@tricoteuses/senat 2.21.0 → 2.21.2

package/lib/src/scripts/retrieve_videos.d.ts

@@ -1,33 +1 @@
-import { Reunion } from "../types/agenda";
-export interface SearchParams {
-    search: string;
-    videotype: string;
-    period?: string;
-    begin?: string;
-    end?: string;
-    organe?: string;
-}
-export declare const SENAT_VIDEOS_SEARCH_AJAX = "https://videos.senat.fr/senat_videos_search.php";
-export declare const SENAT_DATAS_ROOT = "https://videos.senat.fr/Datas/senat";
-export declare function fetchText(url: string): Promise<string | null>;
-export declare function fetchBuffer(url: string): Promise<Buffer | null>;
-export declare function queryString(obj: Record<string, string>): string;
-export declare function parseDataNvs(nvs: string): {
-    epoch?: number;
-    organes: string[];
-    firstChapterLabel?: string;
-    salle?: string;
-};
-export declare function buildSenatVodMasterM3u8FromNvs(nvsText: string): string | null;
-export declare function isAmbiguousTimeOriginal(timeOriginal?: string | null): boolean;
-export declare function getAgendaType(agenda: Reunion): string;
-export declare function fetchAllSearchPages(args: SearchParams, maxPages?: number): Promise<string[]>;
-export declare function getOrgKey(norm: string): string;
-export type MatchResult = {
-    reunionUid: string;
-    picked: null | {
-        m3u8: string;
-        startSec: number;
-        score: number;
-    };
-};
+export {};

package/lib/src/scripts/retrieve_videos.js

@@ -4,387 +4,144 @@ import commandLineArgs from "command-line-args";
 import fs from "fs-extra";
 import fsp from "fs/promises";
 import path from "path";
-import * as cheerio from "cheerio";
-import { AGENDA_FOLDER, DATA_TRANSFORMED_FOLDER, iterLoadSenatAgendas } from "../loaders";
+import { iterLoadSenatAgendas } from "../loaders";
 import { getSessionsFromStart } from "../types/sessions";
 import { commonOptions } from "./shared/cli_helpers";
-import { getAgendaSegmentTimecodes } from "../utils/nvs-timecode";
-import { decodeHtmlEntities } from "../utils/string_cleaning";
-import { matchOneReunion } from "../utils/scoring";
-import { epochToParisDateTime, toFRDate, toTargetEpoch } from "../utils/date";
+import { getAgendaSegmentTimecodes, buildSenatVodMasterM3u8FromNvs } from "../utils/nvs-parsing";
+import { epochToParisDateTime, isAmbiguousTimeOriginal, toTargetEpoch } from "../utils/date";
 import { pathToFileURL } from "url";
-import { weights } from "../utils/weights_scoring_config";
-// ===================== Constants =====================
-const MAX_CANDIDATES = 15;
-const STATS = { total: 0, accepted: 0 };
-const VIDEOS_ROOT_FOLDER = "videos";
-export const SENAT_VIDEOS_SEARCH_AJAX = "https://videos.senat.fr/senat_videos_search.php";
-export const SENAT_DATAS_ROOT = "https://videos.senat.fr/Datas/senat";
-// ===================== CLI =====================
+import { fetchCandidatesForAgenda, fetchText } from "../videos/search";
+import { matchAgendaToVideo } from "../videos/match";
+import { SENAT_DATAS_ROOT, STATS, VIDEOS_ROOT_FOLDER, weights } from "../videos/config";
+import { processBisIfNeeded, processOneReunionMatch, writeIfChanged } from "../videos";
 const optionsDefinitions = [...commonOptions];
 const options = commandLineArgs(optionsDefinitions);
-// ===================== Utils =====================
-export async function fetchText(url) {
-    const res = await fetch(url);
-    if (!res.ok)
-        return null;
-    return await res.text();
-}
-export async function fetchBuffer(url) {
-    const res = await fetch(url);
-    if (!res.ok)
-        return null;
-    const ab = await res.arrayBuffer();
-    return Buffer.from(ab);
-}
-async function writeIfChanged(p, content) {
-    const exists = await fs.pathExists(p);
-    if (exists) {
-        const old = await fsp.readFile(p, "utf-8");
-        if (old === content)
-            return;
-    }
-    await fsp.writeFile(p, content, "utf-8");
-}
-export function queryString(obj) {
-    return Object.entries(obj)
-        .map(([k, v]) => `${encodeURIComponent(k)}=${encodeURIComponent(v)}`)
-        .join("&");
-}
-function extractCandidatesFromSearchHtml(html) {
-    const $ = cheerio.load(html);
-    const out = [];
-    const re = /video\.(\d+)_([a-z0-9]+)/i;
-    $('h3.card-title a.stretched-link[href*="video."]').each((_, a) => {
-        const href = $(a).attr("href") || "";
-        const m = href.match(re);
-        if (!m)
-            return;
-        const id = m[1];
-        const hash = m[2];
-        const pageUrl = `https://videos.senat.fr/video.${id}_${hash}.html`;
-        const title = ($(a).attr("title") || $(a).text() || "").replace(/\s+/g, " ").trim() || undefined;
-        const isSeancePublique = title?.toLowerCase().includes("séance publique") ?? false;
-        out.push({ id, hash, pageUrl, title, isSeancePublique });
-    });
-    // dedupe
-    const seen = new Set();
-    return out.filter((c) => {
-        const k = `${c.id}_${c.hash}`;
-        if (seen.has(k))
-            return false;
-        seen.add(k);
+function shouldSkipAgenda(agenda) {
+    if (!agenda.captationVideo)
         return true;
-    });
-}
-export function parseDataNvs(nvs) {
-    const epochStr = nvs.match(/<metadata\s+name="date"\s+value="(\d+)"/i)?.[1];
-    const epoch = epochStr ? Number(epochStr) : undefined;
-    // There can be multiple organes for one video in meta
-    const organes = [];
-    const organesRegex = /<metadata\b[^>]*\bname="organes"[^>]*>/gi;
-    let m;
-    const salle = decodeHtmlEntities(nvs.match(/<metadata\s+name="salle"\s+value="([^"]+)"/i)?.[1]).trim();
-    while ((m = organesRegex.exec(nvs)) !== null) {
-        const tag = m[0];
-        const label = tag.match(/\blabel="([^"]+)"/i)?.[1];
-        if (label) {
-            const decoded = decodeHtmlEntities(label).trim();
-            if (decoded)
-                organes.push(decoded);
-        }
-    }
-    if (organes.length === 0) {
-        organes.push("Séance publique");
-    }
-    const firstChapterLabelMatch = nvs.match(/<chapter\b[^>]*\blabel="([^"]+)"/i);
-    const firstChapterLabel = firstChapterLabelMatch ? decodeHtmlEntities(firstChapterLabelMatch[1]).trim() : undefined;
-    return { epoch, organes, firstChapterLabel, salle };
-}
-export function buildSenatVodMasterM3u8FromNvs(nvsText) {
-    // serverfiles://senat/2025/10/encoder10_20251022084451_2.mp4
-    const m = nvsText.match(/serverfiles:\/\/senat\/(\d{4})\/(\d{2})\/(encoder\d+)_([0-9]{14})/i);
-    if (!m)
-        return null;
-    const [, yyyy, mm, encoder, stamp] = m;
-    const base = `https://vodsenat.akamaized.net/senat/${yyyy}/${mm}/${encoder}_${stamp}`;
-    return `${base}.smil/master.m3u8`;
-}
-export function isAmbiguousTimeOriginal(timeOriginal) {
-    if (!timeOriginal)
-        return false;
-    const s = timeOriginal.toLowerCase();
-    // Catches "14h", "14 h", "14h30", "14 h 30", "14 heures", "14 heure"
-    const timeRe = /\b([01]?\d|2[0-3])\s*(?:h|heures?|heure)\s*(?:([0-5]\d))?\b/g;
-    const times = new Set();
-    let m;
-    while ((m = timeRe.exec(s))) {
-        const hh = String(m[1]).padStart(2, "0");
-        const mm = m[2] ? String(m[2]).padStart(2, "0") : "00";
-        times.add(`${hh}:${mm}`);
-    }
-    // "midi" / "minuit"
-    if (/\bmidi\b/.test(s))
-        times.add("12:00");
-    if (/\bminuit\b/.test(s))
-        times.add("00:00");
-    if (times.size >= 2)
+    if (!agenda.date || !agenda.startTime)
         return true;
-    const hasDayPeriod = /\b(matin|après-?midi|soir|nuit|journée|toute la journée)\b/.test(s);
-    const hasLinking = /,|\bet\b|\bou\b|\bpuis\b/.test(s);
-    if (times.size === 1 && hasDayPeriod && hasLinking)
+    if (agenda.uid.endsWith("Bis"))
+        return true; // Don't reprocess bis reunions
+    const agendaTs = toTargetEpoch(agenda.startTime, agenda.date);
+    const now = Date.now();
+    if (agendaTs && agendaTs * 1000 > now)
         return true;
     return false;
 }
-export function getAgendaType(agenda) {
-    const o = agenda.organe || "";
-    if (/séance publique/i.test(o))
-        return "Séance publique";
-    return "Commission";
+async function computeContext(agenda, session, dataDir) {
+    const agendaTs = agenda.startTime && agenda.date ? toTargetEpoch(agenda.startTime, agenda.date) : null;
+    const reunionUid = agenda.uid;
+    const baseDir = path.join(dataDir, VIDEOS_ROOT_FOLDER, String(session), reunionUid);
+    await fs.ensureDir(baseDir);
+    return { session, dataDir, baseDir, reunionUid, agendaTs };
 }
-export async function fetchAllSearchPages(args, maxPages = 3) {
-    const pages = [];
-    for (let p = 1; p <= maxPages; p++) {
-        const url = `${SENAT_VIDEOS_SEARCH_AJAX}?${queryString({ ...args, page: String(p) })}`;
-        const html = await fetchText(url);
-        if (!html)
-            break;
-        pages.push(html);
-        if (!/href="\/?video\.\d+_[a-z0-9]+\./i.test(html))
-            break;
-    }
-    return pages;
+function shouldSkipDownload(agenda, baseDir) {
+    if (!options["only-recent"])
+        return false;
+    const now = Date.now();
+    const cutoff = now - options["only-recent"] * 24 * 3600 * 1000;
+    const reunionTs = Date.parse(agenda.date);
+    if (reunionTs >= cutoff)
+        return false;
+    const dataNvsPath = path.join(baseDir, "data.nvs");
+    const finalplayerNvsPath = path.join(baseDir, "finalplayer.nvs");
+    return fs.existsSync(dataNvsPath) && fs.existsSync(finalplayerNvsPath);
 }
-export function getOrgKey(norm) {
-    if (!norm)
-        return "autre";
-    if (norm.includes("seance publique"))
-        return "seance_publique";
-    if (norm.includes("culture"))
-        return "culture";
-    if (norm.includes("finances"))
-        return "finances";
-    if (norm.includes("sociales"))
-        return "affaires_sociales";
-    if (norm.includes("economiques"))
-        return "affaires_economiques";
-    if (norm.includes("europeennes"))
-        return "affaires_europeennes";
-    if (norm.includes("etrangeres") || norm.includes("forces armees") || norm.includes("defense")) {
-        return "affaires_etrangeres_defense";
-    }
-    if (norm.includes("territoire") || norm.includes("durable")) {
-        return "amenagement_territoire_dd";
-    }
-    if (norm.includes("commission des lois"))
-        return "lois";
-    if (norm.includes("delegation aux collectivites territoriales") || norm.includes("delegation a la decentralisation"))
-        return "delegation_collectivites";
-    if (norm.includes("delegation aux droits des femmes") ||
-        norm.includes("egalite des chances entre les hommes et les femmes"))
-        return "delegation_droits_femmes";
-    if (norm.includes("delegation aux entreprises"))
-        return "delegation_entreprises";
-    if (norm.includes("delegation senatoriale aux outre mer") || norm.includes("delegation aux outre mer"))
-        return "delegation_outre_mer";
-    if (norm.includes("delegation a la prospective"))
-        return "delegation_prospective";
-    if (norm.includes("office parlementaire d evaluation des choix scientifiques et technologiques") ||
-        norm.includes("opecst"))
-        return "opecst";
-    return "autre";
+async function writeMatchArtifacts(args) {
+    const { agenda, ctx, best, secondBest } = args;
+    const bestDt = best.epoch ? epochToParisDateTime(best.epoch) : null;
+    const secondBestDt = secondBest && secondBest.epoch ? epochToParisDateTime(secondBest.epoch) : null;
+    const metadata = {
+        reunionUid: ctx.reunionUid,
+        session: ctx.session,
+        accepted: true,
+        threshold: weights.minAccept,
+        agenda: {
+            date: agenda.date,
+            startTime: agenda.startTime,
+            titre: agenda.titre,
+            organe: agenda.organe ?? undefined,
+            uid: agenda.uid,
+        },
+        best: {
+            id: best.id,
+            hash: best.hash,
+            pageUrl: best.pageUrl,
+            epoch: best.epoch ?? null,
+            date: bestDt?.date ?? null,
+            startTime: bestDt?.startTime ?? null,
+            title: best.vtitle ?? null,
+            score: best.score,
+        },
+        secondBest: secondBest
+            ? {
+                id: secondBest.id,
+                hash: secondBest.hash,
+                pageUrl: secondBest.pageUrl,
+                epoch: secondBest.epoch ?? null,
+                date: secondBestDt?.date ?? null,
+                startTime: secondBestDt?.startTime ?? null,
+                title: secondBest.vtitle ?? null,
+                score: secondBest.score,
+            }
+            : null,
+    };
+    await writeIfChanged(path.join(ctx.baseDir, "metadata.json"), JSON.stringify(metadata, null, 2));
+    const dataUrl = `${SENAT_DATAS_ROOT}/${best.id}_${best.hash}/content/data.nvs`;
+    const finalUrl = `${SENAT_DATAS_ROOT}/${best.id}_${best.hash}/content/finalplayer.nvs`;
+    const dataTxt = await fetchText(dataUrl);
+    const finalTxt = await fetchText(finalUrl);
+    if (dataTxt)
+        await fsp.writeFile(path.join(ctx.baseDir, "data.nvs"), dataTxt, "utf-8");
+    if (finalTxt)
+        await fsp.writeFile(path.join(ctx.baseDir, "finalplayer.nvs"), finalTxt, "utf-8");
 }
 async function processGroupedReunion(agenda, session, dataDir) {
     // 1) GuardRails
-    if (!agenda.captationVideo) {
-        // if (!options["silent"]) console.log(`[skip] ${agenda.uid} captationVideo=false`)
-        return;
-    }
-    if (!agenda.date || !agenda.startTime) {
-        // if (!options["silent"]) console.log(`[skip] ${agenda.uid} date/hour missing`)
-        return;
-    }
-    const agendaTs = toTargetEpoch(agenda.startTime, agenda.date);
-    const now = Date.now();
-    if (agendaTs && agendaTs * 1000 > now) {
+    if (shouldSkipAgenda(agenda))
         return;
-    }
-    const reunionUid = agenda.uid;
-    const baseDir = path.join(dataDir, VIDEOS_ROOT_FOLDER, String(session), reunionUid);
-    await fs.ensureDir(baseDir);
-    let skipDownload = false;
-    if (options["only-recent"]) {
-        const now = Date.now();
-        const cutoff = now - options["only-recent"] * 24 * 3600 * 1000;
-        const reunionTs = Date.parse(agenda.date);
-        if (reunionTs < cutoff) {
-            // Check if files already exist
-            const dataNvsPath = path.join(baseDir, "data.nvs");
-            const finalplayerNvsPath = path.join(baseDir, "finalplayer.nvs");
-            if (fs.existsSync(dataNvsPath) && fs.existsSync(finalplayerNvsPath)) {
-                skipDownload = true;
-            }
-        }
-    }
-    let master = null;
-    let dataTxt = null;
-    let finalTxt = null;
-    let accepted = false;
+    const ctx = await computeContext(agenda, session, dataDir);
+    const skipDownload = shouldSkipDownload(agenda, ctx.baseDir);
+    let match = null;
+    let best = null;
+    let secondBest = null;
+    // 2) Match + download artifacts (only if not skipped)
     if (!skipDownload) {
         STATS.total++;
-        const searchParams = {
-            search: "true",
-            videotype: getAgendaType(agenda),
-        };
-        if (agenda.date) {
-            const fr = toFRDate(agenda.date);
-            searchParams.period = "custom";
-            searchParams.begin = fr;
-            searchParams.end = fr;
-        }
-        if (agenda.organe) {
-            searchParams.organe = agenda.organe;
-        }
-        const pages = await fetchAllSearchPages(searchParams);
-        if (!pages.length) {
-            if (!options["silent"]) {
-                console.log(`[miss] ${agenda.uid} no candidates (videotype=${searchParams.videotype}, organe=${searchParams.organe || "-"}, date=${searchParams.begin || "-"})`);
-            }
+        const candidates = await fetchCandidatesForAgenda(agenda, options);
+        if (!candidates) {
+            console.log(`[warn] ${agenda.uid} No candidate found for this reunion. Probably VOD not published yet.`);
             return;
         }
-        const combinedHtml = pages.join("\n<!-- PAGE SPLIT -->\n");
-        const candidates = extractCandidatesFromSearchHtml(combinedHtml).slice(0, MAX_CANDIDATES);
-        if (!candidates.length) {
-            if (!options["silent"]) {
-                console.log(`[miss] ${agenda.uid} no candidates after parse (videotype=${searchParams.videotype}, organe=${searchParams.organe || "-"}, date=${searchParams.begin || "-"})`);
-            }
-            return;
-        }
-        // ==== 2) Enrich via data.nvs + scoring; pick best ====
-        const timeAmbigious = isAmbiguousTimeOriginal(agenda.events[0].timeOriginal);
-        if (timeAmbigious) {
-            console.log(`[match] ${agenda.uid} timeOriginal ambiguous => ignoring time scoring: "${agenda.events[0].timeOriginal}"`);
-        }
-        const best = await matchOneReunion({
-            agenda,
-            agendaTs,
-            timeAmbigious,
-            candidates,
-            weights,
-            fetchDataNvs: (c) => fetchBuffer(`${SENAT_DATAS_ROOT}/${c.id}_${c.hash}/content/data.nvs`),
-        });
-        if (!best) {
-            if (!options["silent"])
-                console.log(`[miss] ${agenda.uid} No candidate found for this reunion`);
+        match = await matchAgendaToVideo({ agenda, agendaTs: ctx.agendaTs, candidates, options });
+        if (!match) {
+            console.log(`[miss] ${agenda.uid} No match found for this reunion`);
             return;
         }
+        ;
+        ({ best, secondBest } = match);
         STATS.accepted++;
-        if (!options["silent"]) {
-            console.log(`[accepted] ${agenda.uid} score=${best.score.toFixed(2)}
-    agenda title="${agenda.titre ?? ""}" agenda organe="${agenda.organe ?? ""}" agenda salle=${agenda.lieu}
-    best title="${best.vtitle ?? ""}" best organe="${best.vorgane ?? ""}" best salle=${best.vsalle}`);
-        }
-        // ==== 3) Write metadata + NVS of the best candidate (always) ====
-        const bestDt = best.epoch ? epochToParisDateTime(best.epoch) : null;
-        const metadata = {
-            reunionUid,
-            session,
-            accepted,
-            threshold: weights.minAccept,
-            agenda: {
-                date: agenda.date,
-                startTime: agenda.startTime,
-                titre: agenda.titre,
-                organe: agenda.organe ?? undefined,
-                uid: agenda.uid,
-            },
-            best: {
-                id: best.id,
-                hash: best.hash,
-                pageUrl: best.pageUrl,
-                epoch: best.epoch ?? null,
-                date: bestDt?.date ?? null,
-                startTime: bestDt?.startTime ?? null,
-                title: best.vtitle ?? null,
-                score: best.score,
-            },
-        };
-        await writeIfChanged(path.join(baseDir, "metadata.json"), JSON.stringify(metadata, null, 2));
-        const dataUrl = `${SENAT_DATAS_ROOT}/${best.id}_${best.hash}/content/data.nvs`;
-        const finalUrl = `${SENAT_DATAS_ROOT}/${best.id}_${best.hash}/content/finalplayer.nvs`;
-        dataTxt = await fetchText(dataUrl);
-        finalTxt = await fetchText(finalUrl);
-        if (dataTxt)
-            await fsp.writeFile(path.join(baseDir, "data.nvs"), dataTxt, "utf-8");
-        if (finalTxt)
-            await fsp.writeFile(path.join(baseDir, "finalplayer.nvs"), finalTxt, "utf-8");
-        if (best.m3u8) {
-            master = best.m3u8;
-        }
-        else {
-            console.log("Cannot download data nvs");
-        }
-    }
-    else {
-        // Skipped download, but need to read data.nvs for urlVideo
-        try {
-            dataTxt = await fsp.readFile(path.join(baseDir, "data.nvs"), "utf-8");
-            finalTxt = await fsp.readFile(path.join(baseDir, "finalplayer.nvs"), "utf-8");
-            master = buildSenatVodMasterM3u8FromNvs(dataTxt);
-        }
-        catch (e) {
-            console.warn(e);
-        }
-    }
-    // ==== 4) Update agenda file (only if accepted + m3u8) ====
-    if (master) {
-        const agendaJsonPath = path.join(dataDir, AGENDA_FOLDER, DATA_TRANSFORMED_FOLDER, String(session), `${agenda.uid}.json`);
-        let timecodeDebutVideo = null;
-        let timecodeFinVideo = null;
-        if (dataTxt && finalTxt) {
-            const agendaKey = agenda.titre || agenda.objet || "";
-            const seg = getAgendaSegmentTimecodes(dataTxt, finalTxt, agendaKey);
-            if (!seg) {
-                console.warn(`[warn] Cannot retrieve agenda segment timecodes from reunion ${reunionUid}`);
-            }
-            else {
-                timecodeDebutVideo = seg.start;
-                timecodeFinVideo = seg.end;
-            }
-        }
-        if (await fs.pathExists(agendaJsonPath)) {
-            const raw = await fsp.readFile(agendaJsonPath, "utf-8");
-            let obj;
-            try {
-                obj = JSON.parse(raw);
-            }
-            catch (e) {
-                console.warn(`[warn] invalid JSON in ${agendaJsonPath}:`, e?.message);
-                obj = null;
-            }
-            if (obj && typeof obj === "object" && !Array.isArray(obj)) {
-                const next = { ...obj, urlVideo: master };
-                if (timecodeDebutVideo != null) {
-                    next.timecodeDebutVideo = timecodeDebutVideo;
-                    next.timecodeFinVideo = timecodeFinVideo;
-                }
-                await writeIfChanged(agendaJsonPath, JSON.stringify(next, null, 2));
-                if (!options["silent"]) {
-                    console.log(`[write] ${agenda.uid} urlVideo ← ${master}` +
-                        (timecodeDebutVideo != null ? ` (timecodeDebutVideo ← ${timecodeDebutVideo}s)` : ""));
-                }
-            }
-            else {
-                console.warn(`[warn] expected an object in ${agendaJsonPath}, got ${Array.isArray(obj) ? "array" : typeof obj}`);
-            }
-        }
-        else {
-            console.warn(`[warn] agenda file not found for update: ${agendaJsonPath}`);
-        }
-    }
-    else {
-        console.warn(`[warn] The video url could not be built for reunion `, reunionUid);
-    }
+        await writeMatchArtifacts({ agenda, ctx, best, secondBest });
+    }
+    if (best && isAmbiguousTimeOriginal(agenda.events[0].timeOriginal)) {
+        if (!options["silent"])
+            console.log("If the time is ambiguous, update agenda startTime from matched video");
+        agenda = { ...agenda, startTime: epochToParisDateTime(best.epoch)?.startTime ?? agenda.startTime };
+    }
+    // 3) Always update BEST agenda JSON from local NVS
+    await processOneReunionMatch({
+        agenda,
+        baseDir: ctx.baseDir,
+        dataDir: ctx.dataDir,
+        session: ctx.session,
+        options,
+        writeIfChanged,
+        getAgendaSegmentTimecodes,
+        buildSenatVodMasterM3u8FromNvs,
+    });
+    // 4) Optional BIS
+    await processBisIfNeeded({ agenda, secondBest, ctx, skipDownload, options });
 }
 async function processAll(dataDir, sessions) {
     console.log("Process all Agendas and fetch video's url");

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

@@ -8,3 +8,4 @@ export declare function epochToParisDateTime(epochSec: number): {
 } | null;
 export declare function toTargetEpoch(time: string | null, date?: string | null): number | null;
 export declare function toFRDate(dateYYYYMMDD: string): string;
+export declare function isAmbiguousTimeOriginal(timeOriginal?: string | null): boolean;

package/lib/src/utils/date.js

@@ -98,3 +98,29 @@ export function toFRDate(dateYYYYMMDD) {
     const [y, m, d] = dateYYYYMMDD.split("-");
     return `${d}/${m}/${y}`; // DD/MM/YYYY
 }
+export function isAmbiguousTimeOriginal(timeOriginal) {
+    if (!timeOriginal)
+        return false;
+    const s = timeOriginal.toLowerCase();
+    // Catches "14h", "14 h", "14h30", "14 h 30", "14 heures", "14 heure"
+    const timeRe = /\b([01]?\d|2[0-3])\s*(?:h|heures?|heure)\s*(?:([0-5]\d))?\b/g;
+    const times = new Set();
+    let m;
+    while ((m = timeRe.exec(s))) {
+        const hh = String(m[1]).padStart(2, "0");
+        const mm = m[2] ? String(m[2]).padStart(2, "0") : "00";
+        times.add(`${hh}:${mm}`);
+    }
+    // "midi" / "minuit"
+    if (/\bmidi\b/.test(s))
+        times.add("12:00");
+    if (/\bminuit\b/.test(s))
+        times.add("00:00");
+    if (times.size >= 2)
+        return true;
+    const hasDayPeriod = /\b(matin|après-?midi|soir|nuit|journée|toute la journée)\b/.test(s);
+    const hasLinking = /,|\bet\b|\bou\b|\bpuis\b/.test(s);
+    if (times.size === 1 && hasDayPeriod && hasLinking)
+        return true;
+    return false;
+}

package/lib/src/utils/nvs-parsing.d.ts

@@ -0,0 +1,24 @@
+export type L1Chapter = {
+    id: string;
+    label: string;
+    index: number;
+};
+export declare function getLevel1Chapters(dataNvs: string): L1Chapter[];
+export declare function pickBestLevel1ChapterForAgenda(chapters: L1Chapter[], agendaTitle: string): {
+    chapter: L1Chapter;
+    score: number;
+} | null;
+export declare function getAgendaSegmentTimecodes(dataNvs: string, finalPlayerNvs: string, agendaTitleOrObjet: string): {
+    start: number;
+    end: number | null;
+    chapterId: string;
+    nextChapterId: string | null;
+    score: number;
+} | null;
+export declare function parseDataNvs(nvs: string): {
+    epoch?: number;
+    organes: string[];
+    firstChapterLabel?: string;
+    salle?: string;
+};
+export declare function buildSenatVodMasterM3u8FromNvs(nvsText: string): string | null;