get-missing-streets.py

#! /usr/bin/python2
# vim: fileencoding=utf-8 encoding=utf-8 et sw=4
#
# Reads an existing-highways.osm file and an addresses.osm file and
# writes an .osm file containing one node per potential missing street
# name -- such that no named highway containing every word from the
# addr:street is found within about ~1.5km
# Output name is hardcoded as missing-streets.osm
#
# The inputs can be generated with something like:
# wget -O existing-highways.osm 'http://overpass-api.de/api/xapi?way[highway=*][name=*][bbox=20.781,52.092,21.281,52.365]'
# wget -O addresses.osm 'http://overpass-api.de/api/xapi?*[addr:street=*][bbox=20.781,52.092,21.281,52.365]'
#
# Not sure about a need to escape the ':' in the second query somehow...
#

import sys
import os
import math

import xml.etree.cElementTree as ElementTree

i = 0
outroot = ElementTree.Element("osm", {"version": "0.6"})
inroot = ElementTree.parse(sys.argv[1]).getroot()

waynodes = {}
existingaddrs = {}
for elem in inroot:
    if "id" not in elem.attrib:
        continue
    id = int(elem.attrib["id"])
    tags = {}
    for sub in elem:
        if sub.tag != "tag":
            continue
        tags[sub.attrib["k"]] = sub.attrib["v"]
    if elem.tag == "node":
        lat = float(elem.attrib["lat"])
        lon = float(elem.attrib["lon"])
        waynodes[id] = (lat, lon)
    if "addr:housenumber" not in tags or "addr:street" not in tags:
        continue
    if elem.tag == "node" and "amenity" in tags:
        continue
    strt = tags["addr:street"].lower()
    if elem.tag not in ["way", "node"]:
        continue
    if elem.tag == "way":
        j = 0
        lat = 0.0
        lon = 0.0
        for sub in elem:
            if sub.tag != "nd":
                continue
            ref = int(sub.attrib["ref"])
            if ref not in waynodes:
                continue
            j += 1
            lat += waynodes[ref][0]
            lon += waynodes[ref][1]
        lat /= j
        lon /= j
    existingaddrs[strt] = (lat, lon)

byword = {}
for elem in inroot:
    if "id" not in elem.attrib:
        continue
    id = int(elem.attrib["id"])
    tags = {}
    nodes = []
    for sub in elem:
        if sub.tag == "nd":
            nodes.append(waynodes[int(sub.attrib["ref"])])
        if sub.tag != "tag":
            continue
        tags[sub.attrib["k"]] = sub.attrib["v"]
    if elem.tag == "node":
        lat = float(elem.attrib["lat"])
        lon = float(elem.attrib["lon"])
        waynodes[id] = (lat, lon)
        continue
    elif elem.tag not in ["way"]:
        continue
    if "highway" not in tags or "name" not in tags:
        continue
    name = tags["name"].lower().split()
    if "alt_name" in tags:
        name += tags["alt_name"].lower().replace(";", " ").split()
    for word in name:
        if word not in byword:
            byword[word] = []
        byword[word].append((name, nodes, id))
inroot = None

i = 0
for strt in existingaddrs:
    lat, lon = existingaddrs[strt]
    all = 0
    hwys = {}
    for word in strt.split():
        all += 1
        if word in byword:
            for hwy in byword[word]:
                hwys[hwy[2]] = hwy
    cont = 0
    for id in hwys:
        name, nodes, _id = hwys[id]
        same = 0
        for word in strt.split():
            if word in [u"Księdza", u"Kapitana", u"Generała"] or word in name:
                same += 1
        if same == all:
            for nlat, nlon in nodes:
                if math.hypot(lat - nlat, lon - nlon) < 0.01:  # ~1km?
                    cont = 1
                    break
        if cont:
            break
    if cont:
        continue
    i += 1
    node = ElementTree.SubElement(
        outroot,
        "node",
        {"lat": str(lat), "lon": str(lon), "version": str(1), "id": str(i)},
    )
    ElementTree.SubElement(node, "tag", {"k": "addr:street", "v": strt})

sys.stdout.write("Writing .osm's\n")
ElementTree.ElementTree(outroot).write("missing-streets.osm", "utf-8")