gistfile1.py

from fileformat_pb2 import Blob
from fileformat_pb2 import BlobHeader

from osmformat_pb2 import *

from zlib import decompress

def map_reader(f, size, url, params):
    data = f.read(4)
    while data != "":
        blobHeaderSize = int(data.encode("hex"), 32)

        data = f.read(blobHeaderSize)

        blobHeader = BlobHeader()
        blobHeader.ParseFromString(data)

        if blobHeader.type == "OSMData":
            #yield f.tell()+1, blobHeader.datasize
            yield f.read(blobHeader.datasize)
        else:
            f.seek(blobHeader.datasize, 1)

        data = f.read(4)

from disco.ddfs import DDFS

#with open("./berlin.osm.pbf") as f:
#    for k, v in map_reader(f, 0, 0, 0):
#        print k
#
#        #p = PrimitiveBlock()
#        #p.ParseFromString(r)
#        #print p.granularity

ddfs = DDFS()
print ddfs
ddfs.chunk("data:osm", ["./berlin.osm.pbf"], reader=map_reader)