[mwm.py] Multilang strings

2 files changed, 44 insertions, 14 deletions

diff --git a/tools/python/mwm/dump_mwm.py b/tools/python/mwm/dump_mwm.py
index 7493f1c423..251c5cd03c 100755
--- a/tools/python/mwm/dump_mwm.py
+++ b/tools/python/mwm/dump_mwm.py
@@ -1,33 +1,34 @@
 #!/usr/bin/python
 import sys, os.path, random
+import json
 from mwm import MWM
 
 if len(sys.argv) < 2:
-  print 'Dumps some MWM structures.'
-  print 'Usage: {0} <country.mwm>'.format(sys.argv[0])
-  sys.exit(1)
+    print 'Dumps some MWM structures.'
+    print 'Usage: {0} <country.mwm>'.format(sys.argv[0])
+    sys.exit(1)
 
 mwm = MWM(open(sys.argv[1], 'rb'))
 mwm.read_types(os.path.join(os.path.dirname(sys.argv[0]), '..', '..', '..', 'data', 'types.txt'))
 print 'Tags:'
 tvv = sorted([(k, v[0], v[1]) for k, v in mwm.tags.iteritems()], key=lambda x: x[1])
 for tv in tvv:
-  print '  {0:<8}: offs {1:9} len {2:8}'.format(tv[0], tv[1], tv[2])
+    print '  {0:<8}: offs {1:9} len {2:8}'.format(tv[0], tv[1], tv[2])
 print 'Version:', mwm.read_version()
 print 'Header:', mwm.read_header()
 print 'Metadata count:', len(mwm.read_metadata())
 
 cross = mwm.read_crossmwm()
 if cross:
-  print 'Outgoing points:', len(cross['out']), 'incoming:', len(cross['in'])
-  print 'Outgoing regions:', set(cross['neighbours'])
+    print 'Outgoing points:', len(cross['out']), 'incoming:', len(cross['in'])
+    print 'Outgoing regions:', set(cross['neighbours'])
 
 print 'Sample features:'
 count = 5
 probability = 1.0 / 1000
 for feature in mwm.iter_features():
-  if random.random() < probability:
-    print feature
-    count -= 1
-    if count <= 0:
-      break
+    if random.random() < probability:
+        print json.dumps(feature, ensure_ascii=False)
+        count -= 1
+        if count <= 0:
+            break
diff --git a/tools/python/mwm/mwm.py b/tools/python/mwm/mwm.py
index 739872abb3..568a2a6d01 100644
--- a/tools/python/mwm/mwm.py
+++ b/tools/python/mwm/mwm.py
@@ -11,7 +11,6 @@ from datetime import datetime
 # - Fix bounds reading in the header
 # - Fix delta point encoding (coords are plausible, but incorrect)
 # - Find why polygon geometry is incorrect in iter_features()
-# - Multilang string reading
 # - Find feature ids in the 'dat' section, or find a way to read the 'offs' section
 
 class MWM:
@@ -388,6 +387,36 @@ class MWM:
         return self.f.read(sz)
 
     def read_multilang(self):
+        def find_multilang_next(s, i):
+            i += 1
+            while i < len(s):
+                c = struct.unpack('B', s[i])[0]
+                if c & 0xC0 == 0x80:
+                    break
+                if c & 0x80 == 0:
+                    pass
+                elif c & 0xFE == 0xFE:
+                    i += 6
+                elif c & 0xFC == 0xFC:
+                    i += 5
+                elif c & 0xF8 == 0xF8:
+                    i += 4
+                elif c & 0xF0 == 0xF0:
+                    i += 3
+                elif c & 0xE0 == 0xE0:
+                    i += 2
+                elif c & 0xC0 == 0xC0:
+                    i += 1
+                i += 1
+            return i
+
         s = self.read_string()
-        # TODO!
-        return s
+        langs = {}
+        i = 0
+        while i < len(s):
+            n = find_multilang_next(s, i)
+            lng = struct.unpack('B', s[i])[0] & 0x3F
+            if lng < len(self.languages):
+                langs[self.languages[lng]] = s[i+1:n]
+            i = n
+        return langs