X-Git-Url: https://git.donarmstrong.com/?p=bin.git;a=blobdiff_plain;f=bibtex_to_paper;h=a6350eb98caedfecffcac5d9b248c5ac41760660;hp=121a85544354bf60f8f2c2336371f1d8f1802713;hb=HEAD;hpb=7f5ebb78dfe45ece2ce9f461aebbdf067381c426

diff --git a/bibtex_to_paper b/bibtex_to_paper
index 121a855..a6350eb 100755
--- a/bibtex_to_paper
+++ b/bibtex_to_paper
@@ -18,6 +18,7 @@ use File::Spec qw(rel2abs);
 use Text::BibTeX;
 use User;
 use Data::Printer;
+use POSIX;
 
 use DBI;
 
@@ -33,8 +34,10 @@ bibtex_to_paper [options] bibtexkey
   --bibtex, -b bibtex file to look up key in
   --bibtex-cache, -c bibtex cache file
   --build-cache, -B build cache using bibtex files
+  --search-by-pmid Search term is a pmid instead of a bibtex key
   --pdf-dir pdf directory
   --pdfviewer, -p pdf viewer to use
+  --only-print Only print PDF file name
   --debug, -d debugging level (Default 0)
   --help, -h display this help
   --man, -m display manual
@@ -56,6 +59,10 @@ Bibtex cache file; rebuilt if bibtex file changes
 PDF viewer to use; defaults to evince unless a .xoj exists, in which
 case xournal is used.
 
+=item B<--only-print>
+
+Only print the PDF file name, don't open it.
+
 =item B<--debug, -d>
 
 Debug verbosity. (Default 0)
@@ -82,6 +89,10 @@ use vars qw($DEBUG);
 my %options = (debug           => 0,
                help            => 0,
                man             => 0,
+               only_print      => 0,
+               search_by_pmid  => 0,
+               search_by_file  => 0,
+               use_git         => 1,
                'bibtex_cache'  => File::Spec->catfile(User->Home,'.bibtex_to_paper_cache'),
               );
 
@@ -90,6 +101,10 @@ GetOptions(\%options,
            'bibtex|b=s@',
            'bibtex_cache|bibtex-cache|c=s',
            'pdfviewer|p=s',
+           'use_git|use-git!',
+           'only_print|only-print!',
+           'search_by_pmid|search-by-pmid!',
+           'search_by_file|search-by-file!',
            'clear_cache|clear-cache!',
            'papers_directory|papers-directory=s@',
            'debug|d+','help|h|?','man|m');
@@ -145,13 +160,17 @@ sub main{
     if (exists $options{papers_directory} and
         defined $dbh
        ) {
+        $dbh->begin_work;
         load_papers_into_database($dbh,$sth,$options{papers_directory});
+        $dbh->commit;
     }
 
     p %entries if $DEBUG;
     if (keys %entries and
         defined $dbh) {
+        $dbh->begin_work;
         load_bibtex_entries_into_database($dbh,$sth,\%entries);
+        $dbh->commit;
     }
 
     p @ARGV if $DEBUG;
@@ -172,21 +191,27 @@ sub load_papers_into_database {
 
     my @dirs = ref($dir)?@{$dir}:$dir;
 
-    my $actually_load_it = sub {
-        if (/\.git/) {
-            $File::Find::prune = 1;
-            return;
-        }
-        return unless /\.pdf$/;
-        my $xoj = 0;
-        if (-e "${_}.xoj") {
-            $xoj = 1;
+    if ($options{use_git}) {
+        my @files = grep /\.pdf\"?$/, split /\n/, qx(git ls-tree HEAD -r --full-name --name-only);
+        for my $file  (@files) {
+            $file =~ s/^\"(.+)\"$/"qq($1)"/gee;
+            insert_or_replace_papers($dbh,$sth,basename($file),File::Spec->rel2abs($file), -e "${file}.xoj");
         }
-        insert_or_replace_papers($dbh,$sth,basename($File::Find::name),File::Spec->rel2abs($_),$xoj);
-    };
-
-    my @pdfs;
-    find($actually_load_it,@dirs);
+    } else {
+        my $actually_load_it = sub {
+            if (/\.git/) {
+                $File::Find::prune = 1;
+                return;
+            }
+            return unless /\.pdf$/;
+            my $xoj = 0;
+            if (-e "${_}.xoj") {
+                $xoj = 1;
+            }
+            insert_or_replace_papers($dbh,$sth,basename($File::Find::name),File::Spec->rel2abs($_),$xoj);
+        };
+        find($actually_load_it,@dirs);
+    }
 }
 
 sub insert_or_replace_papers {
@@ -199,7 +224,7 @@ sub load_bibtex_entries_into_database {
     my ($dbh,$sth,$entries) = @_;
     for my $entry (keys %{$entries}) {
         next unless defined $entries->{$entry};
-        $sth->{insert_bibtex}->execute($entry,@{$entries->{$entry}}{qw(file_name doi html)});
+        $sth->{insert_bibtex}->execute($entry,@{$entries->{$entry}}{qw(file_name pmid doi html)});
         $sth->{insert_bibtex}->finish();
         print STDERR "inserted $entry {".join(',',map {defined $_?"'$_'":"'undef'"} %{$entries->{$entry}})."}\n" if $DEBUG;
     }
@@ -210,7 +235,14 @@ sub open_bibtex_key {
     if (not defined $dbh) {
         open_entry($dbh,$sth,$entries->{$bibtex_key},$options);
     } else {
-        my $entry = select_entry_from_bibtex_key($dbh,$sth,$bibtex_key);
+        my $entry;
+        if ($options->{search_by_pmid}) {
+            $entry = select_entry_from_pmid($dbh,$sth,$bibtex_key);
+        } elsif ($options->{search_by_file}) {
+            $entry = select_entry_from_file($dbh,$sth,$bibtex_key);
+        } else {
+            $entry = select_entry_from_bibtex_key($dbh,$sth,$bibtex_key);
+        }
         p $entry if $DEBUG;
         open_entry($dbh,$sth,$entry,$options);
     }
@@ -223,6 +255,11 @@ sub fork_exec {
         die "Unable to fork for some reason: $!";
     }
     if ($child == 0) {
+        foreach (0 .. (POSIX::sysconf (&POSIX::_SC_OPEN_MAX) || 1024))
+           { POSIX::close $_ }
+        open (STDIN, "</dev/null");
+        open (STDOUT, ">/dev/null");
+        open (STDERR, ">&STDOUT");
         exec(@cmd);
     } else {
         return $child;
@@ -233,11 +270,11 @@ sub fork_exec {
 sub open_pdf {
     my ($file_name,$options,$has_xoj) = @_;
     print STDERR "opening $file_name\n" if $DEBUG;
-    if ($has_xoj) {
-        fork_exec('xournal',$file_name);
-    } else {
-        fork_exec('evince',$file_name)
+    my $pdf_viewer = 'xournal';
+    if (exists $options->{pdfviewer} and defined $options->{pdfviewer}) {
+        $pdf_viewer = $options->{pdfviewer};
     }
+    fork_exec($pdf_viewer,$file_name);
 }
 
 sub open_browser{
@@ -249,6 +286,10 @@ sub open_entry{
     my ($dbh,$sth,$entry,$options) = @_;
 
     return unless defined $entry and ref $entry and keys %{$entry};
+    if ($DEBUG) {
+        print STDERR "Entry: \n";
+        p $entry;
+    }
     if (defined $entry->{file_name} and length $entry->{file_name}) {
         my $paper = select_one($dbh,$sth->{select_papers_by_name},$entry->{file_name});
         if (not defined $paper) {
@@ -260,30 +301,59 @@ sub open_entry{
         p $paper if $DEBUG;
         print STDERR $entry->{file_name} if $DEBUG;
         if (defined $paper) {
+            if ($options->{only_print}) {
+                print $paper->{path};
+                return;
+            }
             open_pdf($paper->{path},$options,$paper->{has_xoj});
             return;
+        } else {
+            print STDERR "Unable to find paper\n" if $DEBUG;
         }
     }
     if (defined $entry->{doi}) {
+        if ($options->{only_print}) {
+            print $entry->{doi};
+            return;
+        }
         my $url = $entry->{doi};
         $url =~ s{^doi://}{http://dx.doi.org/};
         open_browser($url,$options);
         return;
     }
     if (defined $entry->{html}) {
+        if ($options->{only_print}) {
+            print $entry->{html};
+            return;
+        }
         open_browser($entry->{html},$options);
         return;
     }
 }
 
+sub select_entry_from_pmid{
+    my ($dbh,$sth,$pmid) = @_;
+
+    return select_one($dbh,$sth->{select_bibtex_by_pmid},$pmid);
+}
+
+sub select_entry_from_file{
+    my ($dbh,$sth,$filename) = @_;
+
+    return select_one($dbh,$sth->{select_bibtex_by_file_name_like},'%'.$filename.'%');
+}
+
+
 sub select_entry_from_bibtex_key{
     my ($dbh,$sth,$bibtex_key) = @_;
 
     my $entry = select_one($dbh,$sth->{select_bibtex_by_key},$bibtex_key);
     if (not defined $entry) {
+        print STDERR "Unable to find entry by exact search\n" if $DEBUG;
         $bibtex_key =~ s/:.*$//;
         $entry = select_one($dbh,$sth->{select_bibtex_by_approximate_key},$bibtex_key.'%');
     }
+    print STDERR "Found entry\n" if $DEBUG and defined $entry;
     return $entry;
 }
 
@@ -324,11 +394,13 @@ sub parse_bibtex_file {
             my %field_prefix = (doi => 'doi://',
                                 html => 'http://',
                                 file => '',
+                                pmid => '',
                                );
             my %field_name = (doi => 'doi',
                               html => 'html',
+                              pmid => 'pmid',
                               file => 'file_name',);
-            for my $field (qw(file doi html)) {
+            for my $field (qw(file doi html pmid)) {
                 my $field_value = $entry->get($field);
                 if (defined $field_value and $field_value =~ /\S+/) {
                     $entry_data{$field_name{$field}} =
@@ -368,6 +440,7 @@ sub open_cache {
 CREATE TABLE bibtex (
 bibtex_key TEXT PRIMARY KEY,
 file_name TEXT,
+pmid TEXT,
 doi TEXT,
 html TEXT
 );
@@ -377,6 +450,9 @@ CREATE UNIQUE INDEX bibtex_file_name ON bibtex(file_name);
 EOF
         $dbh->do(<<EOF);
 CREATE UNIQUE INDEX bibtex_bibtex_key ON bibtex(bibtex_key);
+EOF
+        $dbh->do(<<EOF);
+CREATE INDEX bibtex_pmid ON bibtex(pmid);
 EOF
         $dbh->do(<<EOF);
 CREATE TABLE papers (
@@ -397,12 +473,15 @@ EOF
 INSERT OR REPLACE INTO papers(file_name,path,has_xoj) VALUES (?,?,?);
 EOF
          insert_bibtex => <<'EOF',
-INSERT OR REPLACE INTO bibtex (bibtex_key,file_name,doi,html) VALUES (?,?,?,?);
+INSERT OR REPLACE INTO bibtex (bibtex_key,file_name,pmid,doi,html) VALUES (?,?,?,?,?);
 EOF
          select_papers_by_name => <<'EOF',
 SELECT * FROM papers WHERE file_name = ?;
 EOF
          select_papers_by_pmid => <<'EOF',
+SELECT * FROM papers JOIN bibtex ON papers.file_name = bibtex.file_name WHERE bibtex.pmid = ?;
+EOF
+         select_papers_by_name_like => <<'EOF',
 SELECT * FROM papers WHERE file_name LIKE ?;
 EOF
          select_papers_by_path => <<'EOF',
@@ -416,6 +495,12 @@ SELECT * FROM bibtex WHERE bibtex_key LIKE ?;
 EOF
          select_bibtex_by_file_name => <<'EOF',
 SELECT * FROM bibtex WHERE file_name = ?;
+EOF
+         select_bibtex_by_file_name_like => <<'EOF',
+SELECT * FROM bibtex WHERE file_name LIKE ?;
+EOF
+         select_bibtex_by_pmid => <<'EOF',
+SELECT * FROM bibtex WHERE pmid = ?;
 EOF
          clear_papers_cache => <<'EOF',
 DELETE FROM papers;