File Coverage

File:lib/CheckSpelling/Apply.pm
Coverage:72.9%

linestmtbrancondsubtimecode
1
1
1
1
1
207387
1
20
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
127
1104
25
use IPC::Open3;
6
1
1
1
1
2
0
22
use File::Basename qw(dirname);
7
1
1
1
1
1
0
16
use File::Path qw(make_path);
8
1
1
1
1
115
266
27
use File::Spec::Functions qw(catfile path);
9
1
1
1
1
2
2
14
use File::Temp qw/ tempfile tempdir /;
10
1
1
1
1
1
1
19
use JSON::PP;
11
1
1
1
1
6
1
2199
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
16
    my @path = path;
34
12
75
    my @pathext = ('');
35
36
12
100
14
    if ($^O eq 'MSWin32') {
37
1
4
3
6
        push @pathext, map { lc } split /;/, $ENV{PATHEXT};
38    }
39
40
12
10
    for my $dir (@path) {
41
51
22
        for my $suffix (@pathext) {
42
63
64
            my $f = catfile $dir, "$command$suffix";
43
63
100
168
            return $f if -x $f;
44        }
45    }
46}
47
48sub needs_command_because {
49
11
11
1831
    my ($command, $reason) = @_;
50
11
100
9
    return if check_exists_command($command);
51
1
5
    CheckSpelling::Util::die_custom $program, 51, 'Please install `'.$command.'` - it is needed to '.$reason;
52}
53
54sub check_basic_tools {
55
3
3
1156
    needs_command_because('git', 'interact with git repositories');
56
3
3
    needs_command_because('curl', 'download other tools');
57
3
15
    $ENV{GH_NO_UPDATE_NOTIFIER}=1;
58
3
8
    $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
1206
    our $token;
64
7
100
67
34
    return $token if defined $token && $token ne '';
65
4
100
9
    $token = $ENV{'GH_TOKEN'} || $ENV{'GITHUB_TOKEN'};
66
4
100
67
10
    return $token if defined $token && $token ne '';
67
1
1
    my ($err, $exit);
68
1
2
    ($token, $err, $exit) = capture_system('gh', 'auth', 'token');
69
1
3
    chomp $token;
70
1
2
    chomp $err;
71
1
5
    return ($token, $err, $exit);
72};
73
74sub download_with_curl {
75
3
3
11
    my ($url, $dest, $flags) = @_;
76
3
50
5
    $flags = '-fsL' unless defined $flags;
77
3
111671
    system('curl',
78        '--connect-timeout', 3,
79        '-A', $ua,
80        $flags,
81        '-o', $dest,
82        $url
83    );
84}
85
86sub tempfile_name {
87
8
8
27
    my ($fh, $filename) = tempfile();
88
8
1347
    close $fh;
89
8
13
    return $filename;
90}
91
92sub strip_comments {
93
8
8
9
    my ($file) = @_;
94
8
32
    my ($fh, $filename) = tempfile();
95
8
1486
    open INPUT, '<', $file;
96
8
89
    while (<INPUT>) {
97
4731
100
2366
        next if /^\s*(?:#.*)/;
98
4661
2924
        print $fh $_;
99    }
100
8
20
    close INPUT;
101
8
55
    close $fh;
102
8
18
    return $filename;
103}
104
105sub capture_system {
106
28
28
77
    my @args = @_;
107
28
116
    my $pid = open3(my $child_in, my $child_out, my $child_err = gensym, @args);
108
28
47300
    my (@err, @out);
109
28
1920422
    while (my $output = <$child_out>) {
110
22
1821
        push @out, $output;
111    }
112
28
570
    while (my $error = <$child_err>) {
113
21
78
        push @err, $error;
114    }
115
28
369
    waitpid( $pid, 0 );
116
28
128
    my $child_exit_status = $?;
117
28
95
    my $output_joined = join '', @out;
118
28
50
    my $error_joined = join '', @err;
119
28
764
    return ($output_joined, $error_joined, $child_exit_status);
120}
121
122sub capture_merged_system {
123
10
10
22
    my ($output_joined, $error_joined, $child_exit_status) = capture_system(@_);
124
10
50
41
    my $joiner = ($output_joined ne '') ? "\n" : '';
125
10
48
    return ($output_joined.$joiner.$error_joined, $child_exit_status);
126}
127
128sub compare_files {
129
4
4
1246
    my ($one, $two) = @_;
130
4
12
    my $one_stripped = strip_comments($one);
131
4
4
    my $two_stripped = strip_comments($two);
132
4
5
    my $exit_code;
133
4
9
    (undef, undef, $exit_code) = capture_system(
134            'diff',
135            '-qwB',
136            $one_stripped, $two_stripped
137        );
138
4
50
11
    if ($? == -1) {
139
0
0
        print "could not compare '$one' and '$two': $!\n";
140
0
0
        return 0;
141    }
142
4
50
7
    if ($? & 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
6
    return 0 if $? == 0;
148
4
15
    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
9739
    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
6
    my $filename = tempfile_name();
174
3
3
    my $source = 'https://raw.githubusercontent.com/check-spelling/check-spelling/prerelease/apply.pl';
175
3
6
    download_with_curl($source, $filename);
176
3
50
67
    if ($? == 0) {
177
3
50
19
        if (compare_files($filename, $0)) {
178
3
43
            print "Current apply script differs from '$source' (locally downloaded to `$filename`). You may wish to upgrade.\n";
179        }
180    }
181}
182
183sub die_with_message {
184
5
5
6
    our $program;
185
5
7
    my ($gh_err_text) = @_;
186
5
50
33
28
    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$gh_err_text";
188
0
0
0
0
0
0
        { CheckSpelling::Util::tear_here(5); die "exiting"; }
189    }
190
5
100
23
    if ($gh_err_text =~ /proxyconnect tcp:.*connect: connection refused/) {
191
1
26
        print "$program: Proxy is not accepting connections.\n";
192
1
5
        for my $proxy (qw(http_proxy HTTP_PROXY https_proxy HTTPS_PROXY)) {
193
4
100
9
            if (defined $ENV{$proxy}) {
194
1
4
                print "  $proxy: '$ENV{$proxy}'\n";
195            }
196        }
197
1
2
        print "\n$gh_err_text";
198
1
1
1
1
6
14
        { CheckSpelling::Util::tear_here(6); die "exiting"; }
199    }
200
4
100
18
    if ($gh_err_text =~ /dial unix .*: connect: .*/) {
201
1
27
        print "$program: Unix http socket is not working.\n";
202
1
27380
        my $gh_http_unix_socket = `gh config get http_unix_socket`;
203
1
26
        print "  http_unix_socket: $gh_http_unix_socket\n";
204
1
5
        print "\n$gh_err_text";
205
1
1
1
5
12
21
        { CheckSpelling::Util::tear_here(7); die "exiting"; }
206    }
207}
208
209sub gh_is_happy_internal {
210
5
5
17
    my ($output, $exit_code) = capture_merged_system(qw(gh api /installation/repositories));
211
5
50
12
    return ($exit_code, $output) if $exit_code == 0;
212
5
23
    ($output, $exit_code) = capture_merged_system(qw(gh api /user));
213
5
16
    return ($exit_code, $output);
214}
215
216sub gh_is_happy {
217
3
3
5
    my ($program) = @_;
218
3
14
    my ($gh_auth_status, $gh_status_lines) = gh_is_happy_internal();
219
3
50
10
    return 1 if $gh_auth_status == 0;
220
3
10
    die_with_message($gh_status_lines);
221
222
1
4
    my @problematic_env_variables;
223
1
3
    for my $variable (qw(GH_TOKEN GITHUB_TOKEN GITHUB_ACTIONS CI)) {
224
4
100
10
        if (defined $ENV{$variable}) {
225
2
26
            delete $ENV{$variable};
226
2
8
            push @problematic_env_variables, $variable;
227
2
3
            ($gh_auth_status, $gh_status_lines) = gh_is_happy_internal();
228
2
50
13
            if ($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
22
    print $gh_status_lines;
236
1
11
    return 0;
237}
238
239sub tools_are_ready {
240
3
3
185726
    my ($program) = @_;
241
3
50
7
    unless (gh_is_happy($program)) {
242
1
5
        $! = 1;
243
1
50
33
6
        my $or_gh_token = (defined $ENV{CI} && $ENV{CI}) ? ' or set the GH_TOKEN environment variable' : '';
244
1
7
        CheckSpelling::Util::die_custom $program, 244, "$program requires a happy gh, please try 'gh auth login'$or_gh_token\n";
245    }
246}
247
248sub run_pipe {
249
8
8
30
    my @args = @_;
250
8
17
    my ($out, undef, $exit_code) = capture_system(@args);
251
8
44
    return $out;
252}
253
254sub unzip_pipe {
255
6
6
9
    my ($artifact, $file) = @_;
256
6
12
    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
6
    return unless $spell_check_this_config =~ /\{.*\}/s;
267
1
1
    my %config;
268
1
1
1
50
0
1
5
    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
347
    my ($repo, $branch, $destination, $path) = ($config{url}, $config{branch}, $config{config}, $config{path});
270
1
4
    my $spell_check_this_dir = tempdir();
271
1
182
    my $exit_code;
272
1
3
    (undef, undef, $exit_code) = capture_system(
273            'git', 'clone',
274            '--depth', '1',
275            '--no-tags',
276            $repo,
277            '--branch', $branch,
278            $spell_check_this_dir
279        );
280
1
50
3
    if ($?) {
281
0
0
        CheckSpelling::Util::die_custom $program, 281, "git clone $repo#$branch failed";
282    }
283
284
1
134
    make_path($destination);
285
1
1867
    system('cp', '-i', '-R', glob("$spell_check_this_dir/$path/*"), $destination);
286
1
1999
    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
2
    my $excludes = $config_ref->{"excludes_file"};
296
1
3
    my $should_exclude_patterns = unzip_pipe($artifact, 'should_exclude.patterns');
297
1
50
3
    unless ($should_exclude_patterns =~ /\w/) {
298
1
2
        $should_exclude_patterns = unzip_pipe($artifact, 'should_exclude.txt');
299
1
50
10
        return unless $should_exclude_patterns =~ /\w/;
300
1
10
        $should_exclude_patterns =~ s{^(.*)}{^\\Q$1\\E\$}gm;
301    }
302
1
1
    my $need_to_add_excludes;
303    my %excludes;
304
1
50
5
    if (-f $excludes) {
305
1
8
        open EXCLUDES, '<', $excludes;
306
1
7
        while (<EXCLUDES>) {
307
1
1
            chomp;
308
1
50
3
            next unless /./;
309
1
4
            $excludes{$_."\n"} = 1;
310        }
311
1
2
        close EXCLUDES;
312    } else {
313
0
0
        $need_to_add_excludes = 1;
314    }
315
1
2
    for $pattern (split /\n/, $should_exclude_patterns) {
316
1
50
2
        next unless $pattern =~ /./;
317
1
2
        $excludes{$pattern."\n"} = 1;
318    }
319
1
18
    open EXCLUDES, '>', $excludes;
320
1
8
    print EXCLUDES join "", sort case_biased keys %excludes;
321
1
30
    close EXCLUDES;
322
1
50
5
    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
3
    return unless @stale;
329
1
1
1
3
    my @expect_files = @{$config_ref->{"expect_files"}};
330    @expect_files = grep {
331
1
1
50
3
6
        print STDERR "Could not find $_\n" unless -f $_;
332
1
2
        -f $_;
333    } @expect_files;
334
1
50
2
    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
1
    my $re = join "|", @stale;
339
1
2
    for my $file (@expect_files) {
340
1
8
        open INPUT, '<', $file;
341
1
2
        my @keep;
342
1
7
        while (<INPUT>) {
343
2
100
46
            next if /^(?:$re)(?:(?:\r|\n)*$|[# ].*)/;
344
1
4
            push @keep, $_;
345        }
346
1
2
        close INPUT;
347
348
1
18
        open OUTPUT, '>', $file;
349
1
4
        print OUTPUT join '', @keep;
350
1
32
        close OUTPUT;
351    };
352}
353
354sub add_expect {
355
1
1
2
    my ($artifact, $config_ref) = @_;
356
1
1
    my @add = split /\s+/s, (unzip_pipe($artifact, 'tokens.txt'));
357
1
50
2
    return unless @add;
358
1
2
    my $new_expect_file = $config_ref->{"new_expect_file"};
359
1
1
    my @words;
360
1
96
    make_path (dirname($new_expect_file));
361
1
50
6
    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
3
    my %items;
368
1
1
    @items{@words} = @words x (1);
369
1
3
    @items{@add} = @add x (1);
370
1
4
    @words = sort case_biased keys %items;
371
1
19
    open FILE, q{>}, $new_expect_file;
372
1
2
    for my $word (@words) {
373
1
50
4
        print FILE "$word\n" if $word =~ /\S/;
374    };
375
1
10
    close FILE;
376
1
1973
    system("git", "add", $new_expect_file);
377}
378
379sub get_artifact_metadata {
380
3
3
3
    my ($url) = @_;
381
3
7
    my $json_file = tempfile_name();
382
3
5
    my ($curl_stdout, $curl_stderr, $curl_result);
383
3
10
    my @curl_args = (
384        'curl',
385        '-L',
386        $url,
387        '-A',
388        $ua,
389        '-s',
390        '--fail-with-body',
391    );
392
3
6
    my ($gh_token) = get_token();
393
3
50
6
    push @curl_args, '-u', "token:$gh_token" if defined $gh_token;
394
3
4
    push @curl_args, (
395        '-o',
396        $json_file
397    );
398
3
5
    ($curl_stdout, $curl_stderr, $curl_result) = capture_system(
399        @curl_args
400    );
401
3
100
39
    unless ($curl_result == 0) {
402
1
50
8
        if ($curl_stdout eq '') {
403
1
7
            local $/;
404
1
33
            open my $error_fh, '<', $json_file;
405
1
15
            $curl_stdout = <$error_fh>;
406
1
4
            close $error_fh;
407        }
408        return (
409
1
16
            out    => $curl_stdout,
410            err    => $curl_stderr,
411            result => $curl_result,
412        );
413    }
414
2
2
    my $link;
415
2
57
    open my $json_file_fh, '<', $json_file;
416
2
2
    my ($id, $download_url, $count);
417    {
418
2
2
2
11
        local $/;
419
2
29
        my $content = <$json_file_fh>;
420
2
16
        my $json = decode_json $content;
421
2
4962
        my $artifact = $json->{'artifacts'}->[0];
422
2
2
        $id = $artifact->{'id'};
423
2
1
        $download_url = $artifact->{'archive_download_url'};
424
2
9
        $count = $json->{'total_count'};
425    }
426
2
8
    close $json_file_fh;
427
2
50
4
    if ($count == 0) {
428        return (
429
0
0
            out => '',
430            err => 'no artifact matches any of the names or patterns provided',
431            result => (3 << 8),
432        );
433    }
434    return (
435
2
18
        id       => $id,
436        download => $download_url,
437        count    => $count,
438    );
439}
440
441sub get_latest_artifact_metadata {
442
2
2
4
    my ($artifact_dir, $repo, $run, $artifact_name) = @_;
443
2
1
    my $page = 1;
444
2
4
    my $url = "$ENV{GITHUB_API_URL}/repos/$repo/actions/runs/$run/artifacts?name=$artifact_name&per_page=1&page=";
445
2
4
    my %first = get_artifact_metadata($url.$page);
446
2
5
    $page = $first{'count'};
447
2
100
11
    if (defined $page) {
448
1
2
        my %second = get_artifact_metadata($url.$page);
449
1
5
        my ($id_1, $id_2) = ($first{'id'}, $second{'id'});
450
1
50
33
10
        if (defined $id_1 && defined $id_2) {
451
1
50
3
            if ($id_2 > $id_1) {
452                return (
453
0
0
                    download => $second{'download'},
454                );
455            }
456        }
457    }
458
2
5
    my $download = $first{'download'};
459
2
100
5
    if (defined $download) {
460        return (
461
1
3
            download => $download,
462        );
463    }
464
1
8
    return %first;
465}
466
467sub download_latest_artifact {
468
2
2
4
    my %maybe_download = get_latest_artifact_metadata(@_);
469
2
3
    my $download = $maybe_download{'download'};
470
2
5
    my $zip_file = tempfile_name();
471
2
100
8
    if (defined $download) {
472
1
3
        my @curl_args = (
473            'curl',
474            $download,
475            '-L',
476            '-A',
477            $ua,
478            '-s',
479            '--fail-with-body',
480        );
481
1
2
        my ($gh_token) = get_token();
482
1
50
2
        push @curl_args, '-u', "token:$gh_token" if defined $gh_token;
483
1
3
        push @curl_args, (
484            '-o',
485            $zip_file
486        );
487
1
2
        ($curl_stdout, $curl_stderr, $curl_result) = capture_system(
488            @curl_args
489        );
490
1
50
9
        if ($curl_result != 0) {
491
1
50
4
            if ($curl_stdout eq '') {
492
1
7
                local $/;
493
1
28
                open my $error_fh, '<', $zip_file;
494
1
9
                $curl_stdout = <$error_fh>;
495
1
5
                close $error_fh;
496            }
497
1
15
            return ("$curl_stdout\n$curl_stderr", $curl_result);
498        }
499
0
0
        my ($artifact_dir, $repo, $run, $artifact_name) = @_;
500
0
0
        ($out, $err, $result) = capture_system(
501            'unzip',
502            '-q',
503            $zip_file,
504            '-d',
505            $artifact_dir,
506            );
507
0
0
        return ("$out\n$err", $result);
508    }
509
1
4
    my ($out, $err, $result) = ($maybe_download{'out'}, $maybe_download{'err'}, $maybe_download{'result'});
510
1
5
    return ("$out\n$err", $result);
511}
512
513sub get_artifacts {
514
2
2
3120
    my ($repo, $run, $suffix) = @_;
515
2
1
    our $program;
516
2
5
    my $artifact_dir = tempdir(CLEANUP => 1);
517
2
265
    my $gh_err_text;
518
2
4
    my $artifact_name = 'check-spelling-comment';
519
2
50
4
    if ($suffix) {
520
0
0
        $artifact_name .= "-$suffix";
521    }
522
2
2
    my $retries_remaining = 3;
523
2
7
    while ($retries_remaining-- > 0) {
524
2
5
        ($gh_err_text, $ret) = download_latest_artifact(
525            $artifact_dir,
526            $repo,
527            $run,
528            $artifact_name
529        );
530
2
50
7
        return glob("$artifact_dir/artifact*.zip") unless ($ret >> 8);
531
532
2
10
        die_with_message($gh_err_text);
533
2
100
21
        if ($gh_err_text =~ /no valid artifacts found to download|"Artifact has expired"/) {
534
1
7
            my $expired_json = run_pipe(
535                'gh', 'api',
536                "/repos/$repo/actions/runs/$run/artifacts",
537                '-q',
538                '[.artifacts.[]|select(.name=="'.$artifact_name.'")][-1]|.expired'
539            );
540
1
50
7
            if ($expired_json ne '') {
541
1
4
                chomp $expired_json;
542
1
2
                my $expired;
543
1
1
50
3
9
                eval { $expired = decode_json $expired_json } || CheckSpelling::Util::die_custom $program, 543, "decode_json failed in update_repository with '$expired_json'";
544
1
50
124
                if ($expired) {
545
1
23
                    print "$program: GitHub Run Artifact expired. You will need to trigger a new run.\n";
546
1
1
1
1
4
13
                    { CheckSpelling::Util::tear_here(1); die "exiting"; }
547                }
548            }
549
0
0
            print "$program: GitHub Run may not have completed. If so, please wait for it to finish and try again.\n";
550
0
0
0
0
0
0
            { CheckSpelling::Util::tear_here(2); die "exiting"; }
551        }
552
1
50
3
        if ($gh_err_text =~ /no artifact matches any of the names or patterns provided/) {
553
0
0
0
            $github_server_url = $ENV{GITHUB_SERVER_URL} || '';
554
0
0
            my $run_link;
555
0
0
0
            if ($github_server_url) {
556
0
0
                $run_link = "[$run]($github_server_url/$repo/actions/runs/$run)";
557            } else {
558
0
0
                $run_link = "$run";
559            }
560
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";
561
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";
562
0
0
0
0
0
0
            { CheckSpelling::Util::tear_here(3); die "exiting"; }
563        }
564
1
50
7
        if ($gh_err_text =~ /HTTP 404: Not Found|"status":"404"/) {
565
1
20
            print "$program: The referenced repository ($repo) may not exist, perhaps you do not have permission to see it. If the repository is hosted by GitHub Enterprise, check-spelling does not know how to integrate with it.\n";
566
1
1
1
1
6
15
            { CheckSpelling::Util::tear_here(8); die "exiting"; }
567        }
568
0
0
0
0
        if ($gh_err_text =~ /HTTP 403: API rate limit exceeded for .*?./) {
569        } elsif ($gh_err_text =~ m{dial tcp \S+:\d+: i/o timeout$}) {
570
0
0
0
            if ($retries_remaining <= 0) {
571
0
0
                print "$program: Timeout connecting to GitHub. This is probably caused by an outage of sorts.\nCheck https://www.githubstatus.com/history\nTry again later.";
572
0
0
0
0
0
0
                { CheckSpelling::Util::tear_here(9); die "exiting"; }
573            }
574        } else {
575
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";
576
0
0
            print $gh_err_text;
577
0
0
0
0
0
0
            { CheckSpelling::Util::tear_here(4); die "exiting"; }
578        }
579
0
0
0
        my $request_id = $1 if ($gh_err_text =~ /\brequest ID\s+(\S+)/);
580
0
0
0
        my $timestamp = $1 if ($gh_err_text =~ /\btimestamp\s+(.*? UTC)/);
581
0
0
0
        my $has_gh_token = defined $ENV{GH_TOKEN} || defined $ENV{GITHUB_TOKEN};
582
0
0
        my $meta_url = 'https://api.github.com/meta';
583
0
0
        while (1) {
584
0
0
            my @curl_args = qw(curl);
585
0
0
0
            unless ($has_gh_token) {
586
0
0
                my ($gh_token) = get_token();
587
0
0
0
                push @curl_args, '-u', "token:$gh_token" if defined $gh_token;
588            }
589
0
0
            push @curl_args, '-I', $meta_url;
590
0
0
            my ($curl_stdout, $curl_stderr, $curl_result);
591
0
0
            ($curl_stdout, $curl_stderr, $curl_result) = capture_system(@curl_args);
592
0
0
            my $delay = 1;
593
0
0
0
0
            if ($curl_stdout =~ m{^HTTP/\S+\s+200}) {
594
0
0
0
                if ($curl_stdout =~ m{^x-ratelimit-remaining:\s+(\d+)$}m) {
595
0
0
                    my $ratelimit_remaining = $1;
596
0
0
0
                    last if ($ratelimit_remaining > 10);
597
598
0
0
                    $delay = 5;
599
0
0
                    print STDERR "Sleeping for $delay seconds because $ratelimit_remaining is close to 0\n";
600                } else {
601
0
0
                    print STDERR "Couldn't find x-ratelimit-remaining, will sleep for $delay\n";
602                }
603            } elsif ($curl_stdout =~ m{^HTTP/\S+\s+403}) {
604
0
0
0
                if ($curl_stdout =~ /^retry-after:\s+(\d+)/m) {
605
0
0
                    $delay = $1;
606
0
0
                    print STDERR "Sleeping for $delay seconds (presumably due to API rate limit)\n";
607                } else {
608
0
0
                    print STDERR "Couldn't find retry-after, will sleep for $delay\n";
609                }
610            } else {
611
0
0
0
                my $response = $1 if $curl_stdout =~ m{^(HTTP/\S+)};
612
0
0
                print STDERR "Unexpected response ($response) from $meta_url; sleeping for $delay\n";
613            }
614
0
0
            sleep $delay;
615        }
616    }
617}
618
619sub update_repository {
620
1
1
3
    my ($artifact) = @_;
621
1
2
    our $program;
622
1
50
6
    CheckSpelling::Util::die_custom $program, 622, "$program: artifact argument contains quote characters" if $artifact =~ /'/;
623
1
2
    my $apply = unzip_pipe($artifact, 'apply.json');
624
1
50
5
    unless ($apply =~ /\{.*\}/s) {
625
0
0
        print STDERR "$program: Could not retrieve valid apply.json from artifact\n";
626
0
0
        $apply = '{
627            "expect_files": [".github/actions/spelling/expect.txt"],
628            "new_expect_file": ".github/actions/spelling/expect.txt",
629            "excludes_file": ".github/actions/spelling/excludes.txt",
630            "spelling_config": ".github/actions/spelling"
631        }';
632    }
633
1
1
    my $config_ref;
634
1
1
50
1
5
    eval { $config_ref = decode_json($apply); } ||
635        CheckSpelling::Util::die_custom $program, 635, "$program: decode_json failed in update_repository with '$apply'";
636
637
1
440
    my $git_repo_root = run_pipe('git', 'rev-parse', '--show-toplevel');
638
1
2
    chomp $git_repo_root;
639
1
50
5
    CheckSpelling::Util::die_custom $program, 639, "$program: Could not find git repo root..." unless $git_repo_root =~ /\w/;
640
1
11
    chdir $git_repo_root;
641
642
1
7
    retrieve_spell_check_this($artifact, $config_ref);
643
1
11
    remove_stale($artifact, $config_ref);
644
1
2
    add_expect($artifact, $config_ref);
645
1
9
    add_to_excludes($artifact, $config_ref);
646
1
1918
    system('git', 'add', '-u', '--', $config_ref->{'spelling_config'});
647}
648
649sub extract_artifacts_from_file {
650
1
1
3
    my ($artifact) = @_;
651
1
1706
    open my $artifact_reader, '-|', 'unzip', '-l', $artifact;
652
1
8
    my ($has_artifact, $only_file) = (0, 0);
653
1
983
    while (my $line = <$artifact_reader>) {
654
6
7
        chomp $line;
655
6
100
15
        if ($line =~ /\s+artifact\.zip$/) {
656
1
1
            $has_artifact = 1;
657
1
3
            next;
658        }
659
5
100
9
        if ($line =~ /\s+1 file$/) {
660
1
2
            $only_file = 1;
661
1
127
            next;
662        }
663
4
50
10
        $only_file = 0 if $only_file;
664    }
665
1
9
    close $artifact_reader;
666
1
0
    my @artifacts;
667
1
50
33
11
    if ($has_artifact && $only_file) {
668
1
5
        my $artifact_dir = tempdir(CLEANUP => 1);
669
1
242
        my ($fh, $gh_err) = tempfile();
670
1
112
        close $fh;
671
1
1833
        system('unzip', '-q', '-d', $artifact_dir, $artifact, 'artifact.zip');
672
1
23
        @artifacts = ("$artifact_dir/artifact.zip");
673    } else {
674
0
0
        @artifacts = ($artifact);
675    }
676
1
11
    return @artifacts;
677}
678
679sub main {
680
2
2
161632
    our $program;
681
2
3
    my ($bash_script, $first, $run);
682
2
4
    ($program, $bash_script, $first, $run) = @_;
683
2
4
    my $syntax = "$program <RUN_URL | OWNER/REPO RUN | ARTIFACT.zip>";
684    # Stages
685    # - 1 check for tools basic
686
2
7
    check_basic_tools();
687    # - 2 check for current
688    # -> 1. download the latest version to a temp file
689    # -> 2. parse current and latest (stripping comments) and compare (whitespace insensitively)
690    # -> 3. offer to update if the latest version is different
691
2
8
    check_current_script($bash_script);
692    # - 4 parse arguments
693
2
50
3
    CheckSpelling::Util::die_custom $program, 693, $syntax unless defined $first;
694
2
50
6
    $ENV{'GITHUB_API_URL'} ||= 'https://api.github.com';
695
2
3
    my $repo;
696    my @artifacts;
697
2
100
16
    if (-s $first) {
698
1
4
        @artifacts = extract_artifacts_from_file($first);
699    } else {
700
1
0
        my $suffix;
701
1
50
1
        if ($first =~ m{^\s*https://.*/([^/]+/[^/]+)/actions/runs/(\d+)(?:/attempts/\d+|)(?:#(\S+)|)\s*$}) {
702
0
0
            ($repo, $run, $suffix) = ($1, $2, $3);
703        } else {
704
1
2
            $repo = $first;
705        }
706
1
50
33
9
        CheckSpelling::Util::die_custom $program, 706, $syntax unless defined $repo && defined $run;
707        # - 3 check for tool readiness (is `gh` working)
708
0
0
        tools_are_ready($program);
709
0
0
        @artifacts = get_artifacts($repo, $run, $suffix);
710    }
711
712    # - 5 do work
713
1
2
    for my $artifact (@artifacts) {
714
1
4
        update_repository($artifact);
715    }
716}
717
718# main($0 ne '-' ? $0 : 'apply.pl', $bash_script, @ARGV);