path: root/tools/merge.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-
#
# Copyright (c) 2014, NORDUnet A/S.
# See LICENSE for licensing information.

import argparse
import json
import base64
import urllib
import urllib2
import sys
import time
import ecdsa
import hashlib
import urlparse
import os
import yaml
import select
import struct
from certtools import build_merkle_tree, create_sth_signature, \
    check_sth_signature, get_eckey_from_file, timing_point, http_request, \
    get_public_key_from_file, get_leaf_hash, decode_certificate_chain, \
    create_ssl_context
from mergetools import parselogrow, get_logorder, read_chain, unpack_entry, \
    verify_entry

parser = argparse.ArgumentParser(description="")
parser.add_argument('--config', help="System configuration", required=True)
parser.add_argument('--localconfig', help="Local configuration", required=True)
parser.add_argument("--nomerge", action='store_true', help="Don't actually do merge")
parser.add_argument("--timing", action='store_true', help="Print timing information")
parser.add_argument("--noverify", action='store_true', help="Don't verify entries")
args = parser.parse_args()

config = yaml.load(open(args.config))
localconfig = yaml.load(open(args.localconfig))

ctbaseurl = config["baseurl"]
frontendnodes = config["frontendnodes"]
storagenodes = config["storagenodes"]
secondaries = localconfig.get("secondary", [])
paths = localconfig["paths"]
mergedb = paths["mergedb"]

signingnodes = config["signingnodes"]
create_ssl_context(cafile=paths["https_cacertfile"])

chainsdir = mergedb + "/chains"
logorderfile = mergedb + "/logorder"

own_key = (localconfig["nodename"], "%s/%s-private.pem" % (paths["privatekeys"], localconfig["nodename"]))

logpublickey = get_public_key_from_file(paths["logpublickey"])

hashed_dir = True

def write_chain(key, value):
    filename = base64.b16encode(key)
    if hashed_dir:
        path = chainsdir + "/" + filename[0:2] + "/" + filename[2:4] + "/" + filename[4:6]
        try:
            os.makedirs(path)
        except Exception, e:
            pass
    else:
        path = chainsdir
    f = open(path + "/" + filename, "w")
    f.write(value)
    f.close()

def add_to_logorder(key):
    f = open(logorderfile, "a")
    f.write(base64.b16encode(key) + "\n")
    f.close()

def fsync_logorder():
    f = open(logorderfile, "a")
    os.fsync(f.fileno())
    f.close()

def get_new_entries(node, baseurl):
    try:
        result = http_request(baseurl + "ct/storage/fetchnewentries", key=own_key, verifynode=node, publickeydir=paths["publickeys"])
        parsed_result = json.loads(result)
        if parsed_result.get(u"result") == u"ok":
            return [base64.b64decode(entry) for entry in parsed_result[u"entries"]]
        print >>sys.stderr, "ERROR: fetchnewentries", parsed_result
        sys.exit(1)
    except urllib2.HTTPError, e:
        print >>sys.stderr, "ERROR: fetchnewentries", e.read()
        sys.exit(1)

def get_entries(node, baseurl, hashes):
    try:
        params = urllib.urlencode({"hash":[base64.b64encode(hash) for hash in hashes]}, doseq=True)
        result = http_request(baseurl + "ct/storage/getentry?" + params, key=own_key, verifynode=node, publickeydir=paths["publickeys"])
        parsed_result = json.loads(result)
        if parsed_result.get(u"result") == u"ok":
            entries = dict([(base64.b64decode(entry["hash"]), base64.b64decode(entry["entry"])) for entry in parsed_result[u"entries"]])
            assert len(entries) == len(hashes)
            assert set(entries.keys()) == set(hashes)
            return entries
        print >>sys.stderr, "ERROR: getentry", parsed_result
        sys.exit(1)
    except urllib2.HTTPError, e:
        print >>sys.stderr, "ERROR: getentry", e.read()
        sys.exit(1)

def get_curpos(node, baseurl):
    try:
        result = http_request(baseurl + "ct/frontend/currentposition", key=own_key, verifynode=node, publickeydir=paths["publickeys"])
        parsed_result = json.loads(result)
        if parsed_result.get(u"result") == u"ok":
            return parsed_result[u"position"]
        print >>sys.stderr, "ERROR: currentposition", parsed_result
        sys.exit(1)
    except urllib2.HTTPError, e:
        print >>sys.stderr, "ERROR: currentposition", e.read()
        sys.exit(1)

def sendlog(node, baseurl, submission):
    try:
        result = http_request(baseurl + "ct/frontend/sendlog",
            json.dumps(submission), key=own_key, verifynode=node, publickeydir=paths["publickeys"])
        return json.loads(result)
    except urllib2.HTTPError, e:
        print >>sys.stderr, "ERROR: sendlog", e.read()
        sys.stderr.flush()
        return None
    except ValueError, e:
        print >>sys.stderr, "==== FAILED REQUEST ===="
        print >>sys.stderr, submission
        print >>sys.stderr, "======= RESPONSE ======="
        print >>sys.stderr, result
        print >>sys.stderr, "========================"
        sys.stderr.flush()
        raise e

def sendentry(node, baseurl, entry, hash):
    try:
        result = http_request(baseurl + "ct/frontend/sendentry",
            json.dumps({"entry":base64.b64encode(entry), "treeleafhash":base64.b64encode(hash)}), key=own_key,
            verifynode=node, publickeydir=paths["publickeys"])
        return json.loads(result)
    except urllib2.HTTPError, e:
        print >>sys.stderr, "ERROR: sendentry", e.read()
        sys.exit(1)
    except ValueError, e:
        print >>sys.stderr, "==== FAILED REQUEST ===="
        print >>sys.stderr, hash
        print >>sys.stderr, "======= RESPONSE ======="
        print >>sys.stderr, result
        print >>sys.stderr, "========================"
        sys.stderr.flush()
        raise e

def sendsth(node, baseurl, submission):
    try:
        result = http_request(baseurl + "ct/frontend/sendsth",
            json.dumps(submission), key=own_key, verifynode=node, publickeydir=paths["publickeys"])
        return json.loads(result)
    except urllib2.HTTPError, e:
        print >>sys.stderr, "ERROR: sendsth", e.read()
        sys.exit(1)
    except ValueError, e:
        print >>sys.stderr, "==== FAILED REQUEST ===="
        print >>sys.stderr, submission
        print >>sys.stderr, "======= RESPONSE ======="
        print >>sys.stderr, result
        print >>sys.stderr, "========================"
        sys.stderr.flush()
        raise e

def get_missingentries(node, baseurl):
    try:
        result = http_request(baseurl + "ct/frontend/missingentries", key=own_key, verifynode=node, publickeydir=paths["publickeys"])
        parsed_result = json.loads(result)
        if parsed_result.get(u"result") == u"ok":
            return parsed_result[u"entries"]
        print >>sys.stderr, "ERROR: missingentries", parsed_result
        sys.exit(1)
    except urllib2.HTTPError, e:
        print >>sys.stderr, "ERROR: missingentries", e.read()
        sys.exit(1)

def chunks(l, n):
    return [l[i:i+n] for i in range(0, len(l), n)]

timing = timing_point()

logorder = get_logorder(logorderfile)

timing_point(timing, "get logorder")

certsinlog = set(logorder)

new_entries_per_node = {}
new_entries = set()
entries_to_fetch = {}

for storagenode in storagenodes:
    print >>sys.stderr, "getting new entries from", storagenode["name"]
    sys.stderr.flush()
    new_entries_per_node[storagenode["name"]] = set(get_new_entries(storagenode["name"], "https://%s/" % storagenode["address"]))
    new_entries.update(new_entries_per_node[storagenode["name"]])
    entries_to_fetch[storagenode["name"]] = []

import subprocess

timing_point(timing, "get new entries")

new_entries -= certsinlog

print >>sys.stderr, "adding", len(new_entries), "entries"
sys.stderr.flush()

if args.nomerge:
    sys.exit(0)

for hash in new_entries:
    for storagenode in storagenodes:
        if hash in new_entries_per_node[storagenode["name"]]:
            entries_to_fetch[storagenode["name"]].append(hash)
            break

if not noverify:
    verifycert = subprocess.Popen([paths["verifycert_bin"], paths["known_roots"]],
                                stdin=subprocess.PIPE, stdout=subprocess.PIPE)

added_entries = 0
for storagenode in storagenodes:
    print >>sys.stderr, "getting %d entries from %s:" % (len(entries_to_fetch[storagenode["name"]]), storagenode["name"]),
    sys.stderr.flush()
    for chunk in chunks(entries_to_fetch[storagenode["name"]], 100):
        entries = get_entries(storagenode["name"], "https://%s/" % storagenode["address"], chunk)
        for hash in chunk:
            entry = entries[hash]
            if not noverify:
                verify_entry(verifycert, entry, hash)
            write_chain(hash, entry)
            add_to_logorder(hash)
            logorder.append(hash)
            certsinlog.add(hash)
            added_entries += 1
        print >>sys.stderr, added_entries,
        sys.stderr.flush()
    print >>sys.stderr
    sys.stderr.flush()
fsync_logorder()
timing_point(timing, "add entries")
print >>sys.stderr, "added", added_entries, "entries"
sys.stderr.flush()

if not noverify:
    verifycert.communicate(struct.pack("I", 0))

tree = build_merkle_tree(logorder)
tree_size = len(logorder)
root_hash = tree[-1][0]
timestamp = int(time.time() * 1000)

for secondary in secondaries:
    remotehost = secondary["host"]
    remotedir = remotehost + ":" + secondary["mergedir"]
    localdir = mergedb
    if localdir[:-1] != '/':
        localdir = localdir + "/"

    print >>sys.stderr, "copying database to secondary:", remotehost
    sys.stderr.flush()
    rsyncstatus = subprocess.call(["rsync", "-r", "--append", "--rsh=ssh", localdir, remotedir])
    if rsyncstatus:
        print >>sys.stderr, "rsync failed:", rsyncstatus
        sys.exit(1)

    print >>sys.stderr, "verifying database at secondary:", remotehost
    sys.stderr.flush()
    verifysecondary = subprocess.Popen(["ssh", remotehost, secondary["verifycommand"], secondary["mergedir"]],
                                    stdout=subprocess.PIPE)

    (verifysecondaryresult, _) = verifysecondary.communicate()

    if root_hash != base64.b16decode(verifysecondaryresult.strip()):
        print >>sys.stderr, "secondary root hash was", verifysecondaryresult.strip()
        print >>sys.stderr, "  expected", base64.b16encode(root_hash)
        sys.exit(1)

tree_head_signature = None
for signingnode in signingnodes:
    try:
        tree_head_signature = create_sth_signature(tree_size, timestamp,
                                                   root_hash, "https://%s/" % signingnode["address"], key=own_key)
        break
    except urllib2.URLError, e:
        print >>sys.stderr, e
        sys.stderr.flush()
if tree_head_signature == None:
    print >>sys.stderr, "Could not contact any signing nodes"
    sys.exit(1)

sth = {"tree_size": tree_size, "timestamp": timestamp,
       "sha256_root_hash": base64.b64encode(root_hash),
       "tree_head_signature": base64.b64encode(tree_head_signature)}

check_sth_signature(ctbaseurl, sth, publickey=logpublickey)

timing_point(timing, "build sth")

if args.timing:
    print >>sys.stderr, timing["deltatimes"]
    sys.stderr.flush()

print base64.b16encode(root_hash)
sys.stdout.flush()

for frontendnode in frontendnodes:
    nodeaddress = "https://%s/" % frontendnode["address"]
    nodename = frontendnode["name"]
    timing = timing_point()
    print >>sys.stderr, "distributing for node", nodename
    sys.stderr.flush()
    curpos = get_curpos(nodename, nodeaddress)
    timing_point(timing, "get curpos")
    print >>sys.stderr, "current position", curpos
    sys.stderr.flush()
    entries = [base64.b64encode(entry) for entry in logorder[curpos:]]
    print >>sys.stderr, "sending log:",
    sys.stderr.flush()
    for chunk in chunks(entries, 1000):
        for trynumber in range(5, 0, -1):
            sendlogresult = sendlog(nodename, nodeaddress, {"start": curpos, "hashes": chunk})
            if sendlogresult == None:
                if trynumber == 1:
                    sys.exit(1)
                select.select([], [], [], 10.0)
                print >>sys.stderr, "tries left:", trynumber
                sys.stderr.flush()
                continue
            break
        if sendlogresult["result"] != "ok":
            print >>sys.stderr, "sendlog:", sendlogresult
            sys.exit(1)
        curpos += len(chunk)
        print >>sys.stderr, curpos,
        sys.stderr.flush()
    print >>sys.stderr
    timing_point(timing, "sendlog")
    print >>sys.stderr, "log sent"
    sys.stderr.flush()
    missingentries = get_missingentries(nodename, nodeaddress)
    timing_point(timing, "get missing")
    print >>sys.stderr, "missing entries:", len(missingentries)
    sys.stderr.flush()
    fetched_entries = 0
    print >>sys.stderr, "fetching missing entries",
    sys.stderr.flush()
    for missingentry in missingentries:
        hash = base64.b64decode(missingentry)
        sendentryresult = sendentry(nodename, nodeaddress, read_chain(chainsdir, hash), hash)
        if sendentryresult["result"] != "ok":
            print >>sys.stderr, "send sth:", sendentryresult
            sys.exit(1)
        fetched_entries += 1
        if added_entries % 1000 == 0:
            print >>sys.stderr, fetched_entries,
            sys.stderr.flush()
    print >>sys.stderr
    sys.stderr.flush()
    timing_point(timing, "send missing")
    sendsthresult = sendsth(nodename, nodeaddress, sth)
    if sendsthresult["result"] != "ok":
        print >>sys.stderr, "send sth:", sendsthresult
        sys.exit(1)
    timing_point(timing, "send sth")
    if args.timing:
        print >>sys.stderr, timing["deltatimes"]
        sys.stderr.flush()