osm2tsvs.py

   1 #! /usr/bin/python
   2 # -* encoding: utf8 -*-
   3
   4 import xml.sax
   5 import sys, gzip
   6
   7 # files
   8 nodes = gzip.open("nodes.tsv.gz", "w")
   9 node_tags = gzip.open("node_tags.tsv.gz", "w")
  10 ways = gzip.open("ways.tsv.gz", "w")
  11 way_tags = gzip.open("way_tags.tsv.gz", "w")
  12 way_nodes = gzip.open("way_nodes.tsv.gz", "w")
  13 relations = gzip.open("relations.tsv.gz", "w")
  14 relation_tags = gzip.open("relation_tags.tsv.gz", "w")
  15 relation_members = gzip.open("relation_members.tsv.gz", "w")
  16
  17 def tab(values):
  18     values = [unicode(x).replace("\t", "\\t") for x in values]
  19     return ("\t".join(values) + "\n").encode("utf8")
  20
  21 class OSMtoTSV(xml.sax.ContentHandler):
  22     id = None
  23     node_offset = None
  24     member_offset = None
  25
  26     def startElement(self, name, attrs):
  27         func_name = "start_%s" % name
  28         if hasattr(self, func_name) and  callable(getattr(self, func_name)):
  29             getattr(self, func_name)(attrs)
  30
  31
  32     def start_node(self, attrs):
  33         nodes.write(tab([attrs.get(x, "NULL") for x in ('id', 'timestamp', 'user', 'lat', 'lon')]))
  34         self.id = attrs['id']
  35         self.type = 'node'
  36
  37     def end_node(self):
  38         self.type = None
  39         self.id = None
  40
  41     def start_way(self, attrs):
  42         ways.write(tab([attrs.get(x, "NULL") for x in ('id', 'timestamp', 'user')]))
  43         self.id = attrs['id']
  44         self.type = "way"
  45         self.node_offset = 0
  46
  47     def end_way(self):
  48         self.type = None
  49         self.node_offset = None
  50         self.id = None
  51
  52     def start_tag(self, attrs):
  53         if self.type == 'node':
  54             node_tags.write(tab([self.id, attrs['k'], attrs['v']]))
  55         elif self.type == 'way':
  56             way_tags.write(tab([self.id, attrs['k'], attrs['v']]))
  57         elif self.type == 'relation':
  58             relation_tags.write(tab([self.id, attrs['k'], attrs['v']]))
  59         else:
  60             assert False, "Got a tag of %s" % self.type
  61
  62     def start_nd(self, attrs):
  63         assert self.type == 'way'
  64         assert self.node_offset is not None
  65         way_nodes.write(tab([self.id, self.node_offset, attrs['ref']]))
  66         self.node_offset += 1
  67
  68     def start_relation(self, attrs):
  69         relations.write(tab([attrs.get(x, "NULL") for x in ('id', 'timestamp', 'user')]))
  70         self.type = "relation"
  71         self.id = attrs['id']
  72         self.member_offset = 0
  73
  74     def end_relation(self):
  75         self.id = None
  76         self.type = None
  77         self.member_offset = None
  78
  79     def start_member(self, attrs):
  80         assert self.member_offset is not None
  81         relation_members.write(tab([self.id, attrs['type'], attrs['ref'], attrs.get('role', "NULL"), self.member_offset]))
  82         self.member_offset += 1
  83
  84
  85     def endElement(self, name):
  86         func_name = "end_%s" % name
  87         if hasattr(self, func_name) and  callable(getattr(self, func_name)):
  88             getattr(self, func_name)()
  89
  90
  91 xml.sax.parse(sys.stdin, OSMtoTSV())