Commit Graph

186 Commits

Author SHA1 Message Date
louispaulb
1673f7cd9a Pont : import basé sur l'ADRESSE DE SERVICE (delivery), pas la facturation (compte)
Problème de fond : quand ticket.delivery_id ne se joignait pas, le pont retombait sur l'adresse de
FACTURATION du compte (résidence) → sujet + géocodage faux. Or les 124 tickets ouverts-3301 ont tous un
delivery côté compte (98 via ticket.delivery_id, 26 via le compte).

- fetchTargoTickets : résolution delivery robuste — COALESCE(ticket.delivery_id, delivery du compte AVEC
  coords [plus récent], delivery du compte [plus récent]) → l'adresse de SERVICE est toujours disponible.
- buildJob préfère déjà svcAddr (delivery) au billAddr → sujet + géocodage utilisent le service.
- Rafraîchit le `subject` des jobs encore au pool (open + non assigné) pour refléter l'adresse de service
  (corrige les anciens sujets basés sur la facturation) ; ne touche pas un job déjà dispatché.

Résultat (re-sync) : delivery 26→37, no_coords 6→1, 0 erreur. Jobs ouverts affichent l'adresse de service
(« Camping Sandysun · 32 Bellevue », « Lac des pins · 25 Érable », « Franklin · 35 rue Hilltop »…).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-07 17:36:17 -04:00
louispaulb
de496f8405 Pont : géoloc camping (fixe) sur les Dispatch Jobs — l'adresse de service ≠ résidence du client
Symptôme : un job de camping (« Lac des pins | Anton Rimerov ») pointait sur la RÉSIDENCE du client
(428 Rue George, Lasalle = 45.58,-73.73) au lieu du camping. Le pont géocodait l'adresse de compte.

- buildJob : détection camping en PRIORITÉ MAX via le registre camping_registry — signal = sujet (label
  explicite, prioritaire) puis ville/adresse de delivery. Garde-fou : le texte doit contenir « camping » OU
  un mot-clé de LIEU spécifique (évite les faux positifs de patronyme, ex. « Daniel Dauphinais »). coord_src='camping'.
  La branche update fait écraser les coords existantes par le camping (comme delivery). 20 jobs ouverts re-coordonnés.
- camping_dispatch_backfill.sql : corrige les jobs DÉJÀ dispatchés (que le sync ne re-traite plus car le ticket
  legacy a quitté le pool ouvert-3301) → 4 Lac des Pins + 2 SandySun. Anton Rimerov/Germaine Thibert → 45.0624,-73.9113 ✓.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-07 17:30:11 -04:00
louispaulb
ab5eceb961 Conformité : repli « centre du code postal / ville » pour les unmatched restants (statut 'area')
Dernier recours quand l'adresse exacte est introuvable : placer le Service Location au CENTROÏDE
(rqa_addresses) de son code postal (préféré) sinon de sa ville → le job apparaît dans le bon secteur.
- Hub : applyAreaFallback() (CTE centroïdes CP/ville, index-friendly) + POST /address/conformity/apply-area.
  Statut 'area', linked_address '≈ centre <CP/ville>'. Hors-QC/junk (absents de rqa_addresses) restent unmatched.
- Ops : carte stat « ≈ Secteur (CP/ville) » + bouton « ≈ Centre CP/ville (reste) » dans la page Conformité.

Exécuté : 317 placés (303 par code postal, 14 par ville) → unmatched 365 → 48 (Toronto/boîtes postales/junk).
État final : validated 16 257 · review 489 · area 317 · unmatched 48 → 99,7 % des services ont une position.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-07 17:14:41 -04:00
louispaulb
ed651d3ce9 Campings : gestion du registre + réapplication self-service depuis Ops
Rend le mécanisme réutilisable (« faire de même pour tous les campings ») :
- Hub (address-conformity.js) : GET /address/conformity/campings (registre + nb lots par camping),
  POST /campings (upsert {keyword,name,address,lat,lon} → applique direct), POST /campings/apply (réappliquer).
  applyCampings() = UPDATE des lots (match ville normalisée) → géoloc fixe du camping.
- Ops (page Conformité adresses) : section « Campings — géoloc de remplacement fixe » : table du registre
  (nom, adresse principale, GPS→Google Maps, nb lots) + formulaire d'ajout (nom/mot-clé/adresse/lat/lon)
  qui ajoute ET applique, + bouton « réappliquer ». api/address.js : campingsList/Upsert/Apply.

→ Pour un nouveau camping : on saisit son adresse principale + GPS, tous ses lots pointent dessus (le tech
navigue au camping). Registre seedé : Lac des Pins, Dauphinais, SandySun, Frontière, Ensoleillé.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-07 17:07:15 -04:00
louispaulb
fbf584f02f Phase 1 (hygiène) : utils partagés + logique pure testable + observabilité erp + 1ers tests
Modularisation / dé-duplication :
- lib/util/text.js : `norm` canonique partagé (remplace 2 ré-implémentations : address-db, legacy-dispatch-sync).
- lib/util/legacy-parse.js : parseurs/mapping PURS du pont (DEPT_JOBTYPE, DUR, jobType, prio, tzDate,
  startTime, coord) extraits hors I/O → testables en isolation, sans pg/mysql/erp.
- legacy-dispatch-sync + address-db importent ces utils (pont vérifié en prod : preview OK, 0 erreur).

Observabilité (sûr, additif, 1 seul point) :
- erp.js create/update/remove : log de l'échec à la SOURCE quand HTTP≥400 → toutes les écritures ERPNext
  silencieuses des 50+ appelants sont désormais tracées, SANS changer aucun flux de contrôle.

Tests (fondation) :
- vitest + npm test ; test/util.test.js : 19 tests verts sur norm + coord(bornes QC)/prio/startTime/jobType/tzDate.
  Tournent sans installer les deps lourdes du hub (modules purs).

Aligné docs/architecture/VISION.md (P0 hygiène). Suite : audit r.ok des appelants financiers (payments/contracts)
en revue supervisée ; CI/CD minimal (Gitea Actions lint+test) ; décomposition des god-files (Phase 2).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-07 10:36:41 -04:00
louispaulb
fc95a6cd4f Optimisation (consolidation helpers address) + doc Vision/modularisation
Optimisation sûre (vérifiée, 0 régression) :
- helpers.js : `cors()` partagé (en-têtes CORS génériques) au lieu de 2 copies locales.
- address-conformity.js : réutilise `pool` (address-db) + `cors` (helpers) au lieu de redéfinir un Pool +
  cors → 1 seul client pg local partagé pour rqa_addresses/fiber.
- address-validate.js : utilise helpers.cors.

docs/architecture/VISION.md (NOUVEAU) — vision + plan de modularisation + roadmap d'optimisation, fondé sur
un audit chiffré (hub 58 modules/23k lignes, Ops 45k lignes, god-files identifiés). Découpe en 9 domaines
(bounded contexts), principe « source de vérité + validation à la saisie + lien stable » (modèle Adresses
généralisé à Client/Device/Service), optimisations P0/P1/P2, métriques de succès. Complète les docs
architecture existants + ENGINEERING_PRACTICES.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-07 01:28:54 -04:00
louispaulb
72d23665b2 Page Ops « Conformité des adresses » — source de vérité unique pour résoudre le backlog
Plus besoin de re-chercher avec un processus complexe : une page liste les adresses de service non
conformes (review/unmatched) avec leur proposition AQ canonique, et permet de RÉSOUDRE une fois (persisté) :
- Approuver : la proposition AQ devient officielle (validated, coords RQA).
- Corriger : recherche AQ locale (rqa_addresses + fibre) → lier la bonne adresse.
- GPS : saisir/coller lat,long (relevé sur map.targointernet.com qui a la géoloc des unités de camping)
  + lien direct « voir sur la carte » par ligne.
- Rejeter : pas d'adresse civique (boîte postale/hors-QC) → 'no_address'.

Tri par type (camping / civique à corriger / à confirmer / non-adresse) + stats + recherche + pagination.

Backend : lib/address-conformity.js (GET stats|list|candidates, POST resolve) sur le Postgres LOCAL,
routé /address/conformity/* (server.js). Front : api/address.js + pages/AddressConformityPage.vue + route
/conformite-adresses + entrée nav « Conformité adresses » (icône MapPinned, requires view_settings).

État courant : validated 15 195 · review 1 366 · unmatched 550 (camping 540 / civique 333 / non-adresse 93).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-07 01:06:55 -04:00
louispaulb
ccad864165 Recherche d'adresses : base LOCALE (Postgres ERPNext) au lieu du Supabase cloud externe
Le hub n'appelle plus rddrjzptzhypltuzmere.supabase.co. La base RQA + dispo fibre est DÉJÀ locale
dans le Postgres ERPNext (rqa_addresses 5,24M + fiber_availability 21,6k jointes par address_id),
le hub y accède (réseau erpnext_erpnext + module pg).

- NOUVEAU lib/address-db.js : recherche locale. Phase 1 (civique présent) = filtre numero btree +
  mots de rue → ~20-150 ms ; Phase 2 (sans civique) = word_similarity (`<%` indexable GIN) au lieu de
  similarity() plein (24-76 s sur 5,24M !) → ~700 ms, dans une txn SET LOCAL (seuil 0.6 + statement_timeout 4s).
  Renvoie 2 formes : searchLocal (mappée, compat historique) + searchRaw (colonnes brutes de la fonction).
- address-search.js : searchAddresses + searchAddressesRpc délèguent à address-db (plus aucun appel Supabase).
  → onboarding (/address/validate), checkout (/api/address-search) ET le pont (géocodage) passent en LOCAL.
- address-validate.js : endpoints PUBLICS pour le site web (CORS) — POST /rpc/search_addresses (compat
  Supabase RPC, tableau direct) + GET /address/search — servis depuis le PG local (fiber_available inclus).
- server.js : route /rpc/ → address-validate.

Résultat pont (vérifié) : couverture 112/125 (vs 109 via Supabase), rqa_geocode 8→25 (table locale plus
complète + search_text désaccentué), Mapbox 37→23, no_coords 16→13, 0 erreur. Le local est meilleur.
Env hub : ADDR_DB_* dans /opt/targo-hub/.env (défauts erpnext-db-1/_eb65bdc0c4b1b2d6).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-06 15:46:47 -04:00
louispaulb
60dc095f56 Pont legacy : géocodage RQA via la recherche TRIGRAM (RPC search_addresses) + garde-fou anti-faux-positif
- address-search.js : expose searchAddressesRpc() → RPC Postgres `search_addresses` (pg_trgm), la MÊME
  recherche que l'autocomplete de disponibilité fibre. Trouve les rues que l'ilike manquait (générique géré
  par la colonne odonyme_recompose_long + phase 2 trigram), priorise les CP J0L/J0S (territoire).
- geocodeRQA() (bridge) bascule de l'ilike vers la RPC. Garde-fou : la phase 2 trigram dérive quand le
  civique est absent du RQA (« 2245 René-Vinet » → « Rue Grenet, Montréal »). On n'accepte un résultat que si
  le civique concorde + un token de nom de rue correspond + (territoire J0L/J0S OU CP/ville legacy concordants).
  Vérifié sur les données réelles : accepte 494 Av Curry / 3055 Routhier / 228 Principale / 61 Jean-François ;
  rejette René-Vinet→Grenet/Panet, chemin Ridge→Ferme, rue West→Perras (bons faux positifs écartés).
- Le faible compte RQA (8) = haute précision (l'ilike comptait 17 dont des faux positifs). Mapbox couvre le
  reste (rues neuves/civiques absents) ; ~109/125 (87 %) coordonnés ; les « aucune » = campings/villes mal écrites.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-06 15:05:14 -04:00
louispaulb
e14026ae23 Pont legacy : coords GPS fiables (delivery→SL→RQA→Mapbox) + routage routier réel (Mapbox Matrix)
Pont (legacy-dispatch-sync.js) :
- Import des coordonnées par job via cascade : table legacy `delivery` (point de service exact,
  JOIN ticket.delivery_id) > Service Location ERPNext > géocodage RQA > géocodage Mapbox.
  Validation bornes Québec (coord()). Couverture 153/172 (89%).
- Géocodage RQA corrigé : retrait du générique de voie (Rue/Rang/Chemin absent de
  odonyme_recompose_normal) + code postal non accolé au terme (sinon ilike ne matche jamais).
- Repli Mapbox geocoding pour rues trop récentes pour le RQA (MAPBOX_TOKEN).
- Backfill + UPGRADE : coords delivery écrasent des coords SL moins précises (jamais l'inverse).
- Comptabilité honnête : vérifie r.ok sur create/update (erp ne throw pas) → errors + error_samples.
- Verrou de sérialisation sync() : tick + runs manuels ne se chevauchent plus (frappe_pg).
- Subject tronqué à 140 (champ Data) → corrige CharacterLengthExceededError sur jobs sans SL.
- Observabilité : coord_src tally + error_samples dans le résumé.

Ops Planification (éditeur de journée) :
- travelBetween() consulte une matrice Mapbox Matrix chargée à l'ouverture (loadDayRoute) →
  temps de trajet ROUTIERS RÉELS ; réordonnancement instantané sans nouvelle requête.
  Repli haversine si Mapbox indispo. Indicateur 🚗 réel vs 📏 vol d'oiseau.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-06 14:43:34 -04:00
louispaulb
22fbb81d3e feat(planif): éditeur journée = planificateur de tournée (heures recalculées, pas d'overlap, RDV verrouillables, détails)
- FIX overlap/ordre : packedDay recalcule les heures depuis la SÉQUENCE (ordre liste) + durées + transport ;
  le timeline et les heures affichées en découlent → réordonner/allonger repousse les suivants, plus d'overlap
- RDV à heure FIXE : verrou par job (🔒, défaut = booking_status 'Confirmé') → garde son heure ; flexibles = enchaînés
- clic sur un job → détails (legacy_detail : dates + description) pour juger urgence/durée ; + tooltip au survol
- save persiste start_time recalculé (+ route_order/priority/duration_h) via reorder-jobs
- hub occupancy renvoie booking_status/legacy_detail/legacy_id ; reorder-jobs accepte start_time
- fix collision fmtH → fmtHM (HH:MM padded)

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-06 13:56:26 -04:00
louispaulb
faea3b0c1a feat(planif): éditeur journée — réordonnancement fiable (flèches+drop) + durée éditable (min) + temps de transport
- FIX réordonnancement : flèches ↑↓ (fiable) + drag-drop basé sur le DROP (au lieu du splice live jittery)
- durée éditable par job en MINUTES (pas de 5, best practice précision) → persistée via reorder-jobs (duration_h)
- temps de transport estimé entre 2 jobs (haversine sur coords Service Location, 40km/h + 5min) affiché entre les lignes
  → en attendant la géoloc live (Capacitor background-geolocation, noté pour plus tard)
- hub : occupancyByTechDay renvoie lat/lon par job ; reorder-jobs accepte duration_h ; total h en pied

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-06 13:41:28 -04:00
louispaulb
d96e7be672 feat(planif): éditeur de JOURNÉE contextuel au clic sur le progressbar (drag-drop réordonner + retirer)
- clic sur le progressbar → q-dialog ciblé sur le tech×jour (garde le contexte de la grille derrière) :
  timeline visuelle (blocs colorés par compétence) + liste éditable des jobs
- réordonnancement par DRAG-DROP (dragstart/dragover/dragend → route_order) + sélecteur de priorité + Enregistrer
- retrait d'un job (✕ → hub POST /roster/unassign-job : assigned_tech null, status open → retour au pool)
- bouton « Dispatch » comme échappatoire vers le tableau complet (gotoDispatch)
- réutilise occupancy/cellBands/cellBlocks/blockStyle + reorderJobs ; best-practice détail-drawer (pas de navigation pleine page)

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-06 13:29:55 -04:00
louispaulb
be222d005e feat(dispatch): date d'ouverture + MàJ en tête du détail ticket (mouseover) pour juger la fermeture
- pont : SELECT date_create+last_update ; legacy_detail préfixé « 🗓 Ouvert <date> · MàJ <date> » puis description.
  Backfill = réécrit legacy_detail si différent (idempotent). Visible dans le mouseover du panneau roster + le détail Dispatch.
  Ex. révélé : LEG-111325 ouvert 2021 jamais touché → candidat fermeture évident.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-06 13:07:15 -04:00
louispaulb
8707cd7c2e fix(dispatch): pont lisait un réplica figé (avril) + auto-fermeture des DJ dont le ticket legacy est closed
- ROOT CAUSE : LEGACY_DB_HOST='legacy-db' = réplica figé au 2026-04-07 → tickets fermés/réassignés depuis
  paraissaient encore ouverts (ex. 244659). Fix infra : .env LEGACY_DB_HOST=10.100.80.100 (DB live, SELECT-only).
- closeResolved() : tout DJ issu du pont (open/assigned/On Hold) dont le ticket legacy est 'closed' → status 'Completed'.
  Appelé à chaque sync + route POST /dispatch/legacy-sync/close-resolved. Résultat 1er run live : 125 ouverts réels,
  102 créés, 44 fermés (dont LEG-244659). NB In Progress non touché.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-06 13:03:10 -04:00
louispaulb
510f8c10e0 feat(planif): blocs cellule colorés par compétence + menu réordonner/re-prioriser au clic sur le progressbar
- Blocs d'occupation = 1 job, coloré par la COULEUR DE SA COMPÉTENCE (palette skills éditable),
  au lieu du dégradé vert→rouge par taux. Hub occupancyByTechDay attache skill+job à chaque bloc
  (skillForJob||deptToSkill) ; blockStyle → getTagColor(blk.skill).
- Clic sur le progressbar (.tl, @click.stop+@mousedown.stop) → menu : liste des jobs du tech×jour avec
  réordonnancement (flèches → route_order) + re-priorisation (select) + Enregistrer.
  Hub POST /roster/reorder-jobs (route_order/priority, séquentiel) ; tri occupancy par route_order.
- Clic HORS du progressbar dans la cellule → menu shift inchangé (créer/modifier).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-06 12:44:22 -04:00
louispaulb
2631ed9199 feat(dispatch): couleur ticket = couleur skill + fil complet du ticket + tri pool (date/ville/priorité)
- Couleurs liées aux skills (éditable/cohérent) : hub deptToSkill() déduit une compétence du type legacy
  → /roster/unassigned-jobs renvoie required_skill ; PlanificationPage colore la carte par getTagColor(required_skill)
  (même couleur que le chip skill) ; bordure 5px
- Fil complet du ticket : hub /dispatch/legacy-sync/ticket-thread (ticket_msg + auteur staff, HTML nettoyé) ;
  api legacyTicketThread ; RightPanel bouton « 💬 Voir le fil / commentaires » (chargé au clic, messages+auteurs+dates)
- Order-by du pool dispatch : useBottomPanel.bottomSort (date|city|priority) + dropdown ⇅ dans BottomPanel
  (ville = 2e segment adresse / token sujet avant |)

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-06 12:37:45 -04:00
louispaulb
b492e30971 feat(dispatch): réconciliation + heartbeat + détails ticket dans Ops + couleurs panneau roster
#1 « ne rien échapper » :
- /dispatch/legacy-sync/reconcile : compare legacy(3301,open) ↔ Dispatch Jobs → missing/orphan (70↔70, 0/0)
- /dispatch/legacy-sync/status : heartbeat (last_sync + stale) pour Uptime-Kuma

Détails ticket dans Ops (bug signalé) :
- pont extrait le 1er message du fil legacy (stripHtml) → champ legacy_detail (backfill 70)
- store legacyDetail ; RightPanel : section « Détails du ticket » ; tooltip dans le panneau roster

Couleurs panneau roster (bug signalé) :
- /roster/unassigned-jobs renvoie job_type/legacy_dept/priority/legacy_* ; PlanificationPage colore
  chaque carte par type (legacyDeptColor partagé) — bordure gauche

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-06 11:50:21 -04:00
louispaulb
7f22647ce4 feat(dispatch): bouton « Activer STB (Ministra) » — lien d'activation TV extrait du ticket legacy (Phase 1)
- pont : extrait le lien connect_ministra.php du ticket_msg (déjà posté par le wizard legacy à la vente)
  → champ legacy_activation_url sur le Dispatch Job (backfill des 10 tickets TV). Read-only legacy, aucun 2e
  chemin d'écriture → zéro risque de double-facturation (cf. analyse processus).
- store _mapJob : expose legacyActivationUrl ; RightPanel : bouton violet « 📺 Activer STB (Ministra) »
  (MÊME lien que le tech reçoit, ouvre connect_ministra.php avec tid/did/sid/cr/m intacts).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-06 11:20:09 -04:00
louispaulb
3c17805770 feat(dispatch): coloriage cartes par type legacy + n° ticket dans le détail
- pont : stocke legacy_dept (dept osTicket) sur le Dispatch Job + backfill des 70 existants
- useHelpers.jobColor/legacyDeptColor : palette comme legacy (Installation vert, Réparation or,
  Télé rose, Téléphonie vert pâle, Désinstallation rouge foncé) ; tech en pause = rouge vif prioritaire
- store _mapJob : expose legacyDept + legacyTicketId
- RightPanel : champ « Ticket legacy » (#id · dept) — le client est déjà un lien cliquable vers la fiche
- doc mise à jour

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-06 10:05:45 -04:00
louispaulb
30d8b0b29c feat(dispatch): pont legacy(osTicket)→Dispatch Job pour les tickets « Tech Targo »
Tire régulièrement les tickets ouverts assignés au compte « Tech Targo » (staff 3301)
de la DB legacy MariaDB et crée/maj un Dispatch Job ERPNext (pool à répartir).

- lib/legacy-dispatch-sync.js : fetch (status=open AND assign_to=3301) + mapping
  customer (legacy_account_id) / Service Location (coords) / job_type (dept) /
  scheduled_date (epoch→America/Toronto) / start_time (am|pm|HH:MM) / priority
- Idempotent via Custom Field Dispatch Job.legacy_ticket_id (lookup avant create) ;
  ne clobbe pas le travail du répartiteur (maj date seulement si encore open+non assigné)
- SÉQUENTIEL (frappe_pg) ; endpoints GET preview (dry-run) + POST run
- Récurrence opt-in : startSync() au boot, LEGACY_DISPATCH_SYNC=on + _MIN=15
- server.js : route /dispatch/legacy-sync + startSync()
- doc docs/features/legacy-dispatch-bridge.md + index

Mise en service : 70 tickets importés (0 erreur), récurrence 15 min active.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-06 09:33:53 -04:00
louispaulb
e33dd8e83a roster(planif): assignation drag-drop + timeline ressource + occupation + nettoyage
Panneau « jobs à assigner » v2 : multi-sélection (cases), groupes parent-enfant
surlignés, heuristique terrain/à distance (activation/netadmin pré-décochés),
pré-total d'heures, aperçu d'occupation PROJETÉE au survol (barre fantôme + badge).

Fix barre d'occupation figée après drop : /roster/assign-job pose désormais un
start_time (premier-trou-libre dans le shift) + garantit duration_h, sinon le job
compte dans les heures mais n'affiche aucun bloc. Nouvel endpoint
/roster/backfill-start-times (idempotent) pour rattraper l'historique.

Infobulle de cellule : nb de jobs + liste triée par priorité (occupancyByTechDay
renvoie jobs[]). Timeline contextuelle par ressource (dialogue, 0 appel réseau).

Lisibilité du drag : fantôme compact semi-transparent décalé sous le curseur
(ne masque plus l'aperçu) + source estompée.

Scoring de priorité : hook proximité (neutre — secteur géré manuellement),
réservé à 20% du score quand la géoloc arrivera.

Refactor hub : helper partagé firstFitStart (assign-job + backfill).
Nettoyage : retrait code mort (onDeleteRosterTag, projUsedH), carte des sections
en tête de PlanificationPage. Doc : docs/features/roster.md + index.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-05 15:50:17 -04:00
louispaulb
9ec829fc6f Planification: marquer une absence d'1 jour depuis la grille (touche A + menu)
Clic sur une case → touche « A » (bascule) ou menu « Marquer absent ce jour / Retirer l'absence » →
crée/supprime une Tech Availability d'1 jour APPROUVÉE (hachurée tout de suite). Multi-sélection
supportée (A marque toutes, re-A retire). Endpoint POST /roster/absence/set {tech,date,type,remove}
(remove ne touche que les absences d'un jour, pas les vacances multi-jours). Type défaut Congé.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-04 22:02:39 -04:00
louispaulb
a60a9bee91 Garde: génération sur un HORIZON multi-semaines (évènement récurrent) au lieu d'une seule semaine
Avant: « Appliquer à la semaine » n'écrivait qu'une semaine → 1 seul tech (rotation hebdo). Maintenant
« Générer la garde » matérialise la rotation sur N semaines (4/8/12/26) d'un coup, écrit direct (publié),
navigable semaine par semaine — comme un évènement récurrent. Endpoint /roster/garde/apply : wipe ciblé
des shifts de garde dans l'horizon + recréation (idempotent, reflète l'édition de la séquence). Saut
d'absent conservé. (source='manuel' car le Select n'autorise que solveur/manuel.)

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-04 21:58:15 -04:00
louispaulb
2e5ef6d001 Roster publish: diff au lieu de wipe+recreate → publication quasi instantanée
Avant: /roster/publish-week supprimait TOUTES les assignations de la semaine puis les recréait toutes
(~2N écritures ERPNext séquentielles → plusieurs secondes/dizaines de s). Maintenant: diff par clé
tech|date|shift — supprime seulement les retirées, crée seulement les nouvelles, ignore les inchangées.
Republier sans changement: 87 assignations → 0 écriture, 37 ms (vs ~174 écritures avant). Une modif
de quelques cellules = quelques écritures seulement.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-04 21:20:03 -04:00
louispaulb
fedaf53a3b Planification: drapeau « longue durée » sur absence (maternité/invalidité = à remplacer)
Tech Availability gagne long_term (Check). Case à cocher « Longue durée » dans le dialogue Congés.
absencesByTechDay encode « (longue durée) » → applyTemplate force « à remplacer » (pas juste sauter
comme des vacances), même si l'absence ne couvre pas toute la semaine. Complète l'intelligence
permanent vs vacances (avant: déduit de « absent toute la semaine »).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-04 20:31:39 -04:00
louispaulb
bb337db14f Planification: hachuré = ABSENT (congé/pause), garde = pointillé ambre (sur appel hors heures)
- Hachuré gris = ABSENT (Tech Availability approuvée + En pause). Nouvel endpoint /roster/absences
  (En pause global + congés approuvés par jour) → la cellule d'un tech absent est hachurée (tooltip = type).
  Remplace l'ancien 'P' pause.
- GARDE = nouveau visuel: bande à CONTOUR POINTILLÉ AMBRE + fond ambre léger (sur appel, hors heures
  d'ouverture) — distinct du travail planifié et de l'absent.
- Légende: dispo (matin→soir) · occupation · absent (hachuré) · garde (pointillé). Retrait 'P pause'.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-04 19:43:47 -04:00
louispaulb
054deb7b97 Roster: la garde ne compte PAS comme heures travaillées (mise en dispo)
Garde (on_call) exclue partout des heures travaillées + du coût:
- hub statsByDay: heures = somme des shifts NON-garde ; nouveau compteur on_call/jour (techs en dispo).
- Ops: hoursOf (heures/tech + alerte heures supp) et costByDate/weekCost excluent la garde.
- nouvelle ligne de pied '🛡️ Garde' = nb de techs en disponibilité/jour (si applicable).
Cohérent avec l'occupation (déjà hors-garde) : la garde = réserve d'urgence, ni offerte ni facturée.
Vérifié 8 juin: 112 h travaillées (garde 6 h exclue), garde=1.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-04 15:52:01 -04:00
louispaulb
ebb440b5d8 Roster: quart de Garde (on_call) = réserve d'urgence, jamais offert au booking
Modèle: champ on_call (Check) sur Shift Template. Un quart garde:
- N'est JAMAIS offert au booking client (techGaps retourne null) — vérifié: tech Jour+Garde
  n'offre que la fenêtre Jour, aucun créneau dans la plage de garde.
- Est EXCLU du dénominateur d'occupation (heures offrables), affiché à part.
- Timeline: bande HACHURÉE (vs neutre pour l'offrable) + 🛡️ dans le label + tag (garde) en infobulle.
- Éditeur de modèles: bascule '🛡️ Garde' pour créer/marquer un quart de garde.
hub: fetchTemplates expose on_call; create/update template le gèrent. Champ ajouté à
setup_dispatch_custom_fields.py (persistance). Démo: Garde 18h-minuit marquée on_call.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-04 15:37:34 -04:00
louispaulb
2f868133d5 Planification: mini-timeline positionnée (fenêtre réelle du shift + blocs pris)
Avant: 'J8' ne distinguait pas 7-15 de 9-17 → mêmes créneaux apparents, dispo réelle différente.
Maintenant chaque cellule affiche: chip (lettre) + intervalle '7–15', et une mini-timeline sur un
axe de journée (06:00→21:00) où la fenêtre du shift est positionnée (donc 7-15 à gauche, 9-17 à
droite = visuellement distinctes) avec les blocs de jobs pris (couleur selon charge) → les TROUS
restants = créneaux offrables. Infobulle = intervalle + h occupées/h (%).

- hub occupancyByTechDay renvoie {h, blocks:[{s,e}]} (heures de début réelles des jobs).
- ops: cellWindow/axisPos/shiftStyle/blockStyle, rendu .tl/.tl-shift/.tl-blk + tick midi.
- démo 8 juin: modèles Matinal 7-15 + Décalé 9-17, techs alignés (7→13.8, 9→18.6 surbooké).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-04 15:21:04 -04:00
louispaulb
fd5c83d367 Planification: taux d'occupation par cellule (jobs assignés / heures du shift)
Chaque cellule tech×jour avec un shift affiche, sous le chip (J8), une mini-barre + % colorés
(vert <70, orange 70-99, rouge >=100 surbooké) + infobulle = intervalle du shift + h occupées/h.
Occupation = Σ duration_h des Dispatch Jobs planifiés assignés ce jour ÷ Σ heures du shift.

- hub: occupancyByTechDay(start,days) + GET /roster/occupancy → map 'TECH|date': heures.
- ops api: getOccupancy ; PlanificationPage: occCells (computed), cellOcc/occColor/cellInterval,
  rendu barre + q-tooltip, chargé dans loadStats. Données démo semaine 8 juin (45/85/120%).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-04 15:13:27 -04:00
louispaulb
96a76812d6 Copilote: agit vraiment sur les absences (gerer_absence + IROPS rematch) + gig dispo
Bug: le copilote comprenait l'absence mais n'agissait pas → aucun impact. Causes:
1) prompt « par défaut analyse » → ne déclenchait pas l'action ;
2) marquer une absence n'excluait PAS des créneaux (techGaps ne testait que le statut
   global En pause, pas les Tech Availability approuvées par jour) ;
3) loadBookingData calculait unavail mais ne le retournait pas (oubli) → garde inerte.

Fixes:
- roster.js: loadBookingData inclut unavail (buildUnavailability = En pause + absence_from/until
  + Tech Availability approuvées) ; techGaps exclut le tech absent ce jour-là ; export bookingSlots/fetchTemplates.
- roster-assistant.js: nouvel outil gerer_absence = crée l'absence (par jour) + trouve les RDV
  impactés + RÉASSIGNE auto à un autre tech libre du même créneau (IROPS), renvoie les
  « à reporter ». Nouvel outil ajouter_disponibilite (tech à l'acte ouvre un créneau). Prompt
  orienté ACTION (signaler une absence = instruction d'exécution).
- Validé prod (lab): copilote crée l'absence ✓, booking exclut le tech absent (109→104) ✓,
  rematch DJ-…→Antoine même créneau ✓ ; données de test nettoyées.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-04 14:43:03 -04:00
louispaulb
d94623e58a Booking #56: politique de créneaux offerts + holds temporaires
- getBookingPolicy() (sous-objet 'booking' du fichier policy): lead_hours, day_start/end,
  days_offered, horizon_days, max_per_day, hold_minutes. Appliquée dans bookingSlots →
  cohérent pour /book + vue agent + fit. ignorePolicy au moment de confirmer (slot encore libre).
- Holds en mémoire (Map TTL): /roster/book/hold {date,start,minutes|release} → fenêtre retirée
  des dispos des autres pendant le hold; libérée à la confirmation. Évite le double-booking
  pendant qu'un agent/client choisit.
- roster-assistant: DEFAULT_POLICY.booking + booking_fields/weekdays (descripteurs UI), fusion fine.
- Testé: plage 10-14h filtre bien; hold 10:00 dispo 12→11.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-04 14:19:42 -04:00
louispaulb
20ba2a327e Booking /book: grille semaine → jour → Matin/Après-midi (se situer dans le temps)
Avant: liste plate jj/mm peu lisible. Maintenant: bandeau de semaine ('Semaine du 8 – 14 juin'),
en-têtes de jour complets (lundi 8 juin), sections Matin/Après-midi. Sélection 1-3 préférences inchangée.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-04 14:04:08 -04:00
louispaulb
fa696fb8e9 Fix reschedule: notify-reschedule désassigne (vide date/heure/tech, status open) → /book repropose des créneaux
Avant: le job gardait scheduled_date → page /book court-circuitait en 'déjà confirmé' sans options.
Maintenant un report = vrai retour au pool → le client choisit un nouveau créneau (vérifié: 50 créneaux proposés, SMS livré).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-04 13:56:20 -04:00
louispaulb
b77e251e8e Reschedule: endpoint aviser-client (lien /book + SMS Twilio + statut 'À reporter') + file 'À reporter' + outil copilote
/roster/job/notify-reschedule (job,phone?) → token /book + SMS + booking_status='À reporter'.
/roster/jobs-to-reschedule → file superviseur. Copilote: outil notifier_client_report.
Testé OK (chemin sans téléphone = sûr). Brique P5 de #55, déclenchable depuis le flux de pause.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-04 13:28:45 -04:00
louispaulb
b2176c54b0 Copilote roster: outils d'action (marquer_indisponibilite + reassigner_job)
Le copilote peut désormais CHANGER les dispos en langage naturel ('marque Simon malade
aujourd'hui' → crée une Tech Availability approuvée) et réassigner un Dispatch Job, sur
instruction explicite, avec confirmation. Testé OK end-to-end. (SMS client + escalade = suite.)

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-04 11:57:33 -04:00
louispaulb
5f7cca60aa Copilote roster (Gemini Flash, function-calling) + politique de reprise configurable
lib/roster-assistant.js : couche conversationnelle sur le roster (distincte du solveur OR-Tools).
Outils data réels (etat_equipe, jobs_du_technicien) via roster.fetchTechnicians + Dispatch Job.
Ex: 'TECH-4776 malade le 16 juin' → résout le nom, liste les RDV impactés, propose des techs
dispos qualifiés. Routes /roster/assistant + /roster/policy (politique persistée fichier).
Réutilise le moteur geminiChat de lib/agent.js (gemini-2.5-flash). Testé OK avec données réelles.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-04 11:08:26 -04:00
louispaulb
70f02c8c85 Store: prix barré par produit via champ Item.store_regular_price
Affiché barré (gris) si > prix de vente (Item Price), sur produits simples + variantes
(les bundles gardent barré=somme des composants). Catalogue lit le champ; page rend <s>.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-03 21:55:57 -04:00
louispaulb
98b43fd36f Store: catalogue live depuis ERPNext (GET /store/catalog)
- Curation = champ Item.show_in_store (case à cocher) ; catégorie=item_group, prix=Item Price, stock=Bin live
- Bundles via Product Bundle (prix vs somme barrée), variantes mono-attribut (delta prix + stock/valeur)
- Fix filtres erp.list (tableaux, pas objets — listUrl teste .length) + child-tables lues via erp.get parent
- Page: fetch /store/catalog au mount, repli démo, bandeau live/démo, stock simple respecté

Prérequis PG corrigé: retrait ORDER BY NULL (MySQLism) dans erpnext utilities/product.py
(validation de variante) — sinon création de variantes impossible sur PostgreSQL.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-03 21:23:01 -04:00
louispaulb
ae9c93d515 Boutique matériel — page modèle /store (staging, self-contained Vue 3)
Vitrine fluide inspirée des meilleurs carts : grille produits, visualisation de
variantes (swatches couleur + pills longueur/modèle, prix delta live, rupture barrée),
panier optimistic (drawer slide, steppers, badge animé, toast, taxes QC, total).
Données démo (caméras/boosters/câbles/TP-Link/bundles). Branchement ERPNext (catalog
+ Product Bundle + pricing) = étape suivante. Validé live au navigateur.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-03 20:09:47 -04:00
louispaulb
c176c51fbb Wizard: 2 cases génériques Prix marketing / Prix original (barré) sur toutes les lignes + aperçu client WYSIWYG
- Prix original (regular_price) disponible sur lignes récurrentes ET ponctuelles (avant: ponctuelles seulement)
- Si Prix original > Prix marketing → barré affiché (sommaire + récap + aperçu live dans l'éditeur)
- Forfait récurrent barré remonte au contrat (monthly_regular) → page d'acceptation affiche <s>99.95</s> 79.95/mois
- Cohérent avec install (install_regular 360→240). Aucun impact CRTC (rabais go-forward, jamais récupéré)

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-03 18:31:54 -04:00
louispaulb
a3679e5b4a Install: prix barré 360$ → 240$ (ancre de valeur) + 'financement crédité chaque mois 24 mois' dans contrat + portail signup
Champ install_regular (Service Contract) = valeur affichée barrée; install_fee reste le montant financé/dû réel. Conforme CRTC: 360 = référence marketing seulement.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-03 18:22:22 -04:00
louispaulb
735de73fde Portail self-service abonnement (staging /signup): stepper épuré Forfait→Coordonnées→Récap→Confirmation, catalogue réel, capture Lead
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-03 18:16:19 -04:00
louispaulb
d05494d4d2 Récap contrat: détail mensuel (forfait + financement install − rabais nouveau client = net), install offerte tant qu'abonné
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-03 18:08:08 -04:00
louispaulb
8c7f14c00a Contrat résidentiel: conformité CRTC 2026-43 (fin du clawback)
Installation = vraie créance financée (champ install_fee : 240$ std → 10$/mois,
120$ simple → 5$/mois). Résiliation résidentielle ne facture que le solde restant
de l'install (service rendu) ; promotions/carte-cadeau jamais récupérées
(termination_fee_benefits=0). Récap reformulé (financement install, plus de
"promotion non étalée"). Conforme avant l'échéance du 12 juin.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-03 18:05:56 -04:00
louispaulb
1a4c45662a Roster AI (planification) + prise de rendez-vous client
Solveur OR-Tools (services/roster-solver) : couverture, compétences,
équité, coût chargé, cadence/efficacité, capacité-par-job ; contraintes
dures/souples façon Timefold.

Hub (lib/roster.js) : génération via solveur, publication par réécriture
de semaine (anti-doublons), demande (effectif ou nb de jobs), cadence/coût/
compétences par tech, pause, congés (Tech Availability + approbation),
booking (slots roster-aware / fit 3-dispos / confirm) + portail public /book.
Réessai sur serialization failures frappe_pg ; appels ERP séquentiels.

Ops : page Planification (grille compacte « J8 », multi-shift, drag-select
+ undo/redo, modèles de semaine, éditeur cadence&coût, congés, SMS opt-in),
page Rendez-vous (répartiteur), jobColor tech en pause → tickets rouges.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-03 16:42:44 -04:00
louispaulb
7f8382496e feat(ops): Email Queue admin page (view/delete/purge ERPNext outbound)
Surfaces ERPNext's Email Queue in Ops (nav « File courriels ») so ops can see
what's queued — important now that mute_emails=1 + scheduler paused mean nothing
flushes — and delete/purge stale entries without the ERPNext desk.

- hub lib/email-queue.js: GET list (by status, recipients read from each row's
  full doc since ERPNext ignores fields on child-doctype REST), DELETE :name,
  POST /purge {status}. Wired in server.js.
- ops: api/emailQueue.js + EmailQueuePage.vue (status filter, recipients,
  reference, error tooltip, per-row delete + « Purger Not Sent »), route + nav.

Verified live: 13 'Not Sent' (old internal test emails, no invoice refs).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-02 17:29:33 -04:00
louispaulb
be81a7ef0b fix(legacy-report): treat accounts with a company name as commercial
98 business accounts had commercial=0 but a company name (Ferme X Inc.,
Ville de Farnham, Les Jardins Sorel…), leaking into the residential report.
Rule is now: commercial = (commercial flag) OR (company present). Residential
@90$ drops 739→654 (0 company-bearing rows left); commercial 244→276.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-02 06:33:25 -04:00
louispaulb
5592ab6871 feat(ops): per-address competitor column via Québec IHV open data
Replaces the reCAPTCHA-blocked Cogeco scraper with the authoritative Québec
"Accès Internet haute vitesse" open ArcGIS data (providers declared to the
gov by the providers themselves — validated to match Cogeco's own popup).

- hub lib/serviceability.js: address → ADR (Adresse_complete → IdAdresse +
  Etat_hiv, civic+postal match w/ JS street disambiguation) → FRN table
  (IdAdresse → FRN_nom providers + signup URLs). Referer-gated proxy, disk
  cache (90d), polite rate limit. Routes /serviceability/lookup[-batch].
- ops ReportInternetCherPage: "Concurrence (FSI)" column — provider chips
  (Cogeco highlighted), batch-fetched on demand with progress; "Cogeco
  disponible" summary card = churn-risk count; manual Cogeco verify icon kept.

Validated live: 37 Chemin Noël → Cogeco+Targo, 147 Montée Richard → Targo
only, Repentigny → Bell+Cogeco. Endpoints documented in
memory/reference_quebec_ihv.md.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-01 22:40:47 -04:00
louispaulb
2cf1558ede fix(cogeco-checker): disable browser cache to rule it out as 401 cause
Tested the hypothesis that a warm Chromium cache (the register GET being
re-served stale) was causing the protected address/search 401. Disabled
the HTTP cache (CDP Network.setCacheDisabled), the on-disk cache
(--disk-cache-size=0) and service workers (serviceWorkers:'block').

Result: identical trace — register=200 (freshly minted, not cached),
autocomplete=200, address/search=401. So cache was NOT the cause; the
401 is a server-side authorization decision on the protected endpoint
(reCAPTCHA Enterprise assertion required). Keeping the cache-disable as
hygiene + to definitively rule it out in future debugging.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-01 21:47:19 -04:00
louispaulb
3811662d66 feat(ops): assisted Cogeco spot-check on overpriced-internet report
Cogeco's address checker is gated by reCAPTCHA Enterprise (risk-score 401
on the protected /boutique/api/address/search call), so per-address
serviceability can't be scraped reliably from a datacenter IP without a
residential proxy. Per product decision, pivot to an assisted spot-check
instead of automated qualification.

- ReportInternetCherPage: add a "Concurrent" column with a one-click
  button that copies the full service address and opens Cogeco's
  availability page in a new tab (human reads the verdict in ~10s, only
  for the leads that matter). fullAddress() builds "addr, city, QC ZIP".
- cogeco-checker: harden the POC anyway — track service-address/search
  responses, retry the verdict call on 401 (re-register cadence), and
  prioritize the authoritative JSON body in interpret(). Recon confirmed
  the wall is reCAPTCHA scoring, not a timing/selector bug.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-01 21:24:36 -04:00
louispaulb
74406d9798 feat(cogeco-checker): POC competitor-serviceability microservice (WIP)
Playwright/Chromium microservice (mirrors modem-bridge: node:20-slim +
Chromium, token auth, port 3302, serialized + rate-limited) that drives
Cogeco's public address checker to determine if a competitor serves a
given address.

What works (proven on prod):
- Anti-bot bypass: vanilla headless gets 403 on /boutique/api/register
  (reCAPTCHA Enterprise blocks datacenter headless). Adding
  playwright-extra + stealth flips it to 200 — register + autocomplete
  succeed.
- Reaches Cogeco's address system and pulls real autocomplete
  suggestions. Confirmed it's Loqate/AddressComplete (id + next:
  Retrieve/Find shape).

What's NOT reliable yet (do not use the verdict for decisions):
- The serviceability verdict. The Loqate flow is multi-step
  (Find → Retrieve → Cogeco serviceability) and a single option click
  doesn't complete it, so the final yes/no API call isn't captured.
- Current interpret() falls back to scanning UI text and produces FALSE
  POSITIVES (a rural out-of-Cogeco address returned available=true off
  generic marketing copy). Needs the real Retrieve+serviceability
  endpoint wired before it can be trusted.

Next: capture the post-selection Retrieve + serviceability call (likely
needs a "continue" step and handling the multi-dwelling "N Addresses"
branch), then parse the real verdict + speeds.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-01 20:56:05 -04:00
louispaulb
bea97e9d45 fix(reports/legacy): freshness from service date, not invoice date
Marc Robidoux flagged as overpriced (129.95$) — he has a loyalty discount
(service id 74448) that should lower it. Investigation: 74448 doesn't
exist in the copy (its max service id is 74393), so the discount was added
after the snapshot. Same freshness issue as Julie Dupuis — not a calc bug.

But this also exposed that the freshness banner was wrong: it read the
newest INVOICE date (Apr 30) while the snapshot actually carries SERVICES
created through May 22 — May's recurring billing run simply hadn't executed
at dump time, so invoices lag services by ~3 weeks. For a report that reads
active services/plans/discounts, the service date is the right freshness
signal.

fetchDataAsOf now returns both {services, invoices}; data_as_of (shown in
the banner) is the service date (May 22), with last_invoice kept for
reference. The copy is ~10 days stale, not ~1 month. Marc's loyalty credit
still won't show until the copy is refreshed (task #38).

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-01 19:57:09 -04:00
louispaulb
417cfb5748 fix(reports/legacy): include Internet equipment cats to capture all discounts
User flagged Claude Bergeron at 99.95$ "including TV". Investigation: the
report already excludes TV (cat 33/34) — his cat-32 Internet subtotal was
94.95. The real issue was the opposite of what it looked like: a -60$
RAB_FTTH_URBA discount on his account lives in cat 26 ("équipement fibre"),
which the report did NOT count. His true net Internet is 44.95$, so he
should drop off the >90$ list entirely (and now does).

Internet equipment categories (26/29 fibre, 7/8 wireless) carry recurring
items that belong on the Internet bill:
  - modem/router rentals: FTTH_LOCMOD +10, LOC_TPL +5, LOCRTHG8245 +6.95
  - Internet discounts:    RAB_FTTH_URBA (hijacked to -60 for Claude)
Added them to CAT_INTERNET_CORE. The existing price_recurr_type=1 filter
still drops one-time install charges (INSTFIBRE -199, etc.) that share
these categories. Verified HVSECTOUR/INSTTELE (odd high-price items in
cat 7/8) have zero active residential services — no aberrations introduced.

Net effect: the report's "net Internet" is now truly net of every
recurring Internet discount, wherever it's categorized. Residential >90$:
554 → 739 (modem rentals legitimately lift borderline bills; deep
equipment-category discounts like Claude's pull others below the line).
TV and téléphonie remain fully excluded.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-01 19:46:58 -04:00
louispaulb
ec6182f4d5 feat(reports/legacy): data-freshness banner + recently-expired-discount column
User correctly spotted that Julie Dupuis shows 114.95$ but actually pays
69.95$ — investigation revealed the legacy COPY (legacy-db container) is a
one-shot snapshot from 2026-05-05 with data through 2026-04-30 and NO
auto-sync. She renegotiated in May (a -50$ discount on service 50999) which
the copy never received. The report was correct vs the copy, but the copy
is ~1 month stale.

Two changes (data-source strategy still pending operator decision —
prod 10.100.80.100:3306 is reachable for a future live/refresh option):

1. data_as_of — the report now reports MAX(invoice.date_orig) from the
   copy and the Ops page shows a banner ("Données legacy au 30 avril —
   copie figée, N jours"). Turns orange past 7 days so nobody acts on
   stale prices unknowingly.

2. recent_expired_discount column — per-address sum of deactivated credit
   lines (status=0, price<0) whose actif_until fell in the last 180 days.
   Surfaces clients whose discount just lapsed (Julie's RAB24M -15 + RAB_X
   -35 expired 2026-03-01), i.e. the prime retention targets whose bill is
   about to jump. Shown in amber with a warning icon + tooltip; included in
   the CSV.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-01 19:31:41 -04:00
louispaulb
383967175a fix(reports/legacy): active clients only — exclude terminated + non-customer accounts
User flagged that several listed accounts are inactive (Or Viande Inc,
Denis Henderson). Root cause: I filtered service.status=1 but NOT the
account, so terminated accounts carrying an orphan active service line
slipped through. The legacy billing job (LEGACY-ACCOUNTING-ANALYSIS.md
§6.1) bills only when BOTH service.status=1 AND account.status=1.

Three account-level filters added:
- account.status = 1   → drops terminated accounts. Or Viande Inc is
  status=4, terminated 2014 (terminate_date set), but still had a
  service.status=1 row. 8602 accounts are status=4 vs 6537 status=1.
- account.group_id = 5 → "Client" per account_group. Drops 6 Prospect,
  7 Fournisseur, 8 Relais (network infra, e.g. Denis Henderson's
  REL_CHRY_CHARLES tower account), 10 Équipement motorisé.
- customer_id NOT LIKE 'PROPRIO%' → 59 landowner-hosts-our-gear accounts
  that live in group 5 but aren't paying customers (Denis Henderson's
  other account PROPRIOH_STCHARLES). A genuine same-name customer
  (Robert Henderson, ROBEH...) correctly stays.

Residential >90$/mo: 983 → 554 (was inflated ~44% by dead/non-customer
accounts). Commercial: 255 → 240.

Ops page note updated to state "comptes clients actifs uniquement" and
list what's excluded.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-01 19:21:24 -04:00
louispaulb
d3620969e2 fix(reports/legacy): exclude expired credits; confirm monthly price model
Reviewed against docs/archive/LEGACY-ACCOUNTING-ANALYSIS.md (the migration
audit) which surfaced two things to check in the overpriced-internet report:

1. service.payment_recurrence (0=annual, 2=monthly, 5=semestrial...) —
   checked whether per-cycle prices needed /N normalization. They do NOT:
   verified a semestrial FTTH1500I carries product.price=109.95, identical
   to the monthly one (billed 6×109.95 every 6 months). Per §6.1
   "prix = quantité × prix_unitaire", product.price is already the monthly
   unit price. The original monthly logic was correct — no division. The
   SKU-LIKE-'%ANN' /12 special-case stays (true annual plans where price
   IS the yearly amount, e.g. FTTH_ANN @ 480$/yr).

2. Promo credits carry an actif_until end date (§10). A discount line whose
   actif_until is past no longer reduces today's bill, so counting it
   understates what the client actually pays. Now excluded.

   NULL-safety: the exclusion needs an explicit `actif_until IS NOT NULL`
   guard — without it, `NOT (price<0 AND actif_until>0 AND actif_until<now)`
   evaluates to NULL for permanent credits (actif_until NULL), which SQL
   treats as not-true and silently DROPS every permanent credit line. That
   briefly inflated the residential count to 3330; with the guard it's a
   correct 1000 (vs 983 before — +17 addresses whose only sub-90 reason was
   a now-expired credit).

Net effect: the report reflects the *current* real monthly Internet bill.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-01 19:12:49 -04:00
louispaulb
37c53019f4 feat(ops/reports): "Internet trop cher" legacy report
New Ops report to surface clients whose net monthly Internet bill
exceeds a threshold — for spotting plans that should be revised.

Hub (lib/legacy-reports.js — new module, read-only MariaDB):
- GET /reports/legacy/overpriced-internet (+ .csv variant)
- Queries the legacy gestionclient DB directly via a small mysql2 pool
  (reuses cfg.LEGACY_DB_* — same vars as auth.js sync-legacy; added
  LEGACY_DB_PASS to the hub .env which was previously unset).
- Grain = delivery (service address), NOT account: a multi-unit
  building (account 13166 has 82 doors / 205 services) would otherwise
  show a single bogus $2117 line instead of ~45 per door.
- Net monthly Internet = SUM of effective per-line price across
  Internet categories (32 fibre, 4 wireless, 23 camping + optional
  add-ons 16/17/21), discounts included (products with price<0 are
  recurring credits like RAB24M -15$).
- Effective price = service.hijack ? hijack_price : product.price.
- Only recurring lines (product.price_recurr_type=1) — excludes
  one-time equipment/install charges.
- Annual plans (SKU LIKE '%ANN', e.g. FTTH_ANN @ 480$/yr) normalized
  /12 so they compare correctly against a monthly threshold (was
  falsely showing $480 → now $40, drops below 90$).
- Excludes TV (33,34) and téléphonie (9) entirely.

Validated counts at 90$/mo: 983 residential, 297 commercial addresses.

Ops UI:
- src/pages/ReportInternetCherPage.vue — threshold/segment/add-ons
  filters, summary cards (count, total monthly, avg, discounts),
  sortable+filterable table (client, address, net, gross, discount,
  plan detail with full tooltip, contact), CSV download.
- Card on the Rapports hub + route /rapports/internet-cher.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-01 19:06:05 -04:00
louispaulb
8455409f2a fix(campaigns/templates): center logos via nested-table pattern
The native-block imageBlock factory was emitting img tags wrapped only
by a td with text-align:center. That doesn't actually center the image
because text-align only affects inline content, and the img has
display:block. The result: top header logo and dark-footer logo were
left-aligned despite the textAlign:"center" prop on the block.

Fix: wrap each img in an inner <table align="<textAlign>"> exactly the
way MJML/Litmus/Mailchimp do it. This is the canonical email-client
pattern that works in Outlook 2007-2019 (which ignores margin:0 auto
on inline tables but respects table align attributes).

Also: the AI converter dumped the entire dark footer band into a
SINGLE htmlBlock with malformed table markup (a stray </td> outside
its row). Split into proper image + text native blocks so:
  1. The logo inherits the new centered nested-table pattern
  2. The URL+copyright text is now individually editable in Unlayer
  3. The {{year}} placeholder is in a text block where it belongs

And one AI hallucination correction: the converter assigned
textAlign:"left" to the top header logo (probably because the
surrounding column had align="left" in the MJML output). Original
design intent was centered — fixed in the spec.

Verified live: both logos (140px top, 120px footer) now render with
align="center" on their nested wrapper table.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-06-01 18:45:30 -04:00
louispaulb
8a3074184a feat(campaigns/templates): Gemini-powered HTML→native converter
Scales the native-block migration from "one template per manual spec"
to "any compiled .html template, one CLI command, ~5 seconds, ~$0.001
per template" via Gemini Flash semantic interpretation.

Pipeline (ai-convert-to-native.js):
  1. Read existing compiled .html
  2. Send inner body to Gemini Flash with a tight JSON schema
     (block.type ∈ text/image/button/divider/html, plus type-specific
     fields like fontSize/color/padding/href).
  3. AI returns { preheader, ariaLabel, blocks: [...] }
  4. Deterministic emit of a templates-spec/<name>-native.js file —
     no AI-touched markup goes into the final compiled output.
  5. Validation: every {{var}} in source MUST survive into the spec;
     warn loudly if any are dropped (the AI occasionally omits minor
     placeholders like {{year}} in the copyright line).

Why deterministic emit matters:
  Gemini understands SEMANTICS reliably ("this paragraph is the
  greeting, this div is the CTA, this span is a chip") but
  hallucinates DETAILS when generating final HTML. Splitting the
  responsibilities means the AI only outputs structured JSON
  describing the layout, and build-native-template.js produces the
  bytes shipped to recipients.

First conversion: gift-email-fr → gift-email-fr-native
  - 15 blocks identified by Gemini in 3006 tokens (Flash, ~5s).
  - 4 row groups: view-in-browser, white card (intro/chips/CTA/
    footer copy), contact info, dark footer band.
  - 7 text + 1 image + 1 button + 6 html blocks (chips, multi-logo
    strip, brand-logo card, expiry section stay as raw HTML —
    correct, those have no native equivalent).
  - HTML payload: 19,664 bytes vs original 39,913 bytes — **-51%**.
  - One AI omission caught by the new sanity check: {{year}} was
    stripped from the © line in the dark footer. Hand-patched in the
    generated spec. Re-running with stricter prompt should reduce
    that occurrence rate.

Hub preview endpoint now defaults vars.year to current year (matches
the test-send endpoint that already did this), so the sample render
shows "© 2026 TARGO Communications" instead of "©  TARGO ...".

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-06-01 18:29:15 -04:00
louispaulb
6dbd1c3de3 feat(campaigns/templates): native-block reminder template (proof of concept)
Until now, every Unlayer-edited template stored as a single giant
"Custom HTML" block (~37 KB). The operator couldn't manipulate the
greeting, the CTA, or the expiry badge independently — they had to
edit raw HTML inside one block.

New scripts/build-native-template.js generates matched .json
(Unlayer design tree) + .html (compiled output) from a JS template
spec under scripts/templates-spec/. Each block becomes a separate
entry in the design tree with its own type:
  - 9 text blocks  : greeting, urgency, body, expiry, prorata,
                     Option 2 text, signature, contact, dark footer
  - 2 image blocks : header logo, footer logo
  - 1 button block : the CTA (🎁 {{amount}})
  - 4 html blocks  : view-in-browser, Option 1 chip, brand-logo
                     card, Option 2 chip (kept as raw HTML — too
                     custom for native equivalents)

gift-email-native-reminder-fr ships as the proof of concept:
- Compiled HTML: 30,867 bytes (vs 39,484 for the MJML-compiled
  reminder-fr — saves 22%)
- JSON: 42,274 bytes (essentially same as before, but now broken into
  16 individually-editable blocks instead of 1 monster Custom HTML)

What this unlocks in Unlayer:
- Click any text → font / color / size / padding / alignment in the
  right panel
- Click the CTA → button-specific controls (corner radius, hover
  color, padding)
- Drag-reorder blocks within the email
- Mobile preview reflects each block's responsive defaults
- Save a block to the personal library for reuse in other campaigns

Limitations on the 4 html blocks:
- Chips (Option 1 / Option 2) require raw HTML edit because the
  rounded badge styling has no native equivalent
- Brand-logo strip needs precise inline img widths Unlayer can't set

Once the operator validates rendering across Gmail/Outlook/Apple
Mail, we'll port the rest: gift-email-fr/en + the existing reminder
templates can all migrate using the same build script.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-06-01 15:31:15 -04:00
louispaulb
6e103ff05b feat(campaigns/reminder): cascade clicks to parent + family banner
A reminder campaign is a deep-copy of its parent's non-clicked
recipients with NEW gift_tokens. Clicks on the reminder were flagging
the CHILD recipient's gift_link_clicked but the parent campaign's
counters never updated — operators had to check two campaigns to see
the cumulative click rate.

Hub:
- New cascadeClickToParent() helper — when a recipient with
  parent_campaign_id is flagged as gift_link_clicked, mirror the flag
  + timestamp onto parent.recipients[parent_row_index] and broadcast
  a recipient-update SSE event so the parent's open page refreshes
  live. Adds a gift_clicked_via_reminder breadcrumb (the child
  campaign id) so the parent UI can show "↩ via la relance XXX".
  Idempotent — already-clicked rows are no-op.
- Three cascade call sites: applyWebhookEvent fast path (CustomID),
  applyWebhookEvent fallback (msgId scan), handleGiftRedirect wrapper.
- handleGiftRedirect also now sets gift_link_clicked=true on first
  successful redirect (Mailjet webhook can lag or drop; the wrapper
  redirect is the most reliable click signal we have).
- GET /campaigns/:id now attaches a "reminders" array with summary
  counters for every reminder child of the campaign.

Ops UI:
- "Cette campagne est une relance" banner on child detail pages with
  a back-link to the parent.
- "N relance(s) envoyée(s)" banner on parent detail pages with
  clickable chips showing each child's gift_clicked/total ratio.
- Recipient table: 🔁 icon next to the gift-click indicator when the
  click came via reminder, plus a "↩ via la relance XXX" line in the
  tooltip so the operator can trace the engagement channel.

One-time backfill applied on prod to mirror clicks that happened
between reminder send and this deploy (1 click cascaded —
cmp-20260522-2d4605 gift_clicked 27 → 28).

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-06-01 15:24:52 -04:00
louispaulb
92ceb54450 style(campaigns/reminder): add "pour te remercier" / "as a thank-you"
Surfaces the WHY of the gift in the reminder body. The original
campaign was sent ~10 days before the reminder fires — recipients
may have forgotten the loyalty/gratitude context, leading to a
"what is this?" reaction when they see the reminder cold.

Adding two words ("pour te remercier" / "as a thank-you") cheaply
reconnects with the original messaging and reinforces TARGO's
relationship framing.

FR: "La carte-cadeau qu'on t'a envoyée pour te remercier peut
     s'utiliser chez des centaines de marques canadiennes..."

EN: "The gift card we sent you as a thank-you can be redeemed at
     hundreds of Canadian brands..."

Greeting kept as "Petit rappel pour {{firstname}}," — first-name
personalization beats generic "toi" on engagement metrics, and the
firstname auto-clean covers ~99% of recipients.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-06-01 15:09:41 -04:00
louispaulb
077262d4e5 fix(campaigns/expiry): format dates in America/Montreal, not container UTC
The targo-hub container runs with TZ=UTC (no override set). Calls to
toLocaleDateString without an explicit timeZone option were rendering
dates in UTC, which meant a wrapper expiring at 23:59 EDT (= 03:59
UTC next day) showed "22 juin 2026" to the recipient instead of the
intended "21 juin".

All 4 date-formatting sites in lib/campaigns.js now pass
timeZone: 'America/Montreal' explicitly:
- worker (sendCampaignAsync) — main send path
- /campaigns/:id/recipients/:i/view — web fallback render
- POST /templates/:name/test-send sample defaults
- POST /templates/:name/preview sample defaults

Verified on prod: stored UTC "2026-06-22T03:59:59Z" now formats
"21 juin 2026" / "June 21, 2026" with the timeZone option, matching
the operator's intent ("expiration en fin de journée le 21 juin EDT").

Also re-patched the relance draft cmp-20260601-f857cd-rem from
2026-06-21T23:59:59Z (= 19:59 EDT, the early-evening cutoff) to
2026-06-22T03:59:59Z (= 23:59:59 EDT, true end of day). Bonus: this
aligns with the original campaign's recipients which expire around
~11:57-12:04 EDT on June 21, so the reminder always works at least
as long as the original — never the inverse confusion.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-06-01 14:58:54 -04:00
louispaulb
f2a1815ed5 fix(hub/cors): allow PATCH in preflight Access-Control-Allow-Methods
Browser CORS preflight (OPTIONS) for PATCH /campaigns/:id was
rejected because PATCH wasn't listed in Access-Control-Allow-Methods.
The browser surfaced this as a generic "Load failed" on the
"Enregistrer" button of the edit-params dialog. curl bypasses CORS
so backend testing missed it.

The header now includes PATCH alongside GET/POST/PUT/DELETE/OPTIONS.
Verified live: OPTIONS preflight now returns the full method list.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-06-01 14:49:11 -04:00
louispaulb
e63f55d864 feat(campaigns/expiry): date picker for explicit cutoff
Operator can now choose an exact date for the wrapper expiry (e.g.
"valid until June 15") instead of computing days from today. Useful
when communicating a specific deadline to recipients.

Worker resolution order:
  1. params.gift_expires_at (full ISO datetime, set by the date picker)
     — all recipients of this campaign get THIS exact date, regardless
     of when the worker fires the send.
  2. Fallback: now() + gift_expiry_days (relative deadline, shifts
     forward by queue lag).

UI in both wizard (new campaign) and edit-params dialog (draft):
- Date picker at the top with cursor-pointer event icon + clear (x)
- Preset toggle (15/30/60/90/180/Custom days) below — auto-disabled
  when explicit date is set so the operator picks ONE mode
- Indicator "≈ N jours à partir d'aujourd'hui" when explicit date is
  active so the operator sees both representations

UI carries the picker value as YYYY-MM-DD (gift_expires_at_display);
launchSend / saveEditParams translate to ISO YYYY-MM-DDT23:59:59Z
before PATCH/POST. Anchoring at end-of-day local means "until June 15"
stays valid through all of June 15, not just the start.

dateAfterToday validator blocks past dates in the picker.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-06-01 14:45:14 -04:00
louispaulb
d06dd53534 style(campaigns/templates): replace em-dashes with periods/commas
Some readers (and several modern style guides) read em-dashes as
"AI-written" feel — the user preferred a mix of period (for full
clauses) and comma (for asides) to keep the copy conversational
without the long pause em-dashes impose.

Period when both sides are independent clauses:
- about fiber. They're about people too. (EN main + reminder)
- on Giftbit. Just click your X. (EN main + reminder)
- pas manqué. La carte-cadeau qu'on t'a envoyée… (FR reminder)
- didn't miss it. The gift card… (EN reminder)

Comma when the second half is an aside or starts with "and":
- something special, for a limited time. (EN main)
- right next door, and we genuinely love… (EN main)
- aucun souci, pas besoin… (FR reminder)
- no worries, no need to reply… (EN reminder)

gift-email-fr unchanged — its user-visible text never had em-dashes
(the 3 detected were inside HTML comments).

No hub restart needed: the send worker reads templates fresh from
disk on every campaign run, so the new copy applies on the very next
"Lancer l'envoi" click.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-06-01 12:26:59 -04:00
louispaulb
0248a195f8 fix(campaigns/reminder): softer tone + render expiry in tests
The reminder copy read as pushy on test sends ("Hâte-toi! ... Tu n'as
encore rien fait, et le délai approche"). Toned down to factual and
friendly: state availability + offer the no-pressure path.

FR before / after:
   Hâte-toi! Ton cadeau de 60 $ expire le ___.       (red bold)
  → 🎁 Ton cadeau de 60 $ reste disponible jusqu'au 1 juillet 2026.
                                                       (brand dark green)

  Tu n'as encore rien fait, et le délai approche. Si tu n'utilises
  pas ton cadeau d'ici là, il ne pourra plus être réclamé.
  → On voulait juste s'assurer que tu ne l'as pas manqué — la carte-
  cadeau qu'on t'a envoyée peut s'utiliser chez des centaines de
  marques canadiennes, en quelques clics.
  Si tu préfères ne pas l'utiliser, aucun souci — pas besoin de
  répondre à ce courriel.

EN copy mirrored.

Also: {{expires_at_date}} was rendering empty in test sends and
previews because neither the test-send endpoint, the preview
endpoint, nor the editor's testSendForm.vars seeded it. Three fixes:
- Hub preview endpoint: compute now+30d as default sample date.
- Hub test-send endpoint: same default + expose view_url='' so the
  Mustache section block collapses cleanly in internal tests.
- Editor test-send dialog: pre-fill expires_at_date (and expires_in_
  days) with the same now+30d value, plus expose both fields as
  editable inputs so the operator can override per-test.

Verified live on prod: the preview endpoint with no vars now renders
"Ton cadeau de 60 $ reste disponible jusqu'au 1 juillet 2026."

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-06-01 11:55:20 -04:00
louispaulb
0f59191fde feat(campaigns): reminder campaign for non-clickers
Adds a "Créer une relance" button on the campaign detail page that
clones the parent campaign into a new draft, targeting only the
recipients who haven't clicked the Giftbit gift link yet.

Backend (POST /campaigns/:id/reminder):
- Filters parent recipients: status sent/opened, not excluded, not
  revoked, wrapper not yet expired, has a gift_url.
- Builds a fresh recipients array — same gift_url (Giftbit shortlink),
  same name/email/language/amount, but cleared gift_token so the worker
  generates a brand-new wrapper at send time. Each campaign owns its
  own click metrics.
- New campaign starts as 'draft' so the operator can review, tweak
  subject/template, and click "Lancer l'envoi" when ready.
- Tracks parent_campaign_id + parent_row_index on each reminder row
  for traceability in CSV reports and debugging.

Templates (gift-email-reminder-fr / gift-email-reminder-en):
- Header swap: "Petit rappel pour {firstname}" / "Quick reminder, X"
- Bold orange urgency line: " Hâte-toi! Ton cadeau de X expire le Y"
  using the existing {{expires_at_date}} and {{amount}} merge vars
- Body shortened — drops the manifesto, focuses on "you have a gift,
  redeem before it's gone"
- Same CTA button + prorata disclaimer + signature + footer as the
  main templates so brand stays consistent.

UI:
- Button visible when campaign is sending/completed AND it's not
  itself a reminder AND there's ≥ 1 eligible non-clicker.
- Confirmation dialog spells out the mechanics: same Giftbit URLs,
  new wrapper tokens, reminder template, sample expiry date pulled
  from the campaign's first recipient with a gift_expires_at.
- On OK, redirects to the new campaign's detail page.

Click stats on the existing campaign (cmp-20260522-2d4605) verified
intact before+after deploy (109 opens, 15 generic clicks, 27 gift CTA
clicks) — saveCampaign persists per-event so the hub restart was a
no-op for accumulated data.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-06-01 11:43:35 -04:00
louispaulb
1774418689 feat(campaigns/send): real SMTP error + auto-retry + one-click Renvoyer
The send worker used to write "SMTP send returned false (see hub logs)"
on every failure, forcing the operator to SSH into the box to find the
actual cause. Now we capture the real reason and surface it in the UI.

Three changes:

1. lib/email.js exposes getLastError() — a side-channel for the most
   recent nodemailer error message, cleared at the start of every
   sendEmail call. Legacy "if (await sendEmail(...))" callers stay on
   the false-return contract; only the campaign worker reads the
   side-channel for detailed error capture.

2. The worker now retries each recipient up to 3 times (initial +
   2 retries with 2s/5s backoff). Most "Unexpected socket close"-style
   transient Mailjet errors recover on the second attempt. We observed
   exactly this case for Myriam Bergevin in cmp-20260522-2d4605 — a
   single socket close interrupted 1 of 202 sends; auto-retry would
   have caught it. retry_count is now stored on the recipient.

3. POST /campaigns/:id/recipients/:row/retry resets a single failed
   row back to pending and re-fires the worker. Surfaced in the
   detail-page table as a small 🔁 button next to the error text on
   any row with status=failed. Useful when auto-retry exhausted its
   3 attempts on a one-off transient.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-22 13:29:25 -04:00
louispaulb
7e0f4d0799 feat(campaigns/wizard): inspectable dropped-row list with one-click recovery
parseMapCsv now collects the actual rows it drops (capped at 200),
each with its skip reason and the raw source-CSV columns (full_name,
email, phone, address, postal). Returned alongside the existing
counters as skipped_rows on the parse response.

Wizard Step 2 adds an "N ligne(s) du Map CSV non importée(s)"
expansion below the imbalance banner, showing:
  Ligne # | Raison | Nom au CSV | Email au CSV | Adresse | CP | →

The action column has a "Ajouter manuellement" button on rows that
have an email (duplicate, multi_skip) — clicking opens the manual-
add dialog pre-filled from the dropped row, so the operator can
recover the contact in two clicks. no_email rows can't be recovered
that way and don't get the button.

The source_row index is the Excel-relative line number (counting the
header) so the operator can cross-reference the actual file.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-22 11:55:48 -04:00
louispaulb
04a2db35c5 fix(campaigns/parse): keep no-name rows + surface skip breakdown
Map CSV rows that had a valid email but no name in the source column
were silently dropped at parsing — that's why a campaign would end up
with N unpaired Giftbit shortlinks for N "missing" contacts that
weren't actually missing, just nameless.

The send worker already handles a missing firstname by substituting
"cher client" / "dear customer", so dropping the row was wasteful.
Now we keep the contact and surface a name_warning on the row so the
operator can either edit the firstname in Step 2 or accept the
default.

Also added counters for previously-silent skip paths:
- duplicate: row's email was already seen above (1 gift / household
  consolidation, depending on the multi setting)
- multi_skip: couple skipped because multi='skip' was selected

Wizard Step 2 imbalance banner now ventilates the skip breakdown so
the operator understands exactly where the N "missing" contacts went:

  Ventilation des contacts droppés au parsing du Map CSV (sur 213
  lignes brutes) : 8 sans email valide · 5 emails en double · 0
  couples ignorés · 3 sans nom (gardés, utilisent "cher client" à
  l'envoi)

Unrelated reassurance on the question that triggered this: language
fallback to French is already in place (matchCustomer returns
language:'fr' on miss, worker reads (r.language || 'fr')) so any
unmatched recipient gets the FR template, never an English one by
default.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-22 11:43:22 -04:00
louispaulb
82195847b5 fix(campaigns/match): handle multi-email + dupe SLs + missing postal
Three independent bugs surfaced while debugging why Alexandre Duval
showed as "non lié" in a campaign:

1. ERPNext Customer.email_id can hold multiple addresses joined by
   ';' or ',' (211 records inherited from Legacy migration). The
   exact-match filter missed them. Now LIKE-searches a window then
   validates locally by splitting on ; , or whitespace.

2. Service Locations have duplicates at the same address — the same
   "7 Rue des Merles" exists 3 times, linked to 3 different customers
   (legacy migration artifact). The civic+postal strategy was taking
   the first hit which could be the wrong household. Added name-aware
   disambiguation: when the recipient has a name, walk the candidates
   and pick the one whose linked Customer name plausibly matches.

3. New 4th matching strategy "name+civic" — kicks in when the CSV
   row has no postal_code (most common Map export failure mode). Does
   a street-word filtered SL search and accepts only candidates whose
   Customer name plausibly matches. Confidence 0.65 (vs 0.85 for
   civic+postal).

Also: SQL filter for both civic+postal and name+civic now includes a
street-word LIKE constraint so the result set isn't dominated by
unrelated "7 ..." addresses, bumped limits to 50/100. The SL
denormalized customer_name field is often empty post-import — we now
fall back to a Customer lookup for the name check.

Verified end-to-end against live ERPNext: Alexandre Duval at
7 Rue des Merles now matches correctly via email (multi-value field),
via civic+postal (despite 3 dupe SLs), and via name+civic (no postal).
Gaëtan David at the same address also matches correctly without
collision.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-22 11:36:51 -04:00
louispaulb
27079bba11 feat(campaigns): one-click Giftbit admin lookup per recipient
Manual workaround for redemption status until /gifts/{uuid} polling
ships (task #25). The trailing path segment of the Giftbit shortlink
is the lookup key for Giftbit's admin search:

  http://gft.link/4kpZMApLK4Bhttps://app.giftbit.com/app/rewards?search=4kpZMApLK4B

Surfaced in three places:
- Inventory page row: 🔗 button next to the copy-URL action
- Campaign detail page recipient table: same button next to the
  Giftbit shortlink
- CSV report: new giftbit_admin_url column for bulk audits in Excel
  (one click per row, no manual concat)

Defensive: only renders if the trailing segment is ≥4 chars (avoids
producing useless searches on malformed/test URLs).

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-22 10:58:33 -04:00
louispaulb
dcc5679fa3 feat(campaigns/templates): visible wrapper-expiry date in the email
Two new template variables are auto-derived from r.gift_expires_at at
render time (separately by the worker and the /view fallback to keep
them consistent):

  {{expires_at_date}}  locale-formatted FR/EN long date — "21 août 2026"
                       / "August 21, 2026". Empty when no wrapper token.
  {{expires_in_days}}  remaining days as string (rounded up). Useful
                       for tight deadlines where a date is too distant
                       to convey urgency.

Templates: a small centered badge appears between the CTA button and
the prorata disclaimer, wrapped in a Mustache section so it disappears
cleanly on campaigns that pre-date the wrapper feature.

   Cadeau valide jusqu'au <strong>21 août 2026</strong>
   Gift valid until <strong>August 21, 2026</strong>

Editor merge-tag panel updated so authors can drop these into custom
copy without remembering the exact variable names. The legacy
{{expiry}} field stays — it's still the right tool for promotion-end
dates that don't track the gift link's own deadline.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-22 10:47:58 -04:00
louispaulb
ec007ed406 feat(campaigns): gifts inventory page + expiry presets
Wizard: gift_expiry_days now lives behind a preset toggle
(15/30/60/90/180 + Custom) instead of a naked number input. Operator
clicks a chip; the value flows back into the existing campaign param.

Inventory page (/campaigns/gifts):
- Cross-campaign view of every wrapper token with status taxonomy
  (active / redeemed / expired / revoked / pending). Each card on
  the counters strip is a click-to-filter shortcut.
- "Réassignables" highlighted in amber when > 0 — these are gifts
  whose wrapper expired or was revoked but the Giftbit URL is still
  unredeemed, ready for a fresh recipient.
- Search across name/email/url/token; per-status and per-campaign
  filter dropdowns.
- One-click copy on the Giftbit URL with a tailored toast that walks
  the operator through the reassignment workflow (paste into manual-
  add dialog of a new campaign).
- Revoke action with confirmation; explicit about what survives
  (the Giftbit URL stays valid on their side) vs what changes (our
  wrapper stops redirecting).

Backend:
- GET /campaigns/gifts flattens every recipient with a gift across
  every campaign — single-shot, no pagination yet (we're under 10k
  gifts total).
- POST /campaigns/:id/recipients/:row/revoke sets gift_revoked=true
  and broadcasts the recipient-update SSE event.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-22 10:21:05 -04:00
louispaulb
6fe6e4df0d feat(campaigns): gift redirect wrapper — own expiry + reusable links
Each campaign recipient now gets a short opaque token (10 base64url
chars, ~60 bits entropy). The email contains

  https://msg.gigafibre.ca/g/<token>

which 302-redirects to the underlying Giftbit shortlink — but ONLY if
the recipient hasn't passed our own expires_at and we haven't revoked
the token. This gives us two new operational capabilities:

1. End-date control independent of Giftbit. The wizard now has a
   "Expiration interne (jours)" field (default 90) that sets our
   own deadline. Useful when the Giftbit gift is valid 12 months
   but the campaign offer should expire in 30 days.

2. Reuse of unredeemed gifts. After our expiry, the old wrapper
   stops working but the Giftbit URL is still valid on their side.
   Pasting that same gift_url into a new campaign (via the manual-add
   dialog) generates a NEW token pointing to the same Giftbit gift —
   the original recipient's old wrapper URL says "expired", the new
   recipient gets a fresh window.

Per-recipient new fields:
- gift_token            short ID used in the wrapper URL
- gift_expires_at       ISO timestamp of our cutoff
- gift_revoked          manual kill-switch (false by default)
- gift_redirected_count clicks that successfully reached Giftbit
- gift_first_redirected_at  first successful redirect timestamp

Routing:
- GET /g/:token  — public, validates and 302s (or expired-page)
- Mailjet click event handler updated to recognise wrapper URLs
  alongside legacy gft.link/giftbit.com URLs.
- /view (browser fallback for in-email rendering) also wraps the
  gift link so expiry/revoke is honoured consistently.

Bootstrap rebuilds the in-memory token→recipient index by scanning
all campaign JSONs on startup — no separate index file to keep in
sync.

CSV report adds gift_token, gift_expires_at, gift_revoked,
gift_redirected_count, gift_first_redirected_at.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-22 10:15:43 -04:00
louispaulb
6c3380d984 feat(campaigns/wizard): per-language template override
Two new dropdowns in Step 1 ("Template français" / "Template anglais")
populated from /campaigns/templates filtered by suffix (-fr / -en).
Selection is stored on campaign.params.template_fr / .template_en
and the worker resolves the actual path via a new resolveTemplatePath
helper:

  1. params.template_<lang>  (per-lang override, set here)
  2. params.template_path    (legacy single-template campaign override)
  3. templateForLanguage()    (default gift-email-<lang>.html)

Defensive name regex inside resolveTemplatePath blocks path traversal —
operator can pick any *-fr / *-en template that exists, nothing else.

The Step 3 summary list now shows which template will actually ship
per language so the operator can sanity-check before launch.

Use cases: seasonal variants (gift-email-2026-summer-fr), A/B tests,
draft templates that aren't ready to be the default yet.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-22 09:46:58 -04:00
louispaulb
9f3a991442 feat(campaigns): delete campaign from the list
DELETE /campaigns/:id removes the JSON from /opt/targo-hub/data/campaigns/.
The Giftbit shortlinks already issued for that campaign live on Giftbit's
side and are unaffected — this is purely about clearing internal tracking
records (typically test runs cluttering the list).

Refuses (409) while the send worker is active for that id so we never
yank the file out from under saveCampaign(). Defensive id regex
(in campaignPath) blocks path-traversal attempts before unlink runs.

UI: red trash icon on each row, disabled while status=sending.
Confirmation dialog spells out what survives the deletion (Giftbit
links) vs what's lost (tracking, opens/clicks, CSV report) so the
operator isn't surprised.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-22 09:33:44 -04:00
louispaulb
fc40e7f684 feat(campaigns): "View in browser" web fallback for failed inbox renders
GET /campaigns/:id/recipients/:i/view re-renders the campaign with the
same vars the worker used at send time — same template, same per-row
amount override, same language pick. Useful when the recipient's mail
client butchers the layout: image-blocking, antique Outlook, niche
third-party apps, accessibility tools.

Templates: Mustache section {{#view_url}}…{{/view_url}} guards a tiny
gray link above the header logo (11px, #94a3b8). The section collapses
to nothing when view_url is empty, so:
  - the /view page itself doesn't show the link (you're already there)
  - wizard previews / test-sends don't show it (no real campaign id)

worker passes view_url = HUB_PUBLIC_URL + /campaigns/<id>/recipients/<i>/view
using the existing cfg.HUB_PUBLIC_URL setting (defaults msg.gigafibre.ca).

Security: campaign-id is a 21-char nanoid (≈10²¹ space). Same level of
exposure as the Giftbit shortlink itself. X-Robots-Tag: noindex on the
response so the URLs don't end up on search engines.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-22 09:30:38 -04:00
louispaulb
02b1b55b7b feat(campaigns): distinguish gift-CTA click from generic email click
Mailjet's click event includes the actual URL the recipient clicked. We
previously bumped every click — CTA button, mailto support, footer link —
to status='clicked' indiscriminately. Now we additionally flag clicks on
the Giftbit shortlink (matched by r.gift_url prefix, fallback to gft.link
or giftbit.com host) as the high-signal "gift_link_clicked" event.

Adds:
- recipient.gift_link_clicked (bool) + gift_clicked_at (ISO timestamp),
  set on first matching click; later non-gift clicks don't unset
- counters.gift_clicked aggregated alongside existing status counters
- "Cadeau cliqué" counter card on detail page (deep-purple, redeem icon)
- 🎁 redeem icon next to status chip when the recipient engaged
- CSV report: new gift_link_clicked + gift_clicked_at columns

Why this matters: "opened" is noisy (Apple Mail Privacy Protection, image
proxies prefetch). A click on the CTA is the only reliable indicator
that the offer landed and the recipient is engaging.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-22 09:22:12 -04:00
louispaulb
1b64e3123d feat(campaigns): CSV report, manual recipients, template polish
Ops UI
- CampaignDetailPage: "CSV" button — downloads per-recipient report
  (shortlinks, status, opened/clicked timestamps, mailjet UUID)
- CampaignNewPage: "Saisie manuelle (sans CSV)" on Step 1 and
  "Ajouter manuellement" on Step 2 — both open the same dialog with
  firstname / email / gift_url / city / postal_code / language /
  amount override. Indigo "manuel" chip in the recipients table.
- New "Ville" column shows city OR postal_code as fallback.

Hub
- GET /campaigns/:id/report.csv — RFC 4180 CSV with UTF-8 BOM so
  Excel auto-detects encoding. 20 columns including new "city".
- Worker honours per-recipient amount override:
  r.amount > derive from r.gift_value_cents > params.amount > "50 $".
  Fixes manual-add showing campaign default instead of typed value.
- Default subject "Un cadeau pour toi" (tutoyer).

Templates
- Order: Intro →  Option 1 → 🎁 marques → CTA → prorata → ⏭️ Option 2.
- New EN intro (manifesto): "Thank you for choosing local. Your
  support helps keep our community connected. / Because great
  connections aren't just about fiber — they're about people too."
- Amazon logo removed (incongruent with "achat local" framing).
- Body paragraphs: text-align justify (greeting/labels stay left).
- Support line: "N'hésite pas à nous écrire / Feel free to email us"
  + dash format 514-448-0773, drop "Support 7j/7" overpromise.
- Logo style fix: inline width:32px to beat Unlayer canvas CSS that
  was rendering brand pills full-width.

Ignore template converter .bak-*.json backups.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-22 09:09:07 -04:00
louispaulb
de8c1c51b8 fix(campaigns/templates): drop 'sans engagement' claim from Option 2
The 'sans engagement ni carte-cadeau' wording in Option 2 was confusing
for customers with an existing multi-month commitment — implies their
subscription is commitment-free, which contradicts their actual contract.

Reworded to make zero claims about the customer's commitment status,
just describes what happens if they ignore the email:

FR:
  before — 'Ne rien faire. Ton abonnement mensuel se poursuit normalement,
            sans engagement ni carte-cadeau.'
  after  — 'Ne rien faire. Aucun changement à ton abonnement actuel.'

EN (Gemini copywriter version was different from earlier templates):
  before — 'Just kick back! Your monthly subscription will continue as
            usual, with no commitment and no gift card.'
  after  — 'Do nothing. No changes to your current subscription.'

Benefits:
  • No false claim about engagement status
  • Shorter, more direct
  • Still preserves the explicit consent UX (customer knows ignoring
    is a valid choice without consequence)
  • No mention of 'no gift card' — that's implicit from not clicking
    the CTA, doesn't need to be stated

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-22 08:09:39 -04:00
louispaulb
972c99ffc1 fix(campaigns/templates): inject inline logos into EN template too
The previous logo injection commit (9415cf3) only matched the FR anchor —
EN was missed because Gemini's copywriter-mode translation rewrote
'at hundreds of brands' as 'to spend at hundreds of your favorite stores'.
Patched EN with the correct anchor + 'and more' caption.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-22 08:03:54 -04:00
louispaulb
9415cf3868 feat(campaigns/templates): inline merchant logos via simple <img> sequence
The previous attempt (commit 6cf9148) tried to inject a nested <table>
of logos, but the anchor selection logic looked for a </p> that didn't
exist — Unlayer renders the amount line inside a <div>, not <p>. As a
result the logos never made it into the templates.

This commit fixes it with a simpler approach: directly append the logo
images to the existing amount-text string. No table nesting, no anchor
hunting — just plain inline-block <img> tags right after the
"🎁 {{amount}} chez des centaines de marques" text.

Markup pattern (inserted right after the amount line, before the
closing </div>):

  <br><br>
  <span style="display:inline-block;">
    <img src="...amazon..." width="32" alt="Amazon" ...>
    <img src="...timhortons..." width="32" alt="Tim Hortons" ...>
    <img src="...walmart..." width="32" alt="Walmart" ...>
    <img src="...homedepot..." width="32" alt="Home Depot" ...>
    <img src="...iga..." width="32" alt="IGA" ...>
    <img src="...homehardware..." width="32" alt="Home Hardware" ...>
    <span style="font-size:13px;color:#64748B;">et plus</span>
  </span>

Each <img> uses display:inline-block + vertical-align:middle so they
sit on the same horizontal line. width=32 attribute set for Outlook;
height:auto in style preserves aspect ratio. margin-right:6px provides
spacing between logos. Caption ("et plus" / "and more") at the end.

Width math (inside 484px-wide pill): 6 × (32 + 6) = 228 px + caption
~50 px = 278 px. Fits with margin to spare.

EN translation auto-detected the equivalent anchor and inserted
"and more" instead of "et plus".

Live test-send verified for both FR + EN.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-22 08:00:44 -04:00
louispaulb
6cf9148eb7 feat(campaigns/templates): inline merchant logos + objective prorata phrasing
Two refinements per user feedback:

1. Objective/factual prorata disclaimer (shorter, conditions-of-service tone)
   FR:
     before — "Si tu annules avant {{commitment_months}} mois, tu rembourses
              seulement au prorata des mois restants."
     after  — "Annulation avant {{commitment_months}} mois : seulement à
              rembourser au prorata des mois restants."
   EN:
     before — "If you cancel before {{commitment_months}} months, you only
              refund the prorated amount for the remaining months."
     after  — "Cancellation before {{commitment_months}} months: only the
              prorated amount for the remaining months is refundable."

   The colon-prefixed structure ("X : Y") reads like a T&C bullet rather
   than a marketing sentence — clearer, less wordy, no subject pronoun.

2. Inline row of 6 merchant logos in the offer info pill
   Inserted between the "60 $ chez des centaines de marques" line and the
   "Instant activation" line. 6 most recognizable QC brands at 32px wide:
     Amazon · Tim Hortons · Walmart · Home Depot · IGA · Home Hardware
   Followed by "et plus" / "and more" caption.

   Uses the existing Mailjet-hosted brand logos (same URLs as the 4×3 grid
   in the older rich variant). 32px width fits comfortably on one line
   (~280px total in a 484px-wide pill). Email-safe single-row table layout
   with vertical-align middle, padding-right 8px for spacing.

   Visual effect: instant recognition for the reader — they see the brands
   they'd actually redeem at, without dropping the full 12-logo grid that
   bloated the previous design.

Applied to .html + .json (both FR + EN) via anchor-based injection:
finds the "🎁 {{amount}} chez/at hundreds of marques/brands" paragraph,
inserts the logo table immediately after its closing </p>. Both files
remain valid + the Unlayer editor will pick up the new table next load.

Verified live via test-send (35-37 KB output, recipient queue ok).

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-22 07:58:55 -04:00
louispaulb
1dda74e8a4 fix(campaigns/templates): clearer prorata phrasing — "mois restants" / "remaining months"
Iterating on the prorata disclaimer per user feedback. The previous
version ("tu rembourses le prorata non utilisé (20 $/mois)") still
read ambiguously — "non utilisé" could mean "the portion you haven't
spent" which is conceptually confusing for a one-time gift card, and
the hardcoded "$20/month" tied the template to the specific
$60/3-month campaign.

New phrasing makes the math explicit: refund only for the months
you're NOT staying.

FR:
  before — "Si tu résilies avant {{commitment_months}} mois,
            tu rembourses le prorata non utilisé (20 $/mois)."
  after  — "Si tu annules avant {{commitment_months}} mois,
            tu rembourses seulement au prorata des mois restants."

EN:
  before — "If you cancel before {{commitment_months}} months,
            you refund the unused pro-rated amount ($20/month)."
  after  — "If you cancel before {{commitment_months}} months,
            you only refund the prorated amount for the remaining months."

Wins:
  • Subject ("tu" / "you") explicit — no ambiguity on who refunds
  • Logic clarified — refund == months NOT STAYED, not "unused
    portion of money" (which doesn't quite map to a one-time gift)
  • Generic over campaign params — no hardcoded "$20/month" so the
    template works at any gift amount + commitment combination
  • "annules" (more common in QC consumer-facing) instead of
    "résilies" (slightly more formal/legal-sounding)

Applied via direct find/replace on .html + .json (FR + EN). Live
test-send queued to confirm rendering.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-22 07:50:51 -04:00
louispaulb
1ee3c00fc4 fix(campaigns/templates): clarify who refunds whom for early cancellation
User feedback: the prorata disclaimer was semantically wrong — both FR
("le prorata du montant est remboursable") and EN ("we'll refund the
pro-rated amount") read as if TARGO would refund the customer, when
actually the customer needs to refund the unused portion of the gift
they received if they cancel within the commitment period.

Plus: add the explicit per-month rate ($20/month at $60 / 3 months) so
the customer knows exactly what they'd owe at any cancellation date.

FR:
  before — "🪂 En cas de départ avant {{commitment_months}} mois,
            le prorata du montant est remboursable."
  after  — "🪂 Si tu résilies avant {{commitment_months}} mois,
            tu rembourses le prorata non utilisé (20 $/mois)."

EN:
  before — "🪂 If you decide to leave before {{commitment_months}}
            months, we'll refund the pro-rated amount."
  after  — "🪂 If you cancel before {{commitment_months}} months,
            you refund the unused pro-rated amount ($20/month)."

Both changes:
  • Subject clarified: customer refunds, not TARGO
  • Added explicit per-month value for transparency
  • Kept warm tone (informal "tu" / "you")
  • Mustache {{commitment_months}} preserved

Applied directly to .html + .json via string substitution (preserves
the Unlayer design tree intact except for that one phrase). The
"$20/month" figure is hardcoded for the current $60/3-month campaign;
a future {{monthly_prorata}} computed variable would generalize but
isn't needed yet.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-22 07:39:01 -04:00
louispaulb
3c9e1e095a feat(campaigns/translate): switch from literal to copywriter-mode AI prompt
User feedback: previous prompt produced robotic word-for-word output that
lost the marketing impact. The system prompt was too restrictive on
preservation, suppressing Gemini's natural rephrasing ability.

Rewrote the system prompt with three structural changes:

1. FRAME shifted from "translator" to "senior marketing copywriter"
   The opening line now says "You are NOT translating words — you are
   rewriting marketing copy that lands the same way for a different
   audience." This unlocks idiomatic rephrasing, sentence reorganization,
   active/passive switching, and cultural metaphor adaptation.

2. FEW-SHOT EXAMPLES showing the desired style
   4 FR→EN pairs in the prompt itself, with explicit "NOT: <literal>"
   anti-examples to show what to avoid:
     • "loyauté envers l'achat local" → "keeping it local"
       (not "loyalty to local shopping")
     • "connexions stables et relations durables" →
       "steady connections — both the fiber kind and the human kind"
       (not "stable connections and lasting relationships")
     • "On est juste à côté" → "We're right next door"
     • "Avec l'arrivée de l'été" → "Summer's here"
   These ground Gemini in the brand voice with concrete examples.

3. TONE constraint explicit
   "Warm, conversational, slightly playful. Like a neighbor explaining
   something — never corporate, never stiff." Use of contractions
   ("we're", "you'll") encouraged.

Plus: temperature bumped from 0.2 → 0.7 so Gemini actually exercises
creative rephrasing instead of staying glued to source word order.

Structural preservation rules (HTML, Mustache vars, brand names, emojis,
URLs, technical values like "3.5 Gbit/s") kept as HARD CONSTRAINTS but
clearly separated from the creative freedom on text content.

Live re-translation of gift-email-fr → gift-email-en applied:
  • 51s response time (similar to literal version)
  • 35,934 → 36,067 bytes (slight expansion, normal for EN)
  • Output markers confirm idiomatic phrasing landed:
    "Thanks for keeping it local", "steady connections — both kinds",
    "right next door", "lending a hand", "Summer's here"
  • Mustache vars + brand names + HTML preserved (verified)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-22 07:29:16 -04:00
louispaulb
43c19048a7 feat(campaigns): AI template translator via Gemini Flash
New "Traduire (AI)" button in the template editor toolbar. One click
translates the current template's HTML to the opposite language
(detected from the -fr/-en suffix), writing the translated content as
the matching companion template.

Backend (lib/campaigns.js):
- New endpoint: POST /campaigns/templates/:name/translate-to/:targetName
- Reads source .html, calls lib/ai.js aiCall() with Gemini Flash
- System prompt enforces 7 strict preservation rules:
  1. Byte-preserve all HTML tags/attributes/styles/Outlook conditionals
  2. Don't translate Mustache {{vars}}
  3. Preserve URLs/emails/phones/hex colors/CSS/brand names (TARGO,
     Gigafibre, Giftbit, Amazon, IGA, Tim Hortons, etc.)
  4. Preserve emojis (🎁  🤝 🪂  ⏭️ )
  5. Keep the warm informal tone (tu in FR, you in EN)
  6. Translate only visible text inside elements (paragraphs, buttons,
     alt attributes, link text)
  7. Output full HTML doc only, no markdown wrapping
- temperature=0.2 for stable output, maxTokens=32768 to fit ~35 KB HTML
- Sanity validates output isn't truncated (>50% of source size)
- Strips defensive markdown fences if AI ignored rule 7
- Auto-backs up existing target before overwrite
- Regenerates Unlayer design JSON from the translated HTML so the
  editor can reload the translated template visually
- Requires { override: true } in body to overwrite existing target
  (409 Conflict otherwise — protects against accidental clobber)

API client (apps/ops/src/api/campaigns.js):
- translateTemplate(srcName, targetName, { override })

Frontend (TemplateEditorPage.vue):
- "Traduire (AI)" button (purple, icon=translate) in toolbar — disabled
  when current template has no -fr/-en suffix
- aiTranslateTargetName computed: detects source lang from suffix,
  flips to opposite (-fr → -en, -en → -fr)
- Confirmation dialog:
  • Shows source → target template names
  • Info banner explaining what's preserved (HTML, vars, brands, emojis)
  • Amber banner + toggle if target exists (must confirm override)
- On success: positive notification with byte counts +
  "Open" action button to jump to the translated template
- Refreshes templates list after translation so the new file appears
  in the selector dropdown

UX: replaces the previous manual translation workflow (where the user
or I had to maintain two parallel templates). One click now does the
whole round-trip. User reviews + adjusts wording in the EN editor if
the AI translation needs polish.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-22 07:21:45 -04:00
louispaulb
dd80da4352 feat(campaigns/templates): mirror user's FR edits to EN + drop legacy .mjml
User edited gift-email-fr in the Unlayer editor with richer marketing
copy (loyalty thanks, brand manifesto, 3.5 Gbit/s upsell, helpful CTA).
Mirror those edits to the EN template via a one-shot translation script
so the bilingual pair stays in sync for the next campaign send.

Translation strategy: plain-string find/replace mapping with FR
phrases in longest-first order to avoid partial matches. Applied to
BOTH the rendered .html (what the recipient sees) AND the .json
(Unlayer design tree — so re-opening the EN editor preserves the
matching structure).

Mapping coverage:
  • Intro paragraphs (greeting, gift announcement, loyalty thanks,
    brand manifesto, speed upsell, "we're around the corner")
  • Offer info pill (amount, instant activation, commitment)
  • CTA button labels (Activer → Redeem, Choisir → Pick)
  • Prorata refund disclaimer
  • Option 2 "do nothing" text
  • Signature ("Merci de faire rouler" → "Thanks for helping...thrive")
  • Footer contact info + "Tous droits réservés" → "All rights reserved"
  • <html lang="fr"> → <html lang="en">

23/28 translation rules matched; the 5 unused ones were for legacy
phrasing not present in the user's latest save (e.g. the old "Tu
choisis local" line that was replaced by the current intro).

Also: drop the obsolete .mjml source files. Now that Unlayer is the
canonical editor, the MJML→HTML compile pipeline is no longer used
on save (Unlayer outputs HTML directly). The .mjml files were stale
copies from the previous MJML-based editor. Removed from disk on
prod and from git history; rollback via git revert if needed.

Verified live: GET /campaigns/templates/gift-email-en returns the
translated content (9 EN markers detected in HTML). Test-send to
louis@targo.ca queued via Mailjet for visual QA.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-22 07:07:05 -04:00
louispaulb
328d13f6f9 fix(hub): templates volume mount must be RW for editor saves
When the Unlayer editor calls PUT /campaigns/templates/:name to save a
design, the hub writes:
  • templates/<name>.html   (compiled email-safe HTML)
  • templates/<name>.json   (Unlayer design tree for editor restore)
  • templates/<name>.bak-<ts>.html  (backup of previous version)

All three need write access to /app/templates inside the container.
The mount was previously declared as :ro, which made these writes
fail with EROFS (read-only filesystem) once the editor was wired up.

Two changes:
  1. Local docker-compose.yml: add ./templates:/app/templates (without
     :ro) and ./uploads:/app/uploads (which was already RW on prod but
     missing from the committed file — local was out of sync).
  2. Prod docker-compose.yml: hot-patched via sed on prod to drop the
     :ro flag, then `docker compose down + up -d` to apply the mount
     change. PUT verified working (returns 200 with size + design_size).

The /app/lib, /app/server.js, /app/public, /app/package.json mounts stay
:ro since the hub never writes to those — keeping the read-only flag
there is defense-in-depth against compromised code overwriting itself.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-22 06:49:48 -04:00
louispaulb
9ad5b46b4c feat(campaigns): create new templates from UI + enable Unlayer template library
Two improvements to the template editor:

1. "+ Nouveau" button + creation dialog
   Users can now create new templates from the editor UI without us
   re-deploying the hub. Click "Nouveau" next to the template selector,
   pick a name + prefix + starter (blank or copy from existing), submit.
   The hub PUTs the new template (existing endpoint, no new code needed
   on the backend — just relaxed validation).

   Form:
     • Type (prefix): gift-email / newsletter / transactional
     • Name suffix: lowercase letters/digits/dashes (e.g. summer-2026)
     • Starter: "Vide" or "Copier depuis <existing template>"

   On submit:
     • If starter != blank: GET source template's html + design
     • PUT new template name with that content
     • Refresh templates list + switch editor to the new one

2. Backend: replace hardcoded EDITABLE_TEMPLATES allow-list with
   regex-validated prefix matching + disk scan
   • EDITABLE_TEMPLATE_PREFIXES = ['gift-email-', 'newsletter-',
     'transactional-'] — bounds what categories users can create
   • TEMPLATE_NAME_RE = /^[a-z0-9-]+$/ — prevents path traversal
   • isValidTemplateName() validates both regex + prefix membership
   • scanEditableTemplates() returns all matching .html/.mjml files
     currently on disk (excludes .bak-* and .legacy-* variants)
   • listEditableTemplates() now scans disk instead of a static list,
     so newly-created templates appear automatically in the dropdown

3. Enable Unlayer's built-in panels
   • templates: true — exposes Unlayer's template library (limited
     free-tier selection but ~10-20 starters available without a
     projectId)
   • stockImages: true — Unsplash search built into image picker
   • imageEditor: true — basic crop/resize on inserted images
   • undoRedo: true — history navigation

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-22 06:39:26 -04:00
louispaulb
d2cb8f4ee0 feat(campaigns): convert existing HTML templates to Unlayer JSON designs
Solve the "editor starts blank" problem by writing a one-time converter
that wraps each compiled .html template into a minimal Unlayer design
JSON (one Custom HTML block containing the entire body content). On
next editor load, Unlayer reads .json and renders the template in the
canvas — instant visual fidelity without manual reconstruction.

Strategy choice: Unlayer's "Import HTML" is a Pro-only feature. Building
a real HTML→Unlayer-blocks parser is several days of work. The minimal-
viable conversion (1 row + 1 Custom HTML block) gets the user 90% there
immediately:

  • Canvas shows the template visually (Unlayer renders the HTML)
  • Variables ({{firstname}}, {{gift_url}}, etc.) preserved as text
  • User can edit the HTML directly via the block's side panel
  • User can incrementally REPLACE the HTML block with native Unlayer
    blocks (Text, Image, Button) for any section they want decomposed —
    on their own schedule, not blocking the campaign send

New file: services/targo-hub/scripts/convert-html-to-unlayer.js
  • CLI: node scripts/convert-html-to-unlayer.js <template-name>
  • Reads templates/<name>.html, extracts <body> inner content, detects
    preheader from a hidden <div style="display:none">, builds Unlayer
    design JSON with brand-appropriate body.values (Targo Green link
    color #00C853, Plus Jakarta Sans font, F5FAF7 page background).
  • Backs up existing .json before overwriting.

Generated outputs (committed):
  templates/gift-email-fr.json — 34 KB (30 KB inner HTML + Unlayer chrome)
  templates/gift-email-en.json — 33 KB

Live verification: GET /campaigns/templates/gift-email-fr now returns
{ design: {...Unlayer JSON...} } alongside html. The editor's
onReady() callback in TemplateEditorPage detects data.design and calls
editor.loadDesign(design) → canvas populated immediately.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-22 06:22:47 -04:00
louispaulb
0cd99b6d44 feat(ops/campaigns): pivot template editor to Unlayer (vue-email-editor)
After honest acknowledgment that easy-email-standard is abandoned and
limited (Chrome-only, no responsive preview, no AMP, no Unsplash, no
file manager), pivoted to Unlayer's vue-email-editor — a Vue 3 native
component giving all the features the user listed for free (internal
use; a small "Powered by Unlayer" badge shows in the sidebar but NOT
in sent emails).

Why drop MJML alongside:
  • MJML was our SERVER-SIDE compilation step because we hand-wrote
    templates. With a visual editor that outputs email-safe HTML
    directly (responsive media queries, Outlook MSO fallbacks, AMP
    where used), the compilation step is redundant.
  • One fewer dependency on the hub (mjml package no longer needed).
  • One fewer file format to persist (.mjml dropped, only .html
    canonical + .json design).

Storage simplification:
  Before: .mjml (source) + .html (compiled) + .json (editor state)
  After:  .html (canonical) + .json (Unlayer design tree)

The hub's send-worker reads .html as before — no changes to send
logic.

Architecture wins:
  • Vue 3 native — zero iframe friction, no postMessage choreography
  • No separate microservice — easy-email container decommissioned
    (docker compose down, code kept under /opt/email-editor/ in case
    of rollback)
  • DNS editor.gigafibre.ca retained but unused — can be removed via
    Cloudflare API cleanup later
  • The editor's mergeTags option exposes our {{firstname}}, {{amount}},
    {{gift_url}}, etc. in Unlayer's native "Merge tags" panel — same
    pattern, more polished UI
  • Features now native: responsive preview (mobile/tablet/desktop
    breakpoints), Unsplash search, file manager, dark mode, design
    history, undo/redo, layers panel, content blocks library

Frontend (TemplateEditorPage.vue):
  • Imports EmailEditor from vue-email-editor
  • onReady() callback: fetch template + loadDesign() to restore canvas
  • saveTemplate(): exportHtml() → PUT { html, design } to hub
  • Top bar kept: template selector, saved chip, preview, test-send,
    save button
  • Removed: iframe-related glue (postMessage listener, iframeKey,
    EDITOR_BASE constant, Cmd-S handling that lived in the iframe)

API client (apps/ops/src/api/campaigns.js):
  • saveTemplate() now accepts opts.design (Unlayer JSON tree) alongside
    content. Legacy opts.format='mjml' still works for backward compat.

Hub (services/targo-hub/lib/campaigns.js):
  • GET /campaigns/templates/:name unconditionally returns
    { name, format, html, design } (+ mjml when format=mjml for
    legacy templates). The design field is null when no .json file
    exists yet.
  • PUT /campaigns/templates/:name HTML save path now accepts
    body.design alongside body.html and persists both with backups.
  • MJML save path (legacy) preserved for any callers using the old
    contract.

Container decommissioned on prod: email-editor container stopped +
removed. The Vue editor lives inside the ops SPA, served from
erp.gigafibre.ca/ops as a normal route.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-22 06:14:06 -04:00
louispaulb
229d450f9b feat(email-editor): persist easy-email JSON state for instant restore on reload
Phase 2.5 — close the load/save loop so the editor isn't broken by a page
refresh.

Problem: easy-email doesn't ship an MJML→JSON parser, so loading an
existing MJML template into the editor canvas isn't possible. First-time
load = empty canvas. Without this fix, every page reload would also reset
to empty (even after saving), making the editor useless past one session.

Solution: persist easy-email's raw JSON tree (editor state) as a third
companion file alongside .mjml + .html. Editor reads .json on load when
present, falls back to empty otherwise.

Three files per template now:
  gift-email-fr.mjml   — MJML source (rendered by send-worker → already done)
  gift-email-fr.html   — compiled HTML (cached output, sent to recipients)
  gift-email-fr.json   — easy-email editor state (UI restoration only)

Backend (lib/campaigns.js):
- New templateJsonPath() helper + EDITABLE_TEMPLATES checks
- GET /campaigns/templates/:name returns { format, mjml, html, json }
  when format=mjml (json null until first easy-email save)
- PUT /campaigns/templates/:name now accepts body.json alongside body.mjml
  (writes both .mjml + .html [compiled] + .json [editor state])
- Backup hook extended to also backup .json before overwrite

Editor (EmailEditorApp.tsx):
- On load: prefer data.json → parse and seed initialValues. If json
  missing but mjml present, show explanatory error banner + empty canvas
  (user reconstructs once; that save fixes future loads).
- On save: send BOTH mjml (compiled via JsonToMjml) AND raw values
  object as json. Hub persists all three artifacts.

First UX flow on next user visit:
  1. Open editor → empty canvas + banner "MJML exists but no JSON
     editor-state yet; reconstruct once to save a JSON snapshot"
  2. User drag-drops blocks to rebuild the template visually
  3. Click save → MJML + HTML + JSON all persist
  4. Subsequent reloads load from JSON instantly with full editor state

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-22 06:04:48 -04:00
louispaulb
e48ce94f26 feat(email-editor): Phase 1 — scaffold easy-email microservice for visual template editing
GrapesJS-mjml proved broken on our content (plugin v1.0.8 incompatible
with MJML v5 — canvas stays empty on load). Pivot to easy-email, a
mature OSS WYSIWYG email editor (MJML-based, MIT license, 4k stars).

Architecture: standalone React+Vite microservice deployed at
editor.gigafibre.ca, iframed from the ops UI's
/campaigns/templates/:name page. Talks to the hub's existing REST
endpoints (/campaigns/templates/*) for load + save. The hub stays the
source of truth — easy-email is purely the editing UI.

Scaffold delivered in this commit (Phase 1):

- services/email-editor/ — new top-level service directory
- package.json: React 18 + easy-email-{core,editor,extensions} 4.16.x
  + Vite 5 + TypeScript 5
- vite.config.ts: standard dev/build config, port 5173 in dev
- tsconfig.json: strict-false to keep iteration fast
- index.html: loads easy-email CSS bundles from unpkg (extensions, editor,
  arco theme)
- src/main.tsx: React entry, mounts EmailEditorApp on #root
- src/EmailEditorApp.tsx:
  • Reads template name from ?name=... URL param (defaults gift-email-fr)
  • GET ${VITE_HUB_URL}/campaigns/templates/:name on mount
  • Renders <EmailEditorProvider> + <StandardLayout> with our merge tags
    map (firstname, amount, gift_url, description, expiry, etc.) so the
    Variables panel shows our Mustache placeholders
  • On save: JsonToMjml() converts easy-email's JSON → MJML, PUT to hub
    → hub compiles to HTML and persists both files
  • postMessage({type: 'email-editor:saved', ...}) to parent window so
    the iframing ops UI knows to refresh
- Dockerfile: multi-stage (Vite build → nginx alpine serve). SPA fallback
  in nginx config so all routes return index.html.
- docker-compose.yml: container behind Traefik at editor.gigafibre.ca
  with Let's Encrypt TLS via the shared proxy network.
- README.md documents the arch, URL params, postMessage protocol, dev
  workflow, and the Phase 1 limitation (no MJML→JSON importer — editor
  starts from empty page until Phase 3).
- .gitignore: standard node/vite/dist exclusions.

Build verified locally: 83 modules transformed, ~2.8 MB bundle (840 KB
gzipped) — large but acceptable since easy-email packages the full
email builder + drag-drop canvas.

Phase 2 (next): Docker deploy on prod + replace GrapesJS in the ops UI
TemplateEditorPage with an iframe pointing here.
Phase 3 (later): MJML → easy-email JSON parser so existing templates
auto-import into the canvas instead of starting blank.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-21 23:52:31 -04:00
louispaulb
829b91eecc feat(campaigns/templates): richer 4-block intro (greeting, hook, gift, upsell)
Expanded the email intro from 3 short paragraphs into 4 semantic blocks,
restoring the marketing-friendly "Tu choisis local..." line that earlier
edits had dropped, plus adding new content about the 3.5 Gbit/s plans
and a "we're right around the corner" CTA framing.

FR intro structure now:
  1. "Bonjour {{firstname}},"
  2. "Tu choisis local, on veut te remercier. / Comme toi, on aime les
     connexions stables et les relations durables." (paired manifesto)
  3. "Avec l'arrivée de l'été, voici un cadeau pour toi, disponible
     pour un temps limité."
  4. "Nous offrons maintenant de nouveaux forfaits, jusqu'à 3.5 Gbit/s.
     Que tu souhaites plus de vitesse, battre une autre offre ou juste
     nous jaser, on est juste à côté."

EN translation mirrors the same 4-block structure.

Editorial rationale for block grouping in MJML:
- Each block is its own <mj-text> for independent drag-drop in GrapesJS
- Lines that always travel together (manifesto pair, upsell + CTA pair)
  share one <mj-text> joined with <br/> to reduce component clutter
- Different styles per block (greeting smaller/secondary, manifesto
  larger/bolder, body paragraphs normal) require separate <mj-text>
  components anyway since MJML inherits styling per-block

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-21 22:44:16 -04:00
louispaulb
1c2fa10f7f feat(campaigns): MJML canonical templates + test-send button
Two big moves:

1. Promote MJML to the canonical template format
   - Move gift-email-fr-mjml.{mjml,html} → gift-email-fr.{mjml,html}
   - Create gift-email-en.mjml (English translation of FR MJML)
   - Compile EN MJML → gift-email-en.html
   - Remove obsolete variants:
     • gift-email-fr-simple.html (now replaced by MJML)
     • gift-email-en-simple.html (same)
     • gift-email-fr-mjml.* (renamed to canonical)
   - The old gift-email-fr.html (rich-with-merchant-grid version) is
     backed up as gift-email-fr.legacy-rich.html.bak — kept on disk
     for reference but not in the editable list.
   - EDITABLE_TEMPLATES is now just ['gift-email-fr', 'gift-email-en'],
     both backed by .mjml source + .html auto-compiled output.

2. Add "Envoyer un test" feature
   Backend:
   - POST /campaigns/templates/:name/test-send accepts { to, vars,
     from?, subject? }. Reads compiled .html, renders Mustache vars,
     sends via Mailjet through email.sendEmail with X-MJ-CustomID
     "test-send:<name>:<timestamp>" so webhook events for tests are
     identifiable. Returns { sent, to, from, message_id, bytes }.
   - Default vars are sensible: firstname="Louis", amount="60 $",
     gift_url="https://gft.link/TEST123", etc. User overrides any
     via the request body.

   Frontend (TemplateEditorPage):
   - Toolbar button "Envoyer un test" (orange) — opens a dialog.
   - Dialog has email input + subject + 7 variable inputs
     (firstname, lastname, amount, commitment_months, gift_url,
     description, expiry) with sensible defaults.
   - "Dirty" banner warning: if the user has unsaved changes, the
     test will use the LAST SAVED version (so save first to test the
     latest). Mentions explicitly in card footer.
   - On send: live notification with the message_id + byte count.
     Errors surface clearly.

Verified live in prod:
  POST /campaigns/templates/gift-email-fr/test-send → 200, message_id
  returned, ~32 KB rendered MJML→HTML output, sent from
  TARGO <support@targointernet.com> (Mailjet-validated sender).

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-21 22:36:35 -04:00