Add BANATIC source to download pipeline

Author: tgrandje

.gitignore

@@ -134,3 +134,13 @@ dmypy.json
 
 *.sqlite
 *.sqlite*
+
+**/emissions.csv
+*.gpkg
+*.topojson
+*.geojson
+*.shp
+*.cpg
+*.dbf
+*.prj
+*.shx

cartiflette/constants.py

@@ -36,6 +36,30 @@
 REFERENCES = gpd.GeoDataFrame(REFERENCES, crs=4326)
 
 DOWNLOAD_PIPELINE_ARGS = {
+    "BANATIC": [
+        "MinistereInterieur",
+        "BANATIC",
+        "PERIMETRE-GROUPEMENTS",
+        [
+            "Auvergne-Rhone-Alpes",
+            "Bourgogne-Franche-Comte",
+            "Bretagne",
+            "Centre-Val-de-Loire",
+            "Corse",
+            "Grand-Est",
+            "Guadeloupe",
+            "Guyane",
+            "Hauts-de-France",
+            "Ile-de-France",
+            "La-Reunion",
+            "Martinique",
+            "Normandie",
+            "Nouvelle-Aquitaine",
+            "Occitanie",
+            "Pays-de-la-Loire",
+            "Provence-Alpes-Cote-d-Azur",
+        ],
+    ],
     "ADMIN-EXPRESS": [
         "IGN",
         "ADMINEXPRESS",

cartiflette/download/scraper.py

@@ -435,7 +435,10 @@ def download_unpack(self, datafile: Dataset, **kwargs) -> DownloadReturn:
                 root_folder, files_locations = datafile.unpack(protocol="7z")
             elif "Zip archive" in filetype:
                 root_folder, files_locations = datafile.unpack(protocol="zip")
-            elif "Unicode text" in filetype:
+            elif (
+                "Unicode text" in filetype
+                or "text, with very long lines" in filetype
+            ):
                 # copy in temp directory without processing
                 root_folder = tempfile.mkdtemp()
                 with open(temp_archive_file_raw, "rb") as f:

cartiflette/utils/sources.yaml

@@ -1,3 +1,50 @@
+MinistereInterieur:
+  BANATIC:
+    PERIMETRE-GROUPEMENTS:
+      structure: "{url_prefix}?zone={territory}&date={date}&format=D"
+      territory:
+        Auvergne-Rhone-Alpes: R84
+        Bourgogne-Franche-Comte: R27
+        Bretagne: R53
+        Centre-Val-de-Loire: R24
+        Corse: R94
+        Grand-Est: R44
+        Guadeloupe: R01
+        Guyane: R03
+        Hauts-de-France: R32
+        Ile-de-France: R11
+        La-Reunion: R04
+        Martinique: R02
+        Normandie: R28
+        Nouvelle-Aquitaine: R75
+        Occitanie: R76
+        Pays-de-la-Loire: R52
+        Provence-Alpes-Cote-d-Azur: R93
+      2023:
+        url_prefix: "https://www.banatic.interieur.gouv.fr/V5/fichiers-en-telechargement/telecharger.php"
+        date: 01/01/2023
+      2022:
+        url_prefix: "https://www.banatic.interieur.gouv.fr/V5/fichiers-en-telechargement/telecharger.php"
+        date: 01/01/2022
+      2021:
+        url_prefix: "https://www.banatic.interieur.gouv.fr/V5/fichiers-en-telechargement/telecharger.php"
+        date: 01/01/2021
+      2020:
+        url_prefix: "https://www.banatic.interieur.gouv.fr/V5/fichiers-en-telechargement/telecharger.php"
+        date: 01/01/2020
+      2019:
+        url_prefix: "https://www.banatic.interieur.gouv.fr/V5/fichiers-en-telechargement/telecharger.php"
+        date: 01/01/2019
+      2018:
+        url_prefix: "https://www.banatic.interieur.gouv.fr/V5/fichiers-en-telechargement/telecharger.php"
+        date: 01/01/2018
+      2017:
+        url_prefix: "https://www.banatic.interieur.gouv.fr/V5/fichiers-en-telechargement/telecharger.php"
+        date: 01/01/2017
+      2016:
+        url_prefix: "https://www.banatic.interieur.gouv.fr/V5/fichiers-en-telechargement/telecharger.php"
+        date: 01/01/2016
+
 IGN:
 
   ADMINEXPRESS: