refactor!: rename hydrovu source to pvacd_hydrovu for multi-tenant layout (ST2DAT-241)

.dlt/config.toml

@@ -1,4 +1,4 @@
-[sources.hydrovu]
+[sources.pvacd_hydrovu]
 api_base_url       = "https://www.hydrovu.com/public-api/v1"
 token_url          = "https://hydrovu.com/public-api/oauth/token"
 gcp_secret         = "hydrovu_pvacd"

AGENTS.md

@@ -20,11 +20,11 @@ Each source is an independent pipeline with three stages:
 API → dlt → GCS (parquet) → Adapter → CanonicalBundle → FROST loader → FROST
 ```
 
-| Stage | Asset (HydroVu) | Asset (CABQ) |
+| Stage | Asset (PVACD HydroVu) | Asset (CABQ) |
 |---|---|---|
-| Ingest (dlt → GCS) | `raw_hydrovu_readings` | `raw_cabq_readings` |
-| Transform (GCS → CanonicalBundles) | `canonical_bundles_hydrovu` | `canonical_bundles_cabq` |
-| Load (CanonicalBundles → FROST) | `frost_load_hydrovu` | `frost_load_cabq` |
+| Ingest (dlt → GCS) | `raw_pvacd_hydrovu_readings` | `raw_cabq_readings` |
+| Transform (GCS → CanonicalBundles) | `canonical_bundles_pvacd_hydrovu` | `canonical_bundles_cabq` |
+| Load (CanonicalBundles → FROST) | `frost_load_pvacd_hydrovu` | `frost_load_cabq` |
 
 HydroVu is live; CABQ is scaffolded (`cabq_*` raise `NotImplementedError`). Use
 HydroVu as the reference implementation when wiring up a new source.
@@ -48,7 +48,7 @@ src/aqueduct_dagster/
 │   ├── pipeline.py        # build_source_pipeline() — shared dlt pipeline factory
 │   ├── http.py            # retry_transient(), TokenManager, BearerAuth, build_authenticated_client()
 │   └── source_registry.py # SOURCE_REGISTRY — single per-source config, read by definitions.py and load.py
-├── sources/        # one folder per agency source (vertical slice) — see hydrovu/ as the reference
+├── sources/        # one folder per source key (vertical slice) — see pvacd_hydrovu/ as the reference
 │   └── <name>/
 │       ├── adapter.py       # raw rows → CanonicalBundle (source-specific)
 │       ├── dlt_pipeline.py  # dlt source/resource/pipeline factory

README.md

@@ -13,17 +13,17 @@ Dagster + dlt + GCS + FROST SensorThings
 Two independent source pipelines, each running on its own schedule:
 
 ```
-HydroVu API  → dlt → GCS (parquet) → HydroVuAdapter → CanonicalBundle → frost_load_hydrovu → FROST
-CABQ API     → dlt → GCS (parquet) → CabqAdapter    → CanonicalBundle → frost_load_cabq    → FROST
+HydroVu API  → dlt → GCS (parquet) → HydroVuAdapter → CanonicalBundle → frost_load_pvacd_hydrovu → FROST
+CABQ API     → dlt → GCS (parquet) → CabqAdapter    → CanonicalBundle → frost_load_cabq          → FROST
 ```
 
 Orchestrated by Dagster. Each pipeline has three assets:
 
-| Asset | HydroVu | CABQ |
-|-------|---------|------|
-| Ingest (dlt → GCS) | `raw_hydrovu_readings` | `raw_cabq_readings` |
-| Transform (GCS → CanonicalBundles) | `canonical_bundles_hydrovu` | `canonical_bundles_cabq` |
-| Load (CanonicalBundles → FROST) | `frost_load_hydrovu` | `frost_load_cabq` |
+| Asset | PVACD HydroVu | CABQ |
+|-------|---------------|------|
+| Ingest (dlt → GCS) | `raw_pvacd_hydrovu_readings` | `raw_cabq_readings` |
+| Transform (GCS → CanonicalBundles) | `canonical_bundles_pvacd_hydrovu` | `canonical_bundles_cabq` |
+| Load (CanonicalBundles → FROST) | `frost_load_pvacd_hydrovu` | `frost_load_cabq` |
 
 
 
@@ -60,21 +60,21 @@ Aqueduct/
 │   │   ├── http.py                 # retry_transient(), TokenManager, BearerAuth, build_authenticated_client()
 │   │   ├── backfill.py             # month_chunks(), BackfillCheckpointStore, ChunkResult — Mode A refetch infra
 │   │   └── source_registry.py      # SOURCE_REGISTRY — single per-source config for definitions.py + load.py
-│   ├── sources/                    # one folder per agency source (vertical slice)
-│   │   ├── hydrovu/
+│   ├── sources/                    # one folder per source key (vertical slice)
+│   │   ├── pvacd_hydrovu/          # PVACD's HydroVu tenant; BernCo's will sit beside it
 │   │   │   ├── adapter.py          # HydroVu → CanonicalBundle mapping
 │   │   │   ├── dlt_pipeline.py     # dlt source + resource + pipeline factory
-│   │   │   ├── ingest.py           # Dagster asset: raw_hydrovu_readings
-│   │   │   ├── transform.py        # Dagster asset: canonical_bundles_hydrovu
+│   │   │   ├── ingest.py           # Dagster asset: raw_pvacd_hydrovu_readings
+│   │   │   ├── transform.py        # Dagster asset: canonical_bundles_pvacd_hydrovu
 │   │   │   └── backfill.py         # Mode A refetch: isolated ingest + transform + load per chunk
-│   │   └── cabq/                   # same shape as hydrovu/ — currently a stub
+│   │   └── cabq/                   # same shape as pvacd_hydrovu/ — currently a stub
 │   │       ├── adapter.py
 │   │       ├── dlt_pipeline.py
 │   │       ├── ingest.py
 │   │       └── transform.py
 │   ├── defs/
 │   │   ├── assets/
-│   │   │   └── load.py             # Dagster assets: frost_load_hydrovu, frost_load_cabq (shared factory)
+│   │   │   └── load.py             # Dagster assets: frost_load_pvacd_hydrovu, frost_load_cabq (shared factory)
 │   │   ├── jobs/
 │   │   │   └── backfill.py         # <source>_backfill_refetch job factory (BackfillRefetchConfig, chunk loop)
 │   │   ├── definitions.py          # Dagster entry point — jobs, schedules, asset registry
@@ -84,7 +84,7 @@ Aqueduct/
 │       └── watermark_store.py      # FrostWatermarkStore — per-run dedup via Dagster context
 └── tests/                          # mirrors src/aqueduct_dagster/'s layout above
     ├── conftest.py                 # cross-file test helpers (e.g. httpx.MockTransport/BearerAuth builders)
-    ├── sources/{hydrovu,cabq}/
+    ├── sources/{pvacd_hydrovu,cabq}/
     ├── shared/
     ├── defs/assets/
     ├── defs/jobs/
@@ -190,7 +190,7 @@ Open the Dagster UI at `http://localhost:3000`.
 
 To run the full HydroVu pipeline end-to-end:
 1. Click **Assets** in the left nav
-2. Select all three `hydrovu` group assets (`raw_hydrovu_readings`, `canonical_bundles_hydrovu`, `frost_load_hydrovu`)
+2. Select all three `pvacd_hydrovu` group assets (`raw_pvacd_hydrovu_readings`, `canonical_bundles_pvacd_hydrovu`, `frost_load_pvacd_hydrovu`)
 3. Click **Materialize selected**
 
 On first run, dlt fetches from `initial_start_date` in `.dlt/config.toml` (currently `2026-05-01`). Subsequent runs are incremental.
@@ -225,6 +225,6 @@ dlt tracks a cursor (`timestamp` field) per source. On first run it fetches from
 `FrostWatermarkStore` tracks the last observation timestamp successfully loaded into FROST per datastream. Each run skips any observation at or before the watermark — FROST has no built-in deduplication.
 
 **Independent pipelines**
-`hydrovu_pipeline` and `cabq_pipeline` are completely independent Dagster jobs. Each has its own schedule and its own terminal load asset (`frost_load_hydrovu` / `frost_load_cabq`). Running one never triggers or blocks the other.
+`pvacd_hydrovu_pipeline` and `cabq_pipeline` are completely independent Dagster jobs. Each has its own schedule and its own terminal load asset (`frost_load_pvacd_hydrovu` / `frost_load_cabq`). Running one never triggers or blocks the other.
 
 ---

deploy/00_config.sh

@@ -50,8 +50,8 @@ export SQL_DB_VERSION="POSTGRES_17"
 # --- Secret Manager ---------------------------------------------------------
 export SECRET_FROST_DB_PW="frost-db-password"
 # HydroVu OAuth client id/secret, read at ingest time via ADC. Matches
-# [sources.hydrovu] gcp_secret in .dlt/config.toml.
-export SECRET_HYDROVU="hydrovu_pvacd"
+# [sources.pvacd_hydrovu] gcp_secret in .dlt/config.toml.
+export SECRET_PVACD_HYDROVU="hydrovu_pvacd"
 
 # --- GCS buckets ------------------------------------------------------------
 # Where Dagster+ writes raw parquet. BUCKET_PROD is the deployment target;

deploy/30_dagster_gcp_auth.sh

@@ -19,7 +19,7 @@
 #
 # Requires: roles/storage.admin (bucket create + IAM), roles/iam.serviceAccountAdmin
 # (only if the SA does not exist yet), roles/iam.serviceAccountKeyAdmin (only with
-# --emit-key), and permission to bind secretAccessor on ${SECRET_HYDROVU}.
+# --emit-key), and permission to bind secretAccessor on ${SECRET_PVACD_HYDROVU}.
 
 set -euo pipefail
 source "$(dirname "${BASH_SOURCE[0]}")/00_config.sh"
@@ -116,8 +116,8 @@ done
 
 # The HydroVu OAuth credentials are themselves fetched through ADC at ingest time,
 # so bucket access alone is not enough to run the pipeline.
-echo "== Grant secretAccessor on ${SECRET_HYDROVU} =="
-gcloud secrets add-iam-policy-binding "${SECRET_HYDROVU}" \
+echo "== Grant secretAccessor on ${SECRET_PVACD_HYDROVU} =="
+gcloud secrets add-iam-policy-binding "${SECRET_PVACD_HYDROVU}" \
   --project="${PROJECT_ID}" \
   --member="serviceAccount:${DAGSTER_SA}" \
   --role="roles/secretmanager.secretAccessor" >/dev/null
@@ -184,7 +184,7 @@ cat <<EOF
 
 Done. Service account : ${DAGSTER_SA}
       Buckets         : gs://${BUCKET_PROD} (prod), gs://${BUCKET_POC} (poc/verify)
-      Secret          : ${SECRET_HYDROVU}
+      Secret          : ${SECRET_PVACD_HYDROVU}
       FROST           : ${FROST_URL:-<not deployed>}
 
 Dagster+ → Deployment → Environment variables (code location aqueduct_dagster_defs_definitions):
@@ -206,6 +206,6 @@ Verify without minting a key (needs roles/iam.serviceAccountTokenCreator on the
 
 Confirm the bindings landed:
   gcloud storage buckets get-iam-policy gs://${BUCKET_PROD} --format=json
-  gcloud secrets get-iam-policy ${SECRET_HYDROVU} --project=${PROJECT_ID}
+  gcloud secrets get-iam-policy ${SECRET_PVACD_HYDROVU} --project=${PROJECT_ID}
   gcloud run services get-iam-policy ${FROST_SERVICE} --project=${PROJECT_ID} --region=${REGION}
 EOF

deploy/31_verify_dagster_auth.py

@@ -104,7 +104,7 @@ def check_secret_manager() -> str:
 
     config = load_config()
     project_number = config["destination"]["filesystem"]["gcp_project_number"]
-    secret_id = config["sources"]["hydrovu"]["gcp_secret"]
+    secret_id = config["sources"]["pvacd_hydrovu"]["gcp_secret"]
 
     client = secretmanager.SecretManagerServiceClient()
     name = client.secret_version_path(project_number, secret_id, "latest")

deploy/README.md

@@ -250,7 +250,7 @@ uv run python deploy/31_verify_dagster_auth.py
 This checks four layers independently: ADC bootstrap, GCS write/read/delete, Secret
 Manager access, and an ID-token GET against FROST. `VERBOSE=1` adds
 tracebacks. For a full end-to-end run instead, `uv run dagster dev` with the same two
-variables set and materialize `hydrovu_pipeline`.
+variables set and materialize `pvacd_hydrovu_pipeline`.
 
 ## How Dagster+ connects to FROST
 

docs/BACKFILL_STRATEGY.md

@@ -147,7 +147,7 @@ regardless of count.
 **Option 3 — Per-source generated jobs, two per source (recommended).**
 
 This follows the pattern already used in this codebase for
-`frost_load_hydrovu` / `frost_load_cabq` and `hydrovu_pipeline` /
+`frost_load_pvacd_hydrovu` / `frost_load_cabq` and `pvacd_hydrovu_pipeline` /
 `cabq_pipeline`: one shared factory function per job type, looped over the
 source registry, so that adding a third source requires one registry entry
 and no new job-wiring code. "Which source" is determined by which job is
@@ -182,7 +182,7 @@ load together:
    (DTW filtering, location join, grouping, adapter mapping), factored into
    a plain function callable from both the normal transform asset and this
    job, mirroring how `_frost_load()` is already factored out separately
-   from the `frost_load_hydrovu` asset wrapper.
+   from the `frost_load_pvacd_hydrovu` asset wrapper.
 3. **Load** — always uses the window-scoped, delete-then-repost mechanism
    described in §4.4, rather than the normal `frost_load` asset's
    single-watermark filter. This mechanism is safe for every situation in
@@ -316,8 +316,8 @@ or list specific ids to scope the run to just those.
 Each source's existing schedule fires its existing pipeline job
 automatically, with no operator involvement:
 
-- `hydrovu_schedule` (`0 6 * * *`) triggers `hydrovu_pipeline`, which runs
-  `raw_hydrovu_readings` → `canonical_bundles_hydrovu` → `frost_load_hydrovu`
+- `pvacd_hydrovu_schedule` (`0 6 * * *`) triggers `pvacd_hydrovu_pipeline`, which runs
+  `raw_pvacd_hydrovu_readings` → `canonical_bundles_pvacd_hydrovu` → `frost_load_pvacd_hydrovu`
   in sequence.
 - `cabq_schedule` (`0 8 * * *`) triggers `cabq_pipeline` independently, with
   its own three assets. Running one source's pipeline never triggers or
@@ -331,7 +331,7 @@ and `<source>_backfill_replay`. Neither has a schedule attached — both are
 launched manually, on demand:
 
 1. Open the Dagster UI and select **Jobs**.
-2. Select the relevant job (for example, `hydrovu_backfill_refetch`).
+2. Select the relevant job (for example, `pvacd_hydrovu_backfill_refetch`).
 3. Open the **Launchpad** and provide the run configuration — entity list,
    start and end dates, and (for Mode A) an isolated pipeline name — as YAML
    or JSON directly in the browser. No command-line access is required.