finished image ocr

bin/autodoc_process.pl

@@ -81,12 +81,12 @@ foreach my $file ( readdir(Q) ) {
 
 				print "\t\tupdating page status to 'ok'\n";
 				update_page_status($pageid, 'ok');
-
+				print "\tdone\n";
 			}
 		}
 		elsif ( $ext =~ /^(jpeg|png)$/ ) {
 			print "\tdetecting image rotation\n";
-			my @res;
+			my %res;
 			for(my $rot=0; $rot<360; $rot+=90) {
 				print "\ttrying $rot degrees rotation\n";
 				my $tempfile = "/tmp/autodoc.$$.jpeg";
@@ -100,7 +100,7 @@ foreach my $file ( readdir(Q) ) {
 
 				print "\t\tfound $dictmatches words in dictionary\n";
 
-				push @res, {
+				$res{$rot} = {
 					lang => $lang,
 					words => $words,
 					dictmatches => $dictmatches
@@ -109,6 +109,37 @@ foreach my $file ( readdir(Q) ) {
 				unlink($tempfile);
 			}
 
+			my $maxwords = 0;
+			my $bestrot;
+			foreach my $rot ( keys %res ) {
+				$bestrot=$rot if !defined $bestrot;
+				if ( $maxwords < $res{$rot}{dictmatches} ) {
+					$maxwords = $res{$rot}{dictmatches};
+					$bestrot = $rot;
+				}
+			}
+
+			print "\tbest OCR results with $bestrot rotation\n";
+
+			my $pageid = get_new_page($docid);
+			print "\t\tcreated page id $pageid\n";
+
+			print "\t\tupdating page status to 'inprogress'\n";
+			update_page_status($pageid, 'inprogress');
+
+			if ( !exists $primary{$docid} ) {
+				print "\t\tsetting document for default primary thumbnail\n";
+				$primary{$docid}=undef;
+				sqlquery($dbh, "CALL set_primary_page(?)",$pageid);
+			}
+			print "\t\tcreating original page jpeg $pageid.jpeg";
+			system(sprintf("convert %s/%s %s/%s.jpeg", $queuedir, $file, $originaldir, $pageid));
+
+			print "\t\tloading extracted words into database\n";
+			create_page_words($pageid, $res{$bestrot}{lang}, $res{$bestrot}{words});
+			print "\t\tupdating page status to 'ok'\n";
+			update_page_status($pageid, 'ok');
+			print "\tdone\n";
 		}
 		else {
 			print "\terror: don't know how to process files of $ext type";
@@ -187,7 +218,7 @@ sub detect_lang {
 		$dictwords++ if $found;
 	}
 
-	print Dumper(\%lcnt);
+	#print Dumper(\%lcnt);
 
 	my $max = 0;
 	my $lmax;