improve missing_in_osm

missing_in_osm/missing_in_osm.sh

@@ -6,13 +6,10 @@ set -euvo pipefail
 
 cd data
 
-## préparation des lignes open data
-xsv join agency_id gtfs_routes.txt agency_id gtfs_agency.txt |xsv select agency_id,agency_name,route_id,route_short_name,route_long_name,route_type,route_color,route_text_color |xsv sort -s route_id|xsv sort -s agency_name > opendata_lignes.csv
-sed -i -e "s/IDFM://g" opendata_lignes.csv
-
-
-## ajout de l'id OSM sur les lignes open data
-xsv join --left route_id opendata_lignes.csv ref:FR:STIF lignes.csv|xsv select 1-8,line_id > opendata_lines_with_osm_id.csv
+## ajout des infos OSM sur les lignes open data
+xsv join --left ID_Line opendata_lines_referential.csv ref:FR:STIF lignes.csv \
+  |xsv select line_id,ID_Line,code,ShortName_Line,network,NetworkName,operator,OperatorName,mode,TransportMode \
+  |xsv sort -s NetworkName,OperatorName,ID_Line > merged_lines.csv
 # /!\ on a parfois plusieurs lignes OSM pour une même ligne opendata
 
 ## extraction des lignes OSM avec un tracé déjà existant

missing_in_osm/missing_lines.py

@@ -10,13 +10,13 @@
         osm_with_shapes.append(elem[0])
 
 lines = []
-with open("../data/opendata_lines_with_osm_id.csv") as csvfile:
+with open("../data/merged_lines.csv") as csvfile:
     csv_ = csv.DictReader(csvfile)
     for tt in csv_:
         line = {}
-        line['opendata_line_id'] = tt['route_id']
-        line['opendata_network'] = tt['agency_name']
-        line['opendata_line_name'] = tt['route_short_name']
+        line['opendata_line_id'] = tt['ID_Line']
+        line['opendata_network'] = tt['NetworkName']
+        line['opendata_line_name'] = tt['ShortName_Line']
         line['found_in_osm'] = tt['line_id'] != ''
         line['osm_line_id'] = tt['line_id']
         line['has_shape_in_osm'] = tt['line_id'] in osm_with_shapes
@@ -39,8 +39,10 @@
 
 
 for a_network in networks:
-    content += "\n# {} \n".format(a_network)
     lines_of_this_network = [line for line in lines if line['opendata_network'] == a_network]
+    if not a_network:
+        a_network = "*Pas de réseau défini côté IDFM*"
+    content += "\n# {} \n".format(a_network)    
     content += "{} lignes open data \n".format(len(lines_of_this_network))
 
     osm_missing_lines_of_this_network = [line for line in lines_of_this_network if not line['found_in_osm']]

run_all.sh

@@ -17,12 +17,12 @@ set -e
 # compute routepoints with GTFS and OSM
 ./audit_routepoints/audit_routepoints.sh
 
-# extract missing lines in OSM with GTFS and OSM
-./missing_in_osm/missing_in_osm.sh
-
 # extract lines ref
 ./lines_ref/init_lines_ref.sh
 
+# extract missing lines in OSM with GTFS and OSM
+./missing_in_osm/missing_in_osm.sh
+
 # extract focus routepoints from GTFS and OSM
 #./focus/focus.sh