Add aiso xml parsing script. Runs over zip files in current directory and adds them...

svn path=/; revision=646

diff --git a/interactome_scripts/aiso_xml_parser.pl b/interactome_scripts/aiso_xml_parser.pl
new file mode 100755 (executable)
index 0000000..858f770
--- /dev/null
+++ b/interactome_scripts/aiso_xml_parser.pl
@@ -0,0 +1,153 @@
+#!/usr/bin/perl
+
+use strict;
+use warnings;
+
+if(-e "$ENV{HOME}/scripts/jaiswallab/interactome_scripts/find_species.pl") {
+       require "$ENV{HOME}/scripts/jaiswallab/interactome_scripts/find_species.pl";
+}elsif(-e "$ENV{HOME}/jaiswallab_svn/interactome_scripts/find_species.pl") {
+       require "$ENV{HOME}/jaiswallab_svn/interactome_scripts/find_species.pl";
+}
+
+use lib "$ENV{HOME}/scripts/jaiswallab/interactome_scripts";
+
+use XML::LibXML;
+use Archive::Zip;
+use Bio::DB::Taxonomy;
+use DbiFloret;
+
+# check for arguments and explain usage
+# if ($#ARGV !=0) {
+       # print "usage: aiso_xml_parser.pl input_zip_file\n";
+       # exit;
+# }
+my @files = glob("*.zip");
+
+my $dbh = DbiFloret::dbconnect_local;
+
+for my $infile (@files){
+               print "working on file $infile\n";
+               
+               my $zip = Archive::Zip->new($infile);
+               
+               my $xml_content = $zip->contents('metadata.xml');
+               
+               my $dom = XML::LibXML->load_xml(string => $xml_content);
+               
+               my $orig_filename = $dom->findvalue('//orig_image_filename');
+               my $species = $dom->findvalue('//species/@name');
+               my $ubio_id = $dom->findvalue('//species/@ubio_id');
+               my $curator = $dom->findvalue('//curator_name');
+               my $comment = $dom->findvalue('//comment');
+               
+               # Get the ncbi taxon id
+               my $db = Bio::DB::Taxonomy->new(-source => 'entrez');
+               my $ncbi_id = $db->get_taxonid($species);
+               
+               #insert the curator info into the curator table
+               # check if curator already in db
+               my $sth_check_curator = $dbh->prepare("select curator_id from curator where firstname = ?");
+               $sth_check_curator->execute($curator);
+               
+               my $curator_id;
+               
+               $curator_id = $sth_check_curator->fetchrow_array and $sth_check_curator->finish;
+               
+               if(!defined($curator_id)) {
+                               my $sth_insert_curator = $dbh->prepare("insert into curator (firstname) values (?)");
+                               $sth_insert_curator->execute($curator);
+                               $curator_id = $sth_insert_curator->{mysql_insertid};
+               }
+               
+               
+               
+               # insert the taxon info into the taxon table
+               my $sth_check_taxon = $dbh->prepare("select taxon_id from taxon where species_id = ?");
+               $sth_check_taxon->execute($ncbi_id);
+               
+               my $taxon_id;
+               
+               $taxon_id = $sth_check_taxon->fetchrow_array and $sth_check_taxon->finish;
+               
+               if(!defined($taxon_id)) {
+                               my $sth_insert_taxon = $dbh->prepare("insert into taxon (species_id, species_name) values (?,?)");
+                               $sth_insert_taxon->execute($ncbi_id, $species);
+                               $taxon_id = $sth_insert_taxon->{mysql_insertid};
+               }
+               
+               
+               #insert the image and data into annotated_image table
+               my @files = $zip->memberNames();
+               my $image;
+               
+               for my $filename (@files) {
+                               if($filename =~ /image/) {
+                                               $image = $zip->contents($filename);
+                               }
+               }
+               
+               # open IMAGE, "$orig_filename" or die $!;
+               # 
+               # my ($image, $buff);
+               # while(read IMAGE, $buff, 16384) {
+                               # $image .= $buff;
+               # }
+               
+               my $sth_insert_image = $dbh->prepare("insert into annotated_image(curator_id, taxon_id, annotated_image) values (?,?,?)");
+               $sth_insert_image->execute($curator_id,$taxon_id,$image);
+               my $annotated_image_id = $sth_insert_image->{mysql_insertid};
+               
+               
+               #set the statement handler to inser the segment info
+               my $sth_insert_segment = $dbh->prepare("insert into segment(annotated_image_id,coordinates) values (?,GeomFromText(?))");
+               
+               #annotated_term table
+               my $sth_insert_annotated_term = $dbh->prepare("insert into annotated_term(ontology_term_id, ontology_term_name) values (?,?)");
+               my $sth_check_annotated_term = $dbh->prepare("select annotated_term_id from annotated_term where ontology_term_id = ?");
+               
+               #annotated_term_image table
+               my $sth_insert_annotated_term_image = $dbh->prepare("insert into annotated_term_image(annotated_term_id, annotated_image_id) values (?,?)");
+               
+               #annotated_term_segment table
+               my $sth_insert_annotated_term_segment = $dbh->prepare("insert into annotated_term_segment(segment_id, annotated_term_id) values (?,?)");
+               
+               foreach my $segment ($dom->findnodes('/image_data/segments/segment')) {
+                               my $layer = $segment->findvalue('./layer');
+                               my $term = $segment->findvalue('./annotation_term');
+                               my $term_id = $segment->findvalue('./annotation_id');
+                               my $coords = $segment->findvalue('./polygon_coords');
+                               
+                               #Need to get the coordinates in WKT (Well known text) format for input
+                               $coords =~ s/(\d+),(\d+)/$1 $2/; # remove the first comma
+                               $coords =~ s/(.*?,.*?),/$1 /mgs; # put the points in (x1 y1, x2 y2,...) format
+                               $coords =~ s/,$//; # remove the last comma
+                               $coords =~ s/,/, /g; # add a space after each comma
+                               $coords = "LineString($coords)";  # need the text to start with LineString and surround points with parantheses
+                                       
+                               $sth_insert_segment->execute($annotated_image_id,$coords);
+                               my $segment_id = $sth_insert_segment->{mysql_insertid};
+                               
+                               #only need to insert if term isn't already in db
+                               $sth_check_annotated_term->execute($term_id);
+                               my $annotated_term_id;
+                               $annotated_term_id = $sth_check_annotated_term->fetchrow_array and $sth_check_annotated_term->finish;
+                               if(!defined($annotated_term_id)){
+                                               $sth_insert_annotated_term->execute($term_id,$term);
+                                               $annotated_term_id = $sth_insert_annotated_term->{mysql_insertid};
+                               }
+                               
+                               $sth_insert_annotated_term_image->execute($annotated_term_id,$annotated_image_id);
+                               
+                               $sth_insert_annotated_term_segment->execute($segment_id,$annotated_term_id);
+                               
+               }
+
+}
+# The following will retrieve the image from the database
+# my $sth_get_image = $dbh->prepare("select annotated_image from annotated_image where annotated_image_id=?");
+# $sth_get_image->execute($annotated_image_id);
+# my $retrieved_image = $sth_get_image->fetch();
+# 
+# open IMAGE, ">retrieved_image.png" or die $!;
+# print IMAGE @$retrieved_image;
+# close IMAGE;