X-Git-Url: https://git.donarmstrong.com/?p=bin.git;a=blobdiff_plain;f=bibtex_to_paper;h=a6350eb98caedfecffcac5d9b248c5ac41760660;hp=7474ccbe6e252dc3e2a55caa7012395e456d1a48;hb=HEAD;hpb=b538bd95ec63124725f440cdbd49475e8e307b15

diff --git a/bibtex_to_paper b/bibtex_to_paper
index 7474ccb..a6350eb 100755
--- a/bibtex_to_paper
+++ b/bibtex_to_paper
@@ -13,7 +13,14 @@ use Getopt::Long;
 use Pod::Usage;
 
 use File::Find;
-use Text::Bibtex;
+use File::Basename qw(basename);
+use File::Spec qw(rel2abs);
+use Text::BibTeX;
+use User;
+use Data::Printer;
+use POSIX;
+
+use DBI;
 
 =head1 NAME
 
@@ -27,8 +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
@@ -50,6 +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)
@@ -76,13 +89,24 @@ 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'),
               );
 
 GetOptions(\%options,
-           'build_cache|build-cache|B!',
-           'bibtex|b=s',
+           'build_cache|build-cache!',
+           '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');
 
 pod2usage() if $options{help};
@@ -100,13 +124,398 @@ if (not exists $options{bibtex} and
 
 pod2usage(join("\n",@USAGE_ERRORS)) if @USAGE_ERRORS;
 
+main();
+
+sub main{
+
+    my $dbh;
+    my $sth;
+    if (exists $options{bibtex_cache}) {
+        my $initialize = 0;
+        if (-e $options{bibtex_cache}) {
+            ($dbh,$sth) = open_cache($options{bibtex_cache});
+        } else {
+            ($dbh,$sth) = initialize_database($options{bibtex_cache});
+        }
+    }
+
+    if (exists $options{clear_cache}) {
+        clear_cache($dbh,$sth);
+    }
+    my %entries;
+    if (exists $options{build_cache}) {
+        $options{bibtex} //= [];
+        $options{bibtex} =
+            [@ARGV,
+             @{ref $options{bibtex}?$options{bibtex}:[$options{bibtex}]},
+            ];
+        @ARGV = ();
+    }
+    if (exists $options{bibtex}) {
+        for my $bibtex_file (@{ref $options{bibtex}?$options{bibtex}:[$options{bibtex}]}) {
+            parse_bibtex_file($bibtex_file,\%entries);
+        }
+    }
+
+    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;
+    for my $bibtex_key (@ARGV) {
+        open_bibtex_key(\%options,$dbh,$sth,\%entries,$bibtex_key);
+    }
+
+}
+
+sub clear_cache {
+    my ($dbh,$sth) = @_;
+    $sth->{clear_papers_cache}->execute();
+    $sth->{clear_bibtex_cache}->execute();
+}
+
+sub load_papers_into_database {
+    my ($dbh,$sth,$dir) = @_;
+
+    my @dirs = ref($dir)?@{$dir}:$dir;
+
+    if ($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");
+        }
+    } 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($actually_load_it,@dirs);
+    }
+}
+
+sub insert_or_replace_papers {
+    my ($dbh,$sth,$file_name,$file_loc,$has_xoj) = @_;
+    $sth->{insert_papers}->execute($file_name,$file_loc,$has_xoj);
+    $sth->{insert_papers}->finish();
+}
+
+sub load_bibtex_entries_into_database {
+    my ($dbh,$sth,$entries) = @_;
+    for my $entry (keys %{$entries}) {
+        next unless defined $entries->{$entry};
+        $sth->{insert_bibtex}->execute($entry,@{$entries->{$entry}}{qw(file_name pmid doi html)});
+        $sth->{insert_bibtex}->finish();
+        print STDERR "inserted $entry {".join(',',map {defined $_?"'$_'":"'undef'"} %{$entries->{$entry}})."}\n" if $DEBUG;
+    }
+}
+
+sub open_bibtex_key {
+    my ($options,$dbh,$sth,$entries,$bibtex_key) = @_;
+    if (not defined $dbh) {
+        open_entry($dbh,$sth,$entries->{$bibtex_key},$options);
+    } else {
+        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);
+    }
+}
+
+sub fork_exec {
+    my (@cmd) = @_;
+    my $child = fork();
+    if (not defined $child) {
+        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 $child;
+    }
+
+}
+
+sub open_pdf {
+    my ($file_name,$options,$has_xoj) = @_;
+    print STDERR "opening $file_name\n" if $DEBUG;
+    my $pdf_viewer = 'xournal';
+    if (exists $options->{pdfviewer} and defined $options->{pdfviewer}) {
+        $pdf_viewer = $options->{pdfviewer};
+    }
+    fork_exec($pdf_viewer,$file_name);
+}
+
+sub open_browser{
+    my ($file) = @_;
+    fork_exec('sensible-browser',$file);
+}
+
+sub open_entry{
+    my ($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 $paper) {
+            my ($pmid) = $entry->{file_name} =~ /pmid_(\d+)/;
+            if (defined $pmid and length $pmid) {
+                $paper = select_one($dbh,$sth->{select_papers_by_pmid},'%pmid_'.$pmid.'.%');
+            }
+        }
+        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 $entry;
+}
+
+sub select_one{
+    my ($dbh,$sth,@bind_vals) = @_;
+    $sth->execute(@bind_vals) or
+        die "Unable to select one: ".$dbh->errstr();
+    my $results = $sth->fetchall_arrayref({});
+    $sth->finish();
+    return ref($results)?$results->[0]:undef;
+}
+
 sub parse_bibtex_file {
-    
+    my ($file,$entries) = @_;
+
+    my $bibfile = Text::BibTeX::File->new($file)
+        or die "Unable to open $file for reading: $!";
+    my @entry_comments;
+    my $entry;
+    while ($entry = Text::BibTeX::Entry->new($bibfile)) {
+        print STDERR "In Entry ".$entry->metatype() if $DEBUG;
+        if ($entry->metatype() == BTE_COMMENT) {
+            push @entry_comments,$entry->value();
+        } elsif ($entry->metatype() == BTE_REGULAR) {
+            my $entry_key = $entry->key();
+            if (not defined $entry_key) {
+                @entry_comments = ();
+                next;
+            }
+            my %entry_data;
+            # if there is a file comment, use it as the file name
+            for my $comment (@entry_comments) {
+                next unless $comment =~ /^\s*file(?:name)?:?\s*(.+?)\s*$/i;
+                next unless length $1;
+                $entry_data{file_name} = $1.'.pdf';
+                last;
+            }
+            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 pmid)) {
+                my $field_value = $entry->get($field);
+                if (defined $field_value and $field_value =~ /\S+/) {
+                    $entry_data{$field_name{$field}} =
+                        $field_prefix{$field}.$field_value if
+                        not defined $entry_data{$field_name{$field}};
+                }
+            }
+            $entries->{$entry_key} = {} if not defined $entries->{$entry_key};
+            for my $field (keys %entry_data) {
+                $entries->{$entry_key}{$field} = $entry_data{$field} if
+                    defined $entry_data{$field};
+            }
+            # reset the entry comments
+            @entry_comments = ();
+        } else {
+            # do nothing
+        }
+        print STDERR "\n" if $DEBUG;
+    }
+    return $entries;
 }
 
 
 sub initialize_database {
-    my 
+    my ($cache) = @_;
+    return open_cache($cache,1);
 }
 
+sub open_cache {
+    my ($cache,$initialize) = @_;
+    my $dbh = DBI->connect("dbi:SQLite:dbname=$cache","","") or
+        die "Unable to open/create database $cache";
+    if ($initialize) {
+        $dbh->do("DROP TABLE IF EXISTS bibtex;");
+        $dbh->do("DROP TABLE IF EXISTS papers;");
+        $dbh->do(<<EOF);
+CREATE TABLE bibtex (
+bibtex_key TEXT PRIMARY KEY,
+file_name TEXT,
+pmid TEXT,
+doi TEXT,
+html TEXT
+);
+EOF
+        $dbh->do(<<EOF);
+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 (
+file_name TEXT PRIMARY KEY,
+path TEXT,
+has_xoj BOOLEAN
+);
+EOF
+        $dbh->do(<<EOF);
+CREATE UNIQUE INDEX papers_path ON papers(path);
+EOF
+        $dbh->do(<<EOF);
+CREATE UNIQUE INDEX papers_file_name ON papers(file_name);
+EOF
+    }
+    my %s =
+        (insert_papers => <<'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,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',
+SELECT * FROM papers WHERE path = ?;
+EOF
+         select_bibtex_by_key => <<'EOF',
+SELECT * FROM bibtex WHERE bibtex_key = ?;
+EOF
+         select_bibtex_by_approximate_key => <<'EOF',
+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;
+EOF
+         clear_bibtex_cache => <<'EOF',
+DELETE FROM bibtex;
+EOF
+        );
+    my $st;
+    for my $key (keys %s) {
+        $st->{$key}=$dbh->prepare($s{$key}) //
+            die "Unable to prepare sql statement: ".$dbh->errstr;
+    }
+    return ($dbh,$st);
+}
+
+
 __END__