feat: update vigilo historical script

.github/workflows/github-pages.yml

@@ -11,7 +11,7 @@ permissions:
 
 concurrency:
   group: pages
-  cancel-in-progress: true
+  cancel-in-progress: false
 
 env:
   JUST_VERSION: "1.51.0"

README.md

@@ -98,21 +98,67 @@ just update             # Convert to Parquet (DuckDB)
 just import             # Load into PostgreSQL
 ```
 
-To process a specific dataset:
-
-```bash
-DATAS_LIST=sirene just update
-DATAS_LIST=sirene just import
-```
-
 > **Note:** `DATAS_LIST` accepts multiple comma-separated dataset names:
 >
 > ```bash
 > DATAS_LIST=sirene,vigilo just update
 > DATAS_LIST=geonames,vigilo,nafrev2,sirene just import
 > ```
 
-Available datasets: `geonames`, `vigilo`, `sirene`, `nafrev2`
+### Per-dataset workflows
+
+Each dataset has two workflows: **update** (full refresh from source → publish to R2) and **import** (consume published Parquet files from R2).
+
+#### GeoNames
+
+```bash
+# Full update (download from geonames.org, convert, publish)
+DATAS_LIST=geonames just download
+DATAS_LIST=geonames just update
+DATAS_LIST=geonames just release
+
+# Import only (download Parquet from R2, load into DuckDB)
+DATAS_LIST=geonames just import
+```
+
+#### Vigilo
+
+```bash
+# Full update (download from Vigilo API, convert, publish)
+DATAS_LIST=vigilo just download
+DATAS_LIST=vigilo just update
+DATAS_LIST=vigilo just release
+
+# Import only (download Parquet from R2, load into DuckDB)
+DATAS_LIST=vigilo just import
+```
+
+#### NAF Rev. 2
+
+```bash
+# Full update (download from INSEE, convert, publish)
+DATAS_LIST=nafrev2 just download
+DATAS_LIST=nafrev2 just update
+DATAS_LIST=nafrev2 just release
+
+# Import only (download Parquet from R2, load into DuckDB)
+DATAS_LIST=nafrev2 just import
+```
+
+#### SIRENE
+
+```bash
+# Full update (download from data.gouv.fr ~4 GB, convert, publish)
+DATAS_LIST=sirene just download
+DATAS_LIST=sirene just update
+DATAS_LIST=sirene just release
+
+# Import only (download Parquet from R2, load into DuckDB)
+DATAS_LIST=sirene just import
+just sirene-etab-parquets   # generate Observable etab/ parquets from APE parquets
+```
+
+> **Note:** SIRENE download is ~4 GB. `just sirene-etab-parquets` must be run after import to generate the per-letter establishment parquets used by the Observable interface.
 
 ---
 

importer/release_download.sh

@@ -56,7 +56,7 @@ for dataset in $datasets; do
 
     echo "[release-download] ${dataset}: syncing from ${tag}..."
     python3 - "$manifest_local" "$dataset" "$FORCE" "$DL_TMP" "$REPO" "$tag" << 'PYEOF'
-import sys, json, os, subprocess
+import sys, json, os, shutil, subprocess
 
 manifest_path, dataset, force, dl_tmp, repo, tag = sys.argv[1:]
 force = force == "1"
@@ -79,7 +79,7 @@ for asset_name, rel_path in manifest.items():
         print(f"[release-download]   ERROR downloading {asset_name}: {result.stderr.strip()}")
         continue
     downloaded = os.path.join(dl_tmp, asset_name)
-    os.rename(downloaded, target)
+    shutil.move(downloaded, target)
     print(f"[release-download]   {asset_name} → {target}")
 PYEOF