2 files changed, 260 insertions, 0 deletions
diff --git a/pkg/hocr/hocr.go b/pkg/hocr/hocr.go
new file mode 100644
index 0000000..dcd0494
--- /dev/null
+++ b/pkg/hocr/hocr.go
@@ -0,0 +1,129 @@
+package hocr
+
+import (
+	"encoding/xml"
+	"errors"
+	"io/ioutil"
+	"regexp"
+	"strconv"
+	"strings"
+)
+
+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 GetText(hocrfn string) (string, error) {
+	var s string
+
+	file, err := ioutil.ReadFile(hocrfn)
+	if err != nil {
+		return s, err
+	}
+
+	h, err := Parse(file)
+	if err != nil {
+		return s, err
+	}
+
+
+	for _, l := range h.Lines {
+		s += getLineText(l)
+	}
+	return s, nil
+}
+
+func GetAvgConf(hocrfn string) (float64, error) {
+	file, err := ioutil.ReadFile(hocrfn)
+	if err != nil {
+		return 0, err
+	}
+
+	h, err := Parse(file)
+	if err != nil {
+		return 0, err
+	}
+
+	var total, num float64
+	for _, l := range h.Lines {
+		for _, w := range l.Words {
+			c, err := wordConf(w.Title)
+			if err != nil {
+				return 0, err
+			}
+			total += c
+			num++
+		}
+	}
+	if num == 0 {
+		return 0, errors.New("No words found")
+	}
+	return total / num, nil
+}
diff --git a/pkg/hocr/lines.go b/pkg/hocr/lines.go
new file mode 100644
index 0000000..e90b0a8
--- /dev/null
+++ b/pkg/hocr/lines.go
@@ -0,0 +1,131 @@
+package hocr
+
+// TODO: Parse line name to zero pad line numbers, so they can
+//       be sorted easily
+
+import (
+	"image"
+	"image/png"
+	"io/ioutil"
+	"log"
+	"os"
+	"path/filepath"
+	"strings"
+
+	"rescribe.xyz/utils/pkg/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)
+
+	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 ln line.Detail
+		ln.Name = l.Id
+		ln.Avgconf = (totalconf / float64(num)) / 100
+		ln.Text = getLineText(l)
+		ln.OcrName = name
+		if i != nil {
+			var imgd line.ImgDirect
+			imgd.Img = i.(*image.Gray).SubImage(image.Rect(coords[0], coords[1], coords[2], coords[3]))
+			ln.Img = imgd
+		}
+		lines = append(lines, ln)
+	}
+	return lines, nil
+}
+
+func GetLineDetails(hocrfn string) (line.Details, error) {
+	var newlines line.Details
+
+	file, err := ioutil.ReadFile(hocrfn)
+	if err != nil {
+		return newlines, err
+	}
+
+	h, err := Parse(file)
+	if err != nil {
+		return newlines, err
+	}
+
+	var img image.Image
+	pngfn := strings.Replace(hocrfn, ".hocr", ".png", 1)
+	pngf, err := os.Open(pngfn)
+	if err != nil {
+		log.Println("Warning: can't open image %s\n", pngfn)
+	} else {
+		defer pngf.Close()
+		img, err = png.Decode(pngf)
+		if err != nil {
+			log.Println("Warning: can't load image %s\n", pngfn)
+		}
+	}
+
+	n := strings.Replace(filepath.Base(hocrfn), ".hocr", "", 1)
+	return parseLineDetails(h, img, n)
+}
+
+func GetLineBasics(hocrfn string) (line.Details, error) {
+	var newlines line.Details
+
+	file, err := ioutil.ReadFile(hocrfn)
+	if err != nil {
+		return newlines, err
+	}
+
+	h, err := Parse(file)
+	if err != nil {
+		return newlines, err
+	}
+
+	n := strings.Replace(filepath.Base(hocrfn), ".hocr", "", 1)
+	return parseLineDetails(h, nil, n)
+}