Add places feed to the repository

scripts/README.md

@@ -0,0 +1,16 @@
+# Scripts
+
+Here are some (one at the moment) scripts to prepare data for the conflator
+or do stuff after conflating.
+
+## pack_places.py
+
+Prepares `places.bin` file for the geocoder. Requires three JSON files:
+
+* places.json
+* regions.json
+* countries.json
+
+These comprise the "places feed" and can be prepared using
+[these scripts](https://github.com/mapsme/geocoding_data). You can
+find a link to a ready-made feed in that repository.

scripts/pack_places.py

@@ -0,0 +1,41 @@
+#!/usr/bin/env python3
+import json
+import struct
+import os
+import sys
+
+
+def pack_coord(coord):
+    data = struct.pack('<l', round(coord * 10000))
+    return data[:-1]
+
+if len(sys.argv) < 2:
+    path = '.'
+else:
+    path = sys.argv[1]
+
+with open(os.path.join(path, 'regions.json'), 'r') as f:
+    regions = [(r, int(rid)) for rid, r in json.load(f).items() if r.get('iso')]
+    reg_idx = {regions[i][1]: i for i in range(len(regions))}
+with open(os.path.join(path, 'countries.json'), 'r') as f:
+    countries = [(r, int(rid)) for rid, r in json.load(f).items() if r.get('iso')]
+    c_idx = {countries[i][1]: i for i in range(len(countries))}
+with open(os.path.join(path, 'places.json'), 'r') as f:
+    places = json.load(f)
+
+out = open('places.bin', 'wb')
+out.write(struct.pack('B', len(countries)))
+for c, _ in countries:
+    out.write(struct.pack('2s', c['iso'].encode('ascii')))
+out.write(struct.pack('<h', len(regions)))
+for r, _ in regions:
+    rname = r['iso'].encode('ascii')
+    out.write(struct.pack('B', len(rname)))
+    out.write(rname)
+for pl in places.values():
+    if pl['country'] not in c_idx:
+        continue
+    out.write(pack_coord(pl['lon']))
+    out.write(pack_coord(pl['lat']))
+    out.write(struct.pack('B', c_idx[pl['country']]))
+    out.write(struct.pack('<h', reg_idx.get(pl.get('region'), -1)))