@tricoteuses/senat 3.3.1 → 3.3.2

package/lib/src/utils/scoring.d.ts

@@ -16,6 +16,15 @@ export declare function dice(a: string, b: string): number;
 export declare function coverage(reference?: string | null, candidate?: string | null): number;
 export declare function diceFiltered(a?: string | null, b?: string | null): number;
 export declare function similarityScore(a?: string | null, b?: string | null): number;
+/**
+ * Org gate used by the video matching pipeline.
+ * sameOrg=true when both organes map to the same known key, or when the strings
+ * are (near) identical even if getOrgKey leaves them unmapped ("autre").
+ */
+export declare function computeOrgMatch(agendaOrgane: string, videoOrganes: string[]): {
+    sameOrg: boolean;
+    bestDice: number;
+};
 export declare function normalize(s?: string | null): string;
 export declare function normalizeSalle(s?: string | null): string | null;
 export declare function scoreVideo(agenda: Reunion, agendaTs: number | null, sameOrg: boolean, w: VideoScoreWeights, videoTitle?: string, videoEpoch?: number, videoOrganes?: string[], timeAmbigious?: boolean, salle?: string, chapterTitles?: L1Chapter[]): {

package/lib/src/utils/scoring.js

@@ -182,10 +182,35 @@ export function diceFiltered(a, b) {
     return (2 * inter) / (A.size + B.size);
 }
 export function similarityScore(a, b) {
-    const cov = coverage(a, b);
+    // Symmetric coverage: a short video title entirely contained in a long agenda
+    // title must score high, whichever direction the extra words are.
+    const cov = Math.max(coverage(a, b), coverage(b, a));
     const d = diceFiltered(a, b);
     return 0.7 * cov + 0.3 * d;
 }
+/**
+ * Org gate used by the video matching pipeline.
+ * sameOrg=true when both organes map to the same known key, or when the strings
+ * are (near) identical even if getOrgKey leaves them unmapped ("autre").
+ */
+export function computeOrgMatch(agendaOrgane, videoOrganes) {
+    const agendaOrgNorm = normalize(agendaOrgane);
+    const agendaKey = getOrgKey(agendaOrgNorm);
+    let bestDice = 0;
+    let hasSameKey = false;
+    for (const vo of videoOrganes) {
+        const videoOrgNorm = normalize(vo);
+        const videoKey = getOrgKey(videoOrgNorm);
+        const d = dice(agendaOrgNorm, videoOrgNorm);
+        if (videoKey === agendaKey && videoKey !== "autre")
+            hasSameKey = true;
+        if (d > bestDice)
+            bestDice = d;
+    }
+    if (bestDice >= 0.999)
+        hasSameKey = true;
+    return { sameOrg: hasSameKey, bestDice };
+}
 export function normalize(s) {
     return (s ?? "")
         .toLowerCase()

package/lib/src/videos/match.d.ts

@@ -1,6 +1,11 @@
 import { CommandLineOptions } from "command-line-args";
 import { Reunion } from "../other_types/agenda.js";
-import { Candidate, MatchResult, MatchWeights } from "./types.js";
+import { BestMatch, Candidate, MatchResult, MatchWeights } from "./types.js";
+/**
+ * Sort comparator for ranked matches: higher score first; on (near-)equal scores,
+ * prefer the candidate temporally closest to the agenda start time.
+ */
+export declare function compareRankedBest(a: BestMatch, b: BestMatch, agendaTs: number | null): number;
 export declare function matchOneReunion(args: {
     agenda: Reunion;
     agendaTs: number | null;

package/lib/src/videos/match.js

@@ -1,8 +1,21 @@
 import { isAmbiguousTimeOriginal } from "../utils/date.js";
 import { buildSenatVodMasterM3u8FromNvs, getLevel1Chapters, parseDataNvs } from "../utils/nvs-parsing.js";
-import { dice, getOrgKey, normalize, scoreVideo } from "../utils/scoring.js";
+import { scoreVideo, computeOrgMatch } from "../utils/scoring.js";
 import { SENAT_DATAS_ROOT, weights } from "./config.js";
 import { fetchBuffer } from "./search.js";
+/**
+ * Sort comparator for ranked matches: higher score first; on (near-)equal scores,
+ * prefer the candidate temporally closest to the agenda start time.
+ */
+export function compareRankedBest(a, b, agendaTs) {
+    const diff = b.score - a.score;
+    if (Math.abs(diff) > 1e-9)
+        return diff;
+    if (agendaTs != null && a.epoch != null && b.epoch != null) {
+        return Math.abs(a.epoch - agendaTs) - Math.abs(b.epoch - agendaTs);
+    }
+    return 0;
+}
 export async function matchOneReunion(args) {
     const { agenda, agendaTs, timeAmbigious, candidates, weights, fetchDataNvs, options } = args;
     if (!options["silent"])
@@ -42,25 +55,12 @@ export async function matchOneReunion(args) {
             continue;
         const meta = parseDataNvs(dataStr);
         let sameOrg = false;
-        // Organe gate (same key OR strong dice)
+        // Organe gate (same key, near-identical strings OR strong dice)
         if (agenda.organe && meta.organes?.length) {
-            const agendaOrgNorm = normalize(agenda.organe);
-            const agendaKey = getOrgKey(agendaOrgNorm);
-            let bestDice = 0;
-            let hasSameKey = false;
-            for (const vo of meta.organes) {
-                const videoOrgNorm = normalize(vo);
-                const videoKey = getOrgKey(videoOrgNorm);
-                const d = dice(agendaOrgNorm, videoOrgNorm);
-                if (videoKey === agendaKey && videoKey !== "autre")
-                    hasSameKey = true;
-                if (d > bestDice)
-                    bestDice = d;
-            }
-            if (hasSameKey) {
-                sameOrg = true;
-            }
-            else if (bestDice < orgSkipDice) {
+            const orgMatch = computeOrgMatch(agenda.organe, meta.organes);
+            sameOrg = orgMatch.sameOrg;
+            const bestDice = orgMatch.bestDice;
+            if (!sameOrg && bestDice < orgSkipDice) {
                 continue;
             }
         }
@@ -92,7 +92,7 @@ export async function matchOneReunion(args) {
     }
     if (!ranked.length)
         return null;
-    ranked.sort((a, b) => b.score - a.score);
+    ranked.sort((a, b) => compareRankedBest(a, b, agendaTs));
     const best = ranked[0];
     const second = ranked[1] ?? null;
     // Accept threshold (best must pass)

package/lib/src/videos/pipeline.js

@@ -85,6 +85,11 @@ export async function processOneReunionMatch(args) {
         return;
     }
     const next = { ...obj, urlVideo: master, startTime: agenda.startTime, urlPageVideo: best?.pageUrl };
+    // Only an accepted match proves a video exists: flip captationVideo
+    // (it defaults to false for anything that is not a séance publique).
+    if (best != null) {
+        next.captationVideo = true;
+    }
     if (timecodeDebutVideo != null) {
         next.timecodeDebutVideo = timecodeDebutVideo;
         if (timecodeFinVideo != null)

package/lib/tests/videoPipelineCaptation.test.d.ts

@@ -0,0 +1 @@
+export {};

package/lib/tests/videoPipelineCaptation.test.js

@@ -0,0 +1,100 @@
+// TDD: when a video match is accepted (best != null), processOneReunionMatch must
+// flip captationVideo to true in the agenda JSON (currently it writes urlVideo
+// but leaves captationVideo false, causing urlVideo + captationVideo=false inconsistency).
+import { describe, it, expect, afterEach } from "vitest";
+import fs from "fs-extra";
+import fsp from "fs/promises";
+import os from "os";
+import path from "path";
+import { processOneReunionMatch, writeIfChanged } from "../src/videos/pipeline.js";
+import { buildSenatVodMasterM3u8FromNvs, getAgendaSegmentTimecodes } from "../src/utils/nvs-parsing.js";
+import { AGENDA_FOLDER, DATA_TRANSFORMED_FOLDER } from "../src/server/loaders.js";
+const DATA_NVS = `<data>
+  <metadata name="date" value="1789041600"/>
+  <metadata name="salle" value="Salle Médicis"/>
+  <chapters>
+    <chapter id="1" label="Table ronde d&#39;élus locaux"/>
+    <chapter id="2" label="Clôture"/>
+  </chapters>
+  <serverfiles>serverfiles://senat/2026/09/encoder6_20260909134255.mp4</serverfiles>
+</data>`;
+const FINALPLAYER_NVS = `<player>
+  <synchro id="1" timecode="120000"/>
+  <synchro id="2" timecode="3600000"/>
+</player>`;
+const BEST_MATCH = {
+    id: "5942500",
+    hash: "6aa13cf51c220",
+    pageUrl: "https://videos.senat.fr/video.5942500_6aa13cf51c220.html",
+    epoch: 1789041600,
+    vtitle: "Table ronde d'élus locaux",
+    score: 1.09,
+    m3u8: "https://vodsenat.akamaized.net/senat/2026/09/encoder6_20260909134255.smil/master.m3u8",
+    signals: { titleScore: 0.7, orgScore: 0.1, salleScore: 0, timeScore: 0.2, sameOrg: true, timeAmbigious: false },
+};
+function makeAgenda(captationVideo) {
+    return {
+        uid: "RUSN20260909IDCCOCU559129",
+        chambre: "SN",
+        date: "2026-09-09",
+        startTime: "14:00:00.000Z",
+        endTime: null,
+        captationVideo,
+        titre: "MI Violences périscolaires - Table ronde d'élus locaux",
+        type: "Commissions",
+        organe: "Comité de contrôle",
+        objet: "Table ronde d'élus locaux",
+        events: [],
+    };
+}
+async function setupDataDir(agenda) {
+    const dataDir = await fs.mkdtemp(path.join(os.tmpdir(), "senat-pipeline-"));
+    const agendaJsonPath = path.join(dataDir, AGENDA_FOLDER, DATA_TRANSFORMED_FOLDER, "2025", `${agenda.uid}.json`);
+    await fs.ensureDir(path.dirname(agendaJsonPath));
+    await fsp.writeFile(agendaJsonPath, JSON.stringify(agenda, null, 2));
+    const baseDir = path.join(dataDir, "videos", "2025", agenda.uid);
+    await fs.ensureDir(baseDir);
+    await fsp.writeFile(path.join(baseDir, "data.nvs"), DATA_NVS);
+    await fsp.writeFile(path.join(baseDir, "finalplayer.nvs"), FINALPLAYER_NVS);
+    return { dataDir, agendaJsonPath, baseDir };
+}
+async function runPipeline(args) {
+    const lastByVideo = new Map();
+    await processOneReunionMatch({
+        agenda: args.agenda,
+        best: args.best,
+        baseDir: args.baseDir,
+        dataDir: args.dataDir,
+        session: 2025,
+        options: { silent: true },
+        writeIfChanged,
+        lastByVideo,
+        getAgendaSegmentTimecodes,
+        buildSenatVodMasterM3u8FromNvs,
+    });
+    return JSON.parse(await fsp.readFile(args.agendaJsonPath, "utf-8"));
+}
+describe("processOneReunionMatch — captationVideo", () => {
+    let dirs;
+    afterEach(async () => {
+        if (dirs?.dataDir)
+            await fs.remove(dirs.dataDir);
+    });
+    it("flips captationVideo to true when a match is accepted", async () => {
+        dirs = await setupDataDir(makeAgenda(false));
+        const result = await runPipeline({ ...dirs, agenda: makeAgenda(false), best: BEST_MATCH });
+        expect(result.captationVideo).toBe(true);
+        expect(result.urlVideo).toBe(BEST_MATCH.m3u8);
+        expect(result.urlPageVideo).toBe(BEST_MATCH.pageUrl);
+    });
+    it("does not flip captationVideo when there is no accepted match (best=null)", async () => {
+        dirs = await setupDataDir(makeAgenda(false));
+        const result = await runPipeline({ ...dirs, agenda: makeAgenda(false), best: null });
+        expect(result.captationVideo).toBe(false);
+    });
+    it("keeps captationVideo true when already true and no accepted match", async () => {
+        dirs = await setupDataDir(makeAgenda(true));
+        const result = await runPipeline({ ...dirs, agenda: makeAgenda(true), best: null });
+        expect(result.captationVideo).toBe(true);
+    });
+});

package/lib/tests/videoScoringSimilarity.test.d.ts

@@ -0,0 +1 @@
+export {};

package/lib/tests/videoScoringSimilarity.test.js

@@ -0,0 +1,68 @@
+// TDD: deux artefacts de scoring font rejeter de vrais matchs malgré minAccept=0.9
+// 1) couverture asymétrique: titre vidéo court entièrement contenu dans le titre agenda long
+// 2) organes strictement identiques mais non mappés par getOrgKey ("autre") => sameOrg=false à tort
+import { describe, it, expect } from "vitest";
+import { similarityScore, computeOrgMatch } from "../src/utils/scoring.js";
+import { compareRankedBest } from "../src/videos/match.js";
+function makeBest(overrides) {
+    return {
+        hash: "h",
+        pageUrl: "https://videos.senat.fr/x.html",
+        score: 1,
+        m3u8: "m",
+        signals: { titleScore: 0, orgScore: 0, salleScore: 0, timeScore: 0, sameOrg: true, timeAmbigious: false },
+        ...overrides,
+    };
+}
+describe("compareRankedBest — tie-break par proximité temporelle", () => {
+    const agendaTs = 1000;
+    it("scores égaux: le candidat le plus proche de l'horaire agenda passe en premier", () => {
+        const apresMidi = makeBest({ id: "apres-midi", score: 0.644, epoch: 910 });
+        const soir = makeBest({ id: "soir", score: 0.644, epoch: 600 });
+        expect(compareRankedBest(soir, apresMidi, agendaTs)).toBeGreaterThan(0);
+        expect(compareRankedBest(apresMidi, soir, agendaTs)).toBeLessThan(0);
+    });
+    it("score différent: le meilleur score passe en premier (comportement inchangé)", () => {
+        const haut = makeBest({ id: "haut", score: 0.8, epoch: 600 });
+        const bas = makeBest({ id: "bas", score: 0.6, epoch: 999 });
+        expect(compareRankedBest(haut, bas, agendaTs)).toBeLessThan(0);
+        expect(compareRankedBest(bas, haut, agendaTs)).toBeGreaterThan(0);
+    });
+    it("scores égaux sans epoch: ordre stable", () => {
+        const a = makeBest({ id: "a", score: 0.6 });
+        const b = makeBest({ id: "b", score: 0.6 });
+        expect(compareRankedBest(a, b, null)).toBe(0);
+    });
+});
+describe("similarityScore — couverture symétrique", () => {
+    it("score haut quand le titre vidéo est entièrement contenu dans le titre agenda (ordre des mots quelconque)", () => {
+        const s = similarityScore("Conférence de presse : Universités : conclusions de la commission d'enquête", "Universités : conférence de presse");
+        expect(s).toBeCloseTo(0.9, 5);
+    });
+    it("cas conférence de presse Prélèvements obligatoires >= 0.9", () => {
+        const s = similarityScore("Conférence de presse : Prélèvements obligatoires sur les entreprises", "Prélèvements obligatoires : conférence de presse");
+        expect(s).toBeGreaterThanOrEqual(0.9);
+    });
+    it("restes ~0 pour des textes sans rapport", () => {
+        const s = similarityScore("Examen du projet de rapport de la commission d'enquête", "Audition de la brigade de protection des mineurs de Paris");
+        expect(s).toBeLessThan(0.3);
+    });
+});
+describe("computeOrgMatch", () => {
+    it("organes identiques non mappés par getOrgKey => sameOrg=true", () => {
+        const r = computeOrgMatch("MI Prélèvements obligatoires", ["MI Prélèvements obligatoires"]);
+        expect(r.sameOrg).toBe(true);
+        expect(r.bestDice).toBe(1);
+    });
+    it("même clé d'organe => sameOrg=true", () => {
+        const r = computeOrgMatch("Commission des lois", ["Commission des lois constitutionnelles"]);
+        expect(r.sameOrg).toBe(true);
+    });
+    it("organes sans rapport => sameOrg=false", () => {
+        const r = computeOrgMatch("CE Universités", [
+            "Commission de la culture, de l'éducation, de la communication et du sport",
+        ]);
+        expect(r.sameOrg).toBe(false);
+        expect(r.bestDice).toBeLessThan(0.8);
+    });
+});

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@tricoteuses/senat",
-  "version": "3.3.1",
+  "version": "3.3.2",
   "description": "Handle French Sénat's open data",
   "keywords": [
     "France",