1 # This module is part of debbugs, and is released
2 # under the terms of the GPL version 2, or any later
3 # version at your option.
4 # See the file README and COPYING for more information.
6 # [Other people have contributed to this file; their copyrights should
8 # Copyright 2004 by Collin Watson <cjwatson@debian.org>
9 # Copyright 2007 by Don Armstrong <don@donarmstrong.com>
18 use vars qw($VERSION $DEBUG @EXPORT @EXPORT_OK %EXPORT_TAGS);
19 use Exporter qw(import);
23 $DEBUG = 0 unless defined $DEBUG;
26 %EXPORT_TAGS = (write => [qw(write_log_records),
28 read => [qw(read_log_records record_text record_regex),
30 misc => [qw(escape_log),
34 Exporter::export_ok_tags(qw(write read misc));
35 $EXPORT_TAGS{all} = [@EXPORT_OK];
40 use Debbugs::Common qw(getbuglocation getbugcomponent make_list);
41 use Params::Validate qw(:types validate_with);
42 use Encode qw(encode encode_utf8 is_utf8);
48 Debbugs::Log - an interface to debbugs .log files
52 The Debbugs::Log module provides a convenient way for scripts to read and
53 write the .log files used by debbugs to store the complete textual records
54 of all bug transactions.
56 Debbugs::Log does not decode utf8 into perl's internal encoding or
57 encode into utf8 from perl's internal encoding. For html records and
58 all recips, this should probably be done. For other records, this should
61 =head2 The .log File Format
63 .log files consist of a sequence of records, of one of the following four
64 types. ^A, ^B, etc. represent those control characters.
74 C<[mail]> must start with /^Received: \(at \S+\) by \S+;/, and is copied to
79 Auto-forwarded messages are recorded like this:
85 C<[mail]> must contain /^X-Debian-Bugs(-\w+)?: This is an autoforward from
86 \S+/. The first line matching that is removed; all lines in the message body
87 that begin with 'X' will be copied to the output, minus the 'X'.
89 Nothing in debbugs actually generates this record type any more, but it may
90 still be in old .logs at some sites.
95 [recip]^D[recip]^D[...] OR -t
100 Each [recip] is output after "Message sent"; C<-t> represents the same
101 sendmail option, indicating that the recipients are taken from the headers
102 of the message itself.
110 [html] is copied unescaped to the output. The record immediately following
111 this one is considered "boring" and only shown in certain output modes.
113 (This is a design flaw in the log format, since it makes it difficult to
114 change the HTML presentation later, or to present the data in an entirely
119 No other types of records are permitted, and the file must end with a ^C
130 7 => 'incoming-recv',
133 =head2 Perl Record Representation
135 Each record is a hash. The C<type> field is C<incoming-recv>, C<autocheck>,
136 C<recips>, or C<html> as above; C<text> contains text from C<[mail]> or
137 C<[html]> as above; C<recips> is a reference to an array of recipients
138 (strings), or undef for C<-t>.
146 Creates a new log reader based on a .log filehandle.
148 my $log = Debbugs::Log->new($logfh);
149 my $log = Debbugs::Log->new(bug_num => $nnn);
150 my $log = Debbugs::Log->new(logfh => $logfh);
156 =item bug_num -- bug number
158 =item logfh -- log filehandle
160 =item log_name -- name of log
164 One of the above options must be passed.
173 ($param{logfh}) = @_;
174 $param{inner_file} = 0;
178 {bug_num => {type => SCALAR,
181 logfh => {type => HANDLE,
184 log_name => {type => SCALAR,
187 inner_file => {type => BOOLEAN,
191 %param = validate_with(params => \@_,
195 if (grep({exists $param{$_} and defined $param{$_}}
196 qw(bug_num logfh log_name)) ne 1) {
197 croak "Exactly one of bug_num, logfh, or log_name ".
198 "must be passed and must be defined";
201 my $class = ref($this) || $this;
205 if (exists $param{logfh}) {
206 $self->{logfh} = $param{logfh}
208 elsif (exists $param{log_name}) {
209 $self->{logfh} = IO::File->new($param{log_name},'r') or
210 die "Unable to open bug log $param{log_name} for reading: $!";
212 elsif (exists $param{bug_num}) {
213 my $location = getbuglocation($param{bug_num},'log');
214 my $bug_log = getbugcomponent($param{bug_num},'log',$location);
215 $self->{logfh} = IO::File->new($bug_log, 'r') or
216 die "Unable to open bug log $bug_log for reading: $!";
219 $self->{state} = 'kill-init';
220 $self->{linenum} = 0;
221 $self->{inner_file} = $param{inner_file};
227 Reads and returns a single record from a log reader object. At end of file,
228 returns undef. Throws exceptions using die(), so you may want to wrap this
236 my $logfh = $this->{logfh};
238 # This comes from bugreport.cgi, but is much simpler since it doesn't
239 # worry about the details of output.
243 while (defined (my $line = <$logfh>)) {
246 if (length($line) == 1 and exists $states{ord($line)}) {
248 my $newstate = $states{ord($line)};
250 # disallowed transitions
251 $_ = "$this->{state} $newstate";
252 unless (/^(go|go-nox|html) kill-end$/ or
253 /^(kill-init|kill-end) (incoming-recv|autocheck|recips|html)$/ or
255 die "transition from $this->{state} to $newstate at $this->{linenum} disallowed";
258 $this->{state} = $newstate;
259 if ($this->{state} =~ /^(autocheck|recips|html|incoming-recv)$/) {
260 $record->{type} = $this->{state};
261 $record->{start} = $logfh->tell;
262 $record->{stop} = $logfh->tell;
263 $record->{inner_file} = $this->{inner_file};
264 } elsif ($this->{state} eq 'kill-end') {
265 if ($this->{inner_file}) {
266 $record->{fh} = IO::InnerFile->new($logfh,$record->{start},$record->{stop} - $record->{start})
273 $record->{stop} = $logfh->tell;
275 if ($this->{state} eq 'incoming-recv') {
277 unless (/^Received: \(at \S+\) by \S+;/) {
278 die "bad line '$pl' in state incoming-recv";
280 $this->{state} = 'go';
281 $record->{text} .= "$_\n" unless $this->{inner_file};
282 } elsif ($this->{state} eq 'html') {
283 $record->{text} .= "$_\n" unless $this->{inner_file};
284 } elsif ($this->{state} eq 'go') {
286 $record->{text} .= "$_\n" unless $this->{inner_file};
287 } elsif ($this->{state} eq 'go-nox') {
288 $record->{text} .= "$_\n" unless $this->{inner_file};
289 } elsif ($this->{state} eq 'recips') {
291 undef $record->{recips};
293 # preserve trailing null fields, e.g. #2298
294 $record->{recips} = [split /\04/, $_, -1];
296 $this->{state} = 'kill-body';
297 $record->{start} = $logfh->tell+2;
298 $record->{stop} = $logfh->tell+2;
299 $record->{inner_file} = $this->{inner_file};
300 } elsif ($this->{state} eq 'autocheck') {
301 $record->{text} .= "$_\n" unless $this->{inner_file};
302 next if !/^X-Debian-Bugs(-\w+)?: This is an autoforward from (\S+)/;
303 $this->{state} = 'autowait';
304 } elsif ($this->{state} eq 'autowait') {
305 $record->{text} .= "$_\n" unless $this->{inner_file};
307 $this->{state} = 'go-nox';
309 die "state $this->{state} at line $this->{linenum} ('$_')";
312 die "state $this->{state} at end" unless $this->{state} eq 'kill-end';
321 =item read_log_records
323 Takes a .log filehandle as input, and returns an array of all records in
324 that file. Throws exceptions using die(), so you may want to wrap this in an
327 Uses exactly the same options as Debbugs::Log::new
335 ($param{logfh}) = @_;
338 %param = validate_with(params => \@_,
339 spec => {bug_num => {type => SCALAR,
342 logfh => {type => HANDLE,
345 log_name => {type => SCALAR,
348 inner_file => {type => BOOLEAN,
354 if (grep({exists $param{$_} and defined $param{$_}} qw(bug_num logfh log_name)) ne 1) {
355 croak "Exactly one of bug_num, logfh, or log_name must be passed and must be defined";
359 my $reader = Debbugs::Log->new(%param);
360 while (defined(my $record = $reader->read_record())) {
361 push @records, $record;
366 =item write_log_records
368 Takes a filehandle and a list of records as input, and prints the .log
369 format representation of those records to that filehandle.
375 sub write_log_records
377 my %param = validate_with(params => \@_,
378 spec => {bug_num => {type => SCALAR,
381 logfh => {type => HANDLE,
384 log_name => {type => SCALAR,
387 records => {type => HASHREF|ARRAYREF,
391 if (grep({exists $param{$_} and defined $param{$_}} qw(bug_num logfh log_name)) ne 1) {
392 croak "Exactly one of bug_num, logfh, or log_name must be passed and must be defined";
395 if (exists $param{logfh}) {
396 $logfh = $param{logfh}
398 elsif (exists $param{log_name}) {
399 $logfh = IO::File->new(">>$param{log_name}") or
400 die "Unable to open bug log $param{log_name} for writing: $!";
402 elsif (exists $param{bug_num}) {
403 my $location = getbuglocation($param{bug_num},'log');
404 my $bug_log = getbugcomponent($param{bug_num},'log',$location);
405 $logfh = IO::File->new($bug_log, 'r') or
406 die "Unable to open bug log $bug_log for reading: $!";
408 my @records = make_list($param{records});
410 for my $record (@records) {
411 my $type = $record->{type};
412 croak "record type '$type' with no text field" unless defined $record->{text};
413 # I am not sure if we really want to croak here; but this is
414 # almost certainly a bug if is_utf8 is on.
415 my $text = $record->{text};
416 if (is_utf8($text)) {
417 carp('Record text was in the wrong encoding (perl internal instead of utf8 octets)');
418 $text = encode_utf8($text)
420 ($text) = escape_log($text);
421 if ($type eq 'autocheck') {
422 print {$logfh} "\01\n$text\03\n" or
423 die "Unable to write to logfile: $!";
424 } elsif ($type eq 'recips') {
425 print {$logfh} "\02\n";
426 my $recips = $record->{recips};
427 if (defined $recips) {
428 croak "recips not undef or array"
429 unless ref($recips) eq 'ARRAY';
430 my $wrong_encoding = 0;
432 map { if (is_utf8($_)) {
438 carp('Recipients was in the wrong encoding (perl internal instead of utf8 octets') if $wrong_encoding;
439 print {$logfh} join("\04", @$recips) . "\n" or
440 die "Unable to write to logfile: $!";
442 print {$logfh} "-t\n" or
443 die "Unable to write to logfile: $!";
445 #$text =~ s/^([\01-\07\030])/\030$1/gm;
446 print {$logfh} "\05\n$text\03\n" or
447 die "Unable to write to logfile: $!";
448 } elsif ($type eq 'html') {
449 print {$logfh} "\06\n$text\03\n" or
450 die "Unable to write to logfile: $!";
451 } elsif ($type eq 'incoming-recv') {
452 #$text =~ s/^([\01-\07\030])/\030$1/gm;
453 print {$logfh} "\07\n$text\03\n" or
454 die "Unable to write to logfile: $!";
456 croak "unknown record type type '$type'";
465 print {$log} escape_log(@log)
467 Applies the log escape regex to the passed logfile.
473 return map {s/^([\01-\07\030])/\030$1/gm; $_ } @log;
479 if ($record->{inner_file}) {
482 my $t = $record->{fh};
484 $record->{fh}->seek(0,0);
487 return $record->{text};
492 my ($record,$regex) = @_;
493 if ($record->{inner_file}) {
495 my $fh = $record->{fh};
497 if (@result = $_ =~ m/$regex/) {
498 $record->{fh}->seek(0,0);
502 $record->{fh}->seek(0,0);
505 my @result = $record->{text} =~ m/$regex/;
507 return $record->{text};
514 This module does none of the formatting that bugreport.cgi et al do. It's
515 simply a means for extracting and rewriting raw records.