From fb77852acbdbbcedcdb9771770cb6771da002851 Mon Sep 17 00:00:00 2001
From: Nick White <git@njw.name>
Date: Wed, 23 Jan 2019 21:54:09 +0000
Subject: Update line-conf-buckets to mostly use package functions too.

Working now, but needs more consolidation to be worth it.
---
 parse/hocr/hocr.go |  5 ++--
 parse/line.go      | 18 +++++++++++++-
 parse/prob/prob.go | 72 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 92 insertions(+), 3 deletions(-)
 create mode 100644 parse/prob/prob.go

(limited to 'parse')

diff --git a/parse/hocr/hocr.go b/parse/hocr/hocr.go
index a281a7a..f7cac05 100644
--- a/parse/hocr/hocr.go
+++ b/parse/hocr/hocr.go
@@ -1,7 +1,8 @@
 package hocr
 
 // TODO: consider making GetLineDetails() a function of Hocr, so could do a
-//       similar thing with prob format files too.
+//       similar thing with prob format files too, and then fire them both
+//       off a generic interface, potentially.
 // TODO: Parse line name to zero pad line numbers, so they come out in the correct order
 
 import (
@@ -137,7 +138,7 @@ func GetLineDetails(h Hocr, i image.Image, name string) (parse.LineDetails, erro
 		}
 		line.Text = strings.TrimRight(linetext, " ")
 		line.Text += "\n"
-		line.Hocrname = name
+		line.OcrName = name
 		var imgd parse.ImgDirect
 		imgd.Img = i.(*image.Gray).SubImage(image.Rect(coords[0], coords[1], coords[2], coords[3]))
 		line.Img = imgd
diff --git a/parse/line.go b/parse/line.go
index 3ddde76..9a2be8e 100644
--- a/parse/line.go
+++ b/parse/line.go
@@ -9,6 +9,7 @@ import (
 	"image"
 	"image/png"
 	"io"
+	"os"
 )
 
 type LineDetail struct {
@@ -16,7 +17,7 @@ type LineDetail struct {
 	Avgconf float64
 	Img CopyableLine
 	Text string
-	Hocrname string
+	OcrName string
 }
 
 type CopyableLine interface {
@@ -37,6 +38,21 @@ func (i ImgDirect) CopyLineTo(w io.Writer) (error) {
 	return nil
 }
 
+type ImgPath struct {
+	Path string
+}
+
+func (i ImgPath) CopyLineTo(w io.Writer) (error) {
+	f, err := os.Open(i.Path)
+	if err != nil {
+		return err
+	}
+	defer f.Close()
+
+	_, err = io.Copy(w, f)
+	return err
+}
+
 type LineDetails []LineDetail
 
 // Used by sort.Sort.
diff --git a/parse/prob/prob.go b/parse/prob/prob.go
new file mode 100644
index 0000000..5a84567
--- /dev/null
+++ b/parse/prob/prob.go
@@ -0,0 +1,72 @@
+package prob
+
+import (
+	"bufio"
+	"io/ioutil"
+	"path/filepath"
+	"strconv"
+	"strings"
+
+	"git.rescribe.xyz/testingtools/parse"
+)
+
+// TODO: probably switch to just relying on io.Reader
+func getLineAvg(r *bufio.Reader) (float64, error) {
+	var err error
+
+	totalconf := float64(0)
+	num := 0
+
+	err = nil
+	for err == nil {
+		var line string
+		line, err = r.ReadString('\n')
+		fields := strings.Fields(line)
+
+		if len(fields) == 2 {
+			conf, converr := strconv.ParseFloat(fields[1], 64)
+			if converr != nil {
+				continue
+			}
+			totalconf += conf
+			num += 1
+		}
+	}
+	if num <= 0 {
+		return 0, nil
+	}
+	avg := totalconf / float64(num)
+	return avg, nil
+}
+
+// TODO: probably switch to just relying on io.Reader
+// Note this only processes one line at a time
+func GetLineDetails(name string, r *bufio.Reader) (parse.LineDetails, error) {
+	var line parse.LineDetail
+	lines := make(parse.LineDetails, 0)
+
+	avg, err := getLineAvg(r)
+	if err != nil {
+		return lines, err
+	}
+
+	filebase := strings.Replace(name, ".prob", "", 1)
+
+	txt, err := ioutil.ReadFile(filebase + ".txt")
+	if err != nil {
+		return lines, err
+	}
+
+	line.Name = name
+	line.Avgconf = avg
+	line.Text = string(txt)
+	line.OcrName = filepath.Dir(filebase)
+
+	var imgfn parse.ImgPath
+	imgfn.Path = filebase + ".bin.png"
+	line.Img = imgfn
+
+	lines = append(lines, line)
+
+	return lines, nil
+}
-- 
cgit v1.2.1-24-ge1ad