[getallhocrs] Skip files which have already been downloaded

2 files changed, 15 insertions, 1 deletions

diff --git a/cmd/bookpipeline/main.go b/cmd/bookpipeline/main.go
index 16ce596..0afda16 100644
--- a/cmd/bookpipeline/main.go
+++ b/cmd/bookpipeline/main.go
@@ -667,7 +667,14 @@ func stopTimer(t *time.Timer) {
 
 // TODO: rather than relying on journald, would be nicer to save the logs
 //       ourselves maybe, so that we weren't relying on a particular systemd
-//       setup.
+//       setup. this can be done by having the conn.Log also append line
+//       to a file (though that would mean everything would have to go through
+//       conn.Log, which we're not consistently doing yet). the correct thing
+//       to do then would be to implement a new interface that covers the part
+//       of log.Logger we use (e.g. Print and Printf), and then have an exported
+//       conn struct that implements those, so that we could pass a log.Logger
+//       or the new conn struct everywhere (we wouldn't be passing a log.Logger,
+//       it's just good to be able to keep the compatibility)
 func savelogs(conn Pipeliner, starttime int64, hostname string) error {
 	cmd := exec.Command("journalctl", "-u", "bookpipeline", "-n", "all")
 	var stdout, stderr bytes.Buffer
diff --git a/cmd/getallhocrs/main.go b/cmd/getallhocrs/main.go
index 136f07e..696a5fc 100644
--- a/cmd/getallhocrs/main.go
+++ b/cmd/getallhocrs/main.go
@@ -68,6 +68,13 @@ func main() {
 			if !strings.HasSuffix(o, ".hocr") {
 				continue
 			}
+			// skip already downloaded items
+			_, err = os.Stat(o)
+			if err == nil || os.IsExist(err) {
+				log.Println("  Skipping already complete download of", o)
+				continue
+			}
+			log.Println("  Downloading", o)
 			err = conn.Download(conn.WIPStorageId(), o, o)
 			if err != nil {
 				log.Fatalln("Failed to download file", o, err)