From d256f967a26ceeb7c3987a1fc447b126a35054f9 Mon Sep 17 00:00:00 2001
From: Nick White <git@njw.name>
Date: Wed, 23 Jan 2019 20:47:33 +0000
Subject: Separate out hocr parts from line parts

---
 parse/hocr/hocr.go | 147 +++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 147 insertions(+)
 create mode 100644 parse/hocr/hocr.go

(limited to 'parse/hocr')

diff --git a/parse/hocr/hocr.go b/parse/hocr/hocr.go
new file mode 100644
index 0000000..a281a7a
--- /dev/null
+++ b/parse/hocr/hocr.go
@@ -0,0 +1,147 @@
+package hocr
+
+// TODO: consider making GetLineDetails() a function of Hocr, so could do a
+//       similar thing with prob format files too.
+// TODO: Parse line name to zero pad line numbers, so they come out in the correct order
+
+import (
+	"encoding/xml"
+	"image"
+	"regexp"
+	"strconv"
+	"strings"
+
+	"git.rescribe.xyz/testingtools/parse"
+)
+
+type Hocr struct {
+	Lines []OcrLine `xml:"body>div>div>p>span"`
+}
+
+type OcrLine struct {
+	Class string `xml:"class,attr"`
+	Id string `xml:"id,attr"`
+	Title string `xml:"title,attr"`
+	Words []OcrWord `xml:"span"`
+	Text string `xml:",chardata"`
+}
+
+type OcrWord struct {
+	Class string `xml:"class,attr"`
+	Id string `xml:"id,attr"`
+	Title string `xml:"title,attr"`
+	Chars []OcrChar `xml:"span"`
+	Text string `xml:",chardata"`
+}
+
+type OcrChar struct {
+	Class string `xml:"class,attr"`
+	Id string `xml:"id,attr"`
+	Title string `xml:"title,attr"`
+	Chars []OcrChar `xml:"span"`
+	Text string `xml:",chardata"`
+}
+
+// Returns the confidence for a word based on its x_wconf value
+func wordConf(s string) (float64, error) {
+	re, err := regexp.Compile(`x_wconf ([0-9.]+)`)
+	if err != nil {
+		return 0.0, err
+	}
+	conf := re.FindStringSubmatch(s)
+	return strconv.ParseFloat(conf[1], 64)
+}
+
+func boxCoords(s string) ([4]int, error) {
+	var coords [4]int
+	re, err := regexp.Compile(`bbox ([0-9]+) ([0-9]+) ([0-9]+) ([0-9]+)`)
+	if err != nil {
+		return coords, err
+	}
+	coordstr := re.FindStringSubmatch(s)
+	for i := range coords {
+		c, err := strconv.Atoi(coordstr[i+1])
+		if err != nil {
+			return coords, err
+		}
+		coords[i] = c
+	}
+	return coords, nil
+}
+
+func noText(s string) bool {
+	t := strings.Trim(s, " \n")
+	return len(t) == 0
+}
+
+func Parse(b []byte) (Hocr, error) {
+	var hocr Hocr
+
+	err := xml.Unmarshal(b, &hocr)
+	if err != nil {
+		return hocr, err
+	}
+
+	return hocr, nil
+}
+
+func GetLineDetails(h Hocr, i image.Image, name string) (parse.LineDetails, error) {
+	lines := make(parse.LineDetails, 0)
+
+	for _, l := range h.Lines {
+		totalconf := float64(0)
+		num := 0
+		for _, w := range l.Words {
+			c, err := wordConf(w.Title)
+			if err != nil {
+				return lines, err
+			}
+			num++
+			totalconf += c
+		}
+
+		coords, err := boxCoords(l.Title)
+		if err != nil {
+			return lines, err
+		}
+
+		var line parse.LineDetail
+		line.Name = l.Id
+		line.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 += " "
+			}
+		}
+		line.Text = strings.TrimRight(linetext, " ")
+		line.Text += "\n"
+		line.Hocrname = name
+		var imgd parse.ImgDirect
+		imgd.Img = i.(*image.Gray).SubImage(image.Rect(coords[0], coords[1], coords[2], coords[3]))
+		line.Img = imgd
+		lines = append(lines, line)
+	}
+	return lines, nil
+}
-- 
cgit v1.2.1-24-ge1ad