83e4b601fca414f90847e42e6fe3396174129ba3
[bugscan.git] / scanlib.pm
1 #! /usr/bin/perl
2 # vim: ts=4 sw=4 nowrap
3 #
4 # General functions for scanning the BTS-database.
5 # Based on bugscan, written by Richard Braakman <dark@debian.org>,
6 # which was based on an unknown other script.
7 #
8 # Global variables:
9 #   %comments       - map from bugnumber to bug description
10 #   %premature      - list of prematurely closed bugreports
11 #   %exclude        - list of bugreports to exclude from the report
12 #   %maintainer     - map from packagename to maintainer
13 #   %section        - map from packagename to section in the FTP-site
14 #   %packagelist    - map from packagename to bugreports
15 #   %NMU            - map with NMU information
16
17 use lib qw(/org/bugs.debian.org/perl);
18 use LWP::UserAgent;
19 use Debbugs::MIME qw(decode_rfc1522 encode_rfc1522);
20 use Debbugs::Packages;
21 use Debbugs::Versions;
22 use Debbugs::Status;
23 use Fcntl qw(O_RDONLY);
24 use strict;
25 use warnings;
26 require bugcfg;
27 package scanlib;
28
29 our (%comments,%premature,%exclude,%maintainer,%section,%packagelist,%NMU,%debbugssection,%bugs);
30
31 sub readcomments() {
32 # Read bug commentary 
33 # It is in paragraph format, with the first line of each paragraph being
34 # the bug number or package name to which the comment applies.
35 # Prefix a bug number with a * to force it to be listed even if it's closed.
36 # (This deals with prematurely closed bugs)
37
38         my $index;                                      # Bug-number for current comment
39         my $file;                                       # Name of comments-file
40
41         %comments = ();                                 # Initialize our data
42         %premature = ();
43         %exclude = ();
44         $file=shift;
45         open(C, $file) or die "open $file: $!\n";
46         while (<C>) {
47                 chomp;
48                 if (m/^\s*$/) {                         # Check for paragraph-breaks
49                         undef $index;
50                 } elsif (defined $index) {
51                         $comments{$index} .= $_ . "\n";
52                 } else {
53                         if (s/^\*//) {                  # Test & remove initial *
54                                 $premature{$_} = 1;
55                         }
56                         if (s/\s+EXCLUDE\s*//) {        # Test & remove EXCLUDE
57                                 $exclude{$_} = 1;
58                                 next;
59                         }
60                         $index = $_;
61                         $comments{$index} = ''; # New comment, initialize data
62                 }
63         }
64         close(C);
65 }
66
67
68 # Read the list of maintainer 
69 sub readmaintainers() {
70         my $pkg;                                        # Name of package
71         my $mnt;                                        # Maintainer name & email
72
73         open(M, $bugcfg::maintainerlist) or die "open $bugcfg::maintainerlist: $!\n";
74         while (<M>) {
75                 chomp;
76                 m/^(\S+)\s+(\S.*\S)\s*$/ or die "Maintainers: $_ ?";
77                 ($pkg, $mnt) = ($1, $2);
78                 $pkg =~ y/A-Z/a-z/;                     # Normalize package-name. why???
79                 $_=$mnt;
80                 if (not m/</) {
81                         $mnt="$2 <$1>" if ( m/(\S+)\s+\(([^)]+)\)/ );
82                 }
83                 $maintainer{$pkg}= $mnt;
84         }
85         close(M);
86 }
87
88
89 sub readsources() {
90         my $root;                                       # Root of archive we are scanning
91         my $archive;                            # Name of archive we are scanning
92         my $sect;                                       # Name of current section
93
94         $root=shift;
95         $archive=shift;
96         for $sect (@bugcfg::sections) {
97                 open(P, "zcat $root/$sect/source/Sources.gz|")
98                         or die open "open: $sect sourcelist: $!\n";
99                 while (<P>) {
100                         chomp;
101                         next unless m/^Package:\s/;
102                         s/^Package:\s*//;                       # Strip the fieldname
103                         $section{$_} = "$archive/$sect";
104                 }
105                 close (P);
106         }
107 }
108
109 sub readpackages() {
110         my $root;                                       # Root of archive we are scanning
111         my $archive;                            # Name of archive we are scanning
112         my $sect;                                       # Name of current section
113         my $arch;                                       # Name of current architecture
114
115         $root=shift;
116         $archive=shift;
117         for $arch ( @bugcfg::architectures ) {
118                 for $sect ( @bugcfg::sections) {
119                         open(P, "zcat $root/$sect/binary-$arch/Packages.gz|")
120                                 or die "open: $root/$sect/binary-$arch/Packages.gz: $!\n";
121                         while (<P>) {
122                                 chomp;
123                                 next unless m/^Package:\s/;     # We're only interested in the packagenames
124                                 s/^Package:\s*//;                       # Strip the fieldname
125                                 $section{$_} = "$archive/$sect";
126                         }
127                         close(P);
128                 }
129         }
130 }
131
132 sub readdebbugssources() {
133         my $file;
134         my $archive;
135
136         $file=shift;
137         $archive=shift;
138         open(P, $file)
139                 or die "open: $file: $!\n";
140         while (<P>) {
141                 chomp;
142                 my ($host, $bin, $sect, $ver, $src) = split /\s+/;
143                 my $sectname = ($sect =~ /^\Q$archive/) ? $sect : "$archive/$sect";
144                 $debbugssection{$bin} = $sectname;
145                 $debbugssection{$src} = $sectname;
146         }
147         close(P);
148 }
149
150 sub readpseudopackages() {
151         open(P, $bugcfg::pseudolist) or die("open $bugcfg::pseudolist: $!\n");
152         while (<P>) {
153                 chomp;
154                 s/\s.*//;
155                 $section{$_} = "pseudo";
156         }
157         close(P);
158 }
159
160
161 sub scanspool() {
162         my @dirs;
163         my $dir;
164
165         chdir($bugcfg::spooldir) or die "chdir $bugcfg::spooldir: $!\n";
166
167         opendir(DIR, $bugcfg::spooldir) or die "opendir $bugcfg::spooldir: $!\n";
168         @dirs=grep(m/^\d+$/,readdir(DIR));
169         closedir(DIR);
170
171         for $dir (@dirs) {
172                 scanspooldir("$bugcfg::spooldir/$dir");
173         }
174
175 }
176
177 sub scanspooldir() {
178         my ($dir)               = @_;
179         my $f;                  # While we're currently processing
180         my @list;               # List of files to process
181         my $skip;               # Flow control
182         my $walk;               # index variable
183         my $taginfo;    # Tag info
184
185         chdir($dir) or die "chdir $dir: $!\n";
186
187         opendir(DIR, $dir) or die "opendir $dir: $!\n";
188         @list = grep { s/\.summary$// }
189                         grep { m/^\d+\.summary$/ } 
190                         readdir(DIR);
191         closedir(DIR);
192
193         for $f (@list) {
194                 next if $exclude{$f};                   # Check the list of bugs to skip
195         
196                 my $bug = Debbugs::Status::read_bug(summary => "$f.summary");
197                 next if (!defined($bug));
198                 
199                 $skip=1;
200                 for $walk (@bugcfg::priorities) {
201                         $skip=0 if $walk eq $bug->{'severity'};
202                 }
203
204                 my @tags = split(' ', $bug->{'keywords'});
205                 for my $tag (@tags) {
206                         for my $s (@bugcfg::skiptags) {
207                                 $skip=1 if $tag eq $s;
208                         }
209                 }
210                 next if $skip==1;
211         
212                 my %disttags = ();      
213                 $disttags{'oldstable'}    = grep(/^woody$/, @tags);
214                 $disttags{'stable'}       = grep(/^sarge$/, @tags);
215                 $disttags{'testing'}      = grep(/^etch$/, @tags);
216                 $disttags{'unstable'}     = grep(/^sid$/, @tags);
217                 $disttags{'experimental'} = grep(/^experimental$/, @tags);
218                 
219                 my $relinfo = "";
220                 warn $section{$bug->{'package'}};
221                 if ($section{$bug->{'package'}} eq 'pseudo') {
222                         # versioning information makes no sense for pseudo packages,
223                         # just use the tags
224                         for my $dist qw(oldstable stable testing unstable experimental) {
225                                 $relinfo .= uc(substr($dist, 0, 1)) if $disttags{$dist};
226                         }
227                 } else {
228                         # default according to dondelelcaro 2006-11-11
229                         if (!$disttags{'oldstable'} && !$disttags{'stable'} && !$disttags{'testing'} && !$disttags{'unstable'} && !$disttags{'experimental'}) {
230                                 $disttags{'testing'} = 1;
231                                 $disttags{'unstable'} = 1;
232                                 $disttags{'experimental'} = 1;
233                         }
234
235                         # only bother to check the versioning status for the distributions indicated by the tags 
236                         for my $dist qw(oldstable stable testing unstable experimental) {
237                                 local $SIG{__WARN__} = sub {};
238
239                                 next if (!$disttags{$dist});
240
241                                 my $presence = Debbugs::Status::bug_presence(
242                                         bug => $f, 
243                                         status => $bug, 
244                                         dist => $dist, 
245                                         arch => [ qw(alpha amd64 arm hppa i386 ia64 mips mipsel powerpc s390 sparc) ]
246                                 );
247
248                                 # ignore bugs that are absent/fixed in this distribution, include everything
249                                 # else (that is, "found" which says that the bug is present, and undef, which
250                                 # indicates that no versioning information is present and it's not closed
251                                 # unversioned)
252                                 if (!defined($presence) || ($presence ne 'absent' && $presence ne 'fixed')) {
253                                         $relinfo .= uc(substr($dist, 0, 1));
254                                 }
255                         }
256                         
257                         next if $relinfo eq '' and not $premature{$f};
258                         $premature{$f}++ if $relinfo eq '';
259                 }
260
261                 $taginfo = "[";
262                 $taginfo .= ($bug->{'keywords'} =~ /\bpending\b/        ? "P" : " ");
263                 $taginfo .= ($bug->{'keywords'} =~ /\bpatch\b/          ? "+" : " ");
264                 $taginfo .= ($bug->{'keywords'} =~ /\bhelp\b/           ? "H" : " ");
265                 $taginfo .= ($bug->{'keywords'} =~ /\bmoreinfo\b/       ? "M" : " ");
266                 $taginfo .= ($bug->{'keywords'} =~ /\bunreproducible\b/ ? "R" : " ");
267                 $taginfo .= ($bug->{'keywords'} =~ /\bsecurity\b/       ? "S" : " ");
268                 $taginfo .= ($bug->{'keywords'} =~ /\bupstream\b/       ? "U" : " ");
269                 $taginfo .= ($bug->{'keywords'} =~ /\betch-ignore\b/    ? "I" : " ");
270                 $taginfo .= "]";
271
272                 if (length($bug->{'mergedwith'})) {
273                         my @merged = split(' ', $bug->{'mergedwith'});
274                         next if ($merged[0] < $f);
275                 }
276
277                 for my $package (split /[,\s]+/, $bug->{'package'}) {
278                         $_= $package; y/A-Z/a-z/; $_= $` if m/[^-+._a-z0-9]/;
279                         if (not defined $section{$_}) {
280                                 if (defined $debbugssection{$_}) {
281                                         $relinfo .= "X";
282                                 } else {
283                                         next;   # Skip unavailable packages
284                                 }
285                         }
286
287                         push @{$packagelist{$_}}, $f;
288                 }
289
290                 if ($relinfo eq "") { # or $relinfo eq "U" # confuses e.g. #210306
291                         $relinfo = "";
292                 } else {
293                         $relinfo = " [$relinfo]";
294                 }
295
296                 $bugs{$f} = "$f $taginfo$relinfo " . $bug->{'subject'};
297         }
298 }
299
300
301 sub readstatus() {
302         my $bug;                # Number of current bug
303         my $subject;    # Subject for current bug
304         my $pkg;                # Name of current package
305         my $file;               # Name of statusfile
306         my $sect;               # Section of current package
307         my $mnt;                # Maintainer of current package
308
309         $file=shift;
310         open(P, $file) or die "open $file: $!";
311         while (<P>) {
312                 chomp;
313                 if (m/^[0-9]+ \[/) {
314                         ($bug,$subject)=split(/ /, $_, 2);
315                         $bugs{$bug}=$subject;
316                         push @{$packagelist{$pkg}}, $bug;
317                 } else {
318                         ($pkg,$sect, $mnt)=split(/ /, $_, 3);
319                         next if (!defined($pkg));
320                         $section{$pkg}=$sect;
321                         $maintainer{$pkg}=$mnt;
322                 }
323         }
324         close P;
325 }
326
327
328 sub readNMUstatus() {
329         my $bug;       # Number of current bug
330         my $source;    # Source upload which closes this bug.
331         my $version;   # Version where this bug was closed.
332         my $flag;      # Whether this paragraph has been processed.
333         my ($field, $value);
334
335         for (split /\n/, LWP::UserAgent->new->request(HTTP::Request->new(GET => shift))->content) {
336                 chomp;
337                 if (m/^$/) {
338                         $NMU{$bug} = 1;
339                         $NMU{$bug, "source"} = $source;
340                         $NMU{$bug, "version"} = $version;
341 #                       $comments{$bug} .= "[FIXED] Fixed package $source is in Incoming\n";
342                         $flag = 0;
343                 } else {
344                         ($field, $value) = split(/: /, $_, 2);
345                         $bug = $value if($field =~ /bug/i);
346                         $source = $value if($field =~ /source/i);
347                         $version = $value if($field =~ /version/i);
348                         $flag = 1;
349                 }
350         }
351         if ($flag) {
352                 $NMU{$bug} = 1;
353                 $NMU{$bug, "source"} = $source;
354                 $NMU{$bug, "version"} = $version;
355 #               $comments{$bug} .= "[FIXED] Fixed package $source in in Incoming\n";
356         }
357         close P;
358 }
359
360
361 sub urlsanit {
362         my $url = shift;
363         $url =~ s/%/%25/g;
364         $url =~ s/\+/%2b/g;
365         my %saniarray = ('<','lt', '>','gt', '&','amp', '"','quot');
366         $url =~ s/([<>&"])/\&$saniarray{$1};/g;
367         return $url;
368 }
369
370 sub htmlsanit {
371     my %saniarray = ('<','lt', '>','gt', '&','amp', '"','quot');
372     my $in = shift || "";
373     $in =~ s/([<>&"])/\&$saniarray{$1};/g;
374     return $in;
375 }
376
377 sub wwwnumber() {
378         my $number = shift;             # Number of bug to html-ize
379 #       my $section);                           # Section for the bug
380
381         "<A HREF=\"http://bugs.debian.org/cgi-bin/bugreport.cgi?archive=no&amp;bug=" .
382                 urlsanit($number) . '">' . htmlsanit($number) . '</A>';
383 #       ($section=$number) =~ s/([0-9]{2}).*/$1/;
384 #       "<A HREF=\"${btsURL}/db/$section/$number.html\">$number</A>";
385 }
386
387 sub wwwname() {
388         my $name = shift;                       # Name of package
389
390         "<A HREF=\"http://bugs.debian.org/cgi-bin/pkgreport.cgi?archive=no&amp;pkg=" .
391                 urlsanit($name) . '">' . htmlsanit($name) . '</A>';
392 #       "<A HREF=\"${btsURL}/db/pa/l$name.html\">$name</A>";
393 }
394
395 sub check_worry {
396         my ($status) = @_;
397
398         if ($status =~ m/^\[[^]]*I/ or
399             $status =~ m/ \[[^]]*X/ or
400             ($status =~ m/ \[[^]]*[OSUE]/ and $status !~ m/ \[[^]]*T/)) {
401                 return 0;
402         }
403         return 1;
404 }
405
406 1;