Track and print bucket stats generically

2 files changed, 72 insertions, 30 deletions

diff --git a/line-conf-buckets/line-conf-buckets.go b/line-conf-buckets/line-conf-buckets.go
index 65c85b8..bc4a1aa 100644
--- a/line-conf-buckets/line-conf-buckets.go
+++ b/line-conf-buckets/line-conf-buckets.go
@@ -6,12 +6,25 @@ import (
 	"fmt"
 	"log"
 	"os"
-	"sort"
 
 	"git.rescribe.xyz/testingtools/parse"
 	"git.rescribe.xyz/testingtools/parse/prob"
 )
 
+func lineDetails(f string) (parse.LineDetails, error) {
+	var newlines parse.LineDetails
+
+	file, err := os.Open(f)
+	if err != nil {
+		return newlines, err
+	}
+	defer file.Close()
+
+	reader := bufio.NewReader(file)
+
+	return prob.GetLineDetails(f, reader)
+}
+
 func main() {
 	flag.Usage = func() {
 		fmt.Fprintf(os.Stderr, "Usage: line-conf-buckets prob1 [prob2] [...]\n")
@@ -28,15 +41,7 @@ func main() {
 	lines := make(parse.LineDetails, 0)
 
 	for _, f := range flag.Args() {
-		file, err := os.Open(f)
-		if err != nil {
-			log.Fatal(err)
-		}
-		defer file.Close()
-
-		reader := bufio.NewReader(file)
-
-		newlines, err := prob.GetLineDetails(f, reader)
+		newlines, err := lineDetails(f)
 		if err != nil {
 			log.Fatal(err)
 		}
@@ -44,14 +49,8 @@ func main() {
                 for _, l := range newlines {
                         lines = append(lines, l)
                 }
-		// explicitly close the file, so we can be sure we won't run out of
-		// handles before defer runs
-		file.Close()
 	}
 
-	sort.Sort(lines)
-
-	//var b parse.BucketSpecs
 	b := parse.BucketSpecs{
 		{ 0, "bad" },
 		{ 0.95, "95to98" },
@@ -59,8 +58,10 @@ func main() {
 	}
 
 	// TODO: set bucket dirname from cmdline
-	err := parse.BucketUp(lines, b, "newbuckets")
+	stats, err := parse.BucketUp(lines, b, "newbuckets")
 	if err != nil {
 		log.Fatal(err)
 	}
+
+	parse.PrintBucketStats(os.Stdout, stats)
 }
diff --git a/parse/bucket.go b/parse/bucket.go
index 3a4cd6a..44b1d24 100644
--- a/parse/bucket.go
+++ b/parse/bucket.go
@@ -1,6 +1,7 @@
 package parse
 
 import (
+	"fmt"
 	"io"
 	"path/filepath"
 	"os"
@@ -17,16 +18,19 @@ 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 {
+func bucketLine(l LineDetail, 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 nil
+		return bucket, nil
 	}
 
 	avgstr := strconv.FormatFloat(l.Avgconf, 'G', -1, 64)
@@ -38,45 +42,82 @@ func bucketLine(l LineDetail, buckets BucketSpecs, dirname string) error {
 
 	err := os.MkdirAll(filepath.Join(dirname, todir), 0700)
 	if err != nil {
-		return err
+		return bucket, err
 	}
 
 	f, err := os.Create(base + ".png")
 	if err != nil {
-		return err
+		return bucket, err
 	}
 	defer f.Close()
 
 	err = l.Img.CopyLineTo(f)
 	if err != nil {
-		return err
+		return bucket, err
 	}
 
 	f, err = os.Create(base + ".txt")
 	if err != nil {
-		return err
+		return bucket, err
 	}
 	defer f.Close()
 
 	_, err = io.WriteString(f, l.Text)
 	if err != nil {
-		return err
+		return bucket, err
 	}
 
-	return nil
+	return bucket, err
+}
+
+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 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 {
+func BucketUp(lines LineDetails, buckets BucketSpecs, dirname string) (BucketStats, error) {
+	var all []string
+	var stats BucketStats
+
+	sort.Sort(lines)
 	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)
+		bname, err := bucketLine(l, buckets, dirname)
 		if err != nil {
-			return err
+			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
+}
+
+func PrintBucketStats(w io.Writer, stats BucketStats) {
+	var total int
+	for _, s := range stats {
+		total += s.num
 	}
 
-	return nil
+	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)
+	}
 }