[storage] Add optional country_name_synonyms parameter to countries.txt to find corresponding CountryId by country name without countries.txt keys modification.

5 files changed, 28 insertions, 2 deletions

diff --git a/tools/python/maps_generator/generator/env.py b/tools/python/maps_generator/generator/env.py
index 7459293c3d..a087a783eb 100644
--- a/tools/python/maps_generator/generator/env.py
+++ b/tools/python/maps_generator/generator/env.py
@@ -209,6 +209,10 @@ class Env:
         return os.path.join(self.user_resource_path, "borders_vs_osm.csv")
 
     @property
+    def countries_synonyms_path(self):
+        return os.path.join(self.user_resource_path, "countries_synonyms.csv")
+
+    @property
     def counties_txt_path(self):
         return os.path.join(self.mwm_path, "countries.txt")
 
diff --git a/tools/python/maps_generator/maps_generator.py b/tools/python/maps_generator/maps_generator.py
index 5e4ef8c215..4018a56863 100644
--- a/tools/python/maps_generator/maps_generator.py
+++ b/tools/python/maps_generator/maps_generator.py
@@ -223,6 +223,7 @@ def stage_descriptions(env):
 def stage_countries_txt(env):
     countries = hierarchy_to_countries(env.old_to_new_path,
                                        env.borders_to_osm_path,
+                                       env.country_synonyms_path,
                                        env.hierarchy_path, env.mwm_path,
                                        env.mwm_version)
     with open(env.counties_txt_path, "w") as f:
diff --git a/tools/python/post_generation/__main__.py b/tools/python/post_generation/__main__.py
index c004c7deb9..deb526186c 100644
--- a/tools/python/post_generation/__main__.py
+++ b/tools/python/post_generation/__main__.py
@@ -66,12 +66,15 @@ The post_generation commands are:
                             help="old_vs_new.csv file")
         parser.add_argument("--osm", required=True,
                             help="borders_vs_osm.csv file")
+        parser.add_argument("--countries_synonyms", required=True,
+                            help="countries_synonyms.csv file")
         parser.add_argument("--mwm_version", type=int, required=True,
                             help="Mwm version")
         parser.add_argument("-o", "--output", required=True,
                             help="Output countries.txt file (default is stdout)")
         args = parser.parse_args(sys.argv[2:])
         countries_json = hierarchy_to_countries_(args.old, args.osm,
+                                                 args.countries_synonyms,
                                                  args.hierarchy,
                                                  args.target,
                                                  args.mwm_version)
diff --git a/tools/python/post_generation/hierarchy_to_countries.py b/tools/python/post_generation/hierarchy_to_countries.py
index e63f33cca2..b2c518e758 100755
--- a/tools/python/post_generation/hierarchy_to_countries.py
+++ b/tools/python/post_generation/hierarchy_to_countries.py
@@ -109,9 +109,24 @@ def parse_borders_vs_osm(borders_vs_osm_csv_path):
                 vsosm[m.group(1)] = [m.group(3)]
     return vsosm
 
+def parse_countries_synonyms(countries_synonyms_csv_path):
+    countries_synonyms = {}
+    if not countries_synonyms_csv_path:
+        return countries_synonyms
+
+    with open(countries_synonyms_csv_path) as f:
+        for line in f:
+            m = re.match(r"(.+)\t(.+)", line.strip())
+            assert m
+            if m.group(1) in countries_synonyms:
+                countries_synonyms[m.group(1)].append(m.group(2))
+            else:
+                countries_synonyms[m.group(1)] = [m.group(2)]
+    return countries_synonyms
 
 def hierarchy_to_countries(old_vs_new_csv_path, borders_vs_osm_csv_path,
-                           hierarchy_path, target_path, version):
+                           countries_synonyms_csv_path, hierarchy_path,
+                           target_path, version):
 
     def fill_last(last, stack):
         name = last["id"]
@@ -124,6 +139,7 @@ def hierarchy_to_countries(old_vs_new_csv_path, borders_vs_osm_csv_path,
 
     oldvs = parse_old_vs_new(old_vs_new_csv_path)
     vsosm = parse_borders_vs_osm(borders_vs_osm_csv_path)
+    countries_synonyms = parse_countries_synonyms(countries_synonyms_csv_path)
     stack = [CountryDict(v=version, nameattr="Countries", g=[])]
     last = None
     with open(hierarchy_path) as f:
@@ -151,6 +167,8 @@ def hierarchy_to_countries(old_vs_new_csv_path, borders_vs_osm_csv_path,
                 last["old"] = oldvs[items[0]]
             if items[0] in vsosm:
                 last["affiliations"] = vsosm[items[0]]
+            if items[0] in countries_synonyms:
+                last["country_name_synonyms"] = countries_synonyms[items[0]]
 
     # the last line is always a file
     del last["d"]
diff --git a/tools/unix/generate_planet.sh b/tools/unix/generate_planet.sh
index 1f5e867bf7..0f642deac9 100755
--- a/tools/unix/generate_planet.sh
+++ b/tools/unix/generate_planet.sh
@@ -636,7 +636,7 @@ if [ "$MODE" == "resources" ]; then
   putmode "Step 8: Updating resource lists"
   # Update countries list
   $PYTHON36 -m $POST_GENERATION_MODULE hierarchy_to_countries --target "$TARGET" --hierarchy "$DATA_PATH/hierarchy.txt" --mwm_version "$COUNTRIES_VERSION" \
-    --old "$DATA_PATH/old_vs_new.csv" --osm "$DATA_PATH/borders_vs_osm.csv" --output "$TARGET/countries.txt" >> "$PLANET_LOG" 2>&1
+    --old "$DATA_PATH/old_vs_new.csv" --osm "$DATA_PATH/borders_vs_osm.csv" --countries_synonyms "$DATA_PATH/countries_synonyms.csv" --output "$TARGET/countries.txt" >> "$PLANET_LOG" 2>&1
 
   # A quick fix: chmodding to a+rw all generated files
   for file in "$TARGET"/*.mwm*; do