From cd1fb1c9f6e1384ac0add8904425e6f92b17a704 Mon Sep 17 00:00:00 2001
From: Nick White <git@njw.name>
Date: Mon, 25 Feb 2019 13:01:28 +0000
Subject: Generalise get text from hocr lines

---
 lib/hocr/hocr.go  | 29 ++-----------------------
 lib/hocr/lines.go | 63 ++++++++++++++++++++++++++++++-------------------------
 2 files changed, 36 insertions(+), 56 deletions(-)

diff --git a/lib/hocr/hocr.go b/lib/hocr/hocr.go
index fbf1523..f6316d8 100644
--- a/lib/hocr/hocr.go
+++ b/lib/hocr/hocr.go
@@ -92,34 +92,9 @@ func GetText(hocrfn string) (string, error) {
 		return s, err
 	}
 
+
 	for _, l := range h.Lines {
-		linetext := l.Text
-		if noText(linetext) {
-			linetext = ""
-			for _, w := range l.Words {
-				if w.Class != "ocrx_word" {
-					continue
-				}
-				linetext += w.Text + " "
-			}
-		}
-		if noText(linetext) {
-			linetext = ""
-			for _, w := range l.Words {
-				if w.Class != "ocrx_word" {
-					continue
-				}
-				for _, c := range w.Chars {
-					if c.Class != "ocrx_cinfo" {
-						continue
-					}
-					linetext += c.Text
-				}
-				linetext += " "
-			}
-		}
-		linetext = strings.TrimRight(linetext, " ") + "\n"
-		s += linetext
+		s += getLineText(l)
 	}
 	return s, nil
 }
diff --git a/lib/hocr/lines.go b/lib/hocr/lines.go
index 00acd1f..c60a619 100644
--- a/lib/hocr/lines.go
+++ b/lib/hocr/lines.go
@@ -14,6 +14,39 @@ import (
 	"rescribe.xyz/go.git/lib/line"
 )
 
+func getLineText(l OcrLine) (string) {
+	linetext := ""
+
+	linetext = l.Text
+	if noText(linetext) {
+		linetext = ""
+		for _, w := range l.Words {
+			if w.Class != "ocrx_word" {
+				continue
+			}
+			linetext += w.Text + " "
+		}
+	}
+	if noText(linetext) {
+		linetext = ""
+		for _, w := range l.Words {
+			if w.Class != "ocrx_word" {
+				continue
+			}
+			for _, c := range w.Chars {
+				if c.Class != "ocrx_cinfo" {
+					continue
+				}
+				linetext += c.Text
+			}
+			linetext += " "
+		}
+	}
+	linetext = strings.TrimRight(linetext, " ")
+	linetext += "\n"
+	return linetext
+}
+
 func parseLineDetails(h Hocr, i image.Image, name string) (line.Details, error) {
 	lines := make(line.Details, 0)
 
@@ -37,35 +70,7 @@ func parseLineDetails(h Hocr, i image.Image, name string) (line.Details, error)
 		var ln line.Detail
 		ln.Name = l.Id
 		ln.Avgconf = (totalconf / float64(num)) / 100
-		linetext := ""
-
-		linetext = l.Text
-		if noText(linetext) {
-			linetext = ""
-			for _, w := range l.Words {
-				if w.Class != "ocrx_word" {
-					continue
-				}
-				linetext += w.Text + " "
-			}
-		}
-		if noText(linetext) {
-			linetext = ""
-			for _, w := range l.Words {
-				if w.Class != "ocrx_word" {
-					continue
-				}
-				for _, c := range w.Chars {
-					if c.Class != "ocrx_cinfo" {
-						continue
-					}
-					linetext += c.Text
-				}
-				linetext += " "
-			}
-		}
-		ln.Text = strings.TrimRight(linetext, " ")
-		ln.Text += "\n"
+		ln.Text = getLineText(l)
 		ln.OcrName = name
 		var imgd line.ImgDirect
 		imgd.Img = i.(*image.Gray).SubImage(image.Rect(coords[0], coords[1], coords[2], coords[3]))
-- 
cgit v1.2.1-24-ge1ad