@tricoteuses/senat/lib/src/videos/match.js

Handle French Sénat's open data

import { isAmbiguousTimeOriginal } from "../utils/date";
import { buildSenatVodMasterM3u8FromNvs, getLevel1Chapters, parseDataNvs } from "../utils/nvs-parsing";
import { dice, getOrgKey, normalize, scoreVideo } from "../utils/scoring";
import { SENAT_DATAS_ROOT, weights } from "./config";
import { fetchBuffer } from "./search";
export async function matchOneReunion(args) {
    const { agenda, agendaTs, timeAmbigious, candidates, weights, fetchDataNvs, options } = args;
    if (!options["silent"])
        console.log(`Matching video for reunion: ${agenda?.uid}`);
    const isSP = (agenda.type ?? "").toLowerCase().includes("séance publique");
    const minAcceptBase = weights?.minAccept;
    const margin = weights?.margin;
    const titleDominance = weights?.titleDominance;
    const orgUncertainPenalty = weights?.orgUncertainPenalty; // 1 disables
    const orgSkipDice = weights?.orgSkipDice;
    const minAccept = minAcceptBase + (candidates.length >= 20 ? 0.08 : candidates.length >= 10 ? 0.05 : 0);
    const vw = weights
        ? {
            wTitle: weights.wTitle,
            wOrg: weights.wOrg,
            wSalle: weights.wSalle,
            wTime: weights.wTime,
            sameOrgBonus: weights.sameOrgBonus,
            titleMin: isSP ? (weights.spTitleMin ?? weights.titleMin) : weights.titleMin,
        }
        : undefined;
    // Keep ranked to compute margin (best vs second best)
    const ranked = [];
    for (const c of candidates) {
        const dataBuf = await fetchDataNvs(c);
        if (!dataBuf)
            continue;
        const dataStr = dataBuf.toString("utf-8");
        const m3u8 = buildSenatVodMasterM3u8FromNvs(dataStr);
        if (!m3u8)
            continue;
        const meta = parseDataNvs(dataStr);
        let sameOrg = false;
        // Organe gate (same key 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) {
                continue;
            }
        }
        let videoTitle = c.title;
        const chapterVideoTitles = getLevel1Chapters(dataStr);
        if (c.isSeancePublique && meta.firstChapterLabel) {
            videoTitle = meta.firstChapterLabel;
        }
        if (!vw)
            throw new Error("matchOneReunion: missing weights (vw)");
        const { score: rawScore, signals } = scoreVideo(agenda, agendaTs, sameOrg, vw, videoTitle, meta.epoch, meta.organes, timeAmbigious, meta.salle, chapterVideoTitles);
        const s = sameOrg ? rawScore : rawScore * orgUncertainPenalty;
        // Optional: title dominance gate (reject if score isn't "driven" enough by title)
        if (titleDominance > 0 && signals.titleScore < titleDominance * s) {
            continue;
        }
        ranked.push({
            id: c.id,
            hash: c.hash,
            pageUrl: c.pageUrl,
            epoch: meta.epoch,
            vtitle: videoTitle,
            score: s,
            vorgane: meta.organes?.[0],
            vsalle: meta.salle,
            m3u8,
            signals,
        });
    }
    if (!ranked.length)
        return null;
    ranked.sort((a, b) => b.score - a.score);
    const best = ranked[0];
    const second = ranked[1] ?? null;
    // Accept threshold (best must pass)
    if (best.score < minAccept) {
        if (!options["silent"])
            console.log(`[threshold] ${agenda.uid} score=${best.score.toFixed(2)}
          agenda title="${agenda.titre ?? ""}"
          best title="${best.vtitle ?? ""}"`);
        return null;
    }
    // Margin (confidence) gate
    if (second && best.score - second.score < margin) {
        if (!options["silent"]) {
            console.log(`[Margin] ${agenda.uid} score=${best.score.toFixed(2)}
          agenda title="${agenda.titre ?? ""}" agenda date="${agenda.startTime ?? ""}"
          best title="${best.vtitle ?? ""}" date="${best.epoch ? new Date(best.epoch * 1000).toISOString() : "null"}"`);
            console.log("second title=", second.vtitle ?? "", " score=", second.score.toFixed(2), "date=", second.epoch ? new Date(second.epoch * 1000).toISOString() : "null");
        }
        if (isSP && timeAmbigious) {
            // Optional safety: second must also be "acceptable"
            if (second.score >= minAccept) {
                console.log(" - special case: séance publique + time ambigious => accepting best + secondBest (bis)");
                return { best, secondBest: second, reason: "margin_ambiguous_time_sp" };
            }
        }
        return null;
    }
    return { best, secondBest: null };
}
export async function matchAgendaToVideo(args) {
    const { agenda, agendaTs, candidates, options } = args;
    const timeAmbigious = isAmbiguousTimeOriginal(agenda.events[0].timeOriginal);
    if (timeAmbigious && !options["silent"]) {
        console.log(`[match] ${agenda.uid} timeOriginal ambiguous => ignoring time scoring: "${agenda.events[0].timeOriginal}"`);
    }
    return matchOneReunion({
        agenda,
        agendaTs,
        timeAmbigious,
        candidates,
        weights,
        options,
        fetchDataNvs: (c) => fetchBuffer(`${SENAT_DATAS_ROOT}/${c.id}_${c.hash}/content/data.nvs`),
    });
}