Don't abort analysis if we encounter a hocr with no words, just skip it

2 files changed, 4 insertions, 1 deletions

diff --git a/bookpipeline/cmd/bookpipeline/main.go b/bookpipeline/cmd/bookpipeline/main.go
index 520311b..97f4109 100644
--- a/bookpipeline/cmd/bookpipeline/main.go
+++ b/bookpipeline/cmd/bookpipeline/main.go
@@ -156,6 +156,9 @@ func analyse(toanalyse chan string, up chan string, errc chan error, logger *log
 			errc <- errors.New(fmt.Sprintf("Error retreiving confidence for %s: %s", path, err))
 			return
 		}
+		if avg == 0 {
+			continue
+		}
 		base := filepath.Base(path)
 		codestart := strings.Index(base, "_bin")
 		name := base[0:codestart]
diff --git a/lib/hocr/hocr.go b/lib/hocr/hocr.go
index dcd0494..052aa19 100644
--- a/lib/hocr/hocr.go
+++ b/lib/hocr/hocr.go
@@ -123,7 +123,7 @@ func GetAvgConf(hocrfn string) (float64, error) {
 		}
 	}
 	if num == 0 {
-		return 0, errors.New("No words found")
+		return 0, nil
 	}
 	return total / num, nil
 }