2 # bibtex_to_paper opens the paper corresponding to a bibtex key
3 # and is released under the terms of the GNU GPL version 3, or any
4 # later version, at your option. See the file README and COPYING for
6 # Copyright 2014 by Don Armstrong <don@donarmstrong.com>.
16 use File::Basename qw(basename);
17 use File::Spec qw(rel2abs);
27 bibtex_to_paper - opens the paper corresponding to a bibtex key
31 bibtex_to_paper [options] bibtexkey
34 --bibtex, -b bibtex file to look up key in
35 --bibtex-cache, -c bibtex cache file
36 --build-cache, -B build cache using bibtex files
37 --search-by-pmid Search term is a pmid instead of a bibtex key
38 --pdf-dir pdf directory
39 --pdfviewer, -p pdf viewer to use
40 --only-print Only print PDF file name
41 --debug, -d debugging level (Default 0)
42 --help, -h display this help
43 --man, -m display manual
51 Bibtex file to look key up in
53 =item B<--bibtex-cache, -c>
55 Bibtex cache file; rebuilt if bibtex file changes
57 =item B<--pdfviewer, -p>
59 PDF viewer to use; defaults to evince unless a .xoj exists, in which
64 Only print the PDF file name, don't open it.
68 Debug verbosity. (Default 0)
72 Display brief usage information.
89 my %options = (debug => 0,
95 'bibtex_cache' => File::Spec->catfile(User->Home,'.bibtex_to_paper_cache'),
99 'build_cache|build-cache!',
101 'bibtex_cache|bibtex-cache|c=s',
104 'only_print|only-print!',
105 'search_by_pmid|search-by-pmid!',
106 'clear_cache|clear-cache!',
107 'papers_directory|papers-directory=s@',
108 'debug|d+','help|h|?','man|m');
110 pod2usage() if $options{help};
111 pod2usage({verbose=>2}) if $options{man};
113 $DEBUG = $options{debug};
116 if (not exists $options{bibtex} and
117 not exists $options{bibtex_cache}) {
119 "You must give at least one of --bibtex".
123 pod2usage(join("\n",@USAGE_ERRORS)) if @USAGE_ERRORS;
131 if (exists $options{bibtex_cache}) {
133 if (-e $options{bibtex_cache}) {
134 ($dbh,$sth) = open_cache($options{bibtex_cache});
136 ($dbh,$sth) = initialize_database($options{bibtex_cache});
140 if (exists $options{clear_cache}) {
141 clear_cache($dbh,$sth);
144 if (exists $options{build_cache}) {
145 $options{bibtex} //= [];
148 @{ref $options{bibtex}?$options{bibtex}:[$options{bibtex}]},
152 if (exists $options{bibtex}) {
153 for my $bibtex_file (@{ref $options{bibtex}?$options{bibtex}:[$options{bibtex}]}) {
154 parse_bibtex_file($bibtex_file,\%entries);
158 if (exists $options{papers_directory} and
162 load_papers_into_database($dbh,$sth,$options{papers_directory});
166 p %entries if $DEBUG;
167 if (keys %entries and
170 load_bibtex_entries_into_database($dbh,$sth,\%entries);
175 for my $bibtex_key (@ARGV) {
176 open_bibtex_key(\%options,$dbh,$sth,\%entries,$bibtex_key);
183 $sth->{clear_papers_cache}->execute();
184 $sth->{clear_bibtex_cache}->execute();
187 sub load_papers_into_database {
188 my ($dbh,$sth,$dir) = @_;
190 my @dirs = ref($dir)?@{$dir}:$dir;
192 if ($options{use_git}) {
193 my @files = grep /\.pdf\"?$/, split /\n/, qx(git ls-tree HEAD -r --full-name --name-only);
194 for my $file (@files) {
195 $file =~ s/^\"(.+)\"$/"qq($1)"/gee;
196 insert_or_replace_papers($dbh,$sth,basename($file),File::Spec->rel2abs($file), -e "${file}.xoj");
199 my $actually_load_it = sub {
201 $File::Find::prune = 1;
204 return unless /\.pdf$/;
209 insert_or_replace_papers($dbh,$sth,basename($File::Find::name),File::Spec->rel2abs($_),$xoj);
211 find($actually_load_it,@dirs);
215 sub insert_or_replace_papers {
216 my ($dbh,$sth,$file_name,$file_loc,$has_xoj) = @_;
217 $sth->{insert_papers}->execute($file_name,$file_loc,$has_xoj);
218 $sth->{insert_papers}->finish();
221 sub load_bibtex_entries_into_database {
222 my ($dbh,$sth,$entries) = @_;
223 for my $entry (keys %{$entries}) {
224 next unless defined $entries->{$entry};
225 $sth->{insert_bibtex}->execute($entry,@{$entries->{$entry}}{qw(file_name pmid doi html)});
226 $sth->{insert_bibtex}->finish();
227 print STDERR "inserted $entry {".join(',',map {defined $_?"'$_'":"'undef'"} %{$entries->{$entry}})."}\n" if $DEBUG;
231 sub open_bibtex_key {
232 my ($options,$dbh,$sth,$entries,$bibtex_key) = @_;
233 if (not defined $dbh) {
234 open_entry($dbh,$sth,$entries->{$bibtex_key},$options);
237 if ($options->{search_by_pmid}) {
238 $entry = select_entry_from_pmid($dbh,$sth,$bibtex_key);
240 $entry = select_entry_from_bibtex_key($dbh,$sth,$bibtex_key);
243 open_entry($dbh,$sth,$entry,$options);
250 if (not defined $child) {
251 die "Unable to fork for some reason: $!";
254 foreach (0 .. (POSIX::sysconf (&POSIX::_SC_OPEN_MAX) || 1024))
256 open (STDIN, "</dev/null");
257 open (STDOUT, ">/dev/null");
258 open (STDERR, ">&STDOUT");
267 my ($file_name,$options,$has_xoj) = @_;
268 print STDERR "opening $file_name\n" if $DEBUG;
269 my $pdf_viewer = 'xournal';
270 if (exists $options->{pdfviewer} and defined $options->{pdfviewer}) {
271 $pdf_viewer = $options->{pdfviewer};
273 fork_exec($pdf_viewer,$file_name);
278 fork_exec('sensible-browser',$file);
282 my ($dbh,$sth,$entry,$options) = @_;
284 return unless defined $entry and ref $entry and keys %{$entry};
286 print STDERR "Entry: \n";
289 if (defined $entry->{file_name} and length $entry->{file_name}) {
290 my $paper = select_one($dbh,$sth->{select_papers_by_name},$entry->{file_name});
291 if (not defined $paper) {
292 my ($pmid) = $entry->{file_name} =~ /pmid_(\d+)/;
293 if (defined $pmid and length $pmid) {
294 $paper = select_one($dbh,$sth->{select_papers_by_pmid},'%pmid_'.$pmid.'.%');
298 print STDERR $entry->{file_name} if $DEBUG;
299 if (defined $paper) {
300 if ($options->{only_print}) {
301 print $paper->{path};
304 open_pdf($paper->{path},$options,$paper->{has_xoj});
307 print STDERR "Unable to find paper\n" if $DEBUG;
310 if (defined $entry->{doi}) {
311 if ($options->{only_print}) {
315 my $url = $entry->{doi};
316 $url =~ s{^doi://}{http://dx.doi.org/};
317 open_browser($url,$options);
320 if (defined $entry->{html}) {
321 if ($options->{only_print}) {
322 print $entry->{html};
325 open_browser($entry->{html},$options);
330 sub select_entry_from_pmid{
331 my ($dbh,$sth,$pmid) = @_;
333 return select_one($dbh,$sth->{select_bibtex_by_pmid},$pmid);
337 sub select_entry_from_bibtex_key{
338 my ($dbh,$sth,$bibtex_key) = @_;
340 my $entry = select_one($dbh,$sth->{select_bibtex_by_key},$bibtex_key);
341 if (not defined $entry) {
342 print STDERR "Unable to find entry by exact search\n" if $DEBUG;
343 $bibtex_key =~ s/:.*$//;
344 $entry = select_one($dbh,$sth->{select_bibtex_by_approximate_key},$bibtex_key.'%');
346 print STDERR "Found entry\n" if $DEBUG and defined $entry;
351 my ($dbh,$sth,@bind_vals) = @_;
352 $sth->execute(@bind_vals) or
353 die "Unable to select one: ".$dbh->errstr();
354 my $results = $sth->fetchall_arrayref({});
356 return ref($results)?$results->[0]:undef;
359 sub parse_bibtex_file {
360 my ($file,$entries) = @_;
362 my $bibfile = Text::BibTeX::File->new($file)
363 or die "Unable to open $file for reading: $!";
366 while ($entry = Text::BibTeX::Entry->new($bibfile)) {
367 print STDERR "In Entry ".$entry->metatype() if $DEBUG;
368 if ($entry->metatype() == BTE_COMMENT) {
369 push @entry_comments,$entry->value();
370 } elsif ($entry->metatype() == BTE_REGULAR) {
371 my $entry_key = $entry->key();
372 if (not defined $entry_key) {
373 @entry_comments = ();
377 # if there is a file comment, use it as the file name
378 for my $comment (@entry_comments) {
379 next unless $comment =~ /^\s*file(?:name)?:?\s*(.+?)\s*$/i;
380 next unless length $1;
381 $entry_data{file_name} = $1.'.pdf';
384 my %field_prefix = (doi => 'doi://',
389 my %field_name = (doi => 'doi',
392 file => 'file_name',);
393 for my $field (qw(file doi html pmid)) {
394 my $field_value = $entry->get($field);
395 if (defined $field_value and $field_value =~ /\S+/) {
396 $entry_data{$field_name{$field}} =
397 $field_prefix{$field}.$field_value if
398 not defined $entry_data{$field_name{$field}};
401 $entries->{$entry_key} = {} if not defined $entries->{$entry_key};
402 for my $field (keys %entry_data) {
403 $entries->{$entry_key}{$field} = $entry_data{$field} if
404 defined $entry_data{$field};
406 # reset the entry comments
407 @entry_comments = ();
411 print STDERR "\n" if $DEBUG;
417 sub initialize_database {
419 return open_cache($cache,1);
423 my ($cache,$initialize) = @_;
424 my $dbh = DBI->connect("dbi:SQLite:dbname=$cache","","") or
425 die "Unable to open/create database $cache";
427 $dbh->do("DROP TABLE IF EXISTS bibtex;");
428 $dbh->do("DROP TABLE IF EXISTS papers;");
430 CREATE TABLE bibtex (
431 bibtex_key TEXT PRIMARY KEY,
439 CREATE UNIQUE INDEX bibtex_file_name ON bibtex(file_name);
442 CREATE UNIQUE INDEX bibtex_bibtex_key ON bibtex(bibtex_key);
445 CREATE INDEX bibtex_pmid ON bibtex(pmid);
448 CREATE TABLE papers (
449 file_name TEXT PRIMARY KEY,
455 CREATE UNIQUE INDEX papers_path ON papers(path);
458 CREATE UNIQUE INDEX papers_file_name ON papers(file_name);
462 (insert_papers => <<'EOF',
463 INSERT OR REPLACE INTO papers(file_name,path,has_xoj) VALUES (?,?,?);
465 insert_bibtex => <<'EOF',
466 INSERT OR REPLACE INTO bibtex (bibtex_key,file_name,pmid,doi,html) VALUES (?,?,?,?,?);
468 select_papers_by_name => <<'EOF',
469 SELECT * FROM papers WHERE file_name = ?;
471 select_papers_by_pmid => <<'EOF',
472 SELECT * FROM papers WHERE file_name LIKE ?;
474 select_papers_by_path => <<'EOF',
475 SELECT * FROM papers WHERE path = ?;
477 select_bibtex_by_key => <<'EOF',
478 SELECT * FROM bibtex WHERE bibtex_key = ?;
480 select_bibtex_by_approximate_key => <<'EOF',
481 SELECT * FROM bibtex WHERE bibtex_key LIKE ?;
483 select_bibtex_by_file_name => <<'EOF',
484 SELECT * FROM bibtex WHERE file_name = ?;
486 select_bibtex_by_pmid => <<'EOF',
487 SELECT * FROM bibtex WHERE pmid = ?;
489 clear_papers_cache => <<'EOF',
492 clear_bibtex_cache => <<'EOF',
497 for my $key (keys %s) {
498 $st->{$key}=$dbh->prepare($s{$key}) //
499 die "Unable to prepare sql statement: ".$dbh->errstr;