@tricoteuses/senat 2.18.9 → 2.18.11

package/lib/scripts/retrieve_videos.js

@@ -10,7 +10,7 @@ import { commonOptions } from "./shared/cli_helpers";
 import { decodeHtmlEntities } from "../model/util";
 import { DateTime } from "luxon";
 // ===================== Constants =====================
-const MATCH_THRESHOLD = 0.6;
+const MATCH_THRESHOLD = 0.5;
 const MAX_CANDIDATES = 15;
 const STATS = { total: 0, accepted: 0 };
 const VIDEOS_ROOT_FOLDER = "videos";
@@ -65,10 +65,19 @@ function epochToParisDateTime(epochSec) {
         startTime: `${hh}:${mi}:${ss}.${ms}${offsetStr}`,
     };
 }
-function toTargetEpoch(time) {
+function toTargetEpoch(time, date) {
     if (!time)
         return null;
-    const dtLocal = DateTime.fromISO(time, { zone: "Europe/Paris" });
+    let dtLocal;
+    if (time.includes("T")) {
+        dtLocal = DateTime.fromISO(time, { zone: "Europe/Paris" });
+    }
+    else if (date) {
+        dtLocal = DateTime.fromISO(`${date}T${time}`, { zone: "Europe/Paris" });
+    }
+    else {
+        return null;
+    }
     if (!dtLocal.isValid)
         return null;
     return Math.floor(dtLocal.toUTC().toSeconds());
@@ -198,6 +207,8 @@ function score(agenda, agendaTs, sameOrg, videoTitle, videoEpoch, videoOrganes)
     const titleScore = Math.max(objetS, titleS);
     let timeScore = 0;
     if (agendaTs && videoEpoch) {
+        console.log("agendaTs", agendaTs);
+        console.log("videoEpoch", videoEpoch);
         const deltaMin = Math.abs(videoEpoch - agendaTs) / 60;
         timeScore = Math.exp(-deltaMin / 60);
     }
@@ -276,7 +287,7 @@ async function processGroupedReunion(agenda, session, dataDir) {
         // if (!options["silent"]) console.log(`[skip] ${agenda.uid} date/hour missing`)
         return;
     }
-    const agendaTs = toTargetEpoch(agenda.startTime);
+    const agendaTs = toTargetEpoch(agenda.startTime, agenda.date);
     const now = Date.now();
     if (agendaTs && agendaTs * 1000 > now) {
         return;
@@ -322,13 +333,6 @@ async function processGroupedReunion(agenda, session, dataDir) {
         if (!dataBuf)
             continue;
         const meta = parseDataNvs(dataBuf.toString("utf-8"));
-        let sessionStart;
-        const finalBuf = await fetchBuffer(finalUrl);
-        if (finalBuf) {
-            const finalMeta = parseFinalNvs(finalBuf.toString("utf-8"));
-            sessionStart = finalMeta.sessionStart;
-        }
-        const videoEpoch = sessionStart ?? meta.epoch;
         let sameOrg = false;
         // If organes are too different, go to next candidates
         if (agenda.organe && meta.organes?.length) {
@@ -349,7 +353,7 @@ async function processGroupedReunion(agenda, session, dataDir) {
             if (hasSameKey) {
                 sameOrg = true; // we are sure this is the same org
             }
-            else if (bestDice < 0.7) {
+            else if (bestDice < 0.8) {
                 // if diff org and dice too low we skip
                 continue;
             }
@@ -358,7 +362,7 @@ async function processGroupedReunion(agenda, session, dataDir) {
         if (c.isSeancePublique && meta.firstChapterLabel) {
             videoTitle = meta.firstChapterLabel;
         }
-        const s = score(agenda, agendaTs, sameOrg, videoTitle, videoEpoch, meta.organes);
+        const s = score(agenda, agendaTs, sameOrg, videoTitle, meta.epoch, meta.organes);
         if (!best || s > best.score) {
             best = {
                 id: c.id,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@tricoteuses/senat",
-  "version": "2.18.9",
+  "version": "2.18.11",
   "description": "Handle French Sénat's open data",
   "keywords": [
     "France",