From 30c088b90e7b6a25d93cbdad7564ff063e62afd3 Mon Sep 17 00:00:00 2001 From: Nick White Date: Fri, 25 Jan 2019 09:55:55 +0000 Subject: Reorganisation and cleanup --- parse/bucket.go | 123 -------------------------------------------------------- 1 file changed, 123 deletions(-) delete mode 100644 parse/bucket.go (limited to 'parse/bucket.go') diff --git a/parse/bucket.go b/parse/bucket.go deleted file mode 100644 index 44b1d24..0000000 --- a/parse/bucket.go +++ /dev/null @@ -1,123 +0,0 @@ -package parse - -import ( - "fmt" - "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) (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, filepath.Base(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 -} - -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) (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 -} - -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) - } -} -- cgit v1.2.1-24-ge1ad