From 632a149df0196f0f057fa4d552aa28d22901bcda Mon Sep 17 00:00:00 2001
From: Nick White <git@njw.name>
Date: Wed, 22 Jan 2020 16:17:05 +0000
Subject: Add GetWordConfs function to hocr pkg
---
pkg/hocr/hocr.go | 29 +++++++++++++++++++++++++++++
1 file changed, 29 insertions(+)
diff --git a/pkg/hocr/hocr.go b/pkg/hocr/hocr.go
index c3c88b3..fa8ae8e 100644
--- a/pkg/hocr/hocr.go
+++ b/pkg/hocr/hocr.go
@@ -127,3 +127,32 @@ func GetAvgConf(hocrfn string) (float64, error) {
}
return total / num, nil
}
+
+// GetWordConfs is a utility function that parses a hocr
+// file and returns an array containing the confidences
+// of each word therein.
+func GetWordConfs(hocrfn string) ([]float64, error) {
+ var confs []float64
+
+ file, err := ioutil.ReadFile(hocrfn)
+ if err != nil {
+ return confs, err
+ }
+
+ h, err := Parse(file)
+ if err != nil {
+ return confs, err
+ }
+
+ for _, l := range h.Lines {
+ for _, w := range l.Words {
+ c, err := wordConf(w.Title)
+ if err != nil {
+ return confs, err
+ }
+ confs = append(confs, c)
+ }
+ }
+
+ return confs, nil
+}
--
cgit v1.2.1-24-ge1ad