feat: sankey diagram WIP

Mermaid ne supporte pas les lien circulaire ce qui rend l'usage impossible pour le moment

.gitignore

@@ -16,4 +16,6 @@
 dist
 test-coverage
 node_modules
+# outptu stats files
 el-stats*.json
+el-stats*.txt

src/index.ts

@@ -13,6 +13,8 @@ import { statsGeneralesDesc } from "./statistiques/v2/generales/StatsGenerales";
 import { computeStatsSociales } from "./statistiques/v2/sociales/computeStatsSociales";
 import { statsSocialesDesc } from "./statistiques/v2/sociales/StatsSociales";
 import { checkDbSchemas } from "./notion/fetch/schemaCheck/checkDbSchemas";
+import { computeSequencEvtPenalSankeyData } from "./statistiques/v2/penales/computeSankeyData";
+import { sankeyDataToMermaidDiagram } from "./statistiques/v2/sankey/sankeyDataToMermaidDiagram";
 
 type ProcessOptions = {
   dryRun: boolean;
@@ -82,6 +84,11 @@ function buildProcessOptions(): ProcessOptions {
   const statsPenales = computeStatsPenales(familles);
   const statsSociales = computeStatsSociales(familles);
 
+  console.log("Sankey Diagram:");
+  writeFileSync(
+    "./el-stats-penal-sankey-diagram.txt",
+    sankeyDataToMermaidDiagram(computeSequencEvtPenalSankeyData(familles))
+  );
   if (options.dryRun) {
     console.log(
       "Dry run => Skip Publishing. Stats are dumped in file el-stats-xxx.json"

src/statistiques/v2/penales/computeSankeyData.ts

@@ -1,3 +1,4 @@
+import { EvenementFamille } from "../../../data/EvenementFamille";
 import { Famille } from "../../../data/Famille";
 import { SankeyData } from "../sankey/SankeyData";
 export function computeSequencEvtPenalSankeyData(familles: Famille[]) {
@@ -5,15 +6,35 @@ export function computeSequencEvtPenalSankeyData(familles: Famille[]) {
 
   familles.forEach((f) => {
     // Compute all transitions, Events are already sorted
-    for (let index = 0; index < f.Evenements.length - 1; index++) {
-      const fromEvt = f.Evenements[index];
-      const toEvt = f.Evenements[index + 1];
-      const fromTransitions = sankeyData[fromEvt.Type] || {};
-      sankeyData[fromEvt.Type] = {
-        ...fromTransitions,
-        [toEvt.Type]: (fromTransitions[toEvt.Type] || 0) + 1,
+    const nonNullEvents = f.Evenements.filter(
+      (evt) => evt.Type !== null && (evt.Type as string) !== "null"
+    );
+
+    // remove duplicate types to avoid circular links that are not supported by mermaid
+    const evenements = removeDuplicateTypes(nonNullEvents);
+
+    for (let index = 0; index < evenements.length - 1; index++) {
+      const sourceEvt = evenements[index];
+      const targetEvt = evenements[index + 1];
+      const sourceSankeyData = sankeyData[sourceEvt.Type] || {};
+      sankeyData[sourceEvt.Type] = {
+        ...sourceSankeyData,
+        [targetEvt.Type]: (sourceSankeyData[targetEvt.Type] || 0) + 1,
       };
     }
   });
   return sankeyData;
 }
+
+function removeDuplicateTypes(events: EvenementFamille[]): EvenementFamille[] {
+  const seenTypes: { [key: string]: boolean } = {};
+  return events.filter((evt) => {
+    const type = evt.Type as string;
+    if (seenTypes[type]) {
+      return false;
+    } else {
+      seenTypes[type] = true;
+      return true;
+    }
+  });
+}

src/statistiques/v2/sankey/SankeyData.ts

@@ -2,7 +2,7 @@
  * Sankey diagram transitions
  */
 export type SankeyData = {
-  [fromKey: string]: {
-    [toKey: string]: number;
+  [source: string]: {
+    [target: string]: number;
   };
 };

src/statistiques/v2/sankey/sankeyDataToMermaidDiagram.ts

@@ -1,5 +1,44 @@
 import { SankeyData } from "./SankeyData";
 
-eport function sankeyDataToMermaidDiagram(sankeyData: SankeyData) {
-    
-}
+export type SankeyDiagramOptions = { showValues: boolean };
+export function sankeyDataToMermaidDiagram(
+  sankeyData: SankeyData,
+  options: SankeyDiagramOptions = { showValues: true }
+) {
+  const header = `---
+config:
+  sankey:
+    showValues: ${options.showValues}
+---
+sankey-beta
+`;
+  return header + arrayToCsv(dataLines(sankeyData));
+}
+
+function dataLines(sankeyData: SankeyData): Array<[string, string, number]> {
+  return Object.entries(sankeyData).flatMap(([source, targets]) => {
+    return Object.entries(targets).map<[string, string, number]>(
+      ([target, count]) => {
+        return [source, target, count];
+      }
+    );
+  });
+}
+
+function arrayToCsv(data: Array<Array<string | number>>): string {
+  return data.map((dataLine) => dataLine.map(toCsvValue).join(",")).join("\n");
+}
+
+function toCsvValue(value: string | number): string {
+  if (typeof value === "number") {
+    return value.toString();
+  } else {
+    // Mermaid does not seem to support diacritics
+    const withoutDiacritics = value
+      .normalize("NFD")
+      .replace(/[\u0300-\u036f]/g, "");
+
+    const escaped = withoutDiacritics.replaceAll('"', '""');
+    return `"${escaped}"`;
+  }
+}