@tricoteuses/senat 2.10.0 → 2.10.2

package/lib/model/compte_rendu.js

@@ -1,69 +1,153 @@
-import { JSDOM } from "jsdom";
+import fs from "fs";
 import * as cheerio from "cheerio";
-const norm = (s) => s.replace(/\u00A0/g, " ").replace(/\s+/g, " ").trim();
-const toTexte = (s) => ({ _: s });
-function extractSommaire($) {
-    const root = $("#wysiwyg").length ? $("#wysiwyg") : $("#cri");
-    const sommaire = {
-        presidentSeance: toTexte(""),
-        sommaire1: [],
-    };
-    // (1) presidency line (e.g., "Présidence de Mme …")
-    const pres = root.find("p.tm2").filter((_, el) => /présidence/i.test($(el).text())).first();
-    if (pres.length) {
-        sommaire.presidentSeance = toTexte(norm(pres.text()));
+import path from "path";
+import { computeIntervalsBySlot } from "../utils/cr_spliting";
+import { norm } 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) {
+    try {
+        const raw = fs.readFileSync(xmlFilePath, "utf8");
+        const $ = cheerio.load(raw, { xml: false });
+        const metadonnees = extractMetadonnees($, xmlFilePath);
+        const order = $("body *").toArray();
+        const idx = new Map(order.map((el, i) => [el, i]));
+        const intervalsAll = computeIntervalsBySlot($, idx, firstSlotOfDay);
+        const intervals = intervalsAll.filter(iv => iv.slot === wantedSlot);
+        if (intervals.length === 0) {
+            console.warn(`[CRI] no intervals for ${path.basename(xmlFilePath)} [${wantedSlot}]`);
+            return null;
+        }
+        metadonnees.sommaire = extractSommaireForIntervals($, idx, intervals);
+        const points = [];
+        let ordre = 0;
+        const addPoint = (p) => points.push({ ...p, ordre_absolu_seance: String(++ordre) });
+        // Titles
+        $("cri\\:titreS1 p.titre_S1").each((_, el) => {
+            if (!elementInAnyInterval(el, idx, intervals))
+                return;
+            const t = normalizeTitle(norm($(el).text() || ""));
+            if (t)
+                addPoint({ code_grammaire: "TITRE_TEXTE_DISCUSSION", texte: { _: t }, code_style: "Titre" });
+        });
+        // Interventions
+        $("div.intervenant").each((_, block) => {
+            if (!elementInAnyInterval(block, idx, intervals))
+                return;
+            const $block = $(block);
+            $block.find([
+                "p[class^='titre_S']",
+                "p.mention_titre",
+                "p.intitule_titre",
+                "p.mention_chapitre",
+                "p.intitule_chapitre",
+                "p.mention_article",
+                "p.intitule_article",
+                "p.mention_section",
+                "p.intitule_section",
+            ].join(",")).remove();
+            const firstP = $block.find("p").first();
+            const speakerLabelRaw = firstP.find(".orateur_nom").text() || firstP.find("a.lien_senfic").text() || "";
+            const speakerLabel = dedupeSpeaker(speakerLabelRaw);
+            const { mat, nom: nomCRI, qua: quaCRI } = readIntervenantMeta($block);
+            const qualFromSpans = extractAndRemoveLeadingQualite($, $block);
+            const qualite = norm(decodeHtmlEntities(quaCRI || "")) || qualFromSpans;
+            const canonicalName = dedupeSpeaker(nomCRI || speakerLabel);
+            const role = roleForSpeaker(speakerLabel) || roleForSpeaker(qualite) || roleForSpeaker(quaCRI || "");
+            const speechHtml = sanitizeInterventionHtml($, $block);
+            if (!norm(cheerio.load(speechHtml).text() || ""))
+                return;
+            addPoint({
+                code_grammaire: "PAROLE_GENERIQUE",
+                roledebat: role,
+                orateurs: { orateur: { nom: canonicalName, id: mat || "", qualite } },
+                texte: { _: speechHtml },
+            });
+        });
+        const contenu = {
+            quantiemes: { journee: metadonnees.dateSeance, session: metadonnees.session },
+            point: points,
+        };
+        return {
+            uid: "CRSSN" + xmlFilePath.replace(/^.*?(\d{8}).*$/i, "$1") + `-${wantedSlot}`,
+            seanceRef: "RUSN" + xmlFilePath.replace(/^.*?(\d{8}).*$/i, "$1") + "IDS-" + wantedSlot,
+            sessionRef: metadonnees.session,
+            metadonnees,
+            contenu,
+        };
     }
-    // (2) extra info lines like "Secrétaires :" (tm5)
-    const paras = [];
-    root.find("p.tm5").each((_, el) => {
-        const t = norm($(el).text());
-        if (t)
-            paras.push(toTexte(t));
-    });
-    if (paras.length) {
-        sommaire.para = paras.length === 1 ? paras[0] : paras;
+    catch (e) {
+        console.error(`[CRI] parseSlot error file=${xmlFilePath} slot=${wantedSlot}:`, e);
+        return null;
     }
-    // (3) first-level items (tm3)
-    const items = [];
-    root.find("p.tm3").each((_, el) => {
-        const $p = $(el);
-        const full = norm($p.text());
-        // try to extract the numeric order at the start: "1. ..." or "2 – ..." etc.
-        const numMatch = full.match(/^(\d+)\s*[.\-–—]/);
-        const valeur = numMatch ? numMatch[1] : undefined;
-        // prefer the linked title text; fallback to full text
-        const a = $p.find("a").first();
-        const intitule = norm(a.text() || full.replace(/^(\d+)\s*[.\-–—]\s*/, ""));
-        // id_syceron = href target without '#' ? TODO verify
-        const href = a.attr("href") || "";
-        const idSyceron = href.startsWith("#") ? href.slice(1) : href;
-        const titreStruct = {
-            id_syceron: idSyceron || "",
-            intitule,
-        };
-        const elem = {
-            valeur_pts_odj: valeur,
-            titreStruct,
-            // sommaire2/3 undefined (first level only)
+}
+export function sessionStartYearFromDate(d) {
+    // Session (1th oct N → 30 sept N+1)
+    const m = d.getMonth();
+    const y = d.getFullYear();
+    return m >= 9 ? y : y - 1;
+}
+export function parseYYYYMMDD(yyyymmdd) {
+    if (!/^\d{8}$/.test(yyyymmdd))
+        return null;
+    const y = Number(yyyymmdd.slice(0, 4));
+    const m = Number(yyyymmdd.slice(4, 6)) - 1;
+    const d = Number(yyyymmdd.slice(6, 8));
+    const dt = new Date(y, m, d);
+    return Number.isFinite(dt.getTime()) ? dt : null;
+}
+export function deriveTitreObjetFromSommaire(sommaire, slot) {
+    const items = extractLevel1Items(sommaire);
+    const meaningful = items.filter(it => !isBoilerplate(it.label));
+    if (meaningful.length === 0) {
+        return {
+            titre: `Séance publique ${slotLabel(slot)}`,
+            objet: "",
         };
-        items.push(elem);
-    });
-    if (items.length) {
-        sommaire.sommaire1 = items;
     }
-    return sommaire;
+    const titre = meaningful[0].label;
+    const objet = meaningful.slice(0, 3).map(it => it.label).join(" ; ");
+    return { titre, objet };
+}
+function slotLabel(slot) {
+    switch (slot) {
+        case "MATIN": return "du matin";
+        case "APRES-MIDI": return "de l’après-midi";
+        case "SOIR": return "du soir";
+        default: return "";
+    }
 }
-function stripTrailingPunct(s) {
-    return s.replace(/\s*([:,.;])\s*$/u, "").trim();
+const BOILERPLATE_PATTERNS = [
+    /proc(?:è|e)s-?verbal/i,
+    /hommages?/i,
+    /désignation des vice-?président/i,
+    /candidatures? aux?/i,
+    /ordre du jour/i,
+    /rappels? au règlement/i,
+    /communications?/i,
+    /dépôts?/i,
+    /proclamation/i,
+    /présidence de/i,
+    /questions? diverses?/i,
+    /ouverture de la séance/i,
+    /clo(?:t|̂)ure de la séance/i,
+];
+const isBoilerplate = (label) => !label?.trim() || BOILERPLATE_PATTERNS.some(rx => rx.test(label));
+function extractLevel1Items(sommaire) {
+    const level1 = asArray(sommaire?.sommaire1);
+    return level1
+        .map(el => ({
+        numero: toInt(el?.valeur_pts_odj),
+        label: String(el?.titreStruct?.intitule ?? "").trim(),
+    }))
+        .filter(it => !!it.label)
+        .sort((a, b) => a.numero - b.numero);
 }
+function stripTrailingPunct(s) { return s.replace(/\s*([:,.;])\s*$/u, "").trim(); }
 function dedupeSpeaker(raw) {
     let s = norm(raw);
     s = stripTrailingPunct(s);
-    const dupPatterns = [
-        /^(.+?)\s*[.]\s*\1$/u,
-        /^(.+?)\s*,\s*\1,?$/u,
-        /^(.+?)\s+\1$/u,
-    ];
+    const dupPatterns = [/^(.+?)\s*[.]\s*\1$/u, /^(.+?)\s*,\s*\1,?$/u, /^(.+?)\s+\1$/u];
     for (const re of dupPatterns) {
         const m = s.match(re);
         if (m) {
@@ -74,12 +158,10 @@ function dedupeSpeaker(raw) {
     return s.replace(/\.\s*$/, "");
 }
 function decodeHtmlEntities(s) {
-    return s
-        .replace(/&#(\d+);/g, (_, d) => String.fromCharCode(parseInt(d, 10)))
+    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) {
-    // Tighten spacing around French apostrophes and punctuation
     let out = s;
     out = out.replace(/\s*’\s*/g, "’");
     out = out.replace(/\b([dljctmsn])\s*’/gi, (_, m) => m + "’");
@@ -87,43 +169,33 @@ function fixApostrophes(s) {
     out = out.replace(/\s+([,;:.!?])/g, "$1");
     return out;
 }
-function normalizeTitle(text) {
-    return text.replace(/^PR[ÉE]SIDENCE DE\b/i, "Présidence de ");
-}
+function normalizeTitle(text) { return text.replace(/^PR[ÉE]SIDENCE DE\b/i, "Présidence de "); }
 function roleForSpeaker(labelOrQualite) {
-    const s = labelOrQualite.toLowerCase();
-    if (/^(m\.|mme)?\s*(le|la)\s+pr[ée]sident(e)?\b/.test(s) ||
-        /\bpr[ée]sident[e]?\s+de\s+séance\b/.test(s)) {
+    const s = (labelOrQualite || "").toLowerCase();
+    if (/^(m\.|mme)?\s*(le|la)\s+pr[ée]sident(e)?\b/.test(s) || /\bpr[ée]sident[e]?\s+de\s+séance\b/.test(s))
         return "président";
-    }
     return "";
 }
-// ---------------- DOM helpers ----------------
-function parseCriIntervenantComment(html) {
-    // From <!-- cri:intervenant mat="..." nom="..." qua="..." ... -->
+function readIntervenantMeta($block) {
+    const int = $block.find('cri\\:intervenant').first();
+    if (int.length)
+        return { mat: int.attr("mat") || undefined, nom: int.attr("nom") || undefined, qua: int.attr("qua") || undefined };
+    const html = $block.html() || "";
     const m = html.match(/<!--\s*cri:intervenant\b([^>]+)-->/i);
     if (!m)
         return {};
-    const attrs = m[1];
     const out = {};
     const re = /(\w+)="([^"]*)"/g;
     let a;
-    while ((a = re.exec(attrs))) {
+    while ((a = re.exec(m[1])))
         out[a[1]] = decodeHtmlEntities(a[2]);
-    }
     return { mat: out["mat"], nom: out["nom"], qua: out["qua"] };
 }
-/**
- * Extract leading .orateur_qualite chunks from the FIRST <p> only,
- * concatenate them, clean punctuation/apostrophes, and REMOVE those nodes
- * (and .orateur_nom) from the first paragraph so the speech starts cleanly.
- */
 function extractAndRemoveLeadingQualite($, $block) {
     const firstP = $block.find("p").first();
     if (firstP.length === 0)
         return "";
     const parts = [];
-    // Iterate over the first <p>'s children from the start
     let stop = false;
     firstP.contents().each((_, node) => {
         if (stop)
@@ -131,7 +203,6 @@ function extractAndRemoveLeadingQualite($, $block) {
         if (node.type === "tag") {
             const $node = $(node);
             if ($node.hasClass("orateur_nom")) {
-                // speaker label node — remove it
                 $node.remove();
                 return;
             }
@@ -140,66 +211,95 @@ function extractAndRemoveLeadingQualite($, $block) {
                 $node.remove();
                 return;
             }
-            // Non-qualite tag: if it has meaningful text, we reached the speech
             const t = norm($node.text() || "");
-            if (t) {
+            if (t)
                 stop = true;
-            }
-            else {
-                // empty-ish node; remove to avoid stray punctuation
+            else
                 $node.remove();
-            }
         }
         else if (node.type === "text") {
             const t = norm(node.data || "");
-            if (!t) {
-                // whitespace only — drop it
-                ;
-                node.data = "";
-                return;
-            }
-            // boundary punctuation like ":" just after label — drop it
-            if (/^[:.,;–—-]+$/.test(t)) {
-                ;
+            if (!t || /^[:.,;–—-]+$/.test(t)) {
                 node.data = "";
                 return;
             }
-            // any other text means speech starts here
             stop = true;
         }
-        else {
-            // comment or others — ignore
-        }
     });
-    const qual = fixApostrophes(norm(parts.join(" ")));
-    return qual;
+    return fixApostrophes(norm(parts.join(" ")));
 }
 function sanitizeInterventionHtml($, $block) {
-    // Clone to avoid mutating outer tree order
     const $clone = $block.clone();
-    // Remove navigation / anchors / images
     $clone.find('a[name]').remove();
     $clone.find('div[align="right"]').remove();
     $clone.find('a.link').remove();
     $clone.find('img').remove();
-    // Remove technical anchors inside interventions
     $clone.find('a#ameli_amendement_cri_phrase, a#ameli_amendement_cra_contenu, a#ameli_amendement_cra_objet').remove();
-    // Remove any remaining speaker label / quality spans anywhere
-    $clone.find(".orateur_nom").remove();
-    $clone.find(".orateur_qualite").remove();
-    // Strip HTML comments
+    $clone.find(".orateur_nom, .orateur_qualite").remove();
     let html = $clone.html() || "";
     html = html.replace(/<!--[\s\S]*?-->/g, "");
     return html.trim();
 }
-function extractMetadonnees($) {
-    const headerText = norm($("h1.page-title").text() || "");
-    const dateMatch = headerText.match(/\b(\d{1,2}\s+\w+\s+\d{4})\b/i);
-    const bodyText = norm($("#cri").text() || "");
-    const sessionMatch = bodyText.match(/\bsession\s+(\d{4}-\d{4})\b/i);
+function extractSommaireForIntervals($, idx, intervals) {
+    const inIv = (el) => elementInAnyInterval(el, idx, intervals);
+    const root = $("body");
+    const sommaire = { presidentSeance: { _: "" }, sommaire1: [] };
+    // (1) Présidence (tm2) — première ligne dans l’intervalle 
+    const pres = root.find("p.tm2").filter((_, el) => inIv(el)).first();
+    if (pres.length)
+        sommaire.presidentSeance = { _: norm(pres.text()) };
+    // (2) Paras tm5 présents dans l’intervalle 
+    const paras = [];
+    root.find("p.tm5").each((_, el) => {
+        if (!inIv(el))
+            return;
+        const t = norm($(el).text());
+        if (t)
+            paras.push({ _: t });
+    });
+    if (paras.length)
+        sommaire.para = paras.length === 1 ? paras[0] : paras;
+    // (3) Items de 1er niveau (tm3) présents dans l’intervalle 
+    const items = [];
+    root.find("p.tm3").each((_, el) => {
+        if (!inIv(el))
+            return;
+        const $p = $(el);
+        const full = norm($p.text() || "");
+        if (!full)
+            return;
+        const numMatch = full.match(/^(\d+)\s*[.\-–—]\s*/);
+        const valeur = numMatch ? numMatch[1] : undefined;
+        // prefere intitule in ancre <a> if present
+        const a = $p.find("a").first();
+        const intituleRaw = a.length ? a.text() : full.replace(/^(\d+)\s*[.\-–—]\s*/, "");
+        const intitule = norm(intituleRaw);
+        // id_syceron from href="#Niv1_SOMx"
+        const href = (a.attr("href") || "").trim();
+        const idSyceron = href.startsWith("#") ? href.slice(1) : href;
+        const titreStruct = { id_syceron: idSyceron || "", intitule };
+        items.push({ valeur_pts_odj: valeur, titreStruct });
+    });
+    if (items.length)
+        sommaire.sommaire1 = items;
+    return sommaire;
+}
+function extractMetadonnees($, filePath) {
+    let dateText = norm($("h1, h2, .page-title").first().text() || "");
+    if (!dateText)
+        dateText = norm($("p").first().text() || "");
+    const dateMatch = dateText.match(/\b(\d{1,2}\s+\w+\s+\d{4})\b/i);
+    const allText = norm($("body").text() || "");
+    const sessionMatch = allText.match(/\bsession\s+(\d{4}-\d{4})\b/i);
+    let dateSeance = dateMatch?.[1] || "";
+    if (!dateSeance) {
+        const m = filePath.match(/d(\d{4})(\d{2})(\d{2})\.xml$/i);
+        if (m)
+            dateSeance = `${m[1]}-${m[2]}-${m[3]}`;
+    }
     return {
-        dateSeance: dateMatch?.[1] || "",
-        dateSeanceJour: dateMatch?.[1] || "",
+        dateSeance,
+        dateSeanceJour: dateSeance,
         numSeanceJour: "",
         numSeance: "",
         typeAssemblee: "SN",
@@ -211,103 +311,15 @@ function extractMetadonnees($) {
         diffusion: "",
         version: "1.0",
         environnement: "",
-        heureGeneration: new Date(),
-        sommaire: extractSommaire($)
+        heureGeneration: new Date()
     };
 }
-// ---------------- main transform ----------------
-export async function parseCompteRenduFromFile(htmlFilePath) {
-    try {
-        const { window } = await JSDOM.fromFile(htmlFilePath, { contentType: "text/html" });
-        const $ = cheerio.load(window.document.documentElement.outerHTML);
-        const metadonnees = extractMetadonnees($);
-        const points = [];
-        let ordre = 0;
-        const addPoint = (p) => points.push({ ...p, ordre_absolu_seance: String(++ordre) });
-        // (1) Global section titles (common high-level headings)
-        let lastTitle = "";
-        $("#cri p[class^='titre_S']").each((_, el) => {
-            const t = normalizeTitle(norm($(el).text() || ""));
-            if (t && t !== lastTitle) {
-                addPoint({ code_grammaire: "TITRE_TEXTE_DISCUSSION", texte: { _: t }, code_style: "Titre" });
-                lastTitle = t;
-            }
-        });
-        // (2) Interventions
-        $("#cri div.intervenant").each((_, block) => {
-            const $block = $(block);
-            // (2.a) Extract internal structural titles inside this block (and remove them)
-            const structuralSel = [
-                "p[class^='titre_S']",
-                "p.mention_titre",
-                "p.intitule_titre",
-                "p.mention_chapitre",
-                "p.intitule_chapitre",
-                "p.mention_article",
-                "p.intitule_article",
-                "p.mention_section",
-                "p.intitule_section",
-            ].join(",");
-            $block.find(structuralSel).each((__, el) => {
-                const title = normalizeTitle(norm($(el).text() || ""));
-                if (title && title !== lastTitle) {
-                    addPoint({ code_grammaire: "TITRE_TEXTE_DISCUSSION", texte: { _: title }, code_style: "Titre" });
-                    lastTitle = title;
-                }
-                $(el).remove();
-            });
-            // (2.b) Speaker label & quality
-            const firstP = $block.find("p").first();
-            const speakerLabelRaw = firstP.find(".orateur_nom").text() ||
-                firstP.find("a.lien_senfic").text() ||
-                "";
-            const speakerLabel = dedupeSpeaker(speakerLabelRaw);
-            // Prefer <!--cri:intervenant ...--> for id/name/qualite when available
-            const rawHtml = $block.html() || "";
-            const { mat, nom: nomFromComment, qua: quaFromCommentRaw } = parseCriIntervenantComment(rawHtml);
-            // Extract and remove leading .orateur_qualite chunks from first <p>
-            const qualFromSpans = extractAndRemoveLeadingQualite($, $block);
-            const qualite = norm(decodeHtmlEntities(quaFromCommentRaw || "")) ||
-                qualFromSpans;
-            const canonicalName = dedupeSpeaker(nomFromComment || speakerLabel);
-            const role = roleForSpeaker(speakerLabel) ||
-                roleForSpeaker(qualite) ||
-                roleForSpeaker(quaFromCommentRaw || "");
-            // (2.c) Build cleaned speech HTML
-            let speechHtml = sanitizeInterventionHtml($, $block);
-            // If nothing meaningful remains, skip
-            if (!norm(cheerio.load(speechHtml).text() || ""))
-                return;
-            addPoint({
-                code_grammaire: "PAROLE_GENERIQUE",
-                roledebat: role,
-                orateurs: {
-                    orateur: {
-                        nom: canonicalName,
-                        id: mat || "",
-                        qualite: qualite,
-                    },
-                },
-                texte: { _: speechHtml },
-            });
-        });
-        const contenu = {
-            quantiemes: {
-                journee: metadonnees.dateSeance,
-                session: metadonnees.session,
-            },
-            point: points,
-        };
-        return {
-            uid: htmlFilePath.replace(/^.*?(\d{8}).*$/i, "$1"),
-            seanceRef: htmlFilePath.replace(/^.*?(\d{8}).*$/i, "$1"),
-            sessionRef: metadonnees.session,
-            metadonnees,
-            contenu,
-        };
-    }
-    catch (e) {
-        console.error("Could not parse compte-rendu with error", e);
-        return null;
-    }
+function elementInAnyInterval(el, idx, intervals) {
+    const p = idx.get(el);
+    if (p == null)
+        return false;
+    for (const iv of intervals)
+        if (p >= iv.start && p < iv.end)
+            return true;
+    return false;
 }

package/lib/model/util.d.ts

@@ -6,3 +6,4 @@ export declare function removeSubstring(expr: Expression<string | null | undefin
 export declare function replace(expr: Expression<string | null | undefined>, pattern: Expression<string>, replacement: Expression<string>): import("kysely").RawBuilder<string>;
 export declare function rtrim(expr: Expression<string | null | undefined>): import("kysely").RawBuilder<string>;
 export declare function toDateString(expr: Expression<Date | null | undefined>, format?: Expression<string>): import("kysely").RawBuilder<string>;
+export declare function norm(s?: string | null): string;

package/lib/model/util.js

@@ -21,3 +21,6 @@ export function rtrim(expr) {
 export function toDateString(expr, format = sql.val(STANDARD_DATE_FORMAT)) {
     return sql `to_char(${expr}, ${format})`;
 }
+export function norm(s) {
+    return (s || "").replace(/\u00A0/g, " ").replace(/\s+/g, " ").trim();
+}

package/lib/scripts/retrieve_agenda.js

@@ -8,7 +8,8 @@ import { parseAgendaFromFile } from "../model/agenda";
 import { getSessionsFromStart } from "../types/sessions";
 import { ID_DATE_FORMAT } from "./datautil";
 import { commonOptions } from "./shared/cli_helpers";
-import { ensureAndClearDir } from "./shared/util";
+import { ensureAndClearDir, fetchWithRetry } from "./shared/util";
+import { groupNonSPByTypeOrganeHour, groupSeancePubliqueBySlot } from "../utils/reunion_grouping";
 const optionsDefinitions = [
     ...commonOptions,
     {
@@ -64,7 +65,7 @@ async function downloadAgenda(agendaName, agendaPath) {
     if (!options["silent"]) {
         console.log(`Downloading Agenda ${agendaUrl}…`);
     }
-    const response = await fetch(agendaUrl);
+    const response = await fetchWithRetry(agendaUrl);
     if (!response.ok) {
         if (response.status === 404) {
             console.warn(`Agenda ${agendaUrl} not found`);
@@ -80,15 +81,33 @@ async function downloadAgenda(agendaName, agendaPath) {
     }
     fs.writeFileSync(agendaPath, Buffer.from(agendaContent));
 }
+function writeGroupsAsFiles(dir, groups) {
+    for (const g of groups) {
+        const outPath = path.join(dir, `${g.uid}.json`);
+        fs.writeJSONSync(outPath, g, { spaces: 2 });
+    }
+}
 async function parseAgenda(transformedAgendaSessionDir, agendaFileName, agendaPath) {
-    if (!options["silent"]) {
+    if (!options["silent"])
         console.log(`Parsing Agenda ${agendaPath}…`);
-    }
     const parsedAgendaEvents = await parseAgendaFromFile(agendaPath);
-    if (!parsedAgendaEvents || parsedAgendaEvents.length === 0) {
+    if (!parsedAgendaEvents?.length)
         return;
+    const flatPath = path.join(transformedAgendaSessionDir, `${agendaFileName}.json`);
+    fs.writeJSONSync(flatPath, parsedAgendaEvents, { spaces: 2 });
+    // 1) SP → groubed by (date, slot)
+    const spGrouped = groupSeancePubliqueBySlot(parsedAgendaEvents);
+    if (spGrouped.length > 0) {
+        writeGroupsAsFiles(transformedAgendaSessionDir, spGrouped);
+    }
+    // 2) NON-SP → groubed by (date, organe, hour)
+    const groupedBySuffix = groupNonSPByTypeOrganeHour(parsedAgendaEvents);
+    for (const suffix of ["IDC", "IDM", "IDO", "IDI"]) {
+        const groups = groupedBySuffix[suffix] || [];
+        if (groups.length > 0) {
+            writeGroupsAsFiles(transformedAgendaSessionDir, groups);
+        }
     }
-    fs.writeJSONSync(path.join(transformedAgendaSessionDir, `${agendaFileName}.json`), parsedAgendaEvents, { spaces: 2 });
 }
 async function main() {
     const dataDir = options["dataDir"];

package/lib/scripts/retrieve_comptes_rendus.d.ts

@@ -1 +1,6 @@
-export {};
+/**
+ * Needs to be run after retrieve_agenda.ts !
+ * - downloads the ZIP of comptes-rendus des débats (CRI) from data.senat.fr
+ * - extracts XML files, distributes them by session/year
+ */
+export declare function retrieveCriXmlDump(dataDir: string, options?: Record<string, any>): Promise<void>;