merge: PC6 journal n8n + JournalList (round 2)

This commit is contained in:
Jules Neny
2026-05-09 01:15:10 +02:00
6 changed files with 521 additions and 8 deletions

View File

@@ -4,3 +4,7 @@
# 3) Recupere les feed IDs et copie ce fichier vers .env.local puis remplis ci-dessous # 3) Recupere les feed IDs et copie ce fichier vers .env.local puis remplis ci-dessous
PUBLIC_BEHOLD_AEP= PUBLIC_BEHOLD_AEP=
PUBLIC_BEHOLD_JULESNENY= PUBLIC_BEHOLD_JULESNENY=
# Journal unifie (PC6) - URL JSON agrege par n8n cron nocturne
# Override en local : pointer vers un mock /public/data/journal.json par exemple
PUBLIC_JOURNAL_URL=https://data.trans-former.fr/journal.json

View File

@@ -0,0 +1,147 @@
# PC6 — Setup journal unifié + n8n workflow
Spec : `0 INBOX/PROMPTS/page-cerveau-build/PROMPT-PC6-journal-n8n.md`
Pilote : `0 INBOX/PROMPTS/page-cerveau-build/PILOTE-PC.md` (delta 5, delta 15)
## TL;DR
Le journal de la colonne G est alimenté par un cron n8n nocturne qui agrège plusieurs sources publiques (Gitéa Atom, Behold @aep, Behold @julesneny) et écrit un JSON statique servi par Caddy sur `data.trans-former.fr/journal.json`.
```
n8n cron (3h00 UTC)
-> fetch Gitéa Atom + Behold @aep + Behold @julesneny
-> normalisation Code Node
-> tri desc + top 100
-> écrit /home/node/.n8n/journal/journal.json (volume Docker)
-> Caddy data.trans-former.fr file_server expose ce fichier
-> JournalList.vue fetch côté client (no rebuild Astro requis)
```
## Sources V1 actives
| Plateforme | Hashtag | URL feed | Statut |
|---|---|---|---|
| Gitéa | `#stack` | `https://git.trans-former.fr/jules.atom` | ACTIF |
| Behold @aep | `#aep-politique` | `https://feeds.behold.so/{PUBLIC_BEHOLD_AEP}` | conditionnel (skip si feed ID absent) |
| Behold @julesneny | `#peinture` | `https://feeds.behold.so/{PUBLIC_BEHOLD_JULESNENY}` | conditionnel (skip si feed ID absent) |
## Sources skipped (V1 -> V1.5/V2)
| Plateforme | Hashtag | Raison |
|---|---|---|
| GitHub.com | `#stack` | username `julesneny` n'existe pas (HTTP 404). Pivot Gitéa pour le MVP. À reconfirmer si Jules a un autre handle GitHub public. |
| Substack | `#politique` | `transformations.substack.com` est pris par "WoodHorse" (pas Jules). Handle Substack à confirmer avant V1.5. |
| LinkedIn | `#building-public` | V2 (RSS via service tiers ou scrape) |
| Castopod | `#podcast` | V2 (Castopod RSS prêt mais hors scope MVP) |
| Blog `trans-former.fr` | `#manifeste` | V2 (post-PC8 deploy) |
## Format JSON
```json
{
"generatedAt": "2026-05-09T03:00:00Z",
"items": [
{
"id": "gitea-2026-05-09-pc6",
"platform": "gitea",
"hashtag": "#stack",
"date": "2026-05-09T01:01:00Z",
"titre": "PC6 journal unifié + n8n agrégateur",
"extrait": "...",
"url": "https://git.trans-former.fr/jules/astro-site-cerveau/commit/...",
"thumbnail": null
}
],
"counts": { "total": N, "gitea": N, "instagram": N }
}
```
## Composant Vue
`src/components/vue/JournalList.vue` :
- fetch `import.meta.env.PUBLIC_JOURNAL_URL` (défaut `https://data.trans-former.fr/journal.json`)
- écoute `window.addEventListener('hashtag-filter-change', ...)` émis par ColJournal.astro
- filtre par hashtag (vide ou tous cochés -> tout afficher ; tous décochés -> rien)
- tri desc déjà fait côté n8n, le composant respecte l'ordre
Cabling : `src/components/astro/ColJournal.astro` importe et rend `<JournalList client:visible />` dans `#journal-list`.
## Variable d'env
`PUBLIC_JOURNAL_URL=https://data.trans-former.fr/journal.json` (`.env.example`)
Override possible en local pointant vers `/data/journal.json` (mock fourni dans `public/data/journal.json`).
## Setup VPS — étapes (ops, à valider Jules)
### 1. DNS
Dans OVH zone DNS `trans-former.fr` :
```
data A 178.104.106.195 TTL 600
```
Attendre propagation (~5min).
### 2. Volume Docker partagé n8n -> Caddy
Le container n8n monte `vps-kit_n8n_data:/home/node/.n8n`. On va simplement lire un fichier dans ce volume depuis Caddy.
Path source : `/var/lib/docker/volumes/vps-kit_n8n_data/_data/journal/journal.json`
```bash
ssh vps-hetzner "mkdir -p /var/lib/docker/volumes/vps-kit_n8n_data/_data/journal && \
chown 1000:1000 /var/lib/docker/volumes/vps-kit_n8n_data/_data/journal"
```
### 3. Caddyfile bloc
Ajouter dans le Caddyfile (probablement `/etc/caddy/Caddyfile` ou `/opt/vps-kit/configs/Caddyfile`) :
```caddy
data.trans-former.fr {
root * /var/lib/docker/volumes/vps-kit_n8n_data/_data/journal
file_server {
index journal.json
}
encode gzip
header {
Cache-Control "public, max-age=300"
Access-Control-Allow-Origin "https://trans-former.fr"
}
log {
output file /var/log/caddy/data.log
}
}
```
Backup + reload :
```bash
ssh vps-hetzner "cp /etc/caddy/Caddyfile /etc/caddy/Caddyfile.bak.$(date +%Y%m%d-%H%M%S) && \
systemctl reload caddy && \
systemctl status caddy --no-pager | head -10"
```
### 4. Workflow n8n
Importer `docs/n8n-workflow-journal-aggregate.json` dans https://automate.trans-former.fr (UI -> Import from file).
Activer le toggle, vérifier le cron (`0 3 * * *`).
Configurer les credentials env n8n si besoin (Behold feed IDs) -> non bloquants si absents (workflow skip).
### 5. Smoke test
```bash
# Run manuel (UI n8n -> Execute Workflow)
ssh vps-hetzner "ls -la /var/lib/docker/volumes/vps-kit_n8n_data/_data/journal/"
curl -sf https://data.trans-former.fr/journal.json | jq '.counts'
```
## Backlog (hors scope PC6)
- Trigger rebuild Astro Coolify webhook (PC8)
- Sources V2 : LinkedIn, Castopod, Blog, Substack (post handle confirmé)
- Storage archivage long-terme (V1 = écrasement quotidien)
- Real-time updates (V3)

View File

@@ -0,0 +1,152 @@
{
"name": "journal-aggregate",
"nodes": [
{
"parameters": {
"rule": {
"interval": [
{
"field": "cronExpression",
"expression": "0 3 * * *"
}
]
}
},
"id": "schedule-trigger",
"name": "Cron-3h",
"type": "n8n-nodes-base.scheduleTrigger",
"typeVersion": 1.1,
"position": [240, 320]
},
{
"parameters": {
"url": "https://git.trans-former.fr/jules.atom",
"options": {
"response": {
"response": {
"responseFormat": "text"
}
},
"timeout": 15000
}
},
"id": "fetch-gitea",
"name": "Fetch-gitea",
"type": "n8n-nodes-base.httpRequest",
"typeVersion": 4.1,
"position": [460, 200]
},
{
"parameters": {
"url": "=https://feeds.behold.so/{{ $env.PUBLIC_BEHOLD_AEP || 'NOT_SET' }}",
"options": {
"response": {
"response": {
"neverError": true,
"responseFormat": "json"
}
},
"timeout": 15000
}
},
"id": "fetch-behold-aep",
"name": "Fetch-behold-aep",
"type": "n8n-nodes-base.httpRequest",
"typeVersion": 4.1,
"position": [460, 320]
},
{
"parameters": {
"url": "=https://feeds.behold.so/{{ $env.PUBLIC_BEHOLD_JULESNENY || 'NOT_SET' }}",
"options": {
"response": {
"response": {
"neverError": true,
"responseFormat": "json"
}
},
"timeout": 15000
}
},
"id": "fetch-behold-julesneny",
"name": "Fetch-behold-julesneny",
"type": "n8n-nodes-base.httpRequest",
"typeVersion": 4.1,
"position": [460, 440]
},
{
"parameters": {
"jsCode": "// Normalisation des 3 sources V1 vers le format JSON unifié\n// Sources : Gitéa Atom (XML), Behold @aep (JSON), Behold @julesneny (JSON)\n\nconst items = [];\n\n// ---- Helper parse Atom XML (Gitéa) ----\nfunction parseAtomGitea(xml) {\n const out = [];\n if (!xml || typeof xml !== 'string') return out;\n const entries = xml.split(/<entry>/i).slice(1);\n for (const raw of entries) {\n const block = '<entry>' + raw.split(/<\\/entry>/i)[0] + '</entry>';\n const title = (block.match(/<title>([\\s\\S]*?)<\\/title>/i) || [])[1] || '';\n const updated = (block.match(/<updated>([^<]+)<\\/updated>/i) || [])[1] || '';\n const link = (block.match(/<link[^>]*href=\"([^\"]+)\"/i) || [])[1] || '';\n const summary = (block.match(/<summary[^>]*>([\\s\\S]*?)<\\/summary>/i) || [])[1] || '';\n const id = (block.match(/<id>([^<]+)<\\/id>/i) || [])[1] || link || updated;\n if (!updated) continue;\n // Decode HTML entities basique + strip tags\n const decode = (s) => s\n .replace(/&lt;/g, '<')\n .replace(/&gt;/g, '>')\n .replace(/&quot;/g, '\"')\n .replace(/&#34;/g, '\"')\n .replace(/&#xA;/g, ' ')\n .replace(/&amp;/g, '&')\n .replace(/<[^>]+>/g, ' ')\n .replace(/\\s+/g, ' ')\n .trim();\n const titreClean = decode(title);\n const extrait = decode(summary).slice(0, 280);\n out.push({\n id: 'gitea-' + (id.slice(0, 80) || updated),\n platform: 'gitea',\n hashtag: '#stack',\n date: new Date(updated).toISOString(),\n titre: titreClean.slice(0, 140),\n extrait,\n url: link,\n thumbnail: null,\n });\n }\n return out;\n}\n\n// ---- Récupère payloads depuis les 3 nodes amont ----\nlet giteaXml = '';\ntry {\n const giteaItems = $('Fetch-gitea').all();\n if (giteaItems.length && giteaItems[0].json) {\n // HTTP node typeVersion 4.1 met le body brut dans .data si responseFormat=text\n giteaXml = giteaItems[0].json.data || giteaItems[0].json.body || '';\n if (typeof giteaXml !== 'string') giteaXml = String(giteaXml);\n }\n} catch (e) {\n console.log('Gitéa fetch missing:', e.message);\n}\n\nlet beholdAep = [];\ntry {\n const aep = $('Fetch-behold-aep').all();\n if (aep.length && aep[0].json && Array.isArray(aep[0].json.posts)) {\n beholdAep = aep[0].json.posts;\n } else if (aep.length && Array.isArray(aep[0].json)) {\n beholdAep = aep[0].json;\n }\n} catch (e) {\n console.log('Behold AEP fetch missing:', e.message);\n}\n\nlet beholdJulesneny = [];\ntry {\n const j = $('Fetch-behold-julesneny').all();\n if (j.length && j[0].json && Array.isArray(j[0].json.posts)) {\n beholdJulesneny = j[0].json.posts;\n } else if (j.length && Array.isArray(j[0].json)) {\n beholdJulesneny = j[0].json;\n }\n} catch (e) {\n console.log('Behold Julesneny fetch missing:', e.message);\n}\n\n// ---- Normalisation ----\nfor (const it of parseAtomGitea(giteaXml)) items.push(it);\n\nfor (const post of beholdAep) {\n if (!post || !post.id) continue;\n const ts = post.timestamp || post.taken_at || post.date;\n const caption = (post.caption || post.captionWithEmojis || '').toString();\n items.push({\n id: 'insta-aep-' + post.id,\n platform: 'instagram',\n hashtag: '#aep-politique',\n date: ts ? new Date(ts).toISOString() : new Date().toISOString(),\n titre: caption.slice(0, 100) || '@aep.politique',\n extrait: caption.slice(0, 280),\n url: post.permalink || post.url || 'https://instagram.com/aep.politique',\n thumbnail: post.thumbnailUrl || post.mediaUrl || (post.sizes && post.sizes.medium && post.sizes.medium.mediaUrl) || null,\n });\n}\n\nfor (const post of beholdJulesneny) {\n if (!post || !post.id) continue;\n const ts = post.timestamp || post.taken_at || post.date;\n const caption = (post.caption || post.captionWithEmojis || '').toString();\n items.push({\n id: 'insta-julesneny-' + post.id,\n platform: 'instagram',\n hashtag: '#peinture',\n date: ts ? new Date(ts).toISOString() : new Date().toISOString(),\n titre: caption.slice(0, 100) || '@julesneny',\n extrait: caption.slice(0, 280),\n url: post.permalink || post.url || 'https://instagram.com/julesneny',\n thumbnail: post.thumbnailUrl || post.mediaUrl || (post.sizes && post.sizes.medium && post.sizes.medium.mediaUrl) || null,\n });\n}\n\n// ---- Tri desc + cap top 100 ----\nitems.sort((a, b) => b.date.localeCompare(a.date));\nconst top = items.slice(0, 100);\n\nconst counts = {\n total: top.length,\n gitea: top.filter((i) => i.platform === 'gitea').length,\n instagram: top.filter((i) => i.platform === 'instagram').length,\n};\n\nreturn [\n {\n json: {\n generatedAt: new Date().toISOString(),\n items: top,\n counts,\n },\n },\n];"
},
"id": "normalise",
"name": "Normalise",
"type": "n8n-nodes-base.code",
"typeVersion": 2,
"position": [720, 320]
},
{
"parameters": {
"operation": "toJson",
"fieldName": "data",
"options": {
"format": true
}
},
"id": "to-json",
"name": "To-json-string",
"type": "n8n-nodes-base.set",
"typeVersion": 3.4,
"position": [940, 320],
"notesInFlow": false,
"notes": "Transforme l'objet JS en string JSON pour Write Binary File"
},
{
"parameters": {
"operation": "write",
"fileName": "/home/node/.n8n/journal/journal.json",
"dataPropertyName": "data",
"options": {}
},
"id": "write-file",
"name": "Write-journal-json",
"type": "n8n-nodes-base.readWriteFile",
"typeVersion": 1,
"position": [1160, 320]
}
],
"connections": {
"Cron-3h": {
"main": [
[
{ "node": "Fetch-gitea", "type": "main", "index": 0 },
{ "node": "Fetch-behold-aep", "type": "main", "index": 0 },
{ "node": "Fetch-behold-julesneny", "type": "main", "index": 0 }
]
]
},
"Fetch-gitea": {
"main": [[{ "node": "Normalise", "type": "main", "index": 0 }]]
},
"Fetch-behold-aep": {
"main": [[{ "node": "Normalise", "type": "main", "index": 0 }]]
},
"Fetch-behold-julesneny": {
"main": [[{ "node": "Normalise", "type": "main", "index": 0 }]]
},
"Normalise": {
"main": [[{ "node": "To-json-string", "type": "main", "index": 0 }]]
},
"To-json-string": {
"main": [[{ "node": "Write-journal-json", "type": "main", "index": 0 }]]
}
},
"settings": {
"executionOrder": "v1",
"saveExecutionProgress": true,
"saveManualExecutions": true
},
"tags": [
{ "name": "page-cerveau" },
{ "name": "PC6" }
]
}

52
public/data/journal.json Normal file
View File

@@ -0,0 +1,52 @@
{
"generatedAt": "2026-05-09T01:00:00Z",
"fallback": true,
"note": "Mock local — agrégateur n8n live pousse sur data.trans-former.fr/journal.json. Ce fichier sert de fallback dev tant que data.trans-former.fr DNS/Caddy ne sont pas en place.",
"items": [
{
"id": "gitea-mock-pc6-feat",
"platform": "gitea",
"hashtag": "#stack",
"date": "2026-05-09T01:01:00Z",
"titre": "PC6 journal unifié + n8n agrégateur (mock)",
"extrait": "Composant JournalList Vue + workflow n8n cron 3h00. Sources V1 : Gitéa Atom + Behold @aep + Behold @julesneny.",
"url": "https://git.trans-former.fr/jules/astro-site-cerveau",
"thumbnail": null
},
{
"id": "gitea-mock-pc3-mindmap",
"platform": "gitea",
"hashtag": "#stack",
"date": "2026-05-09T00:59:41Z",
"titre": "PC3 mindmap Carte O (D3 force-directed)",
"extrait": "Scrape AEP/Articles + tabs centre HAUT.",
"url": "https://git.trans-former.fr/jules/astro-site-cerveau/commit/32bdc9a",
"thumbnail": null
},
{
"id": "insta-mock-aep-1",
"platform": "instagram",
"hashtag": "#aep-politique",
"date": "2026-05-07T18:30:00Z",
"titre": "Mock carrousel @aep.politique",
"extrait": "Placeholder carrousel manifeste écologie politique. Cron n8n live remplace ce mock par la vraie API Behold.",
"url": "https://instagram.com/aep.politique",
"thumbnail": null
},
{
"id": "insta-mock-julesneny-1",
"platform": "instagram",
"hashtag": "#peinture",
"date": "2026-05-05T14:00:00Z",
"titre": "Mock peinture @julesneny",
"extrait": "Placeholder art / poésie / Corse. Cron n8n live remplace ce mock par la vraie API Behold.",
"url": "https://instagram.com/julesneny",
"thumbnail": null
}
],
"counts": {
"total": 4,
"gitea": 2,
"instagram": 2
}
}

View File

@@ -1,6 +1,8 @@
--- ---
// ColJournal - colonne gauche : CTA Manifeste + Hashtags accordeon + Journal skeleton // ColJournal - colonne gauche : CTA Manifeste + Hashtags accordeon + Journal (PC6)
// 7 hashtags = 7 plateformes (cf delta 15 du PILOTE-PC.md) // 7 hashtags = 7 plateformes (cf delta 15 du PILOTE-PC.md)
import JournalList from '../vue/JournalList.vue';
const hashtags = [ const hashtags = [
{ tag: '#manifeste', plateforme: 'Blog trans-former.fr', canal: 'ecriture longue' }, { tag: '#manifeste', plateforme: 'Blog trans-former.fr', canal: 'ecriture longue' },
{ tag: '#building-public', plateforme: 'LinkedIn', canal: 'journal pro' }, { tag: '#building-public', plateforme: 'LinkedIn', canal: 'journal pro' },
@@ -55,10 +57,7 @@ const hashtags = [
<span class="text-xs text-neutral-400 font-normal">chrono</span> <span class="text-xs text-neutral-400 font-normal">chrono</span>
</h2> </h2>
<div id="journal-list" class="space-y-3 text-sm"> <div id="journal-list" class="space-y-3 text-sm">
<!-- PC6 remplit ce slot via <JournalList client:visible /> --> <JournalList client:visible />
<p class="text-neutral-400 italic text-xs">
Chargement du journal...
</p>
</div> </div>
</section> </section>
</div> </div>

View File

@@ -1,9 +1,168 @@
<script setup lang="ts"> <script setup lang="ts">
// Placeholder journal list — PC6 lit public/data/journal.json import { ref, computed, onMounted, onUnmounted } from 'vue'
interface JournalItem {
id: string
platform: 'substack' | 'gitea' | 'github' | 'instagram' | 'castopod' | 'blog' | 'linkedin'
hashtag: string
date: string
titre: string
extrait: string
url: string
thumbnail: string | null
}
interface JournalPayload {
generatedAt: string
items: JournalItem[]
counts?: Record<string, number>
}
const items = ref<JournalItem[]>([])
const filters = ref<Record<string, boolean>>({})
const loading = ref(true)
const errored = ref(false)
const empty = ref(false)
const JOURNAL_URL =
(import.meta as unknown as { env: Record<string, string | undefined> }).env
.PUBLIC_JOURNAL_URL || 'https://data.trans-former.fr/journal.json'
const STORAGE_KEY = 'tf-hashtag-filters'
const loadFilters = () => {
try {
filters.value = JSON.parse(localStorage.getItem(STORAGE_KEY) || '{}')
} catch {
filters.value = {}
}
}
const onFilterChange = (e: Event) => {
const ce = e as CustomEvent
if (ce.detail && typeof ce.detail === 'object') {
filters.value = { ...ce.detail }
}
}
onMounted(async () => {
loadFilters()
window.addEventListener('hashtag-filter-change', onFilterChange as EventListener)
try {
const res = await fetch(JOURNAL_URL, { cache: 'no-store' })
if (!res.ok) {
errored.value = true
return
}
const data = (await res.json()) as JournalPayload
items.value = Array.isArray(data.items) ? data.items : []
if (items.value.length === 0) empty.value = true
} catch (e) {
errored.value = true
console.error('JournalList fetch failed', e)
} finally {
loading.value = false
}
})
onUnmounted(() => {
window.removeEventListener('hashtag-filter-change', onFilterChange as EventListener)
})
const visibleItems = computed(() => {
// Tous les filtres faux = aucun affiche ; tous true ou aucune cle = tout afficher
const keys = Object.keys(filters.value)
if (keys.length === 0) return items.value
const activeKeys = keys.filter((k) => filters.value[k])
// Si l'utilisateur a explicitement décoché tous les hashtags, on n'affiche rien
if (activeKeys.length === 0 && keys.some((k) => filters.value[k] === false)) return []
if (activeKeys.length === 0) return items.value
return items.value.filter((it) => activeKeys.includes(it.hashtag))
})
const formatDate = (iso: string) => {
try {
const d = new Date(iso)
if (isNaN(d.getTime())) return ''
return `${String(d.getDate()).padStart(2, '0')}/${String(d.getMonth() + 1).padStart(2, '0')}`
} catch {
return ''
}
}
const platformLabel = (p: string) => {
switch (p) {
case 'gitea':
case 'github':
return 'code'
case 'substack':
return 'substack'
case 'instagram':
return 'insta'
case 'castopod':
return 'pod'
case 'blog':
return 'blog'
case 'linkedin':
return 'linkedin'
default:
return p
}
}
</script> </script>
<template> <template>
<div class="h-full w-full flex items-center justify-center text-sm text-neutral-400"> <div class="space-y-3">
Journal list placeholder (PC6) <p v-if="loading" class="text-neutral-400 italic text-xs">
Chargement du journal...
</p>
<p v-else-if="errored" class="text-neutral-400 italic text-xs">
Journal en cours d'agregation. Reviens dans quelques heures.
</p>
<p v-else-if="empty" class="text-neutral-400 italic text-xs">
Aucun item agrege. Le cron nocturne tourne a 3h.
</p>
<p
v-else-if="!visibleItems.length"
class="text-neutral-400 italic text-xs"
>
Aucun item ; ajuste les filtres.
</p>
<article
v-for="item in visibleItems"
:key="item.id"
class="border-b border-neutral-100 pb-3 last:border-b-0"
>
<a
:href="item.url"
target="_blank"
rel="noopener noreferrer"
class="block group"
>
<div class="flex items-baseline gap-2 text-[11px] text-neutral-500 mb-1">
<time>{{ formatDate(item.date) }}</time>
<span class="font-mono text-neutral-700">{{ item.hashtag }}</span>
<span class="text-neutral-400">{{ platformLabel(item.platform) }}</span>
</div>
<h4
class="text-sm text-neutral-900 group-hover:text-neutral-600 leading-snug"
>
{{ item.titre }}
</h4>
<p
v-if="item.extrait"
class="text-xs text-neutral-500 mt-1 line-clamp-2"
>
{{ item.extrait }}
</p>
<img
v-if="item.thumbnail"
:src="item.thumbnail"
:alt="item.titre"
loading="lazy"
class="mt-2 w-full max-h-32 object-cover rounded"
/>
</a>
</article>
</div> </div>
</template> </template>