2 files changed, 220 insertions, 0 deletions
diff --git a/cmd/bucket-lines/bucket.go b/cmd/bucket-lines/bucket.go
new file mode 100644
index 0000000..7b6fc4f
--- /dev/null
+++ b/cmd/bucket-lines/bucket.go
@@ -0,0 +1,131 @@
+package main
+
+import (
+	"fmt"
+	"io"
+	"os"
+	"path/filepath"
+	"sort"
+	"strconv"
+
+	"rescribe.xyz/utils/pkg/line"
+)
+
+type BucketSpec struct {
+	Min  float64
+	Name string
+}
+type BucketSpecs []BucketSpec
+
+func (b BucketSpecs) Len() int           { return len(b) }
+func (b BucketSpecs) Swap(i, j int)      { b[i], b[j] = b[j], b[i] }
+func (b BucketSpecs) Less(i, j int) bool { return b[i].Min < b[j].Min }
+
+type BucketStat struct {
+	name string
+	num  int
+}
+type BucketStats []BucketStat
+
+func (b BucketStats) Len() int           { return len(b) }
+func (b BucketStats) Swap(i, j int)      { b[i], b[j] = b[j], b[i] }
+func (b BucketStats) Less(i, j int) bool { return b[i].num < b[j].num }
+
+// Copies the image and text for a line into a directory based on
+// the line confidence, as defined by the buckets struct
+func bucketLine(l line.Detail, buckets BucketSpecs, dirname string) (string, error) {
+	var bucket string
+
+	todir := ""
+	for _, b := range buckets {
+		if l.Avgconf >= b.Min {
+			todir = b.Name
+			bucket = b.Name
+		}
+	}
+
+	if todir == "" {
+		return bucket, nil
+	}
+
+	avgstr := strconv.FormatFloat(l.Avgconf, 'G', -1, 64)
+	if len(avgstr) > 2 {
+		avgstr = avgstr[2:]
+	}
+
+	base := filepath.Join(dirname, todir, l.OcrName+"_"+l.Name+"_"+avgstr)
+
+	err := os.MkdirAll(filepath.Join(dirname, todir), 0700)
+	if err != nil {
+		return bucket, err
+	}
+
+	f, err := os.Create(base + ".png")
+	if err != nil {
+		return bucket, err
+	}
+	defer f.Close()
+
+	err = l.Img.CopyLineTo(f)
+	if err != nil {
+		return bucket, err
+	}
+
+	f, err = os.Create(base + ".txt")
+	if err != nil {
+		return bucket, err
+	}
+	defer f.Close()
+
+	_, err = io.WriteString(f, l.Text)
+	if err != nil {
+		return bucket, err
+	}
+
+	return bucket, err
+}
+
+// Copies line images and text into directories based on their
+// confidence, as defined by the buckets struct, and returns
+// statistics of whire lines went in the process.
+func BucketUp(lines line.Details, buckets BucketSpecs, dirname string) (BucketStats, error) {
+	var all []string
+	var stats BucketStats
+
+	sort.Sort(lines)
+	sort.Sort(buckets)
+	for _, l := range lines {
+		bname, err := bucketLine(l, buckets, dirname)
+		if err != nil {
+			return stats, err
+		}
+		all = append(all, bname)
+	}
+
+	for _, b := range all {
+		i := sort.Search(len(stats), func(i int) bool { return stats[i].name == b })
+		if i == len(stats) {
+			newstat := BucketStat{b, 0}
+			stats = append(stats, newstat)
+			i = len(stats) - 1
+		}
+		stats[i].num++
+	}
+
+	return stats, nil
+}
+
+// Prints statistics of where lines went when bucketing
+func PrintBucketStats(w io.Writer, stats BucketStats) {
+	var total int
+	for _, s := range stats {
+		total += s.num
+	}
+
+	fmt.Fprintf(w, "Copied %d lines\n", total)
+	fmt.Fprintf(w, "---------------------------------\n")
+	sort.Sort(stats)
+	for _, s := range stats {
+		fmt.Fprintf(w, "Lines in %7s: %2d%%\n", s.name, 100*s.num/total)
+	}
+}
diff --git a/cmd/bucket-lines/main.go b/cmd/bucket-lines/main.go
new file mode 100644
index 0000000..af81b44
--- /dev/null
+++ b/cmd/bucket-lines/main.go
@@ -0,0 +1,89 @@
+package main
+
+import (
+	"encoding/json"
+	"flag"
+	"fmt"
+	"io/ioutil"
+	"log"
+	"os"
+	"path/filepath"
+
+	"rescribe.xyz/utils/pkg/hocr"
+	"rescribe.xyz/utils/pkg/line"
+	"rescribe.xyz/utils/pkg/prob"
+)
+
+func main() {
+	b := BucketSpecs{
+		// minimum confidence, name
+		{0, "bad"},
+		{0.95, "95to98"},
+		{0.98, "98plus"},
+	}
+
+	flag.Usage = func() {
+		fmt.Fprintf(os.Stderr, "Usage: bucket-lines [-d dir] [-s specs.json] [hocr1] [prob1] [hocr2] [...]\n")
+		fmt.Fprintf(os.Stderr, "Copies image-text line pairs into different directories according\n")
+		fmt.Fprintf(os.Stderr, "to the average character probability for the line.\n")
+		fmt.Fprintf(os.Stderr, "Both .hocr and .prob files can be processed.\n")
+		fmt.Fprintf(os.Stderr, "For .hocr files, the x_wconf data is used to calculate confidence.\n")
+		fmt.Fprintf(os.Stderr, "The .prob files are generated using ocropy-rpred's --probabilities\n")
+		fmt.Fprintf(os.Stderr, "option.\n")
+		fmt.Fprintf(os.Stderr, "The .prob and .hocr files are assumed to be in the same directory\n")
+		fmt.Fprintf(os.Stderr, "as the line's image and text files.\n\n")
+		flag.PrintDefaults()
+		fmt.Fprintf(os.Stderr, "\nAn example specs.json file would be the following:\n")
+		fmt.Fprintf(os.Stderr, "[{\"min\": 0, \"name\": \"terrible\"}, {\"min\": 0.80, \"name\": \"ok\"}, {\"min\": 0.98, \"name\": \"great\"}]\n")
+	}
+	dir := flag.String("d", "buckets", "Directory to store the buckets")
+	specs := flag.String("s", "", "JSON file describing specs to bucket into")
+	flag.Parse()
+	if flag.NArg() < 1 {
+		flag.Usage()
+		os.Exit(1)
+	}
+
+	if *specs != "" {
+		js, err := ioutil.ReadFile(*specs)
+		if err != nil {
+			log.Fatal(err)
+		}
+		err = json.Unmarshal(js, &b)
+		if err != nil {
+			log.Fatal(err)
+		}
+	}
+
+	var err error
+	lines := make(line.Details, 0)
+
+	for _, f := range flag.Args() {
+		var newlines line.Details
+		switch ext := filepath.Ext(f); ext {
+		case ".prob":
+			newlines, err = prob.GetLineDetails(f)
+		case ".hocr":
+			newlines, err = hocr.GetLineDetails(f)
+		default:
+			log.Printf("Skipping file '%s' as it isn't a .prob or .hocr\n", f)
+			continue
+		}
+		if err != nil {
+			log.Fatal(err)
+		}
+
+		for _, l := range newlines {
+			if l.Img != nil {
+				lines = append(lines, l)
+			}
+		}
+	}
+
+	stats, err := BucketUp(lines, b, *dir)
+	if err != nil {
+		log.Fatal(err)
+	}
+
+	PrintBucketStats(os.Stdout, stats)
+}