File Coverage

File:lib/CheckSpelling/Apply.pm
Coverage:72.7%

linestmtbrancondsubtimecode
1
1
1
1
1
247334
1
21
package CheckSpelling::Apply; use CheckSpelling::Util; #!/usr/bin/env perl
2":" || q@<<"=END_OF_PERL"@;
3
4
1
1
1
1
2
0
27
use Symbol 'gensym';
5
1
1
1
1
172
1222
24
use IPC::Open3;
6
1
1
1
1
2
0
24
use File::Basename qw(dirname);
7
1
1
1
1
2
0
17
use File::Path qw(make_path);
8
1
1
1
1
160
302
32
use File::Spec::Functions qw(catfile path);
9
1
1
1
1
2
1
18
use File::Temp qw/ tempfile tempdir /;
10
1
1
1
1
2
1
21
use JSON::PP;
11
1
1
1
1
4
0
2537
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 $ENV{SYSTEMROOT};
29
30sub check_exists_command {
31
12
12
12
    my ($command) = @_;
32
33
12
25
    my @path = path;
34
12
91
    my @pathext = ('');
35
36
12
100
16
    if ($^O eq 'MSWin32') {
37
1
4
4
6
        push @pathext, map { lc } split /;/, $ENV{PATHEXT};
38    }
39
40
12
16
    for my $dir (@path) {
41
51
24
        for my $suffix (@pathext) {
42
63
79
            my $f = catfile $dir, "$command$suffix";
43
63
100
281
            return $f if -x $f;
44        }
45    }
46}
47
48sub needs_command_because {
49
11
11
2306
    my ($command, $reason) = @_;
50
11
100
9
    return if check_exists_command($command);
51
1
8
    CheckSpelling::Util::die_custom $program, 51, 'Please install `'.$command.'` - it is needed to '.$reason;
52}
53
54sub check_basic_tools {
55
3
3
1507
    needs_command_because('git', 'interact with git repositories');
56
3
5
    needs_command_because('curl', 'download other tools');
57
3
19
    $ENV{GH_NO_UPDATE_NOTIFIER}=1;
58
3
10
    $ENV{GH_NO_EXTENSION_UPDATE_NOTIFIER}=1;
59
3
4
    needs_command_because('gh', 'interact with github');
60}
61
62sub get_token {
63
7
7
1507
    our $token;
64
7
100
67
46
    return $token if defined $token && $token ne '';
65
4
100
12
    $token = $ENV{'GH_TOKEN'} || $ENV{'GITHUB_TOKEN'};
66
4
100
67
18
    return $token if defined $token && $token ne '';
67
1
2
    my ($err, $exit);
68
1
3
    ($token, $err, $exit) = capture_system('gh', 'auth', 'token');
69
1
5
    chomp $token;
70
1
3
    chomp $err;
71
1
5
    return ($token, $err, $exit);
72};
73
74sub download_with_curl {
75
3
3
4
    my ($url, $dest, $flags) = @_;
76
3
50
13
    $flags = '-fsL' unless defined $flags;
77
3
124558
    system('curl',
78        '--connect-timeout', 3,
79        '-A', $ua,
80        $flags,
81        '-o', $dest,
82        $url
83    );
84}
85
86sub tempfile_name {
87
8
8
37
    my ($fh, $filename) = tempfile();
88
8
1711
    close $fh;
89
8
17
    return $filename;
90}
91
92sub strip_comments {
93
8
8
8
    my ($file) = @_;
94
8
42
    my ($fh, $filename) = tempfile();
95
8
1858
    open INPUT, '<', $file;
96
8
132
    while (<INPUT>) {
97
4722
100
3321
        next if /^\s*(?:#.*)/;
98
4652
3890
        print $fh $_;
99    }
100
8
29
    close INPUT;
101
8
97
    close $fh;
102
8
23
    return $filename;
103}
104
105sub capture_system {
106
28
28
80
    my @args = @_;
107
28
161
    my $pid = open3(my $child_in, my $child_out, my $child_err = gensym, @args);
108
28
59477
    my (@err, @out);
109
28
1374119
    while (my $output = <$child_out>) {
110
22
2780
        push @out, $output;
111    }
112
28
790
    while (my $error = <$child_err>) {
113
21
129
        push @err, $error;
114    }
115
28
450
    waitpid( $pid, 0 );
116
28
159
    my $child_exit_status = $?;
117
28
97
    my $output_joined = join '', @out;
118
28
55
    my $error_joined = join '', @err;
119
28
957
    return ($output_joined, $error_joined, $child_exit_status);
120}
121
122sub capture_merged_system {
123
10
10
48
    my ($output_joined, $error_joined, $child_exit_status) = capture_system(@_);
124
10
50
54
    my $joiner = ($output_joined ne '') ? "\n" : '';
125
10
47
    return ($output_joined.$joiner.$error_joined, $child_exit_status);
126}
127
128sub compare_files {
129
4
4
1589
    my ($one, $two) = @_;
130
4
7
    my $one_stripped = strip_comments($one);
131
4
6
    my $two_stripped = strip_comments($two);
132
4
3
    my $exit_code;
133
4
13
    (undef, undef, $exit_code) = capture_system(
134            'diff',
135            '-qwB',
136            $one_stripped, $two_stripped
137        );
138
4
50
10
    if ($? == -1) {
139
0
0
        print "could not compare '$one' and '$two': $!\n";
140
0
0
        return 0;
141    }
142
4
50
8
    if ($? & 127) {
143
0
0
0
        printf "child died with signal %d, %s core dump\n",
144        ($? & 127),  ($? & 128) ? 'with' : 'without';
145
0
0
        return 0;
146    }
147
4
50
7
    return 0 if $? == 0;
148
4
19
    return 1;
149}
150
151my $bash_script=q{
152=END_OF_PERL@
153# bash
154set -e
155if [ "$OUTPUT" = "$ERROR" ]; then
156    ("$@" 2>&1) > "$OUTPUT"
157else
158    "$@" > "$OUTPUT" 2> "$ERROR"
159fi
160exit
161};
162
163sub check_current_script {
164
3
50
3
13812
    if ("$0" eq '-') {
165
0
0
        my ($bash_script) = @_;
166
0
0
        my $fh;
167
0
0
        ($fh, $0) = tempfile();
168
0
0
        $bash_script =~ s/^=.*\@$//m;
169
0
0
        print $fh $bash_script;
170
0
0
        close $fh;
171
0
0
        return;
172    }
173
3
9
    my $filename = tempfile_name();
174
3
6
    my $source = 'https://raw.githubusercontent.com/check-spelling/check-spelling/prerelease/apply.pl';
175
3
7
    download_with_curl($source, $filename);
176
3
50
84
    if ($? == 0) {
177
3
50
25
        if (compare_files($filename, $0)) {
178
3
45
            print "Current apply script differs from '$source' (locally downloaded to `$filename`). You may wish to upgrade.\n";
179        }
180    }
181}
182
183sub die_with_message {
184
5
5
6
    our $program;
185
5
10
    my ($gh_err_text) = @_;
186
5
50
33
30
    if ($gh_err_text =~ /error connecting to / && $gh_err_text =~ /check your internet connection/) {
187
0
0
        print "$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$gh_err_text";
188
0
0
0
0
0
0
        { CheckSpelling::Util::tear_here(5); die "exiting"; }
189    }
190
5
100
20
    if ($gh_err_text =~ /proxyconnect tcp:.*connect: connection refused/) {
191
1
33
        print "$program: Proxy is not accepting connections.\n";
192
1
2
        for my $proxy (qw(http_proxy HTTP_PROXY https_proxy HTTPS_PROXY)) {
193
4
100
13
            if (defined $ENV{$proxy}) {
194
1
4
                print "  $proxy: '$ENV{$proxy}'\n";
195            }
196        }
197
1
3
        print "\n$gh_err_text";
198
1
1
1
1
7
19
        { CheckSpelling::Util::tear_here(6); die "exiting"; }
199    }
200
4
100
15
    if ($gh_err_text =~ /dial unix .*: connect: .*/) {
201
1
33
        print "$program: Unix http socket is not working.\n";
202
1
28733
        my $gh_http_unix_socket = `gh config get http_unix_socket`;
203
1
31
        print "  http_unix_socket: $gh_http_unix_socket\n";
204
1
9
        print "\n$gh_err_text";
205
1
1
1
7
14
27
        { CheckSpelling::Util::tear_here(7); die "exiting"; }
206    }
207}
208
209sub gh_is_happy_internal {
210
5
5
18
    my ($output, $exit_code) = capture_merged_system(qw(gh api /installation/repositories));
211
5
50
14
    return ($exit_code, $output) if $exit_code == 0;
212
5
22
    ($output, $exit_code) = capture_merged_system(qw(gh api /user));
213
5
18
    return ($exit_code, $output);
214}
215
216sub gh_is_happy {
217
3
3
5
    my ($program) = @_;
218
3
9
    my ($gh_auth_status, $gh_status_lines) = gh_is_happy_internal();
219
3
50
13
    return 1 if $gh_auth_status == 0;
220
3
10
    die_with_message($gh_status_lines);
221
222
1
1
    my @problematic_env_variables;
223
1
3
    for my $variable (qw(GH_TOKEN GITHUB_TOKEN GITHUB_ACTIONS CI)) {
224
4
100
9
        if (defined $ENV{$variable}) {
225
2
33
            delete $ENV{$variable};
226
2
7
            push @problematic_env_variables, $variable;
227
2
6
            ($gh_auth_status, $gh_status_lines) = gh_is_happy_internal();
228
2
50
16
            if ($gh_auth_status == 0) {
229
0
0
                print STDERR "$0: gh program did not like these environment variables: ".join(', ', @problematic_env_variables)." -- consider unsetting them.\n";
230
0
0
                return 1;
231            }
232        }
233    }
234
235
1
32
    print $gh_status_lines;
236
1
18
    return 0;
237}
238
239sub tools_are_ready {
240
3
3
99305
    my ($program) = @_;
241
3
50
9
    unless (gh_is_happy($program)) {
242
1
6
        $! = 1;
243
1
50
33
10
        my $or_gh_token = (defined $ENV{CI} && $ENV{CI}) ? ' or set the GH_TOKEN environment variable' : '';
244
1
5
        CheckSpelling::Util::die_custom $program, 244, "$program requires a happy gh, please try 'gh auth login'$or_gh_token\n";
245    }
246}
247
248sub run_pipe {
249
8
8
33
    my @args = @_;
250
8
18
    my ($out, undef, $exit_code) = capture_system(@args);
251
8
59
    return $out;
252}
253
254sub unzip_pipe {
255
6
6
16
    my ($artifact, $file) = @_;
256
6
9
    return run_pipe(
257        'unzip',
258        '-p', $artifact,
259        $file
260    );
261}
262
263sub retrieve_spell_check_this {
264
1
1
2
    my ($artifact, $config_ref) = @_;
265
1
2
    my $spell_check_this_config = unzip_pipe($artifact, 'spell_check_this.json');
266
1
50
9
    return unless $spell_check_this_config =~ /\{.*\}/s;
267
1
3
    my %config;
268
1
1
1
50
2
0
7
    eval { %config = %{decode_json $spell_check_this_config}; } || CheckSpelling::Util::die_custom $program, 268, "decode_json failed in retrieve_spell_check_this with '$spell_check_this_config'";
269
1
453
    my ($repo, $branch, $destination, $path) = ($config{url}, $config{branch}, $config{config}, $config{path});
270
1
5
    my $spell_check_this_dir = tempdir();
271
1
230
    my $exit_code;
272
1
5
    (undef, undef, $exit_code) = capture_system(
273            'git', 'clone',
274            '--depth', '1',
275            '--no-tags',
276            $repo,
277            '--branch', $branch,
278            $spell_check_this_dir
279        );
280
1
50
4
    if ($?) {
281
0
0
        CheckSpelling::Util::die_custom $program, 281, "git clone $repo#$branch failed";
282    }
283
284
1
187
    make_path($destination);
285
1
2542
    system('cp', '-i', '-R', glob("$spell_check_this_dir/$path/*"), $destination);
286
1
2890
    system('git', 'add', '-f', $destination);
287}
288
289sub case_biased {
290
1
1
6
    lc($a)."-".$a cmp lc($b)."-".$b;
291}
292
293sub add_to_excludes {
294
1
1
4
    my ($artifact, $config_ref) = @_;
295
1
1
    my $excludes = $config_ref->{"excludes_file"};
296
1
5
    my $should_exclude_patterns = unzip_pipe($artifact, 'should_exclude.patterns');
297
1
50
5
    unless ($should_exclude_patterns =~ /\w/) {
298
1
4
        $should_exclude_patterns = unzip_pipe($artifact, 'should_exclude.txt');
299
1
50
14
        return unless $should_exclude_patterns =~ /\w/;
300
1
10
        $should_exclude_patterns =~ s{^(.*)}{^\\Q$1\\E\$}gm;
301    }
302
1
2
    my $need_to_add_excludes;
303    my %excludes;
304
1
50
7
    if (-f $excludes) {
305
1
10
        open EXCLUDES, '<', $excludes;
306
1
10
        while (<EXCLUDES>) {
307
1
3
            chomp;
308
1
50
3
            next unless /./;
309
1
4
            $excludes{$_."\n"} = 1;
310        }
311
1
3
        close EXCLUDES;
312    } else {
313
0
0
        $need_to_add_excludes = 1;
314    }
315
1
3
    for $pattern (split /\n/, $should_exclude_patterns) {
316
1
50
3
        next unless $pattern =~ /./;
317
1
3
        $excludes{$pattern."\n"} = 1;
318    }
319
1
54
    open EXCLUDES, '>', $excludes;
320
1
14
    print EXCLUDES join "", sort case_biased keys %excludes;
321
1
33
    close EXCLUDES;
322
1
50
7
    system('git', 'add', '--', $excludes) if $need_to_add_excludes;
323}
324
325sub remove_stale {
326
1
1
3
    my ($artifact, $config_ref) = @_;
327
1
2
    my @stale = split /\s+/s, unzip_pipe($artifact, 'remove_words.txt');
328
1
50
4
    return unless @stale;
329
1
1
1
2
    my @expect_files = @{$config_ref->{"expect_files"}};
330    @expect_files = grep {
331
1
1
50
3
10
        print STDERR "Could not find $_\n" unless -f $_;
332
1
3
        -f $_;
333    } @expect_files;
334
1
50
1
    unless (@expect_files) {
335
0
0
        CheckSpelling::Util::die_custom $program, 335, "Could not find any of the processed expect files, are you on the wrong branch?";
336    }
337
338
1
4
    my $re = join "|", @stale;
339
1
2
    for my $file (@expect_files) {
340
1
10
        open INPUT, '<', $file;
341
1
1
        my @keep;
342
1
11
        while (<INPUT>) {
343
2
100
53
            next if /^(?:$re)(?:(?:\r|\n)*$|[# ].*)/;
344
1
5
            push @keep, $_;
345        }
346
1
3
        close INPUT;
347
348
1
31
        open OUTPUT, '>', $file;
349
1
3
        print OUTPUT join '', @keep;
350
1
33
        close OUTPUT;
351    };
352}
353
354sub add_expect {
355
1
1
1
    my ($artifact, $config_ref) = @_;
356
1
1
    my @add = split /\s+/s, (unzip_pipe($artifact, 'tokens.txt'));
357
1
50
5
    return unless @add;
358
1
3
    my $new_expect_file = $config_ref->{"new_expect_file"};
359
1
1
    my @words;
360
1
151
    make_path (dirname($new_expect_file));
361
1
50
8
    if (-s $new_expect_file) {
362
0
0
        open FILE, q{<}, $new_expect_file;
363
0
0
        local $/ = undef;
364
0
0
        @words = split /\s+/, <FILE>;
365
0
0
        close FILE;
366    }
367
1
1
    my %items;
368
1
2
    @items{@words} = @words x (1);
369
1
6
    @items{@add} = @add x (1);
370
1
3
    @words = sort case_biased keys %items;
371
1
24
    open FILE, q{>}, $new_expect_file;
372
1
3
    for my $word (@words) {
373
1
50
7
        print FILE "$word\n" if $word =~ /\S/;
374    };
375
1
14
    close FILE;
376
1
2949
    system("git", "add", $new_expect_file);
377}
378
379sub get_artifact_metadata {
380
3
3
5
    my ($url) = @_;
381
3
6
    my $json_file = tempfile_name();
382
3
6
    my ($curl_stdout, $curl_stderr, $curl_result);
383
3
11
    my @curl_args = (
384        'curl',
385        $url,
386        '-A',
387        $ua,
388        '-s',
389        '--fail-with-body',
390    );
391
3
5
    my ($gh_token) = get_token();
392
3
50
11
    push @curl_args, '-u', "token:$gh_token" if defined $gh_token;
393
3
4
    push @curl_args, (
394        '-o',
395        $json_file
396    );
397
3
8
    ($curl_stdout, $curl_stderr, $curl_result) = capture_system(
398        @curl_args
399    );
400
3
100
18
    unless ($curl_result == 0) {
401
1
50
9
        if ($curl_stdout eq '') {
402
1
10
            local $/;
403
1
33
            open my $error_fh, '<', $json_file;
404
1
12
            $curl_stdout = <$error_fh>;
405
1
5
            close $error_fh;
406        }
407        return (
408
1
24
            out    => $curl_stdout,
409            err    => $curl_stderr,
410            result => $curl_result,
411        );
412    }
413
2
5
    my $link;
414
2
94
    open my $json_file_fh, '<', $json_file;
415
2
4
    my ($id, $download_url, $count);
416    {
417
2
2
1
13
        local $/;
418
2
30
        my $content = <$json_file_fh>;
419
2
22
        my $json = decode_json $content;
420
2
6299
        my $artifact = $json->{'artifacts'}->[0];
421
2
2
        $id = $artifact->{'id'};
422
2
2
        $download_url = $artifact->{'archive_download_url'};
423
2
9
        $count = $json->{'total_count'};
424    }
425
2
7
    close $json_file_fh;
426
2
50
4
    if ($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
27
        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
2
    my $page = 1;
443
2
6
    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
5
    $page = $first{'count'};
446
2
100
8
    if (defined $page) {
447
1
4
        my %second = get_artifact_metadata($url.$page);
448
1
6
        my ($id_1, $id_2) = ($first{'id'}, $second{'id'});
449
1
50
33
11
        if (defined $id_1 && defined $id_2) {
450
1
50
6
            if ($id_2 > $id_1) {
451                return (
452
1
6
                    download => $second{'download'},
453                );
454            }
455        }
456    }
457
1
3
    my $download = $first{'download'};
458
1
50
2
    if (defined $download) {
459        return (
460
0
0
            download => $download,
461        );
462    }
463
1
15
    return %first;
464}
465
466sub download_latest_artifact {
467
2
2
7
    my %maybe_download = get_latest_artifact_metadata(@_);
468
2
4
    my $download = $maybe_download{'download'};
469
2
21
    my $zip_file = tempfile_name();
470
2
100
6
    if (defined $download) {
471
1
6
        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
4
        push @curl_args, '-u', "token:$gh_token" if defined $gh_token;
482
1
2
        push @curl_args, (
483            '-o',
484            $zip_file
485        );
486
1
2
        ($curl_stdout, $curl_stderr, $curl_result) = capture_system(
487            @curl_args
488        );
489
1
50
8
        if ($curl_result != 0) {
490
1
50
4
            if ($curl_stdout eq '') {
491
1
7
                local $/;
492
1
27
                open my $error_fh, '<', $zip_file;
493
1
11
                $curl_stdout = <$error_fh>;
494
1
5
                close $error_fh;
495            }
496
1
16
            return ("$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", $result);
507    }
508
1
5
    my ($out, $err, $result) = ($maybe_download{'out'}, $maybe_download{'err'}, $maybe_download{'result'});
509
1
6
    return ("$out\n$err", $result);
510}
511
512sub get_artifacts {
513
2
2
3967
    my ($repo, $run, $suffix) = @_;
514
2
3
    our $program;
515
2
5
    my $artifact_dir = tempdir(CLEANUP => 1);
516
2
385
    my $gh_err_text;
517
2
3
    my $artifact_name = 'check-spelling-comment';
518
2
50
3
    if ($suffix) {
519
0
0
        $artifact_name .= "-$suffix";
520    }
521
2
2
    my $retries_remaining = 3;
522
2
9
    while ($retries_remaining-- > 0) {
523
2
5
        ($gh_err_text, $ret) = download_latest_artifact(
524            $artifact_dir,
525            $repo,
526            $run,
527            $artifact_name
528        );
529
2
50
9
        return glob("$artifact_dir/artifact*.zip") unless ($ret >> 8);
530
531
2
12
        die_with_message($gh_err_text);
532
2
100
26
        if ($gh_err_text =~ /no valid artifacts found to download|"Artifact has expired"/) {
533
1
9
            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
11
            if ($expired_json ne '') {
540
1
4
                chomp $expired_json;
541
1
2
                my $expired;
542
1
1
50
3
12
                eval { $expired = decode_json $expired_json } || CheckSpelling::Util::die_custom $program, 542, "decode_json failed in update_repository with '$expired_json'";
543
1
50
160
                if ($expired) {
544
1
52
                    print "$program: GitHub Run Artifact expired. You will need to trigger a new run.\n";
545
1
1
1
1
5
21
                    { CheckSpelling::Util::tear_here(1); die "exiting"; }
546                }
547            }
548
0
0
            print "$program: GitHub Run may not have completed. If so, please wait for it to finish and try again.\n";
549
0
0
0
0
0
0
            { CheckSpelling::Util::tear_here(2); die "exiting"; }
550        }
551
1
50
4
        if ($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
9
        if ($gh_err_text =~ /HTTP 404: Not Found|"status":"404"/) {
564
1
20
            print "$program: The referenced repository ($repo) may not exist, perhaps you do not have permission to see it. If the repository is hosted by GitHub Enterprise, check-spelling does not know how to integrate with it.\n";
565
1
1
1
1
7
17
            { CheckSpelling::Util::tear_here(8); die "exiting"; }
566        }
567
0
0
0
0
        if ($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 $delay;
614        }
615    }
616}
617
618sub update_repository {
619
1
1
3
    my ($artifact) = @_;
620
1
2
    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
9
    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
0
    my $config_ref;
633
1
1
50
18
9
    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
668
    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
6
    chdir $git_repo_root;
640
641
1
12
    retrieve_spell_check_this($artifact, $config_ref);
642
1
17
    remove_stale($artifact, $config_ref);
643
1
6
    add_expect($artifact, $config_ref);
644
1
14
    add_to_excludes($artifact, $config_ref);
645
1
2760
    system('git', 'add', '-u', '--', $config_ref->{'spelling_config'});
646}
647
648sub extract_artifacts_from_file {
649
1
1
3
    my ($artifact) = @_;
650
1
1500
    open my $artifact_reader, '-|', 'unzip', '-l', $artifact;
651
1
12
    my ($has_artifact, $only_file) = (0, 0);
652
1
950
    while (my $line = <$artifact_reader>) {
653
6
8
        chomp $line;
654
6
100
18
        if ($line =~ /\s+artifact\.zip$/) {
655
1
0
            $has_artifact = 1;
656
1
4
            next;
657        }
658
5
100
10
        if ($line =~ /\s+1 file$/) {
659
1
0
            $only_file = 1;
660
1
158
            next;
661        }
662
4
50
10
        $only_file = 0 if $only_file;
663    }
664
1
11
    close $artifact_reader;
665
1
1
    my @artifacts;
666
1
50
33
11
    if ($has_artifact && $only_file) {
667
1
9
        my $artifact_dir = tempdir(CLEANUP => 1);
668
1
280
        my ($fh, $gh_err) = tempfile();
669
1
139
        close $fh;
670
1
2681
        system('unzip', '-q', '-d', $artifact_dir, $artifact, 'artifact.zip');
671
1
27
        @artifacts = ("$artifact_dir/artifact.zip");
672    } else {
673
0
0
        @artifacts = ($artifact);
674    }
675
1
19
    return @artifacts;
676}
677
678sub main {
679
2
2
163554
    our $program;
680
2
2
    my ($bash_script, $first, $run);
681
2
5
    ($program, $bash_script, $first, $run) = @_;
682
2
3
    my $syntax = "$program <RUN_URL | OWNER/REPO RUN | ARTIFACT.zip>";
683    # Stages
684    # - 1 check for tools basic
685
2
6
    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
8
    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
8
    $ENV{'GITHUB_API_URL'} ||= 'https://api.github.com';
694
2
3
    my $repo;
695    my @artifacts;
696
2
100
18
    if (-s $first) {
697
1
7
        @artifacts = extract_artifacts_from_file($first);
698    } else {
699
1
1
        my $suffix;
700
1
50
1
        if ($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
4
    for my $artifact (@artifacts) {
713
1
2
        update_repository($artifact);
714    }
715}
716
717# main($0 ne '-' ? $0 : 'apply.pl', $bash_script, @ARGV);