Create general BucketUp function, use it for line-conf-buckets

3 files changed, 92 insertions, 89 deletions

diff --git a/line-conf-buckets/line-conf-buckets.go b/line-conf-buckets/line-conf-buckets.go
index 1c33ba4..65c85b8 100644
--- a/line-conf-buckets/line-conf-buckets.go
+++ b/line-conf-buckets/line-conf-buckets.go
@@ -4,62 +4,14 @@ import (
 	"bufio"
 	"flag"
 	"fmt"
-	"io"
 	"log"
 	"os"
-	"path/filepath"
 	"sort"
-	"strconv"
-	"strings"
 
 	"git.rescribe.xyz/testingtools/parse"
 	"git.rescribe.xyz/testingtools/parse/prob"
 )
 
-// TODO: this is just a placeholder, do this more sensibly, as -tess does (hint: full txt should already be in the LineDetail)
-func copyline(filebase string, dirname string, basename string, avgconf string, outdir string, todir string, l parse.LineDetail) (err error) {
-	outname := filepath.Join(outdir, todir, filepath.Base(dirname) + "_" + basename + "_" + avgconf)
-	//log.Fatalf("I'd use '%s' as outname, and '%s' as filebase\n", outname, filebase)
-
-	for _, extn := range []string{".txt"} {
-		infile, err := os.Open(filebase + extn)
-		if err != nil {
-			fmt.Fprintf(os.Stderr, "Failed to open %s\n", filebase + extn)
-			return err
-		}
-		defer infile.Close()
-
-		err = os.MkdirAll(filepath.Join(outdir, todir), 0700)
-		if err != nil {
-			return err
-		}
-	
-		outfile, err := os.Create(outname + extn)
-		if err != nil {
-			fmt.Fprintf(os.Stderr, "Failed to create %s\n", outname + extn)
-			return err
-		}
-		defer outfile.Close()
-	
-		_, err = io.Copy(outfile, infile)
-		if err != nil {
-			return err
-		}
-	}
-
-	f, err := os.Create(outname + ".bin.png")
-	if err != nil {
-		return err
-	}
-	defer f.Close()
-	err = l.Img.CopyLineTo(f)
-	if err != nil {
-		return err
-	}
-
-	return err
-}
-
 func main() {
 	flag.Usage = func() {
 		fmt.Fprintf(os.Stderr, "Usage: line-conf-buckets prob1 [prob2] [...]\n")
@@ -99,47 +51,16 @@ func main() {
 
 	sort.Sort(lines)
 
-	worstnum := 0
-	mediumnum := 0
-	bestnum := 0
-
-	outdir := "buckets" // TODO: set this from cmdline
-	todir := ""
-
-	for _, l := range lines {
-		switch {
-		case l.Avgconf < 0.95: 
-			todir = "bad"
-			worstnum++
-		case l.Avgconf < 0.98:
-			todir = "95to98"
-			mediumnum++
-		default:
-			todir = "98plus"
-			bestnum++
-		}
-
-		avgstr := strconv.FormatFloat(l.Avgconf, 'G', -1, 64)
-		if len(avgstr) > 2 {
-			avgstr = avgstr[2:]
-		}
-		filebase := strings.Replace(l.Name, ".prob", "", 1)
-		basename := filepath.Base(filebase)
-		err := copyline(filebase, l.OcrName, basename, avgstr, outdir, todir, l)
-		if err != nil {
-			log.Fatal(err)
-		}
+	//var b parse.BucketSpecs
+	b := parse.BucketSpecs{
+		{ 0, "bad" },
+		{ 0.95, "95to98" },
+		{ 0.98, "98plus" },
 	}
 
-	total := worstnum + mediumnum + bestnum
-
-	if total == 0 {
-		log.Fatal("No lines copied")
+	// TODO: set bucket dirname from cmdline
+	err := parse.BucketUp(lines, b, "newbuckets")
+	if err != nil {
+		log.Fatal(err)
 	}
-
-	fmt.Printf("Copied lines to %s\n", outdir)
-	fmt.Printf("---------------------------------\n")
-	fmt.Printf("Lines 98%%+ quality:     %d%%\n", 100 * bestnum / total)
-	fmt.Printf("Lines 95-98%% quality:   %d%%\n", 100 * mediumnum / total)
-	fmt.Printf("Lines <95%% quality:     %d%%\n", 100 * worstnum / total)
 }
diff --git a/parse/bucket.go b/parse/bucket.go
new file mode 100644
index 0000000..3a4cd6a
--- /dev/null
+++ b/parse/bucket.go
@@ -0,0 +1,82 @@
+package parse
+
+import (
+	"io"
+	"path/filepath"
+	"os"
+	"sort"
+	"strconv"
+)
+
+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 }
+
+func bucketLine(l LineDetail, buckets BucketSpecs, dirname string) error {
+	todir := ""
+	for _, b := range buckets {
+		if l.Avgconf >= b.Min {
+			todir = b.Name
+		}
+	}
+
+	if todir == "" {
+		return nil
+	}
+
+	avgstr := strconv.FormatFloat(l.Avgconf, 'G', -1, 64)
+	if len(avgstr) > 2 {
+		avgstr = avgstr[2:]
+	}
+
+	base := filepath.Join(dirname, todir, filepath.Base(l.OcrName) + "_" + l.Name + "_" + avgstr)
+
+	err := os.MkdirAll(filepath.Join(dirname, todir), 0700)
+	if err != nil {
+		return err
+	}
+
+	f, err := os.Create(base + ".png")
+	if err != nil {
+		return err
+	}
+	defer f.Close()
+
+	err = l.Img.CopyLineTo(f)
+	if err != nil {
+		return err
+	}
+
+	f, err = os.Create(base + ".txt")
+	if err != nil {
+		return err
+	}
+	defer f.Close()
+
+	_, err = io.WriteString(f, l.Text)
+	if err != nil {
+		return err
+	}
+
+	return nil
+}
+
+// Copies line images and text into directories based on their
+// confidence, as defined by the buckets struct
+func BucketUp(lines LineDetails, buckets BucketSpecs, dirname string) error {
+	sort.Sort(buckets)
+	// TODO: record and print out summary of % in each bucket category (see how tools did it)
+	for _, l := range lines {
+		err := bucketLine(l, buckets, dirname)
+		if err != nil {
+			return err
+		}
+	}
+
+	return nil
+}
diff --git a/parse/prob/prob.go b/parse/prob/prob.go
index 5a84567..6781a1a 100644
--- a/parse/prob/prob.go
+++ b/parse/prob/prob.go
@@ -57,7 +57,7 @@ func GetLineDetails(name string, r *bufio.Reader) (parse.LineDetails, error) {
 		return lines, err
 	}
 
-	line.Name = name
+	line.Name = filepath.Base(filebase)
 	line.Avgconf = avg
 	line.Text = string(txt)
 	line.OcrName = filepath.Dir(filebase)