gistfile1.txt

#!/usr/bin/python

from pymongo import Connection
import subprocess
import sys

def advance(cur)
    try:
        return cur.next()
    except StopIteration:
        return None

def main():
    # Master ought to come first.
    (host1, host2, dbname1) = sys.argv[1:5]

    # These are hooey that, by murphy's law, I'd have to write if I
    # didn't write them.

    if (len(sys.argv)) > 4:
            dbname2 = sys.argv[5]
    else:
            dbname2 = dbname1
    if (len(sys.argv)) > 5:
            port1 = sys.argv[6]
    else:
            port1="27017"
    if (len(sys.argv)) > 6:
            port2 = sys.argv[7]
    else:
            port2="27017"

    # Make two connections, 
    conn1 = Connection(host1, int(port1))
    conn2 = Connection(host2, int(port2))

    # Get two dbs
    db1 = conn1[dbname1]
    db2 = conn2[dbname2]

    # TODO: check for matching collections

    for collname in db1.collection_names():
        msg1="Missing from %s: %s %%s" % (host1, collname)
        msg2="Missing from %s: %s %%s" % (host2, collname)
        # Get two collections.
        coll1=db1[collname]
        coll2=db2[collname]

        # Get two cursors. 
        cur1=iter(coll1.find(snapshot=True).sort("_id"))
        cur2=iter(coll2.find(snapshot=True).sort("_id"))

        while True:
            doc1 = advance(cur1)
            if not doc1: # no more in cur1, exhaust cur2
                for doc2 in cur2:
                    print msg1 % doc2["_id"]
                break
            doc2 = advance(cur2)
            if not doc2:
                for doc1 in cur1:
                    print msg2 % doc1["_id"]
                break

            id1=doc1["_id"]
            id2=doc2["_id"]

            if id1 == id2: # same obj
                if doc1 != doc2:
                    print "Skew: %s %s" % (collname, id1)
            elif (id1 < id2): 
                print msg2 % id1
                for doc1 in cur1:
                    id1 = doc1["_id"]
                    if id1 == id2:
                        break
                    else:
                        print msg2 % id1
            else: # id1 > id2
                print msg1 % id2
                for doc2 in cur2:
                    id2 = doc2["_id"]
                    if id1 == id2:
                        break
                    else:
                        print msg1 % id2

if __name__ == "__main__":
    main()