Develop (#53)

* feat(web): make submenu tabs responsive and full-width

- Add horizontal scroll navigation with chevron buttons for AdminTabs
- Implement flex-1 layout for tabs to fill available width on large screens
- Add smooth scrolling and scroll state detection for small screens
- Ensure tab borders align with main content container
- Update all tab components (AdminTabs, ApiDocsTabs, ConsumptionTabs) for consistency

🤖 Generated with Claude Code

Co-Authored-By: Claude <[email protected]>

* feat(web): make submenu tabs responsive and full-width (#41)

- Add horizontal scroll navigation with chevron buttons for AdminTabs
- Implement flex-1 layout for tabs to fill available width on large screens
- Add smooth scrolling and scroll state detection for small screens
- Ensure tab borders align with main content container
- Update all tab components (AdminTabs, ApiDocsTabs, ConsumptionTabs)

🤖 Generated with Claude Code

Co-authored-by: Clément VALENTIN <[email protected]>
Co-authored-by: Claude <[email protected]>

* fix(admin): add missing cache_service import and improve stats loading UI (#43)

Fixed NameError in /admin/stats endpoint caused by missing cache_service import.
Added proper loading state using LoadingOverlay component with blur effect.
Extended LoadingOverlay to support 'admin' data type with appropriate messages.

🤖 Generated with Claude Code

Co-authored-by: Clément VALENTIN <[email protected]>
Co-authored-by: Claude <[email protected]>

* fix(web): regenerate package-lock.json for npm ci compatibility (#45)

The previous package-lock.json was missing tree-sitter dependencies
([email protected] and [email protected]), causing `npm ci` to fail
during Docker builds with the error "package.json and package-lock.json
are not in sync".

This regenerates the lock file to include all required dependencies.

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

Co-authored-by: Clément VALENTIN <[email protected]>
Co-authored-by: Claude <[email protected]>

* feat(web): add password visibility toggle on login page

Add eye icon button to show/hide password in the login form for better UX.

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

Co-Authored-By: Claude <[email protected]>

* feat(k8s): add automatic Kubernetes deployment script with dev mode (#47)

Add complete Kubernetes deployment automation via rancher-desktop with dev mode support (local volume mounts for hot-reload). Includes Helm chart modifications for dev volumes, readiness probes with Host header, Vite allowedHosts configuration, and environment variable loading from .env files. Integrates with Conductor for automatic project startup/shutdown.

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

Co-authored-by: Clément VALENTIN <[email protected]>
Co-authored-by: Claude <[email protected]>

* fix(simulator): handle undefined or non-array offersData (#52)

Fix "offersData.filter is not a function" error by ensuring offersData
is an array before calling filter. Add validation checks and improve
error messages when offers fail to load.

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

Co-authored-by: Clément VALENTIN <[email protected]>
Co-authored-by: Claude <[email protected]>

* fix: ensure yesterday's data is fetched from Enedis API (#54)

Fix date range handling to ensure consumption and production detail data from yesterday (J-1) is properly fetched. The batch endpoints now automatically extend date ranges when a single day is requested, ensuring minimum 2-day constraint for Enedis API calls.

Key changes:
- adjust_date_range() caps end_date to yesterday (J-1) instead of today
- Batch endpoints extend start backwards when start == end to guarantee 2-day range
- Updated API documentation with date constraints and automatic handling details

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

Co-authored-by: Clément VALENTIN <[email protected]>
Co-authored-by: Claude <[email protected]>

* fix: regenerate package-lock.json with all dependencies

Fixes npm ci build failure caused by missing tree-sitter dependencies.
Regenerated package-lock.json using npm install.

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

Co-Authored-By: Claude <[email protected]>

---------

Co-authored-by: Clément VALENTIN <[email protected]>
Co-authored-by: Claude <[email protected]>

Author: 3 people

apps/api/src/routers/enedis.py

@@ -110,7 +110,7 @@ async def verify_pdl_ownership(usage_point_id: str, user: User, db: AsyncSession
 def adjust_date_range(start: str, end: str) -> tuple[str, str]:
     """
     Adjust date range to ensure it's valid for Enedis API:
-    1. Cap end date to today if it's in the future
+    1. Cap end date to yesterday (J-1) since Enedis only provides data up to J-1
     2. If start == end, move start to 1 day before
     Returns (adjusted_start, adjusted_end) as strings (YYYY-MM-DD).
     """
@@ -123,11 +123,13 @@ def adjust_date_range(start: str, end: str) -> tuple[str, str]:
         paris_tz = ZoneInfo("Europe/Paris")
         today_paris = datetime.now(paris_tz).replace(hour=0, minute=0, second=0, microsecond=0)
         today = today_paris.replace(tzinfo=None)
+        # Enedis data is only available up to yesterday (J-1)
+        yesterday = today - timedelta(days=1)
 
-        # 1. Cap end date to today if it's in the future
-        if end_date > today:
-            logger.info(f"[DATE ADJUST] End date {end} is in the future, capping to today {today.strftime('%Y-%m-%d')}")
-            end_date = today
+        # 1. Cap end date to yesterday if it's after yesterday (including today)
+        if end_date > yesterday:
+            logger.info(f"[DATE ADJUST] End date {end} is after yesterday (J-1), capping to {yesterday.strftime('%Y-%m-%d')}")
+            end_date = yesterday
 
         # 2. If start == end, move start to 1 day before
         if start_date >= end_date:
@@ -1094,16 +1096,21 @@ async def get_consumption_detail_batch(
                     # Use chunk_end + 1 day to get last 23:30 reading
                     fetch_end_date = chunk_end_date + timedelta(days=1)
 
-                # CRITICAL: Never request data beyond yesterday (Enedis only has J-1 data)
-                if fetch_end_date > yesterday:
-                    fetch_end_date = yesterday
+                # CRITICAL: Never request data beyond today (Enedis returns J-1 data when end=today)
+                # We use TODAY (not yesterday) as the cap because Enedis API requires end > start
+                # and returns data up to J-1 of the end date
+                if fetch_end_date > today:
+                    fetch_end_date = today
 
                 fetch_end = fetch_end_date.strftime("%Y-%m-%d")
 
                 # CRITICAL: Enedis requires at least 2 days (start != end)
+                # If we would skip, extend start backwards to ensure we have a valid range
                 if current_start_str == fetch_end:
-                    log_with_pdl("warning", usage_point_id, f"[BATCH SKIP] Cannot fetch single day {current_start_str} (Enedis requires min 2 days). Skipping.")
-                    break  # Skip this chunk
+                    # Extend start 1 day backwards to get a 2-day range
+                    extended_start = current_start - timedelta(days=1)
+                    current_start_str = extended_start.strftime("%Y-%m-%d")
+                    log_with_pdl("info", usage_point_id, f"[BATCH EXTEND] Extended start from {current_start.strftime('%Y-%m-%d')} to {current_start_str} to ensure min 2-day range")
 
                 try:
                     if is_demo:
@@ -1703,16 +1710,21 @@ async def get_production_detail_batch(
                     # Use chunk_end + 1 day to get last 23:30 reading
                     fetch_end_date = chunk_end_date + timedelta(days=1)
 
-                # CRITICAL: Never request data beyond yesterday (Enedis only has J-1 data)
-                if fetch_end_date > yesterday:
-                    fetch_end_date = yesterday
+                # CRITICAL: Never request data beyond today (Enedis returns J-1 data when end=today)
+                # We use TODAY (not yesterday) as the cap because Enedis API requires end > start
+                # and returns data up to J-1 of the end date
+                if fetch_end_date > today:
+                    fetch_end_date = today
 
                 fetch_end = fetch_end_date.strftime("%Y-%m-%d")
 
                 # CRITICAL: Enedis requires at least 2 days (start != end)
+                # If we would skip, extend start backwards to ensure we have a valid range
                 if current_start_str == fetch_end:
-                    log_with_pdl("warning", usage_point_id, f"[BATCH PRODUCTION SKIP] Cannot fetch single day {current_start_str} (Enedis requires min 2 days). Skipping.")
-                    break  # Skip this chunk
+                    # Extend start 1 day backwards to get a 2-day range
+                    extended_start = current_start - timedelta(days=1)
+                    current_start_str = extended_start.strftime("%Y-%m-%d")
+                    log_with_pdl("info", usage_point_id, f"[BATCH PRODUCTION EXTEND] Extended start from {current_start.strftime('%Y-%m-%d')} to {current_start_str} to ensure min 2-day range")
 
                 try:
                     if is_demo:

apps/web/src/pages/Simulator.tsx

@@ -319,20 +319,27 @@ export default function Simulator() {
       const selectedPdlData = Array.isArray(pdlsData) ? pdlsData.find((p) => p.usage_point_id === selectedPdl) : undefined
       const subscribedPower = selectedPdlData?.subscribed_power
 
+      // Ensure offersData is an array
+      const offersArray = Array.isArray(offersData) ? offersData : []
+
       // Filter offers by subscribed power if available
-      const filteredOffers = subscribedPower && offersData
-        ? offersData.filter((offer) => {
+      const filteredOffers = subscribedPower && offersArray.length > 0
+        ? offersArray.filter((offer) => {
             const match = offer.name.match(/(\d+)\s*kVA/i)
             if (match) {
               const offerPower = parseInt(match[1])
               return offerPower === subscribedPower
             }
             return true
           })
-        : offersData || []
+        : offersArray
+
+      if (offersArray.length === 0) {
+        throw new Error('Aucune offre disponible. Veuillez patienter le temps du chargement des offres.')
+      }
 
       if (filteredOffers.length === 0) {
-        throw new Error('Aucune offre disponible pour la puissance souscrite de votre PDL')
+        throw new Error(`Aucune offre disponible pour la puissance souscrite de ${subscribedPower} kVA`)
       }
 
       // Calculer les simulations pour chaque offre
@@ -875,8 +882,12 @@ export default function Simulator() {
     }
 
     // Don't auto-launch if PDL data, offers, or providers are not loaded yet
-    if (!pdlsData || !offersData || !providersData) {
-      logger.log('[Auto-launch] Skipping auto-launch - data not loaded yet')
+    if (!pdlsData || !Array.isArray(offersData) || offersData.length === 0 || !providersData) {
+      logger.log('[Auto-launch] Skipping auto-launch - data not loaded yet', {
+        pdlsData: !!pdlsData,
+        offersData: Array.isArray(offersData) ? offersData.length : 'not array',
+        providersData: !!providersData
+      })
       return
     }
 

docs/enedis-api/endpoint.md

@@ -31,6 +31,61 @@ Le parcours OAuth (autorisation + recuperation du token) est documente dans `doc
 | Customer                     | Informations titulaires (identite, contacts).                   | `https://gw.ext.prod-sandbox.api.enedis.fr/customers_i/v5`        | `https://gw.ext.prod.api.enedis.fr/customers_i/v5`        | `docs/enedis-api/openapi/10-customer.json`                     |
 | Contact                      | Coordonnees de contact client.                                  | `https://gw.ext.prod-sandbox.api.enedis.fr/customers_cd/v5`       | `https://gw.ext.prod.api.enedis.fr/customers_cd/v5`       | `docs/enedis-api/openapi/11-contact.json`                      |
 
+## Contraintes de dates de l'API Enedis
+
+L'API Enedis impose plusieurs contraintes sur les plages de dates. Ces contraintes sont gerees automatiquement par le backend (`apps/api/src/routers/enedis.py`).
+
+### Disponibilite des donnees (J-1)
+
+**Les donnees Enedis ne sont disponibles que jusqu'a la veille (J-1) de la date actuelle.**
+
+- Si aujourd'hui est le 04/12/2025, les donnees les plus recentes disponibles sont celles du 03/12/2025
+- Les donnees sont mises a disposition chaque jour a partir de 8h (heure de Paris)
+- Certains compteurs peuvent remonter leurs donnees plus tard ; elles deviennent alors accessibles le lendemain
+
+### Plages de dates par type d'endpoint
+
+| Endpoint                | Plage max par appel | Historique max        | Notes                                      |
+|-------------------------|--------------------|-----------------------|--------------------------------------------|
+| Courbe de charge detail | **7 jours**        | 24 mois + 15 jours    | Donnees au pas 30min (ou 10/15/60min)      |
+| Consommation journaliere| 365 jours          | 36 mois + 15 jours    | Une valeur par jour                        |
+| Puissance max           | 365 jours          | 36 mois + 15 jours    | Une valeur par jour                        |
+| Production detail       | **7 jours**        | 24 mois + 15 jours    | Donnees au pas 30min                       |
+| Production journaliere  | 365 jours          | 36 mois + 15 jours    | Une valeur par jour                        |
+
+### Contrainte minimum 2 jours (start < end)
+
+**L'API Enedis exige que la date de debut soit strictement inferieure a la date de fin.**
+
+- Une requete avec `start=2025-12-03` et `end=2025-12-03` sera rejetee
+- Pour obtenir les donnees d'un seul jour, il faut demander une plage d'au moins 2 jours
+- Exemple : pour le 03/12, demander `start=2025-12-02, end=2025-12-04`
+
+### Gestion automatique par le backend
+
+Le backend (`apps/api/src/routers/enedis.py`) gere automatiquement ces contraintes :
+
+1. **`adjust_date_range()`** : Ajuste automatiquement la date de fin a J-1 si elle depasse
+2. **Endpoints batch** : Decoupent les grandes plages en chunks de 7 jours max
+3. **Extension automatique** : Si une requete ne contient qu'un seul jour (ex: hier), le backend etend la plage vers le passe pour garantir min 2 jours
+4. **Cache granulaire** : Chaque jour est mis en cache individuellement pour eviter les re-telechargements
+
+### Exemple de flux pour recuperer les donnees d'hier
+
+```
+Frontend demande: start=2025-12-03, end=2025-12-03 (hier)
+                           ↓
+Backend adjust_date_range(): end reste 2025-12-03 (< today)
+                           ↓
+Backend batch: detecte start == end → etend start a 2025-12-02
+                           ↓
+Appel Enedis: start=2025-12-02, end=2025-12-04 (today)
+                           ↓
+Enedis retourne: donnees du 02/12 et 03/12 (J-1 de end)
+                           ↓
+Backend filtre: garde uniquement les donnees demandees (03/12)
+```
+
 ## Bonnes pratiques
 
 - Chaque endpoint doit etre invoque via la passerelle interne (voir `docs/features-spec/05-gateway.md`).