diff options
author | Nick White <git@njw.name> | 2019-08-28 17:35:20 +0100 |
---|---|---|
committer | Nick White <git@njw.name> | 2019-08-28 17:35:20 +0100 |
commit | 49b87574ba54b7450f7c0520153f5d4a4695076f (patch) | |
tree | 0c1c67475e9d5441d6b034aeae68c1690a9ecf56 /bookpipeline | |
parent | c692d23044f914abffcd21302c3e152e19ab82af (diff) |
Move graph function to its own file, and further improve layout
Diffstat (limited to 'bookpipeline')
-rw-r--r-- | bookpipeline/graph.go | 140 | ||||
-rw-r--r-- | bookpipeline/main.go | 140 |
2 files changed, 140 insertions, 140 deletions
diff --git a/bookpipeline/graph.go b/bookpipeline/graph.go new file mode 100644 index 0000000..27ffd39 --- /dev/null +++ b/bookpipeline/graph.go @@ -0,0 +1,140 @@ +package main + +import ( + "fmt" + "io" + "path/filepath" + "sort" + "strconv" + "strings" + + "github.com/wcharczuk/go-chart" +) + +const maxticks = 20 +const cutoff = 70 + +type GraphConf struct { + pgnum, conf float64 +} + +func graph(confs map[string]*Conf, bookname string, w io.Writer) (error) { + // Organise confs to sort them by page + var graphconf []GraphConf + for _, conf := range confs { + name := filepath.Base(conf.path) + numend := strings.Index(name, "_") + pgnum, err := strconv.ParseFloat(name[0:numend], 64) + if err != nil { + continue + } + var c GraphConf + c.pgnum = pgnum + c.conf = conf.conf + graphconf = append(graphconf, c) + } + sort.Slice(graphconf, func(i, j int) bool { return graphconf[i].pgnum < graphconf[j].pgnum }) + + // Create main xvalues and yvalues, annotations and ticks + var xvalues, yvalues []float64 + var annotations []chart.Value2 + var ticks []chart.Tick + i := 0 + tickevery := len(graphconf) / maxticks + for _, c := range graphconf { + i = i + 1 + xvalues = append(xvalues, c.pgnum) + yvalues = append(yvalues, c.conf) + if c.conf < cutoff { + annotations = append(annotations, chart.Value2{Label: fmt.Sprintf("%.0f", c.pgnum), XValue: c.pgnum, YValue: c.conf}) + } + if tickevery % i == 0 { + ticks = append(ticks, chart.Tick{c.pgnum, fmt.Sprintf("%.0f", c.pgnum)}) + } + } + mainSeries := chart.ContinuousSeries{ + XValues: xvalues, + YValues: yvalues, + } + + // Create 70% line + yvalues = []float64{} + for _ = range xvalues { + yvalues = append(yvalues, cutoff) + } + cutoffSeries := chart.ContinuousSeries{ + XValues: xvalues, + YValues: yvalues, + Style: chart.Style{ + Show: true, + StrokeColor: chart.ColorAlternateGreen, + StrokeDashArray: []float64{10.0, 5.0}, + }, + } + + // Create lines marking top and bottom 10% confidence + sort.Slice(graphconf, func(i, j int) bool { return graphconf[i].conf < graphconf[j].conf }) + lowconf := graphconf[int(len(graphconf) / 10)].conf + highconf := graphconf[int((len(graphconf) / 10) * 9)].conf + yvalues = []float64{} + for _ = range graphconf { + yvalues = append(yvalues, lowconf) + } + minSeries := &chart.ContinuousSeries{ + Style: chart.Style{ + Show: true, + StrokeColor: chart.ColorAlternateGray, + StrokeDashArray: []float64{5.0, 5.0}, + }, + XValues: xvalues, + YValues: yvalues, + } + yvalues = []float64{} + for _ = range graphconf { + yvalues = append(yvalues, highconf) + } + maxSeries := &chart.ContinuousSeries{ + Style: chart.Style{ + Show: true, + StrokeColor: chart.ColorAlternateGray, + StrokeDashArray: []float64{5.0, 5.0}, + }, + XValues: xvalues, + YValues: yvalues, + } + + graph := chart.Chart{ + Title: bookname, + TitleStyle: chart.StyleShow(), + Width: 1920, + Height: 1080, + XAxis: chart.XAxis{ + Name: "Page number", + NameStyle: chart.StyleShow(), + Style: chart.StyleShow(), + Range: &chart.ContinuousRange{ + Min: 0.0, + }, + Ticks: ticks, + }, + YAxis: chart.YAxis{ + Name: "Confidence", + NameStyle: chart.StyleShow(), + Style: chart.StyleShow(), + Range: &chart.ContinuousRange{ + Min: 0.0, + Max: 100.0, + }, + }, + Series: []chart.Series{ + mainSeries, + minSeries, + maxSeries, + cutoffSeries, + chart.AnnotationSeries{ + Annotations: annotations, + }, + }, + } + return graph.Render(chart.PNG, w) +} diff --git a/bookpipeline/main.go b/bookpipeline/main.go index de38ab4..b7b01dd 100644 --- a/bookpipeline/main.go +++ b/bookpipeline/main.go @@ -7,19 +7,14 @@ import ( "errors" "flag" "fmt" - "io" "log" "os" "os/exec" "path/filepath" "regexp" - "sort" - "strconv" "strings" "time" - "github.com/wcharczuk/go-chart" - "rescribe.xyz/go.git/lib/hocr" "rescribe.xyz/go.git/preproc" ) @@ -41,9 +36,6 @@ one is found this general process is followed: ` -const maxticks = 20 -const cutoff = 70 - // null writer to enable non-verbose logging to be discarded type NullWriter bool @@ -143,138 +135,6 @@ type Conf struct { path, code string conf float64 } -type GraphConf struct { - pgnum, conf float64 -} - -func graph(confs map[string]*Conf, bookname string, w io.Writer) (error) { - // Organise confs to sort them by page - var graphconf []GraphConf - for _, conf := range confs { - name := filepath.Base(conf.path) - numend := strings.Index(name, "_") - pgnum, err := strconv.ParseFloat(name[0:numend], 64) - if err != nil { - continue - } - var c GraphConf - c.pgnum = pgnum - c.conf = conf.conf - graphconf = append(graphconf, c) - } - sort.Slice(graphconf, func(i, j int) bool { return graphconf[i].pgnum < graphconf[j].pgnum }) - - // Create main xvalues and yvalues, annotations and ticks - var xvalues, yvalues []float64 - var annotations []chart.Value2 - var ticks []chart.Tick - i := 0 - tickevery := len(graphconf) / maxticks - for _, c := range graphconf { - i = i + 1 - xvalues = append(xvalues, c.pgnum) - yvalues = append(yvalues, c.conf) - if c.conf < cutoff { - annotations = append(annotations, chart.Value2{Label: fmt.Sprintf("%.0f", c.pgnum), XValue: c.pgnum, YValue: c.conf}) - } - if tickevery % i == 0 { - ticks = append(ticks, chart.Tick{c.pgnum, fmt.Sprintf("%.0f", c.pgnum)}) - } - } - mainSeries := chart.ContinuousSeries{ - XValues: xvalues, - YValues: yvalues, - } - - // Create 70% line - yvalues = []float64{} - for _, _ = range xvalues { - yvalues = append(yvalues, cutoff) - } - cutoffSeries := chart.ContinuousSeries{ - XValues: xvalues, - YValues: yvalues, - Style: chart.Style{ - Show: true, - StrokeColor: chart.ColorAlternateGreen, - StrokeDashArray: []float64{10.0, 5.0}, - }, - } - - // Create lines marking top and bottom 10% confidence - sort.Slice(graphconf, func(i, j int) bool { return graphconf[i].conf < graphconf[j].conf }) - cutoff := int(len(graphconf) / 10) - mostconf := graphconf[cutoff:len(graphconf)-cutoff] - sort.Slice(mostconf, func(i, j int) bool { return mostconf[i].pgnum < mostconf[j].pgnum }) - xvalues = []float64{} - yvalues = []float64{} - for _, c := range mostconf { - xvalues = append(xvalues, c.pgnum) - yvalues = append(yvalues, c.conf) - } - mostSeries := chart.ContinuousSeries{ - XValues: xvalues, - YValues: yvalues, - } - minSeries := &chart.MinSeries{ - Style: chart.Style{ - Show: true, - StrokeColor: chart.ColorAlternateGray, - StrokeDashArray: []float64{5.0, 5.0}, - }, - InnerSeries: mostSeries, - } - maxSeries := &chart.MaxSeries{ - Style: chart.Style{ - Show: true, - StrokeColor: chart.ColorAlternateGray, - StrokeDashArray: []float64{5.0, 5.0}, - }, - InnerSeries: mostSeries, - } - - graph := chart.Chart{ - Title: fmt.Sprintf("Confidence of pages from %s", bookname), - TitleStyle: chart.StyleShow(), - Width: 1920, - Height: 1080, - XAxis: chart.XAxis{ - Name: "Page number", - NameStyle: chart.StyleShow(), - Style: chart.StyleShow(), - Range: &chart.ContinuousRange{ - Min: 0.0, - }, - Ticks: ticks, - }, - YAxis: chart.YAxis{ - Name: "Confidence", - NameStyle: chart.StyleShow(), - Style: chart.StyleShow(), - Range: &chart.ContinuousRange{ - Min: 0.0, - Max: 100.0, - }, - }, - Series: []chart.Series{ - mainSeries, - minSeries, - maxSeries, - cutoffSeries, - chart.LastValueAnnotation(minSeries), - chart.LastValueAnnotation(maxSeries), - chart.AnnotationSeries{ - Annotations: annotations, - }, - //chart.ContinuousSeries{ - // YAxis: chart.YAxisSecondary, - // XValues: xvalues, - // YValues: yvalues, - //}, - }, - } - return graph.Render(chart.PNG, w) -} func analyse(toanalyse chan string, up chan string, errc chan error, logger *log.Logger) { confs := make(map[string][]*Conf) |