@tricoteuses/senat 2.15.6 → 2.16.0

package/lib/datasets.js

@@ -220,7 +220,6 @@ export const datasets = {
             ],
             votsen: [
                 { name: "idx_scrnum", columns: ["scrnum"] },
-                { name: "idx_sesann", columns: ["sesann"] },
                 { name: "idx_titsencod", columns: ["titsencod"] },
                 { name: "idx_stavotidt", columns: ["stavotidt"] },
                 { name: "idx_posvotcod", columns: ["posvotcod"] },

package/lib/model/agenda.js

@@ -7,9 +7,7 @@ function eventIsSeance(eventElement) {
     return eventElement.classList.contains("evt-seance");
 }
 function getEventType(eventClasses) {
-    const typeClass = [...eventClasses]
-        .find(className => className.startsWith("evt-"))
-        || null;
+    const typeClass = [...eventClasses].find((className) => className.startsWith("evt-")) || null;
     switch (typeClass) {
         case "evt-seance":
             return "Séance publique";
@@ -25,8 +23,7 @@ function getEventType(eventClasses) {
     return null;
 }
 function getUrlDossierSenat(lienElements) {
-    const urlElement = [...lienElements]
-        .find(lienElement => lienElement.textContent?.includes("dossier législatif"));
+    const urlElement = [...lienElements].find((lienElement) => lienElement.textContent?.includes("dossier législatif"));
     return urlElement ? urlElement.getAttribute("href") : null;
 }
 function getQuantieme(eventElement, seancesElements) {
@@ -60,17 +57,16 @@ function normalizeTime(timeStr) {
         ?.replace(/^(?:l')?après-midi/i, "16h00") // We chose "après-midi" to mean 16h00
         ?.replace(/^(?:le )?soir/i, "20h00") // We chose "soir" to mean 20h00
         ?.replace(/^(?:la )?nuit/i, "22h00") // We chose "nuit" to mean 22h00
-        ?.replace(/^à\s/ig, "")
-        ?.replace(/heures/ig, "h00")
+        ?.replace(/^à\s/gi, "")
+        ?.replace(/heures/gi, "h00")
         ?.replace(/\set.*/i, "")
         ?.replace(/,.*/, "")
         ?.replace(/\s\(hors hémicycle\)/i, "")
-        ?.replace(/\s*h\s*/ig, "h");
+        ?.replace(/\s*h\s*/gi, "h");
 }
 function getStartAndEndTimes(timeStr) {
     const normalizedTime = normalizeTime(timeStr);
-    const timeMatches = normalizedTime
-        ?.match(/^de (?<startTime>\d{2}h\d{2}) à (?<endTime>\d{2}h\d{2})$/i);
+    const timeMatches = normalizedTime?.match(/^de (?<startTime>\d{2}h\d{2}) à (?<endTime>\d{2}h\d{2})$/i);
     if (timeMatches?.groups) {
         const { startTime, endTime } = timeMatches.groups;
         return {
@@ -86,7 +82,7 @@ function getStartAndEndTimes(timeStr) {
 function transformAgenda(document, fileName) {
     const agendaEvents = [];
     const eventElements = document.querySelectorAll(".evt");
-    const seanceElements = Array.from(eventElements).filter(eventElement => eventIsSeance(eventElement));
+    const seanceElements = Array.from(eventElements).filter((eventElement) => eventIsSeance(eventElement));
     for (const eventElement of eventElements) {
         const id = eventElement.previousElementSibling?.getAttribute("name") || null;
         if (!id) {
@@ -96,12 +92,9 @@ function transformAgenda(document, fileName) {
         const date = DateTime.fromFormat(fileName, ID_DATE_FORMAT).toFormat(STANDARD_DATE_FORMAT);
         const timeOriginal = eventElement.querySelector(".time")?.textContent || null;
         const { startTime, endTime } = getStartAndEndTimes(timeOriginal);
-        const titre = eventElement.querySelector(".titre")?.textContent?.trim() || null;
+        const titre = eventElement.querySelector(".titre")?.textContent?.trim() || "";
         const organe = eventElement.querySelector(".organe")?.textContent?.trim() || null;
-        const objet = eventElement.querySelector(".objet")?.textContent
-            ?.trim()
-            ?.replace(/^- /, "")
-            || null;
+        const objet = eventElement.querySelector(".objet")?.textContent?.trim()?.replace(/^- /, "") || null;
         const lieu = eventElement.querySelector(".lieu")?.textContent || null;
         const videoElement = eventElement.querySelector(".video");
         const urlDossierSenat = getUrlDossierSenat(eventElement.querySelectorAll(".lien a"));

package/lib/model/commission.d.ts

@@ -2,8 +2,16 @@ import * as cheerio from "cheerio";
 import { CompteRendu } from "../types/compte_rendu";
 import { GroupedReunion } from "../types/agenda";
 export declare function getRemainingTextAfterSpeakerHeader($: cheerio.CheerioAPI, $p: cheerio.Cheerio<any>): string;
-export declare function parseCommissionCRFromFile(htmlFilePath: string, best?: GroupedReunion, fallback?: {
+export type DaySection = {
+    title: string;
+    $start: cheerio.Cheerio<any>;
+};
+export declare function cleanTitle(t: string): string;
+export declare function extractDayH3Sections($: cheerio.CheerioAPI, dateISO: string): DaySection[];
+export declare function parseCommissionCRSectionFromDom($: cheerio.CheerioAPI, htmlFilePath: string, opts: {
     dateISO: string;
     hourShort: string | null;
     organe?: string | null;
+    section: DaySection;
+    matched?: GroupedReunion;
 }): CompteRendu | null;

package/lib/model/commission.js

@@ -1,6 +1,5 @@
 import * as cheerio from "cheerio";
 import path from "path";
-import fs from "fs";
 import { norm, toCRDate } from "./util";
 import { makeTypeGroupUid } from "../utils/reunion_grouping";
 import { frDateToISO, hourShortToStartTime } from "../utils/cr_spliting";
@@ -162,47 +161,68 @@ function buildPointsFromParagraphs($, paras) {
     flush();
     return points;
 }
-export function parseCommissionCRFromFile(htmlFilePath, best, fallback) {
-    try {
-        if (!best && !fallback) {
-            console.warn(`[COM-CR][parse] missing both 'best' and 'fallback' for ${path.basename(htmlFilePath)}`);
-            return null;
+export function cleanTitle(t) {
+    return (t || "").replace(/\s+/g, " ").trim();
+}
+export function extractDayH3Sections($, dateISO) {
+    const sections = [];
+    const $dayRoot = findDayRoot($, dateISO);
+    if ($dayRoot.length === 0)
+        return sections;
+    let $cursor = $dayRoot.next();
+    while ($cursor.length && !$cursor.is("h2")) {
+        if ($cursor.is("h3")) {
+            const title = cleanTitle($cursor.text());
+            if (title)
+                sections.push({ title, $start: $cursor });
+            $cursor = $cursor.next();
+            continue;
+        }
+        const $h3 = $cursor.find("h3").first();
+        if ($h3.length) {
+            const title = cleanTitle($h3.text());
+            if (title)
+                sections.push({ title, $start: $h3 });
+            $cursor = $cursor.next();
+            continue;
         }
-        const raw = fs.readFileSync(htmlFilePath, "utf8");
-        const $ = cheerio.load(raw, { xmlMode: false });
-        const dateISO = best?.date ?? fallback.dateISO;
-        const startTime = best?.startTime ?? hourShortToStartTime(fallback.hourShort);
-        const organe = best?.organe ?? fallback?.organe ?? undefined;
-        const seanceRef = best?.uid ?? makeTypeGroupUid(dateISO, "COM", fallback.hourShort ?? "NA", organe);
+        $cursor = $cursor.next();
+    }
+    return sections;
+}
+export function parseCommissionCRSectionFromDom($, htmlFilePath, opts) {
+    try {
+        const { dateISO, hourShort, organe, section, matched } = opts;
+        const seanceRef = matched?.uid ?? makeTypeGroupUid(dateISO, "COM", matched?.events[0].id ?? hourShort ?? "", organe ?? undefined);
         const uid = seanceRef.replace(/^RU/, "CRC");
-        const dateSeance = toCRDate(dateISO, startTime);
+        const dateSeance = toCRDate(dateISO, matched?.startTime ?? hourShortToStartTime(hourShort));
         const $dayRoot = findDayRoot($, dateISO);
         if ($dayRoot.length === 0) {
             console.warn(`[COM-CR][parse] day root not found for ${dateISO} in ${path.basename(htmlFilePath)}`);
             return null;
         }
-        // --- Collect paragraphes/h3 until next h2 ---
-        const dayParas = [];
-        let $cursor = $dayRoot.next();
-        while ($cursor.length && !$cursor.is("h2")) {
-            if ($cursor.is("h3"))
-                dayParas.push($cursor);
+        const paras = [];
+        let $cursor = section.$start;
+        // Jump title if we do not want to add it to paragraphes
+        $cursor = $cursor.next();
+        while ($cursor.length && !$cursor.is("h2") && !$cursor.is("h3")) {
             if ($cursor.is(PARA_h3_SEL)) {
-                dayParas.push($cursor);
+                paras.push($cursor);
             }
             else {
                 const $ps = $cursor.find(PARA_h3_SEL);
                 if ($ps.length)
                     $ps.each((_, p) => {
-                        dayParas.push($(p));
+                        paras.push($(p));
                     });
             }
             $cursor = $cursor.next();
         }
-        // --- points ---
-        const allDayPoints = buildPointsFromParagraphs($, dayParas);
-        const points = allDayPoints.length > 0 ? allDayPoints : [];
-        // --- session ---
+        const points = buildPointsFromParagraphs($, paras);
+        if (points.length < 4) {
+            console.warn(`[COM-CR][parse] Insufficient points found for section="${section.title}" date=${dateISO} in ${path.basename(htmlFilePath)}`);
+            return null;
+        }
         const session = dateISO.slice(5, 7) >= "10" ? `${dateISO.slice(0, 4)}` : `${Number(dateISO.slice(0, 4)) - 1}`;
         const contenu = {
             quantiemes: { journee: dateISO, session },
@@ -224,16 +244,10 @@ export function parseCommissionCRFromFile(htmlFilePath, best, fallback) {
             environnement: "prod",
             heureGeneration: new Date(),
         };
-        return {
-            uid,
-            seanceRef,
-            sessionRef: session,
-            metadonnees,
-            contenu,
-        };
+        return { uid, seanceRef, sessionRef: session, metadonnees, contenu };
     }
     catch (e) {
-        console.error(`[COM-CR][parse] error file=${path.basename(htmlFilePath)}:`, e);
+        console.error(`[COM-CR][parse] error section file=${path.basename(htmlFilePath)}:`, e);
         return null;
     }
 }

package/lib/model/scrutins.js

@@ -1,12 +1,13 @@
 import { jsonArrayFrom } from "kysely/helpers/postgres";
 import { dbSenat } from "../databases";
 import { rtrim, toDateString } from "./util";
-function votes(scrutinNum, scrutinSession, scrutinDate) {
+function votes(scrutinNum, scrutinDate) {
     return jsonArrayFrom(dbSenat
         .selectFrom("dosleg.votsen")
         .leftJoin("dosleg.titsen", "dosleg.titsen.titsencod", "dosleg.votsen.titsencod")
         .leftJoin("dosleg.stavot", "dosleg.stavot.stavotidt", "dosleg.votsen.stavotidt")
         .leftJoin("dosleg.posvot", "dosleg.posvot.posvotcod", "dosleg.votsen.posvotcod")
+        .leftJoin("sens.sen", "dosleg.votsen.senmat", "sens.sen.senmat")
         .leftJoin("sens.memgrppol", (join) => join
         .onRef("sens.memgrppol.senmat", "=", "dosleg.votsen.senmat")
         .onRef("sens.memgrppol.memgrppoldatdeb", "<=", scrutinDate)
@@ -15,7 +16,7 @@ function votes(scrutinNum, scrutinSession, scrutinDate) {
         eb("sens.memgrppol.memgrppoldatfin", "is", null)
     ])))
         .where("dosleg.votsen.scrnum", "=", scrutinNum)
-        .where("dosleg.votsen.sesann", "=", scrutinSession)
+        .where("sens.sen.etasencod", "=", "ACTIF")
         .select([
         "dosleg.votsen.senmat as matricule_votant",
         "dosleg.votsen.senmatdel as matricule_delegant",
@@ -64,7 +65,7 @@ const findAllScrutinsQuery = dbSenat
     "scr.scrconsea as nombre_contre_seance",
     "scr.scrpou as nombre_pour",
     "scr.scrpousea as nombre_pour_seance",
-    votes(ref("scr.scrnum"), ref("scr.sesann"), ref("scr.scrdat")).as("votes"),
+    votes(ref("scr.scrnum"), ref("scr.scrdat")).as("votes"),
     misesAuPoint(ref("scr.scrnum")).as("mises_au_point"),
 ])
     .$narrowType();

package/lib/model/seance.js

@@ -2,7 +2,7 @@ import fs from "fs";
 import * as cheerio from "cheerio";
 import path from "path";
 import { computeIntervalsBySlot } from "../utils/cr_spliting";
-import { norm, toCRDate } from "./util";
+import { decodeHtmlEntities, norm, toCRDate } from "./util";
 const asArray = (x) => (x == null ? [] : Array.isArray(x) ? x : [x]);
 const toInt = (s) => (Number.isFinite(Number(s)) ? Number(s) : Number.POSITIVE_INFINITY);
 export async function parseCompteRenduSlotFromFile(xmlFilePath, wantedSlot, firstSlotOfDay) {
@@ -166,11 +166,6 @@ function dedupeSpeaker(raw) {
     }
     return s.replace(/\.\s*$/, "");
 }
-function decodeHtmlEntities(s) {
-    return s
-        .replace(/&#(\d+);/g, (_, d) => String.fromCharCode(parseInt(d, 10)))
-        .replace(/&#x([0-9a-fA-F]+);/g, (_, h) => String.fromCharCode(parseInt(h, 16)));
-}
 function fixApostrophes(s) {
     let out = s;
     out = out.replace(/\s*’\s*/g, "’");

package/lib/model/util.d.ts

@@ -8,3 +8,6 @@ export declare function rtrim(expr: Expression<string | null | undefined>): impo
 export declare function toDateString(expr: Expression<Date | null | undefined>, format?: Expression<string>): import("kysely").RawBuilder<string>;
 export declare function norm(s?: string | null): string;
 export declare function toCRDate(dateISO: string, startTime?: string | null): string;
+export declare function normalizeTitle(t: string): string;
+export declare function jaccardTokenSim(a: string, b: string): number;
+export declare function decodeHtmlEntities(s?: string | null): string;

package/lib/model/util.js

@@ -42,3 +42,35 @@ export function toCRDate(dateISO, startTime) {
     }
     return `${yyyymmdd}${hh}${mm}${ss}${SSS}`;
 }
+export function normalizeTitle(t) {
+    return (t || "")
+        .toLowerCase()
+        .normalize("NFD")
+        .replace(/\p{Diacritic}/gu, "")
+        .replace(/[^a-z0-9\s]/g, " ")
+        .replace(/\s+/g, " ")
+        .trim();
+}
+export function jaccardTokenSim(a, b) {
+    const A = new Set(normalizeTitle(a).split(" ").filter(Boolean));
+    const B = new Set(normalizeTitle(b).split(" ").filter(Boolean));
+    if (A.size === 0 || B.size === 0)
+        return 0;
+    let inter = 0;
+    for (const x of A)
+        if (B.has(x))
+            inter++;
+    return inter / (A.size + B.size - inter);
+}
+export function decodeHtmlEntities(s) {
+    if (!s)
+        return "";
+    return s
+        .replace(/&#x([0-9a-fA-F]+);/g, (_, h) => String.fromCodePoint(parseInt(h, 16)))
+        .replace(/&#(\d+);/g, (_, d) => String.fromCodePoint(parseInt(d, 10)))
+        .replace(/&/g, "&")
+        .replace(/&lt;/g, "<")
+        .replace(/&gt;/g, ">")
+        .replace(/&quot;/g, '"')
+        .replace(/&apos;/g, "'");
+}

package/lib/scripts/retrieve_cr_commission.js

@@ -4,12 +4,13 @@ import path from "path";
 import * as cheerio from "cheerio";
 import { COMMISSION_FOLDER, DATA_ORIGINAL_FOLDER, DATA_TRANSFORMED_FOLDER } from "../loaders";
 import { loadAgendaForDate, parseCommissionMetadataFromHtml, linkCRtoCommissionGroup } from "../utils/cr_spliting";
-import { parseCommissionCRFromFile } from "../model/commission";
+import { cleanTitle, extractDayH3Sections, parseCommissionCRSectionFromDom } from "../model/commission";
 import commandLineArgs from "command-line-args";
 import { commonOptions } from "./shared/cli_helpers";
 import { sessionStartYearFromDate } from "../model/seance";
 import { getSessionsFromStart } from "../types/sessions";
 import { ensureAndClearDir, fetchWithRetry } from "./shared/util";
+import { jaccardTokenSim } from "../model/util";
 class CommissionCRDownloadError extends Error {
     constructor(message, url) {
         super(`An error occurred while retrieving Commission CR ${url}: ${message}`);
@@ -27,35 +28,35 @@ const optionsDefinitions = [
 ];
 const options = commandLineArgs(optionsDefinitions);
 const COMMISSION_HUBS = {
-    "affaires-etrangeres": [
+    "Commission des affaires étrangères": [
         "https://www.senat.fr/compte-rendu-commissions/affaires-etrangeres.html",
         "https://www.senat.fr/compte-rendu-commissions/affaires-etrangeres_archives.html",
     ],
-    "affaires-economiques": [
+    "Commission des affaires économiques": [
         "https://www.senat.fr/compte-rendu-commissions/economie.html",
         "https://www.senat.fr/compte-rendu-commissions/economie_archives.html",
     ],
-    "amenagement-developpement-durable": [
+    "Commission de l'amenagement du territoire et du développement durable": [
         "https://www.senat.fr/compte-rendu-commissions/developpement-durable.html",
         "https://www.senat.fr/compte-rendu-commissions/developpement-durable_archives.html",
     ],
-    culture: [
+    "Commission de la culture": [
         "https://www.senat.fr/compte-rendu-commissions/culture.html",
         "https://www.senat.fr/compte-rendu-commissions/culture_archives.html",
     ],
-    finances: [
+    "Commission des finances": [
         "https://www.senat.fr/compte-rendu-commissions/finances.html",
         "https://www.senat.fr/compte-rendu-commissions/finances_archives.html",
     ],
-    lois: [
+    "Commission des lois": [
         "https://www.senat.fr/compte-rendu-commissions/lois.html",
         "https://www.senat.fr/compte-rendu-commissions/lois_archives.html",
     ],
-    "affaires-sociales": [
+    "Commission des affaires sociales": [
         "https://www.senat.fr/compte-rendu-commissions/affaires-sociales.html",
         "https://www.senat.fr/compte-rendu-commissions/affaires-sociales_archives.html",
     ],
-    "affaires-europeennes": [
+    "Commission des affaires européennes": [
         "https://www.senat.fr/compte-rendu-commissions/affaires-europeennes.html",
         "https://www.senat.fr/compte-rendu-commissions/affaires-europeennes_archives.html",
     ],
@@ -170,6 +171,15 @@ function timeProximityScore(h, openHHMM, maxDeltaMin) {
         return 0;
     return 1 - d / maxDeltaMin; // 0..1 (1 = même heure)
 }
+function titleSimilarity(reunion, sectionTitle) {
+    const t = reunion.titre ?? "";
+    const o = reunion.objet ?? "";
+    if (!sectionTitle.trim())
+        return 0;
+    const sTit = jaccardTokenSim(t, sectionTitle);
+    const sObj = jaccardTokenSim(o, sectionTitle);
+    return Math.max(sTit, sObj);
+}
 async function retrieveCommissionCRs(options = {}) {
     const dataDir = options["dataDir"];
     const fromSession = Number(options["fromSession"]);
@@ -245,86 +255,94 @@ async function retrieveCommissionCRs(options = {}) {
             for (const f of htmlFiles) {
                 const htmlPath = path.join(commissionDir, f);
                 let meta;
+                let raw = "";
                 try {
-                    const raw = await fs.readFile(htmlPath, "utf8");
+                    raw = await fs.readFile(htmlPath, "utf8");
                     meta = parseCommissionMetadataFromHtml(raw, f);
                 }
                 catch (e) {
                     console.warn(`[COM-CR][PRE][${session}] Cannot read/parse ${f}:`, e);
                     continue;
                 }
-                const organeLabel = meta?.organeTitleRaw;
-                for (const day of meta.days ?? []) {
+                if (!meta?.days?.length)
+                    continue;
+                const $ = cheerio.load(raw, { xmlMode: false });
+                for (const day of meta.days) {
                     const yyyymmdd = day.date.replace(/-/g, "");
                     const dt = new Date(Number(day.date.slice(0, 4)), Number(day.date.slice(5, 7)) - 1, Number(day.date.slice(8, 10)));
                     const daySession = sessionStartYearFromDate(dt);
                     let hits = await loadAgendaForDate(dataDir, yyyymmdd, daySession);
-                    let best = null;
-                    let reason = "fallback-none";
-                    let deltaMin;
-                    // gate + scoring combined
+                    console.log(`[COM-CR][TRANSFORM] ${f} → ${hits.length} agenda events on ${day.date} :`);
+                    const sections = extractDayH3Sections($, day.date);
+                    if (sections.length === 0) {
+                        console.warn(`[COM-CR][TRANSFORM] no sections found for ${f} on ${day.date}, skipping.`);
+                        continue;
+                    }
                     const MAX_TIME_DELTA_MIN = 120;
-                    const ORGANE_GATE = 0.55; // minimum similarity organe to be considered
-                    const W_ORG = 0.7, W_TIM = 0.3;
-                    if (hits.length) {
-                        // 1) Gate organe : only keep those above gate, then score with combined organe+time
-                        const gated = hits
-                            .map((h) => {
-                            const sOrg = organeSimilarity(h, commissionKey); // 0..1
-                            const sTim = timeProximityScore(h, day.openTime ?? null, MAX_TIME_DELTA_MIN); // 0..1
-                            const total = W_ORG * sOrg + +W_TIM * sTim;
-                            return { h, sOrg, sTim, total };
-                        })
-                            .filter((x) => x.sOrg >= ORGANE_GATE)
-                            .sort((a, b) => b.total - a.total);
-                        if (gated[0]) {
-                            best = gated[0].h;
-                            reason = gated[0].sOrg >= ORGANE_GATE ? "organe" : "fallback-none";
-                            if (day.openTime && best?.startTime) {
-                                deltaMin = Math.abs(timeToMinutes(best.startTime) - timeToMinutes(day.openTime));
+                    const ORGANE_GATE = 0.55;
+                    const TITLE_GATE = 0.2;
+                    const W_ORG = 0.5;
+                    const W_TIM = 0.2;
+                    const W_TIT = 0.3;
+                    for (let sIdx = 0; sIdx < sections.length; sIdx++) {
+                        const sec = sections[sIdx];
+                        let best = null;
+                        let reason = "fallback-none";
+                        if (hits.length) {
+                            const scored = hits
+                                .map((h) => {
+                                const sOrg = organeSimilarity(h, commissionKey); // 0..1
+                                const sTim = timeProximityScore(h, day.openTime ?? null, MAX_TIME_DELTA_MIN); // 0..1
+                                const sTit = titleSimilarity(h, sec.title); // 0..1
+                                const total = W_ORG * sOrg + W_TIM * sTim + W_TIT * sTit;
+                                return { h, sOrg, sTim, sTit, total };
+                            })
+                                .filter((x) => x.sOrg >= ORGANE_GATE && x.sTit >= TITLE_GATE)
+                                .sort((a, b) => b.total - a.total);
+                            if (scored[0]) {
+                                best = scored[0].h;
+                                reason =
+                                    scored[0].sTit >= Math.max(scored[0].sOrg, scored[0].sTim)
+                                        ? "title"
+                                        : scored[0].sOrg >= scored[0].sTim
+                                            ? "organe"
+                                            : "time";
                             }
                         }
+                        const hourShort = toHourShort(day.openTime) ?? "NA";
+                        const cr = parseCommissionCRSectionFromDom($, htmlPath, {
+                            dateISO: day.date,
+                            hourShort,
+                            organe: commissionKey,
+                            section: sec,
+                            matched: best ?? undefined,
+                        });
+                        if (!cr) {
+                            console.warn(`[COM-CR][TRANSFORM] parse failed for section#${sIdx} ${path.basename(htmlPath)} → ${best ? best.uid : "NO-GROUP"} (${commissionKey})`);
+                            continue;
+                        }
+                        const fileUid = cr.uid;
+                        const transformedSessionDir = path.join(transformedRoot, String(daySession));
+                        fs.ensureDirSync(transformedSessionDir);
+                        const outPath = path.join(transformedSessionDir, `${fileUid}.json`);
+                        await fs.writeJSON(outPath, cr, { spaces: 2 });
+                        const titreGuess = cleanTitle(sections[sIdx].title) || "Commission du " + day.date;
+                        const up = await linkCRtoCommissionGroup({
+                            dataDir,
+                            dateISO: day.date,
+                            organeDetected: commissionKey,
+                            hourShort,
+                            crUid: fileUid,
+                            titreGuess,
+                            groupUid: best ? best.uid : undefined,
+                        });
+                        totalFiles++;
+                        if (up.created || up.updated)
+                            linkedFiles++;
                         else {
-                            best = null;
-                            reason = "fallback-none";
+                            console.warn(`[COM-CR][AGENDA][WARN] CR ${fileUid} (section#${sIdx}) not linked (reason=${reason})`);
                         }
                     }
-                    // Parse CR
-                    const hourShort = toHourShort(day.openTime) ?? "NA";
-                    const cr = parseCommissionCRFromFile(htmlPath, best ?? undefined, {
-                        dateISO: day.date,
-                        hourShort,
-                        organe: commissionKey,
-                    });
-                    if (!cr) {
-                        console.warn(`[COM-CR][TRANSFORM] parse failed for ${f} → ${best ? best.uid : "NO-GROUP"} (${commissionKey})`);
-                        continue;
-                    }
-                    const fileUid = cr.uid;
-                    const transformedSessionDir = path.join(transformedRoot, String(daySession));
-                    fs.ensureDirSync(transformedSessionDir);
-                    const outPath = path.join(transformedSessionDir, `${fileUid}.json`);
-                    await fs.writeJSON(outPath, cr, { spaces: 2 });
-                    const npts = Array.isArray(cr.contenu.point) ? cr.contenu.point.length : cr.contenu.point ? 1 : 0;
-                    if (!options["silent"]) {
-                        console.log(`[COM-CR][TRANSFORM] saved ${path.basename(outPath)} (points=${npts}) [${commissionKey}]`);
-                    }
-                    const titreGuess = organeLabel || "Commission";
-                    const up = await linkCRtoCommissionGroup({
-                        dataDir,
-                        dateISO: day.date,
-                        organeDetected: best?.organe ?? null,
-                        hourShort,
-                        crUid: fileUid,
-                        titreGuess,
-                        groupUid: best ? best.uid : undefined,
-                    });
-                    totalFiles++;
-                    if (up.created || up.updated)
-                        linkedFiles++;
-                    else {
-                        console.warn(`[COM-CR][AGENDA][WARN] CR ${fileUid} not linked to any agenda group (reason=${reason}, delta=${deltaMin ?? "NA"}m)`);
-                    }
                 }
             }
             if (!options["silent"]) {

package/lib/scripts/retrieve_videos.d.ts

@@ -1,6 +1,4 @@
 import { GroupedReunion } from "../types/agenda";
-export declare function buildSenatVodMasterM3u8FromNvs(nvsText: string, finalText: string): string | null;
-export declare function score(agenda: GroupedReunion, agendaTs: number | null, videoTitle?: string, videoEpoch?: number): number;
 /**
  * Build search strategies for senat's videos
  */