_comments($two);
131
4
8
    my $exit_code;
132
4
8
    (undef, undef, $exit_code) = capture_system(
133            'diff',
134            '-qwB',
135            $one_stripped, $two_stripped
136        );
137
4
50
8
    if ($? == -1) {
138
0
0
        print "could not compare '$one' and '$two': $!\n";
139
0
0
        return 0;
140    }
141
4
50
8
    if ($? & 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
19
    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
13016
    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
6
    my $filename = tempfile_name();
173
3
2
    my $source = 'https://raw.githubusercontent.com/check-spelling/check-spelling/prerelease/apply.pl';
174
3
10
    download_with_curl($source, $filename);
175
3
50
116
    if ($? == 0) {
176
3
50
26
        if (compare_files($filename, $0)) {
177
3
51
            print "Current apply script differs from '$source' (locally downloaded to `$filename`). You may wish to upgrade.\n";
178        }
179    }
180}
181
182sub die_with_message {
183
5
5
2
    our $program;
184
5
8
    my ($gh_err_text) = @_;
185
5
50
33
25
    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
30
        print "$program: Proxy is not accepting connections.\n";
191
1
5
        for my $proxy (qw(http_proxy HTTP_PROXY https_proxy HTTPS_PROXY)) {
192
4
100
12
            if (defined $ENV{$proxy}) {
193
1
5
                print "  $proxy: '$ENV{$proxy}'\n";
194            }
195        }
196
1
3
        print "\n$gh_err_text";
197
1
1
1
1
8
15
        { CheckSpelling::Util::tear_here(6); die "exiting"; }
198    }
199
4
100
14
    if ($gh_err_text =~ /dial unix .*: connect: .*/) {
200
1
28
        print "$program: Unix http socket is not working.\n";
201
1
28331
        my $gh_http_unix_socket = `gh config get http_unix_socket`;
202
1
29
        print "  http_unix_socket: $gh_http_unix_socket\n";
203
1
7
        print "\n$gh_err_text";
204
1
1
1
8
17
22
        { CheckSpelling::Util::tear_here(7); die "exiting"; }
205    }
206}
207
208sub gh_is_happy_internal {
209
5
5
43
    my ($output, $exit_code) = capture_merged_system(qw(gh api /installation/repositories));
210
5
50
14
    return ($exit_code, $output) if $exit_code == 0;
211
5
12
    ($output, $exit_code) = capture_merged_system(qw(gh api /user));
212
5
21
    return ($exit_code, $output);
213}
214
215sub gh_is_happy {
216
3
3
3
    my ($program) = @_;
217
3
9
    my ($gh_auth_status, $gh_status_lines) = gh_is_happy_internal();
218
3
50
9
    return 1 if $gh_auth_status == 0;
219
3
13
    die_with_message($gh_status_lines);
220
221
1
2
    my @problematic_env_variables;
222
1
4
    for my $variable (qw(GH_TOKEN GITHUB_TOKEN GITHUB_ACTIONS CI)) {
223
4
100
12
        if (defined $ENV{$variable}) {
224
2
26
            delete $ENV{$variable};
225
2
6
            push @problematic_env_variables, $variable;
226
2
4
            ($gh_auth_status, $gh_status_lines) = gh_is_happy_internal();
227
2
50
13
            if ($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
36
    print $gh_status_lines;
235
1
21
    return 0;
236}
237
238sub tools_are_ready {
239
3
3
97722
    my ($program) = @_;
240
3
50
12
    unless (gh_is_happy($program)) {
241
1
4
        $! = 1;
242
1
50
33
6
        my $or_gh_token = (defined $ENV{CI} && $ENV{CI}) ? ' or set the GH_TOKEN environment variable' : '';
243
1
13
        CheckSpelling::Util::die_custom $program, 243, "$program requires a happy gh, please try 'gh auth login'$or_gh_token\n";
244    }
245}
246
247sub run_pipe {
248
8
8
29
    my @args = @_;
249
8
19
    my ($out, undef, $exit_code) = capture_system(@args);
250
8
45
    return $out;
251}
252
253sub unzip_pipe {
254
6
6
13
    my ($artifact, $file) = @_;
255
6
9
    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
2
    my $spell_check_this_config = unzip_pipe($artifact, 'spell_check_this.json');
265
1
50
10
    return unless $spell_check_this_config =~ /\{.*\}/s;
266
1
1
    my %config;
267
1
1
1
50
2
0
7
    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
418
    my ($repo, $branch, $destination, $path) = ($config{url}, $config{branch}, $config{config}, $config{path});
269
1
4
    my $spell_check_this_dir = tempdir();
270
1
209
    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            $spell_check_this_dir
278        );
279
1
50
5
    if ($?) {
280
0
0
        CheckSpelling::Util::die_custom $program, 280, "git clone $repo#$branch failed";
281    }
282
283
1
168
    make_path($destination);
284
1
2429
    system('cp', '-i', '-R', glob("$spell_check_this_dir/$path/*"), $destination);
285
1
2708
    system('git', 'add', '-f', $destination);
286}
287
288sub case_biased {
289
1
1
7
    lc($a)."-".$a cmp lc($b)."-".$b;
290}
291
292sub add_to_excludes {
293
1
1
5
    my ($artifact, $config_ref) = @_;
294
1
3
    my $excludes = $config_ref->{"excludes_file"};
295
1
3
    my $should_exclude_patterns = unzip_pipe($artifact, 'should_exclude.patterns');
296
1
50
4
    unless ($should_exclude_patterns =~ /\w/) {
297
1
5
        $should_exclude_patterns = unzip_pipe($artifact, 'should_exclude.txt');
298
1
50
13
        return unless $should_exclude_patterns =~ /\w/;
299
1
11
        $should_exclude_patterns =~ s{^(.*)}{^\\Q$1\\E\$}gm;
300    }
301
1
1
    my $need_to_add_excludes;
302    my %excludes;
303
1
50
7
    if (-f $excludes) {
304
1
9
        open EXCLUDES, '<', $excludes;
305
1
11
        while (<EXCLUDES>) {
306
1
2
            chomp;
307
1
50
3
            next unless /./;
308
1
5
            $excludes{$_."\n"} = 1;
309        }
310
1
3
        close EXCLUDES;
311    } else {
312
0
0
        $need_to_add_excludes = 1;
313    }
314
1
3
    for $pattern (split /\n/, $should_exclude_patterns) {
315
1
50
2
        next unless $pattern =~ /./;
316
1
3
        $excludes{$pattern."\n"} = 1;
317    }
318
1
25
    open EXCLUDES, '>', $excludes;
319
1
12
    print EXCLUDES join "", sort case_biased keys %excludes;
320
1
30
    close EXCLUDES;
321
1
50
4
    system('git', 'add', '--', $excludes) if $need_to_add_excludes;
322}
323
324sub remove_stale {
325
1
1
3
    my ($artifact, $config_ref) = @_;
326
1
4
    my @stale = split /\s+/s, unzip_pipe($artifact, 'remove_words.txt');
327
1
50
6
    return unless @stale;
328
1
1
1
3
    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
2
    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
11
        open INPUT, '<', $file;
340
1
1
        my @keep;
341
1
11
        while (<INPUT>) {
342
2
100
113
            next if /^(?:$re)(?:(?:\r|\n)*$|[# ].*)/;
343
1
3
            push @keep, $_;
344        }
345
1
4
        close INPUT;
346
347
1
28
        open OUTPUT, '>', $file;
348
1
9
        print OUTPUT join '', @keep;
349
1
31
        close OUTPUT;
350    };
351}
352
353sub add_expect {
354
1
1
1
    my ($artifact, $config_ref) = @_;
355
1
3
    my @add = split /\s+/s, (unzip_pipe($artifact, 'tokens.txt'));
356
1
50
5
    return unless @add;
357
1
1
    my $new_expect_file = $config_ref->{"new_expect_file"};
358
1
2
    my @words;
359
1
127
    make_path (dirname($new_expect_file));
360
1
50
10
    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
2
    my %items;
367
1
2
    @items{@words} = @words x (1);
368
1
5
    @items{@add} = @add x (1);
369
1
5
    @words = sort case_biased keys %items;
370
1
39
    open FILE, q{>}, $new_expect_file;
371
1
5
    for my $word (@words) {
372
1
50
7
        print FILE "$word\n" if $word =~ /\S/;
373    };
374
1
14
    close FILE;
375
1
2743
    system("git", "add", $new_expect_file);
376}
377
378sub get_artifact_metadata {
379
3
3
3
    my ($url) = @_;
380
3
9
    my $json_file = tempfile_name();
381
3
5
    my ($curl_stdout, $curl_stderr, $curl_result);
382
3
14
    my @curl_args = (
383        'curl',
384        $url,
385        '-A',
386        $ua,
387        '-s',
388        '--fail-with-body',
389    );
390
3
7
    my ($gh_token) = get_token();
391
3
50
9
    push @curl_args, '-u', "token:$gh_token" if defined $gh_token;
392
3
3
    push @curl_args, (
393        '-o',
394        $json_file
395    );
396
3
4
    ($curl_stdout, $curl_stderr, $curl_result) = capture_system(
397        @curl_args
398    );
399
3
100
17
    unless ($curl_result == 0) {
400
1
50
8
        if ($curl_stdout eq '') {
401
1
5
            local $/;
402
1
24
            open my $error_fh, '<', $json_file;
403
1
11
            $curl_stdout = <$error_fh>;
404
1
4
            close $error_fh;
405        }
406        return (
407
1
14
            out    => $curl_stdout,
408            err    => $curl_stderr,
409            result => $curl_result,
410        );
411    }
412
2
6
    my $link;
413
2
52
    open my $json_file_fh, '<', $json_file;
414
2
34
    my ($id, $download_url, $count);
415    {
416
2
2
0
14
        local $/;
417
2
26
        my $content = <$json_file_fh>;
418
2
25
        my $json = decode_json $content;
419
2
5548
        my $artifact = $json->{'artifacts'}->[0];
420
2
2
        $id = $artifact->{'id'};
421
2
2
        $download_url = $artifact->{'archive_download_url'};
422
2
9
        $count = $json->{'total_count'};
423    }
424
2
8
    close $json_file_fh;
425
2
50
2
    if ($count == 0) {
426        return (
427
0
0
            out => '',
428            err => 'no artifact matches any of the names or patterns provided',
429            result => (3 << 8),
430        );
431    }
432    return (
433
2
22
        id       => $id,
434        download => $download_url,
435        count    => $count,
436    );
437}
438
439sub get_latest_artifact_metadata {
440
2
2
5
    my ($artifact_dir, $repo, $run, $artifact_name) = @_;
441
2
2
    my $page = 1;
442
2
9
    my $url = "$ENV{GITHUB_API_URL}/repos/$repo/actions/runs/$run/artifacts?name=$artifact_name&per_page=1&page=";
443
2
6
    my %first = get_artifact_metadata($url.$page);
444
2
4
    $page = $first{'count'};
445
2
100
10
    if (defined $page) {
446
1
3
        my %second = get_artifact_metadata($url.$page);
447
1
5
        my ($id_1, $id_2) = ($first{'id'}, $second{'id'});
448
1
50
33
10
        if (defined $id_1 && defined $id_2) {
449
1
50
2
            if ($id_2 > $id_1) {
450                return (
451
0
0
                    download => $second{'download'},
452                );
453            }
454        }
455    }
456
2
2
    my $download = $first{'download'};
457
2
100
3
    if (defined $download) {
458        return (
459
1
5
            download => $download,
460        );
461    }
462
1
8
    return %first;
463}
464
465sub download_latest_artifact {
466
2
2
4
    my %maybe_download = get_latest_artifact_metadata(@_);
467
2
4
    my $download = $maybe_download{'download'};
468
2
2
    my $zip_file = tempfile_name();
469
2
100
4
    if (defined $download) {
470
1
5
        my @curl_args = (
471            'curl',
472            $download,
473            '-L',
474            '-A',
475            $ua,
476            '-s',
477            '--fail-with-body',
478        );
479
1
4
        my ($gh_token) = get_token();
480
1
50
5
        push @curl_args, '-u', "token:$gh_token" if defined $gh_token;
481
1
3
        push @curl_args, (
482            '-o',
483            $zip_file
484        );
485
1
2
        ($curl_stdout, $curl_stderr, $curl_result) = capture_system(
486            @curl_args
487        );
488
1
50
7
        if ($curl_result != 0) {
489
1
50
3
            if ($curl_stdout eq '') {
490
1
10
                local $/;
491
1
28
                open my $error_fh, '<', $zip_file;
492
1
11
                $curl_stdout = <$error_fh>;
493
1
7
                close $error_fh;
494            }
495
1
11
            return ("$curl_stdout\n$curl_stderr", $curl_result);
496        }
497
0
0
        my ($artifact_dir, $repo, $run, $artifact_name) = @_;
498
0
0
        ($out, $err, $result) = capture_system(
499            'unzip',
500            '-q',
501            $zip_file,
502            '-d',
503            $artifact_dir,
504            );
505
0
0
        return ("$out\n$err", $result);
506    }
507
1
2
    my ($out, $err, $result) = ($maybe_download{'out'}, $maybe_download{'err'}, $maybe_download{'result'});
508
1
32
    return ("$out\n$err", $result);
509}
510
511sub get_artifacts {
512
2
2
3636
    my ($repo, $run, $suffix) = @_;
513
2
3
    our $program;
514
2
4
    my $artifact_dir = tempdir(CLEANUP => 1);
515
2
368
    my $gh_err_text;
516
2
2
    my $artifact_name = 'check-spelling-comment';
517
2
50
4
    if ($suffix) {
518
0
0
        $artifact_name .= "-$suffix";
519    }
520
2
2
    my $retries_remaining = 3;
521
2
8
    while ($retries_remaining-- > 0) {
522
2
5
        ($gh_err_text, $ret) = download_latest_artifact(
523            $artifact_dir,
524            $repo,
525            $run,
526            $artifact_name
527        );
528
2
50
8
        return glob("$artifact_dir/artifact*.zip") unless ($ret >> 8);
529
530
2
9
        die_with_message($gh_err_text);
531
2
100
24
        if ($gh_err_text =~ /no valid artifacts found to download|"Artifact has expired"/) {
532
1
6
            my $expired_json = run_pipe(
533                'gh', 'api',
534                "/repos/$repo/actions/runs/$run/artifacts",
535                '-q',
536                '[.artifacts.[]|select(.name=="'.$artifact_name.'")][-1]|.expired'
537            );
538
1
50
9
            if ($expired_json ne '') {
539
1
3
                chomp $expired_json;
540
1
2
                my $expired;
541
1
1
50
4
10
                eval { $expired = decode_json $expired_json } || CheckSpelling::Util::die_custom $program, 541, "decode_json failed in update_repository with '$expired_json'";
542
1
50
137
                if ($expired) {
543
1
47
                    print "$program: GitHub Run Artifact expired. You will need to trigger a new run.\n";
544
1
1
1
1
6
15
                    { CheckSpelling::Util::tear_here(1); die "exiting"; }
545                }
546            }
547
0
0
            print "$program: GitHub Run may not have completed. If so, please wait for it to finish and try again.\n";
548
0
0
0
0
0
0
            { CheckSpelling::Util::tear_here(2); die "exiting"; }
549        }
550
1
50
2
        if ($gh_err_text =~ /no artifact matches any of the names or patterns provided/) {
551
0
0
0
            $github_server_url = $ENV{GITHUB_SERVER_URL} || '';
552
0
0
            my $run_link;
553
0
0
0
            if ($github_server_url) {
554
0
0
                $run_link = "[$run]($github_server_url/$repo/actions/runs/$run)";
555            } else {
556
0
0
                $run_link = "$run";
557            }
558
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";
559
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";
560
0
0
0
0
0
0
            { CheckSpelling::Util::tear_here(3); die "exiting"; }
561        }
562
1
50
8
        if ($gh_err_text =~ /HTTP 404: Not Found|"status":"404"/) {
563
1
18
            print "$program: The referenced repository ($repo) may not exist, perhaps you do not have permission to see it. If the repository is hosted by GitHub Enterprise, check-spelling does not know how to integrate with it.\n";
564
1
1
1
1
5
12
            { CheckSpelling::Util::tear_here(8); die "exiting"; }
565        }
566
0
0
0
0
        if ($gh_err_text =~ /HTTP 403: API rate limit exceeded for .*?./) {
567        } elsif ($gh_err_text =~ m{dial tcp \S+:\d+: i/o timeout$}) {
568
0
0
0
            if ($retries_remaining <= 0) {
569
0
0
                print "$program: Timeout connecting to GitHub. This is probably caused by an outage of sorts.\nCheck https://www.githubstatus.com/history\nTry again later.";
570
0
0
0
0
0
0
                { CheckSpelling::Util::tear_here(9); die "exiting"; }
571            }
572        } else {
573
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";
574
0
0
            print $gh_err_text;
575
0
0
0
0
0
0
            { CheckSpelling::Util::tear_here(4); die "exiting"; }
576        }
577
0
0
0
        my $request_id = $1 if ($gh_err_text =~ /\brequest ID\s+(\S+)/);
578
0
0
0
        my $timestamp = $1 if ($gh_err_text =~ /\btimestamp\s+(.*? UTC)/);
579
0
0
0
        my $has_gh_token = defined $ENV{GH_TOKEN} || defined $ENV{GITHUB_TOKEN};
580
0
0
        my $meta_url = 'https://api.github.com/meta';
581
0
0
        while (1) {
582
0
0
            my @curl_args = qw(curl);
583
0
0
0
            unless ($has_gh_token) {
584
0
0
                my ($gh_token) = get_token();
585
0
0
0
                push @curl_args, '-u', "token:$gh_token" if defined $gh_token;
586            }
587
0
0
            push @curl_args, '-I', $meta_url;
588
0
0
            my ($curl_stdout, $curl_stderr, $curl_result);
589
0
0
            ($curl_stdout, $curl_stderr, $curl_result) = capture_system(@curl_args);
590
0
0
            my $delay = 1;
591
0
0
0
0
            if ($curl_stdout =~ m{^HTTP/\S+\s+200}) {
592
0
0
0
                if ($curl_stdout =~ m{^x-ratelimit-remaining:\s+(\d+)$}m) {
593
0
0
                    my $ratelimit_remaining = $1;
594
0
0
0
                    last if ($ratelimit_remaining > 10);
595
596
0
0
                    $delay = 5;
597
0
0
                    print STDERR "Sleeping for $delay seconds because $ratelimit_remaining is close to 0\n";
598                } else {
599
0
0
                    print STDERR "Couldn't find x-ratelimit-remaining, will sleep for $delay\n";
600                }
601            } elsif ($curl_stdout =~ m{^HTTP/\S+\s+403}) {
602
0
0
0
                if ($curl_stdout =~ /^retry-after:\s+(\d+)/m) {
603
0
0
                    $delay = $1;
604
0
0
                    print STDERR "Sleeping for $delay seconds (presumably due to API rate limit)\n";
605                } else {
606
0
0
                    print STDERR "Couldn't find retry-after, will sleep for $delay\n";
607                }
608            } else {
609
0
0
0
                my $response = $1 if $curl_stdout =~ m{^(HTTP/\S+)};
610
0
0
                print STDERR "Unexpected response ($response) from $meta_url; sleeping for $delay\n";
611            }
612
0
0
            sleep $delay;
613        }
614    }
615}
616
617sub update_repository {
618
1
1
3
    my ($artifact) = @_;
619
1
1
    our $program;
620
1
50
8
    CheckSpelling::Util::die_custom $program, 620, "$program: artifact argument contains quote characters" if $artifact =~ /'/;
621
1
4
    my $apply = unzip_pipe($artifact, 'apply.json');
622
1
50
7
    unless ($apply =~ /\{.*\}/s) {
623
0
0
        print STDERR "$program: Could not retrieve valid apply.json from artifact\n";
624
0
0
        $apply = '{
625            "expect_files": [".github/actions/spelling/expect.txt"],
626            "new_expect_file": ".github/actions/spelling/expect.txt",
627            "excludes_file": ".github/actions/spelling/excludes.txt",
628            "spelling_config": ".github/actions/spelling"
629        }';
630    }
631
1
0
    my $config_ref;
632
1
1
50
2
24
    eval { $config_ref = decode_json($apply); } ||
633        CheckSpelling::Util::die_custom $program, 633, "$program: decode_json failed in update_repository with '$apply'";
634
635
1
470
    my $git_repo_root = run_pipe('git', 'rev-parse', '--show-toplevel');
636
1
3
    chomp $git_repo_root;
637
1
50
8
    CheckSpelling::Util::die_custom $program, 637, "$program: Could not find git repo root..." unless $git_repo_root =~ /\w/;
638
1
5
    chdir $git_repo_root;
639
640
1
2
    retrieve_spell_check_this($artifact, $config_ref);
641
1
16
    remove_stale($artifact, $config_ref);
642
1
2
    add_expect($artifact, $config_ref);
643
1
11
    add_to_excludes($artifact, $config_ref);
644
1
2631
    system('git', 'add', '-u', '--', $config_ref->{'spelling_config'});
645}
646
647sub extract_artifacts_from_file {
648
1
1
4
    my ($artifact) = @_;
649
1
1314
    open my $artifact_reader, '-|', 'unzip', '-l', $artifact;
650
1
11
    my ($has_artifact, $only_file) = (0, 0);
651
1
907
    while (my $line = <$artifact_reader>) {
652
6
4
        chomp $line;
653
6
100
19
        if ($line =~ /\s+artifact\.zip$/) {
654
1
1
            $has_artifact = 1;
655
1
2
            next;
656        }
657
5
100
10
        if ($line =~ /\s+1 file$/) {
658
1
1
            $only_file = 1;
659
1
139
            next;
660        }
661
4
50
10
        $only_file = 0 if $only_file;
662    }
663
1
11
    close $artifact_reader;
664
1
1
    my @artifacts;
665
1
50
33
9
    if ($has_artifact && $only_file) {
666
1
8
        my $artifact_dir = tempdir(CLEANUP => 1);
667
1
280
        my ($fh, $gh_err) = tempfile();
668
1
143
        close $fh;
669
1
2392
        system('unzip', '-q', '-d', $artifact_dir, $artifact, 'artifact.zip');
670
1
29
        @artifacts = ("$artifact_dir/artifact.zip");
671    } else {
672
0
0
        @artifacts = ($artifact);
673    }
674
1
14
    return @artifacts;
675}
676
677sub main {
678
2
2
198220
    our $program;
679
2
2
    my ($bash_script, $first, $run);
680
2
8
    ($program, $bash_script, $first, $run) = @_;
681
2
2
    my $syntax = "$program <RUN_URL | OWNER/REPO RUN | ARTIFACT.zip>";
682    # Stages
683    # - 1 check for tools basic
684
2
7
    check_basic_tools();
685    # - 2 check for current
686    # -> 1. download the latest version to a temp file
687    # -> 2. parse current and latest (stripping comments) and compare (whitespace insensitively)
688    # -> 3. offer to update if the latest version is different
689
2
4
    check_current_script($bash_script);
690    # - 4 parse arguments
691
2
50
5
    CheckSpelling::Util::die_custom $program, 691, $syntax unless defined $first;
692
2
50
4
    $ENV{'GITHUB_API_URL'} ||= 'https://api.github.com';
693
2
4
    my $repo;
694    my @artifacts;
695
2
100
16
    if (-s $first) {
696
1
4
        @artifacts = extract_artifacts_from_file($first);
697    } else {
698
1
2
        my $suffix;
699
1
50
2
        if ($first =~ m{^\s*https://.*/([^/]+/[^/]+)/actions/runs/(\d+)(?:/attempts/\d+|)(?:#(\S+)|)\s*$}) {
700
0
0
            ($repo, $run, $suffix) = ($1, $2, $3);
701        } else {
702
1
2
            $repo = $first;
703        }
704
1
50
33
10
        CheckSpelling::Util::die_custom $program, 704, $syntax unless defined $repo && defined $run;
705        # - 3 check for tool readiness (is `gh` working)
706
0
0
        tools_are_ready($program);
707
0
0
        @artifacts = get_artifacts($repo, $run, $suffix);
708    }
709
710    # - 5 do work
711
1
2
    for my $artifact (@artifacts) {
712
1
4
        update_repository($artifact);
713    }
714}
715
716# main($0 ne '-' ? $0 : 'apply.pl', $bash_script, @ARGV);