feat: debug information when constraint fail (#29)

* style: fix typo

* feat: wip! debug info

* feat: wip! debug info

* feat: debug info

* test: debug info when constraint fail

* chore: add pimo in ci

.devcontainer/Dockerfile.ci

@@ -7,3 +7,6 @@ RUN apk add --update --progress --no-cache make gomplate
 ARG VERSION_MILLER=6.2.0
 RUN    wget -nv -O- https://github.com/johnkerl/miller/releases/download/v${VERSION_MILLER}/miller-${VERSION_MILLER}-linux-amd64.tar.gz | tar xz --strip-components 1 -C /usr/bin miller-${VERSION_MILLER}-linux-amd64/mlr \
     && chmod +x /usr/bin/mlr
+
+ARG VERSION_PIMO=1.19.0
+RUN wget -O- https://github.com/CGI-FR/PIMO/releases/download/v${VERSION_PIMO}/pimo_${VERSION_PIMO}_linux_amd64.tar.gz | tar xz -C /usr/bin pimo

CHANGELOG.md

@@ -16,6 +16,7 @@ Types of changes
 
 ## [0.4.1]
 
+- `Added` debug information in logs when a constraint fail.
 - `Fixed` error handling, fatal errors will not print mimo help.
 - `Fixed` all counters are now persisted (with persist option).
 

cmd/mimo/main.go

@@ -21,6 +21,7 @@ import (
 	"fmt"
 	"os"
 	"runtime"
+	"slices"
 	"sort"
 	"strings"
 
@@ -236,11 +237,31 @@ func appendColumnMetric(report mimo.Report, colname string, haserror bool) bool
 			Float64("rate-identifiable", metrics.Identifiant.Rate()).
 			Msg("summmary for column " + colname)
 		haserror = true
+
+		logSamples("coherence", "real-value", "pseudonyms", metrics.GetInvalidSamplesForCoherentRate(10))      //nolint:gomnd
+		logSamples("identifiant", "pseudonym", "real-values", metrics.GetInvalidSamplesForIdentifiantRate(10)) //nolint:gomnd
 	}
 
 	return haserror
 }
 
+func logSamples(target, labelForValue, labelForAssigned string, samples []mimo.Sample) {
+	for _, sample := range samples {
+		lenMax := fmt.Sprintf("%d", len(sample.AssignedValues))
+
+		if len(sample.AssignedValues) > 10 { //nolint:gomnd
+			sample.AssignedValues = sample.AssignedValues[:10]
+		}
+
+		slices.Sort(sample.AssignedValues)
+
+		log.Error().
+			Str(labelForValue, sample.OriginalValue).
+			Strs(labelForAssigned, sample.AssignedValues).
+			Msg("sample value that failed " + target + " because it was attributed " + lenMax + " " + labelForAssigned)
+	}
+}
+
 func initLog() {
 	color := false
 

internal/infra/config_loader.go

@@ -110,7 +110,7 @@ func CreateConfig(yamlconfig *YAMLStructure) (mimo.Config, error) {
 				case "maskingRate":
 					constraint.Target = mimo.MaskingRate
 				case "coherentRate":
-					constraint.Target = mimo.CohenrentRate
+					constraint.Target = mimo.CoherentRate
 				case "identifiantRate":
 					constraint.Target = mimo.IdentifiantRate
 				default:

internal/infra/pebble_multimap.go

@@ -23,6 +23,7 @@ import (
 	"errors"
 	"fmt"
 	"os"
+	"strings"
 
 	"github.com/cgi-fr/mimo/pkg/mimo"
 	"github.com/rs/zerolog/log"
@@ -153,6 +154,70 @@ func (b PebbleMultimapBackend) GetSize(key string) int {
 	return int(count)
 }
 
+func (b PebbleMultimapBackend) GetSamplesMulti(maxlen int) []mimo.Sample {
+	samples := []mimo.Sample{}
+
+	iter, _ := b.db.NewIter(b.prefixIterOptions([]byte(KeyPrefix)))
+
+	for valid := iter.First(); valid; valid = iter.Next() {
+		key := strings.TrimPrefix(string(iter.Key()), KeyPrefix)
+		array := map[string]int{}
+
+		_ = json.Unmarshal(iter.Value(), &array)
+
+		if len(array) > 1 {
+			assignedValues := []string{}
+
+			for assignedValue := range array {
+				assignedValues = append(assignedValues, assignedValue)
+			}
+
+			samples = append(samples, mimo.Sample{
+				OriginalValue:  key,
+				AssignedValues: assignedValues,
+			})
+		}
+
+		if len(samples) == maxlen {
+			break
+		}
+	}
+
+	return samples
+}
+
+func (b PebbleMultimapBackend) GetSamplesMono(maxlen int) []mimo.Sample {
+	samples := []mimo.Sample{}
+
+	iter, _ := b.db.NewIter(b.prefixIterOptions([]byte(KeyPrefix)))
+
+	for valid := iter.First(); valid; valid = iter.Next() {
+		key := strings.TrimPrefix(string(iter.Key()), KeyPrefix)
+		array := map[string]int{}
+
+		_ = json.Unmarshal(iter.Value(), &array)
+
+		if len(array) == 1 {
+			assignedValues := []string{}
+
+			for assignedValue := range array {
+				assignedValues = append(assignedValues, assignedValue)
+			}
+
+			samples = append(samples, mimo.Sample{
+				OriginalValue:  key,
+				AssignedValues: assignedValues,
+			})
+		}
+
+		if len(samples) == maxlen {
+			break
+		}
+	}
+
+	return samples
+}
+
 func (b PebbleMultimapBackend) NewSizeIterator() mimo.SizeIterator { //nolint: ireturn
 	iter, _ := b.db.NewIter(b.prefixIterOptions([]byte(CountPrefix)))
 

pkg/mimo/driven.go

@@ -32,6 +32,8 @@ type MultimapBackend interface {
 	SetKey(key string, value map[string]int) error
 	GetSize(key string) int
 	NewSizeIterator() SizeIterator
+	GetSamplesMono(n int) []Sample
+	GetSamplesMulti(n int) []Sample
 }
 
 type SizeIterator interface {

pkg/mimo/in_memory_multimap.go

@@ -43,6 +43,56 @@ func (m InMemoryMultimapBackend) GetSize(key string) int {
 	return len(m[key])
 }
 
+func (m InMemoryMultimapBackend) GetSamplesMulti(maxlen int) []Sample {
+	samples := []Sample{}
+
+	for value, array := range m {
+		if len(array) > 1 {
+			assignedValues := []string{}
+
+			for assignedValue := range array {
+				assignedValues = append(assignedValues, assignedValue)
+			}
+
+			samples = append(samples, Sample{
+				OriginalValue:  value,
+				AssignedValues: assignedValues,
+			})
+		}
+
+		if len(samples) == maxlen {
+			break
+		}
+	}
+
+	return samples
+}
+
+func (m InMemoryMultimapBackend) GetSamplesMono(maxlen int) []Sample {
+	samples := []Sample{}
+
+	for value, array := range m {
+		if len(array) == 1 {
+			assignedValues := []string{}
+
+			for assignedValue := range array {
+				assignedValues = append(assignedValues, assignedValue)
+			}
+
+			samples = append(samples, Sample{
+				OriginalValue:  value,
+				AssignedValues: assignedValues,
+			})
+		}
+
+		if len(samples) == maxlen {
+			break
+		}
+	}
+
+	return samples
+}
+
 // CountMin returns the minimum count of values associated to a key across the map.
 func (m InMemoryMultimapBackend) NewSizeIterator() SizeIterator { //nolint: ireturn
 	sizes := []int{}

pkg/mimo/model.go

@@ -188,7 +188,7 @@ func (m Metrics) CoherenceRateValidate() int {
 	result := 0
 
 	for _, constraint := range m.Constraints {
-		if constraint.Target == CohenrentRate {
+		if constraint.Target == CoherentRate {
 			if !validate(constraint.Type, constraint.Value, m.Coherence.Rate()) {
 				return -1
 			}
@@ -247,6 +247,72 @@ func (m Metrics) Validate() int {
 	return 0
 }
 
+// GetInvalidSamplesForCoherentRate will return at most n invalid sample if a constraint on coherent rate failed.
+func (m Metrics) GetInvalidSamplesForCoherentRate(maxlen int) []Sample {
+	constraint := m.findFailedCoherentConstraint()
+	samples := []Sample{}
+
+	if constraint != nil {
+		if (constraint.Type == ShouldEqual && constraint.Value > m.Coherence.Rate()) ||
+			constraint.Type == ShouldBeGreaterThan || constraint.Type == ShouldBeGreaterThanOrEqualTo {
+			samples = append(samples, m.Coherence.Backend.GetSamplesMulti(maxlen)...)
+		}
+
+		if (constraint.Type == ShouldEqual && constraint.Value < m.Coherence.Rate()) ||
+			constraint.Type == ShouldBeLessThanOrEqualTo || constraint.Type == ShouldBeLowerThan {
+			samples = append(samples, m.Coherence.Backend.GetSamplesMono(maxlen)...)
+		}
+	}
+
+	return samples
+}
+
+func (m Metrics) findFailedCoherentConstraint() *Constraint {
+	for _, c := range m.Constraints {
+		c := c
+		if c.Target == CoherentRate {
+			if !validate(c.Type, c.Value, m.Coherence.Rate()) {
+				return &c
+			}
+		}
+	}
+
+	return nil
+}
+
+// GetInvalidSamplesForIdentifiantRate will return at most n invalid sample if a constraint on identifiant rate failed.
+func (m Metrics) GetInvalidSamplesForIdentifiantRate(maxlen int) []Sample {
+	constraint := m.findFailedIdentifiantConstraint()
+	samples := []Sample{}
+
+	if constraint != nil {
+		if (constraint.Type == ShouldEqual && constraint.Value > m.Identifiant.Rate()) ||
+			constraint.Type == ShouldBeGreaterThan || constraint.Type == ShouldBeGreaterThanOrEqualTo {
+			samples = append(samples, m.Identifiant.Backend.GetSamplesMulti(maxlen)...)
+		}
+
+		if (constraint.Type == ShouldEqual && constraint.Value < m.Identifiant.Rate()) ||
+			constraint.Type == ShouldBeLessThanOrEqualTo || constraint.Type == ShouldBeLowerThan {
+			samples = append(samples, m.Identifiant.Backend.GetSamplesMono(maxlen)...)
+		}
+	}
+
+	return samples
+}
+
+func (m Metrics) findFailedIdentifiantConstraint() *Constraint {
+	for _, c := range m.Constraints {
+		c := c
+		if c.Target == IdentifiantRate {
+			if !validate(c.Type, c.Value, m.Identifiant.Rate()) {
+				return &c
+			}
+		}
+	}
+
+	return nil
+}
+
 type Report struct {
 	Metrics         map[string]Metrics
 	subs            Suscribers
@@ -451,3 +517,8 @@ func isExcluded(exclude []any, value any, valueStr string) bool {
 
 	return false
 }
+
+type Sample struct {
+	OriginalValue  string
+	AssignedValues []string
+}

pkg/mimo/model_config.go

@@ -46,7 +46,7 @@ type ConstraintTarget int
 
 const (
 	MaskingRate ConstraintTarget = iota
-	CohenrentRate
+	CoherentRate
 	IdentifiantRate
 )
 

test/configs/config_debug_constraint.yaml

@@ -0,0 +1,8 @@
+version: "1"
+metrics:
+  - name: "name"
+    constraints:
+      coherentRate:
+        shouldBeGreaterThan: 0
+      identifiantRate:
+        shouldBeGreaterThan: 0

test/reports/report_debug_constraints_no.html

@@ -0,0 +1,40 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+  <meta charset="UTF-8" />
+  <title>MIMO Report</title>
+  <meta name="viewport" content="width=device-width,initial-scale=1" />
+  <meta name="description" content="MIMO Report" />
+</head>
+<body>
+  <h1>MIMO Report</h1>
+  <table border="1" cellspacing="0" cellpadding="5">
+    <thead>
+      <th>Field</th>
+      <th>Nil</th>
+      <th>Ignored</th>
+      <th>Masked</th>
+      <th>Missed</th>
+      <th>Masking Rate</th>
+      <th>Coherent Rate</th>
+      <th>Identifiable Rate</th>
+      <th>K</th>
+    </thead>
+    <tbody>
+
+      <tr>
+        <td>name</td>
+        <td>0</td>
+        <td>0</td>
+        <td>56</td>
+        <td>44</td>
+        <td style="background-color: orange">56.00 %</td>
+        <td style="background-color: orange">0.00 %</td>
+        <td style="background-color: orange">0.00 %</td>
+        <td style="background-color: lightgreen">2</td>
+      </tr>
+
+    </tbody>
+  </table>
+</body>
+</html>

test/reports/report_debug_constraints_yes.html

@@ -0,0 +1,40 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+  <meta charset="UTF-8" />
+  <title>MIMO Report</title>
+  <meta name="viewport" content="width=device-width,initial-scale=1" />
+  <meta name="description" content="MIMO Report" />
+</head>
+<body>
+  <h1>MIMO Report</h1>
+  <table border="1" cellspacing="0" cellpadding="5">
+    <thead>
+      <th>Field</th>
+      <th>Nil</th>
+      <th>Ignored</th>
+      <th>Masked</th>
+      <th>Missed</th>
+      <th>Masking Rate</th>
+      <th>Coherent Rate</th>
+      <th>Identifiable Rate</th>
+      <th>K</th>
+    </thead>
+    <tbody>
+
+      <tr>
+        <td>name</td>
+        <td>0</td>
+        <td>0</td>
+        <td>56</td>
+        <td>44</td>
+        <td style="background-color: orange">56.00 %</td>
+        <td style="background-color: red">0.00 %</td>
+        <td style="background-color: red">0.00 %</td>
+        <td style="background-color: lightgreen">2</td>
+      </tr>
+
+    </tbody>
+  </table>
+</body>
+</html>