File Coverage

File:lib/CheckSpelling/Sarif.pm
Coverage:86.5%

linestmtbrancondsubtimecode
1#! -*-perl-*-
2
3package CheckSpelling::Sarif;
4
5our $VERSION='0.1.0';
6our $flatten=0;
7
8
1
1
1
1
108017
1
2
use utf8;
9
10
1
1
1
1
15
0
27
use File::Basename;
11
1
1
1
1
206
1237
72
use Digest::SHA qw($errmsg);
12
1
1
1
1
2
1
25
use JSON::PP;
13
1
1
1
1
173
3722
26
use Hash::Merge qw( merge );
14
1
1
1
1
142
1
18
use CheckSpelling::Util;
15
1
1
1
1
148
1
1760
use CheckSpelling::GitSources;
16
17sub encode_low_ascii {
18
9
9
224
    $_ = shift;
19
9
1
12
3
    s/([\x{0}-\x{9}\x{0b}\x{1f}#%])/"\\u".sprintf("%04x",ord($1))/eg;
20
9
5
    return $_;
21}
22
23sub url_encode {
24
10
10
10
    $_ = shift;
25
10
0
8
0
    s<([^-!\$&'()*+,/:;=?\@A-Za-z0-9_.~])><"%".sprintf("%02x",ord($1))>eg;
26
10
19
    return $_;
27}
28
29sub double_slash_escape {
30
8
8
2
    $_ = shift;
31
8
30
    s/(["()\]\\])/\\\\$1/g;
32
8
7
    return $_;
33}
34
35sub fingerprintLocations {
36
6
6
8
    my ($locations, $encoded_files_ref, $line_hashes_ref, $hashes_needed_for_files_ref, $message, $hashed_message) = @_;
37
6
3
    my @locations_json = ();
38
6
3
    my @fingerprints = ();
39
6
6
    for my $location (@$locations) {
40
9
3
        my $encoded_file = $location->{uri};
41
9
6
        my $line = $location->{startLine};
42
9
5
        my $column = $location->{startColumn};
43
9
5
        my $endColumn = $location->{endColumn};
44
9
9
        my $partialFingerprint = '';
45
9
4
        my $file = $encoded_files_ref->{$encoded_file};
46
9
100
7
        if (defined $line_hashes_ref->{$file}) {
47
8
10
            my $line_hash = $line_hashes_ref->{$file}{$line};
48
8
100
6
            if (defined $line_hash) {
49
4
4
2
6
                my @instances = sort keys %{$hashes_needed_for_files_ref->{$file}{$line}{$hashed_message}};
50
4
2
                my $hit = scalar @instances;
51
4
4
                while (--$hit > 0) {
52
0
0
0
                    last if $instances[$hit] == $column;
53                }
54
4
9
                $partialFingerprint = Digest::SHA::sha1_base64("$line_hash:$message:$hit");
55            }
56        }
57
9
7
        push @fingerprints, $partialFingerprint;
58
9
50
8
        my $startColumn = $column ? qq<, "startColumn": $column> : '';
59
9
50
9
        $endColumn = $endColumn ? qq<, "endColumn": $endColumn> : '';
60
9
50
4
        $line = 1 unless $line;
61
9
8
        my $json_fragment = qq<{ "physicalLocation": { "artifactLocation": { "uri": "$encoded_file", "uriBaseId": "%SRCROOT%" }, "region": { "startLine": $line$startColumn$endColumn } } }>;
62
9
7
        push @locations_json, $json_fragment;
63    }
64
6
15
    return { locations_json => \@locations_json, fingerprints => \@fingerprints };
65}
66
67sub hashFiles {
68
2
2
3
    my ($hashes_needed_for_files_ref, $line_hashes_ref, $directoryToRepo_ref, $used_hashes_ref) = @_;
69
2
10
    for my $file (sort keys %$hashes_needed_for_files_ref) {
70
3
4
        $line_hashes_ref->{$file} = ();
71
3
100
21
        unless (-e $file) {
72
1
2
            delete $hashes_needed_for_files_ref->{$file};
73
1
1
            next;
74        }
75
2
2
1
6
        my @lines = sort (keys %{$hashes_needed_for_files_ref->{$file}});
76
2
50
70
        unless (defined $directoryToRepo_ref->{dirname($file)}) {
77
2
10
            my ($parsed_file, $git_base_dir, $prefix, $remote_url, $rev, $branch) = CheckSpelling::GitSources::git_source_and_rev($file);
78        }
79
2
29
        open $file_fh, '<', $file;
80
2
2
        my $line = shift @lines;
81
2
100
4
        $line = 2 if $line == 1;
82
2
3
        my $buffer = '';
83
2
14
        while (<$file_fh>) {
84
108
100
54
            if ($line == $.) {
85
5
3
                my $sample = substr $buffer, -100, 100;
86
5
6
                my $hash = Digest::SHA::sha1_base64($sample);
87
5
5
                for (; $line == $.; $line = shift @lines) {
88
6
8
                    my $hit = $used_hashes_ref->{$hash}++;
89
6
100
4
                    $hash = "$hash:$hit" if $hit;
90
6
6
                    $line_hashes_ref->{$file}{$line} = $hash;
91
6
100
9
                    last unless @lines;
92                }
93            }
94
108
54
            $buffer .= $_;
95
108
355
            $buffer =~ s/\s+/ /g;
96
108
85
            $buffer = substr $buffer, -100, 100;
97        }
98
2
7
        close $file_fh;
99    }
100}
101
102sub addToHashesNeededForFiles {
103
9
9
24
    my ($file, $line, $column, $message, $hashes_needed_for_files_ref) = @_;
104
9
27
    my $hashed_message = Digest::SHA::sha1_base64($message);
105
9
100
9
    $hashes_needed_for_files_ref->{$file} = () unless defined $hashes_needed_for_files_ref->{$file};
106
9
100
20
    $hashes_needed_for_files_ref->{$file}{$line} = () unless defined $hashes_needed_for_files_ref->{$file}{$line};
107
9
100
20
    $hashes_needed_for_files_ref->{$file}{$line}{$hashed_message} = () unless defined $hashes_needed_for_files_ref->{$file}{$line}{$hashed_message};
108
9
35
    $hashes_needed_for_files_ref->{$file}{$line}{$hashed_message}{$column} = '1';
109}
110
111sub parse_warnings {
112
1
1
2
    my ($warnings) = @_;
113
1
0
    our $flatten;
114
1
1
    our %directoryToRepo;
115
1
0
    our $provenanceInsertion;
116
1
1
    our %provenanceStringToIndex;
117
1
0
    our %directoryToProvenanceInsertion;
118
1
1
    my @results;
119
1
50
14
    unless (open WARNINGS, '<', $warnings) {
120
0
0
        print STDERR "Could not open $warnings\n";
121
0
0
        return [];
122    }
123
1
1
    my $rules = ();
124
1
1
    my %encoded_files = ();
125
1
2
    my %hashes_needed_for_files = ();
126
1
15
    while (<WARNINGS>) {
127
10
100
10
        next if m{^https://};
128
9
100
35
        next unless m{^(.+):(\d+):(\d+) \.\.\. (\d+),\s(Error|Warning|Notice)\s-\s(.+\s\((.+)\))$};
129
8
27
        my ($file, $line, $column, $endColumn, $severity, $message, $code) = ($1, $2, $3, $4, $5, $6, $7);
130
8
106
        my $directory = dirname($file);
131
8
100
8
        unless (defined $directoryToProvenanceInsertion{$directory}) {
132
2
3
            my $provenanceString = collectVersionControlProvenance($file);
133
2
50
155
            if ($provenanceString) {
134
2
50
3
                if (defined $provenanceStringToIndex{$provenanceString}) {
135
0
0
                    $directoryToProvenanceInsertion{$directory} = $provenanceStringToIndex{$provenanceString};
136                } else {
137
2
5
                    $provenanceStringToIndex{$provenanceString} = $provenanceInsertion;
138
2
3
                    $directoryToProvenanceInsertion{$directory} = $provenanceInsertion;
139
2
2
                    ++$provenanceInsertion;
140                }
141            }
142        }
143        # single-slash-escape `"` and `\`
144
8
8
        $message =~ s/(["\\])/\\$1/g;
145
8
9
        $message = encode_low_ascii $message;
146        # double-slash-escape `"`, `(`, `)`, `]`
147
8
9
        $message = double_slash_escape $message;
148        # encode `message` and `file` to protect against low ascii`
149
8
4
        my $encoded_file = url_encode $file;
150
8
8
        $encoded_files{$encoded_file} = $file;
151        # hack to make the first `...` identifier a link (that goes nowhere, but is probably blue and underlined) in GitHub's SARIF view
152
8
100
8
        if ($message =~ /(`{2,})/) {
153
1
1
            my $backticks = $1;
154
1
31
            while ($message =~ /($backticks`+)(?=[`].*?\g{-1})/gs) {
155
0
0
0
                $backticks = $1 if length($1) > length($backticks);
156            }
157
1
14
            $message =~ s/(^|[^\\])$backticks(.+?)$backticks/${1}[${2}](#security-tab)/;
158        } else {
159
7
20
            $message =~ s/(^|[^\\])\`([^`]+[^`\\])\`/${1}[${2}](#security-tab)/;
160        }
161        # replace '`' with `\`+`'` because GitHub's SARIF parser doesn't like them
162
8
8
        $message =~ s/\`/'/g;
163
8
100
7
        unless (defined $rules->{$code}) {
164
2
4
            $rules->{$code} = {};
165        }
166
8
3
        my $rule = $rules->{$code};
167
8
100
6
        unless (defined $rule->{$message}) {
168
5
10
            $rule->{$message} = [];
169        }
170
8
9
        addToHashesNeededForFiles($file, $line, $column, $message, \%hashes_needed_for_files);
171
8
5
        my $locations = $rule->{$message};
172
8
17
        my $physicalLocation = {
173            'uri' => $encoded_file,
174            'startLine' => $line,
175            'startColumn' => $column,
176            'endColumn' => $endColumn,
177        };
178
8
3
        push @$locations, $physicalLocation;
179
8
26
        $rule->{$message} = $locations;
180    }
181
1
0
    my %line_hashes = ();
182
1
2
    my %used_hashes = ();
183
1
3
    hashFiles(\%hashes_needed_for_files, \%line_hashes, \%directoryToRepo, \%used_hashes);
184
1
1
1
3
    for my $code (sort keys %{$rules}) {
185
2
2
        my $rule = $rules->{$code};
186
2
2
2
15
        for my $message (sort keys %{$rule}) {
187
5
10
            my $hashed_message = Digest::SHA::sha1_base64($message);
188
5
3
            my $locations = $rule->{$message};
189
5
7
            my $fingerprintResults = fingerprintLocations($locations, \%encoded_files, \%line_hashes, \%hashes_needed_for_files, $message, $hashed_message);
190
5
5
2
4
            my @locations_json = @{$fingerprintResults->{locations_json}};
191
5
5
3
2
            my @fingerprints = @{$fingerprintResults->{fingerprints}};
192
5
50
5
            if ($flatten) {
193
0
0
                my $locations_json_flat = join ',', @locations_json;
194
0
0
                my $partialFingerprints;
195
0
0
                my $partialFingerprint = (sort @fingerprints)[0];
196
0
0
0
                if ($partialFingerprint ne '') {
197
0
0
                    $partialFingerprints = qq<"partialFingerprints": { "cs0" : "$partialFingerprint" },>;
198                }
199
0
0
                $message =~ s/\s\\\\\([^()]+?\\\)$//g;
200
0
0
                my $result_json = qq<{"ruleId": "$code", $partialFingerprints "message": { "text": "$message" }, "locations": [ $locations_json_flat ] }>;
201
0
0
                my $result = decode_json $result_json;
202
0
0
                push @results, $result;
203            } else {
204
5
1
                my $limit = scalar @locations_json;
205
5
5
                for (my $i = 0; $i < $limit; ++$i) {
206
8
8
                    my $locations_json_flat = $locations_json[$i];
207
8
1
                    my $partialFingerprints = '';
208
8
4
                    my $partialFingerprint = $fingerprints[$i];
209
8
100
4
                    if ($partialFingerprint ne '') {
210
4
4
                        $partialFingerprints = qq<"partialFingerprints": { "cs0" : "$partialFingerprint" },>;
211                    }
212
8
14
                    $message =~ s/\s\\\\\([^()]+?\\\)$//g;
213
8
9
                    my $result_json = qq<{"ruleId": "$code", $partialFingerprints "message": { "text": "$message" }, "locations": [ $locations_json_flat ] }>;
214
8
7
                    my $result = decode_json $result_json;
215
8
8229
                    push @results, $result;
216                }
217            }
218        }
219    }
220
1
7
    close WARNINGS;
221
1
9
    return \@results;
222}
223
224sub get_runs_from_sarif {
225
2
2
2
    my ($sarif_json) = @_;
226
2
2
    my %runs_view;
227
2
50
3
    return %runs_view unless $sarif_json->{'runs'};
228
2
2
2
2
    my @sarif_json_runs=@{$sarif_json->{'runs'}};
229
2
3
    foreach my $sarif_json_run (@sarif_json_runs) {
230
2
2
2
5
        my %sarif_json_run_hash=%{$sarif_json_run};
231
2
50
2
        next unless defined $sarif_json_run_hash{'tool'};
232
233
2
2
1
3
        my %sarif_json_run_tool_hash = %{$sarif_json_run_hash{'tool'}};
234
2
50
2
        next unless defined $sarif_json_run_tool_hash{'driver'};
235
236
2
2
1
4
        my %sarif_json_run_tool_driver_hash = %{$sarif_json_run_tool_hash{'driver'}};
237        next unless defined $sarif_json_run_tool_driver_hash{'name'} &&
238
2
50
33
7
            defined $sarif_json_run_tool_driver_hash{'rules'};
239
240
2
3
        my $driver_name = $sarif_json_run_tool_driver_hash{'name'};
241
2
2
2
4
        my @sarif_json_run_tool_driver_rules = @{$sarif_json_run_tool_driver_hash{'rules'}};
242
2
1
        my %driver_view;
243
2
2
        for my $driver_rule (@sarif_json_run_tool_driver_rules) {
244
39
50
26
            next unless defined $driver_rule->{'id'};
245
39
46
            $driver_view{$driver_rule->{'id'}} = $driver_rule;
246        }
247
2
5
        $runs_view{$sarif_json_run_tool_driver_hash{'name'}} = \%driver_view;
248    }
249
2
3
    return %runs_view;
250}
251
252sub collectVersionControlProvenance {
253
2
2
2
    my ($file) = @_;
254
2
19
    my ($parsed_file, $git_base_dir, $prefix, $remote_url, $rev, $branch) = CheckSpelling::GitSources::git_source_and_rev($file);
255
2
50
6
    return '' unless $remote_url;
256
2
3
    my $base = substr $parsed_file, 0, length($file);
257
2
4
    my $provenance = [$remote_url, $rev, $branch, $git_base_dir];
258
2
7
    return JSON::PP::encode_json($provenance);
259}
260
261sub generateVersionControlProvenance {
262
1
1
1
    my ($versionControlProvenanceList, $run) = @_;
263
1
1
    my %provenance;
264    sub buildVersionControlProvenance {
265
1
1
0
        my $d = $_;
266
1
1
1
1
        my ($remote_url, $rev, $branch, $git_base_dir) = @{JSON::PP::decode_json($d)};
267
1
50
325
        my $dir = $git_base_dir eq '.' ? '%SRCROOT%' : "DIR_$provenanceStringToIndex{$d}";
268
1
1
        my $mappedTo = {
269            "uriBaseId" => $dir
270        };
271
1
1
        my $versionControlProvenance = {
272            "mappedTo" => $mappedTo
273        };
274
1
50
2
        $versionControlProvenance->{"revisionId"} = $rev if defined $rev;
275
1
50
3
        $versionControlProvenance->{"branch"} = $branch if defined $branch;
276
1
50
2
        $versionControlProvenance->{"repositoryUri"} = $remote_url if defined $remote_url;
277
1
1
        return $versionControlProvenance;
278    }
279
1
2
    @provenanceList = map(buildVersionControlProvenance,@$versionControlProvenanceList);
280
1
1
    $run->{"versionControlProvenance"} = \@provenanceList;
281}
282
283my $provenanceInsertion = 0;
284my %provenanceStringToIndex = ();
285my %directoryToProvenanceInsertion = ();
286
287sub main {
288
1
1
18479
    my ($sarif_template_file, $sarif_template_overlay_file, $category) = @_;
289
1
50
6
    unless (-f $sarif_template_file) {
290
0
0
        warn "Could not find sarif template";
291
0
0
        return '';
292    }
293
294
1
50
3
    $ENV{GITHUB_SERVER_URL} = '' unless defined $ENV{GITHUB_SERVER_URL};
295
1
50
1
    $ENV{GITHUB_REPOSITORY} = '' unless defined $ENV{GITHUB_REPOSITORY};
296
1
4
    my $sarif_template = CheckSpelling::Util::read_file $sarif_template_file;
297
1
50
1
    die "sarif template is empty" unless $sarif_template;
298
299
1
0
0
9
0
    my $json = JSON::PP->new->utf8->pretty->sort_by(sub { $JSON::PP::a cmp $JSON::PP::b });
300
1
82
    my $sarif_json = $json->decode($sarif_template);
301
302
1
50
33
104784
    if (defined $sarif_template_overlay_file && -s $sarif_template_overlay_file) {
303
1
11
        my $merger = Hash::Merge->new();
304
1
116
        my $merge_behaviors = $merger->{'behaviors'}->{$merger->get_behavior()};
305
1
7
        my $merge_arrays = $merge_behaviors->{'ARRAY'}->{'ARRAY'};
306
307        $merge_behaviors->{'ARRAY'}->{'ARRAY'} = sub {
308
37
50
37
4243
            return $merge_arrays->(@_) if ref($_[0][0]).ref($_[1][0]);
309
37
37
11
47
            return [@{$_[1]}];
310
1
5
        };
311
312
1
3
        my $sarif_template_overlay = CheckSpelling::Util::read_file $sarif_template_overlay_file;
313
1
6
        my %runs_base = get_runs_from_sarif($sarif_json);
314
315
1
24
        my $sarif_template_hash = $json->decode($sarif_template_overlay);
316
1
1800
        my %runs_overlay = get_runs_from_sarif($sarif_template_hash);
317
1
1
        for my $run_id (keys %runs_overlay) {
318
1
50
2
            if (defined $runs_base{$run_id}) {
319
1
7
                my $run_base_hash = $runs_base{$run_id};
320
1
1
                my $run_overlay_hash = $runs_overlay{$run_id};
321
1
1
                for my $overlay_id (keys %$run_overlay_hash) {
322                    $run_base_hash->{$overlay_id} = $merger->merge(
323                        $run_overlay_hash->{$overlay_id},
324
1
3
                        $run_base_hash->{$overlay_id}
325                    );
326                }
327            } else {
328
0
0
                $runs_base{$run_id} = $runs_overlay{$run_id};
329            }
330        }
331        #$sarif_json->
332
1
1
22
1
        my @sarif_json_runs = @{$sarif_json->{'runs'}};
333
1
1
        foreach my $sarif_json_run (@sarif_json_runs) {
334
1
1
1
2
            my %sarif_json_run_hash=%{$sarif_json_run};
335
1
50
1
            next unless defined $sarif_json_run_hash{'tool'};
336
337
1
1
1
1
            my %sarif_json_run_tool_hash = %{$sarif_json_run_hash{'tool'}};
338
1
50
1
            next unless defined $sarif_json_run_tool_hash{'driver'};
339
340
1
1
1
3
            my %sarif_json_run_tool_driver_hash = %{$sarif_json_run_tool_hash{'driver'}};
341
1
1
            my $driver_name = $sarif_json_run_tool_driver_hash{'name'};
342            next unless defined $driver_name &&
343
1
50
33
4
                defined $sarif_json_run_tool_driver_hash{'rules'};
344
345
1
0
            my $driver_view_hash = $runs_base{$driver_name};
346
1
50
3
            next unless defined $driver_view_hash;
347
348
1
1
1
5
            my @sarif_json_run_tool_driver_rules = @{$sarif_json_run_tool_driver_hash{'rules'}};
349
1
2
            for my $driver_rule_number (0 .. scalar @sarif_json_run_tool_driver_rules) {
350
39
2651
                my $driver_rule = $sarif_json_run_tool_driver_rules[$driver_rule_number];
351
39
20
                my $driver_rule_id = $driver_rule->{'id'};
352                next unless defined $driver_rule_id &&
353
39
100
67
53
                    defined $driver_view_hash->{$driver_rule_id};
354
38
26
                $sarif_json_run_tool_driver_hash{'rules'}[$driver_rule_number] = $merger->merge($driver_view_hash->{$driver_rule_id}, $driver_rule);
355            }
356        }
357
1
1
        delete $sarif_template_hash->{'runs'};
358
1
1
        $sarif_json = $merger->merge($sarif_json, $sarif_template_hash);
359    }
360    {
361
1
1
1
585
1
1
        my @sarif_json_runs = @{$sarif_json->{'runs'}};
362
1
1
        foreach my $sarif_json_run (@sarif_json_runs) {
363
1
1
            my %sarif_json_run_automationDetails;
364
1
1
            $sarif_json_run_automationDetails{id} = $category;
365
1
2
            $sarif_json_run->{'automationDetails'} = \%sarif_json_run_automationDetails;
366        }
367    }
368
369
1
1
42
2
    my %sarif = %{$sarif_json};
370
371
1
2
    $sarif{'runs'}[0]{'tool'}{'driver'}{'version'} = $ENV{CHECK_SPELLING_VERSION};
372
373
1
2
    my $results = parse_warnings $ENV{warning_output};
374
1
50
2
    if ($results) {
375
1
1
        $sarif{'runs'}[0]{'results'} = $results;
376
1
1
        our %provenanceStringToIndex;
377
1
1
        my @provenanceList = keys %provenanceStringToIndex;
378
1
3
        generateVersionControlProvenance(\@provenanceList, $sarif{'runs'}[0]);
379
1
1
        my %codes;
380
1
2
        for my $result_ref (@$results) {
381
8
8
4
8
            my %result = %{$result_ref};
382
8
6
            $codes{$result{'ruleId'}} = 1;
383        }
384
1
1
        my $rules_ref = $sarif{'runs'}[0]{'tool'}{'driver'}{'rules'};
385
1
1
1
3
        my @rules = @{$rules_ref};
386
1
2
        my $missing_rule_definition_id = 'missing-rule-definition';
387
1
38
1
23
        my ($missing_rule_definition_ref) = grep { $_->{'id'} eq $missing_rule_definition_id } @rules;
388
1
38
1
20
        @rules = grep { defined $codes{$_->{'id'}} } @rules;
389
1
0
        my $code_index = 0;
390
1
1
1
2
        my %defined_codes = map { $_->{'id'} => $code_index++ } @rules;
391
1
2
2
3
        my @missing_codes = grep { !defined $defined_codes{$_}} keys %codes;
392
1
1
        my $missing_rule_definition_index;
393
1
50
1
        if (@missing_codes) {
394
1
0
            push @rules, $missing_rule_definition_ref;
395
1
2
            $missing_rule_definition_index = $defined_codes{$missing_rule_definition_id} = $code_index++;
396
1
33
28
            my $spellchecker = $ENV{spellchecker} || dirname(dirname(dirname(__FILE__)));
397
1
1
            my %hashes_needed_for_files = ();
398
1
0
            my %line_hashes = ();
399
1
1
            my %used_hashes = ();
400
1
0
            our %directoryToRepo;
401
1
1
            for my $missing_code (@missing_codes) {
402
1
1
                my $message = "No rule definition for `$missing_code`";
403
1
158096
                my $code_locations = `find '$spellchecker' -name '.git*' -prune -type f -name '*.sh' -o -name '*.pl' -o -name '*.pm' -o -name '*.t' -print0|xargs -0 grep -n '$missing_code' | perl -pe 's<^\./><>'`;
404
1
7
                my @locations;
405
1
10
                for my $line (split /\n/, $code_locations) {
406
1
5
                    chomp $line;
407
1
21
                    my ($file, $lineno, $code) = $line =~ /^(.+?):(\d+):(.+)$/;
408
1
50
6
                    next unless defined $file;
409
1
32
                    $code =~ /^(.*?)\b$missing_code\b/;
410
1
4
                    my $startColumn = length($1) + 1;
411
1
8
                    my $location = {
412                        'uri' => url_encode($file),
413                        'startLine' => $lineno,
414                        'startColumn' => $startColumn,
415                        'endColumn' => length($1) + length($missing_code) + 1,
416                    };
417
1
2
                    push @locations, $location;
418
1
3
                    my $encoded_file = url_encode $file;
419
1
3
                    $encoded_files{$encoded_file} = $file;
420
1
4
                    addToHashesNeededForFiles($file, $lineno, $startColumn, $message, \%hashes_needed_for_files);
421                }
422
1
4
                hashFiles(\%hashes_needed_for_files, \%line_hashes, \%directoryToRepo, \%used_hashes);
423
1
6
                my $fingerprintResults = fingerprintLocations(\@locations, \%encoded_files, \%encoded_files, \%line_hashes, $message, Digest::SHA::sha1_base64($message));
424
1
1
1
1
                my @locations_json = @{$fingerprintResults->{locations_json}};
425
1
1
1
1
                my @fingerprints = @{$fingerprintResults->{fingerprints}};
426
1
2
                my $locations_json_flat = join ',', @locations_json;
427
1
1
                my $partialFingerprints = '';
428
1
50
2
                my $locations = $locations_json_flat ? qq<, "locations": [ $locations_json_flat ]> : '';
429
1
3
                my $result_json = qq<{"ruleId": "$missing_rule_definition_id", $partialFingerprints "message": { "text": "$message" }$locations }>;
430
1
7
                my $result = decode_json $result_json;
431
1
1
1024
5
                push @{$results}, $result;
432            }
433        }
434
1
6
        $sarif{'runs'}[0]{'tool'}{'driver'}{'rules'} = \@rules;
435
1
1
0
2
        for my $result_index (0 .. scalar @{$results}) {
436
10
8
            my $result = $results->[$result_index];
437
10
11
            my $ruleId = $result->{'ruleId'};
438
10
100
100
16
            next if defined $ruleId && defined $defined_codes{$ruleId};
439
2
142
            $result->{'ruleId'} = $missing_rule_definition_id;
440        }
441    }
442
443
1
3
    return encode_json \%sarif;
444}
445
4461;