1#!/usr/bin/python
2
3import cgi, os, sys, urllib2
4import common
5from multiprocessing import pool
6from autotest_lib.frontend import setup_django_environment
7
8from autotest_lib.client.common_lib import global_config
9from autotest_lib.client.bin import utils
10from autotest_lib.frontend.afe.json_rpc import serviceHandler
11from autotest_lib.server import system_utils
12from autotest_lib.server import utils as server_utils
13
14
15_PAGE = """\
16Status: 302 Found
17Content-Type: text/plain
18Location: %s\r\n\r
19"""
20
21GOOGLE_STORAGE_PATTERN = 'storage.cloud.google.com/'
22
23# Define function for retrieving logs
24def _retrieve_logs_dummy(job_path):
25    pass
26
27site_retrieve_logs = utils.import_site_function(__file__,
28    "autotest_lib.tko.site_retrieve_logs", "site_retrieve_logs",
29    _retrieve_logs_dummy)
30
31site_find_repository_host = utils.import_site_function(__file__,
32    "autotest_lib.tko.site_retrieve_logs", "site_find_repository_host",
33    _retrieve_logs_dummy)
34
35form = cgi.FieldStorage(keep_blank_values=True)
36# determine if this is a JSON-RPC request. we support both so that the new TKO
37# client can use its RPC client code, but the old TKO can still use simple GET
38# params.
39_is_json_request = form.has_key('callback')
40
41# if this key exists, we check if requested log exists in local machine,
42# and do not return Google Storage URL when the log doesn't exist.
43_local_only = form.has_key('localonly')
44
45
46def _get_requested_path():
47    if _is_json_request:
48        request_data = form['request'].value
49        request = serviceHandler.ServiceHandler.translateRequest(request_data)
50        parameters = request['params'][0]
51        return parameters['path']
52
53    return form['job'].value
54
55
56def _check_result(args):
57    host = args['host']
58    job_path = args['job_path']
59    shard = args['shard']
60    if shard:
61        http_path = 'http://%s/tko/retrieve_logs.cgi?localonly&job=%s' % (
62                host, job_path)
63    else:
64        http_path = 'http://%s%s' % (host, job_path)
65
66    try:
67        utils.urlopen(http_path)
68
69        # On Vms the shard name is set to the default gateway but the
70        # browser used to navigate frontends (that runs on the host of
71        # the vms) is immune to the same NAT routing the vms have, so we
72        # need to replace the gateway with 'localhost'.
73        if utils.DEFAULT_VM_GATEWAY in host:
74            normalized_host = host.replace(utils.DEFAULT_VM_GATEWAY, 'localhost')
75        else:
76            normalized_host = utils.normalize_hostname(host)
77        return 'http', normalized_host, job_path
78    except urllib2.URLError:
79        return None
80
81
82def _get_tpool_args(hosts, job_path, is_shard, host_set):
83    """Get a list of arguments to be passed to multiprocessing.pool.ThreadPool.
84
85    @param hosts: a list of host names.
86    @param job_path: a requested job path.
87    @param is_shard: True if hosts are shards, False otherwise.
88    @param host_set: a Set to filter out duplicated hosts.
89
90    @return: a list of dictionaries to be used as input of _check_result().
91    """
92    args = []
93    for host in hosts:
94        host = host.strip()
95        if host and host != 'localhost' and host not in host_set:
96            host_set.add(host)
97            arg = {'host': host, 'job_path': job_path, 'shard': is_shard}
98            args.append(arg)
99    return args
100
101
102def find_repository_host(job_path):
103    """Find the machine holding the given logs and return a URL to the logs"""
104    site_repo_info = site_find_repository_host(job_path)
105    if site_repo_info is not None:
106        return site_repo_info
107
108    # This cgi script is run only in master (cautotest) and shards.
109    # Drones do not run this script when receiving '/results/...' request.
110    # Only master should check drones and shards for the requested log.
111    # Also restricted users do not have access to drones or shards,
112    # always point them to localhost or google storage.
113    if (not server_utils.is_shard() and
114        not server_utils.is_restricted_user(os.environ.get('REMOTE_USER'))):
115        drones = system_utils.get_drones()
116        shards = system_utils.get_shards()
117
118        host_set = set()
119        tpool_args = _get_tpool_args(drones, job_path, False, host_set)
120        tpool_args += _get_tpool_args(shards, job_path, True, host_set)
121
122        tpool = pool.ThreadPool()
123        for result_path in tpool.imap_unordered(_check_result, tpool_args):
124            if result_path:
125                return result_path
126
127    # If the URL requested is a test result, it is now either on the local
128    # host or in Google Storage.
129    if job_path.startswith('/results/'):
130        # We only care about the path after '/results/'.
131        job_relative_path = job_path[9:]
132        if not _local_only and not os.path.exists(
133                    os.path.join('/usr/local/autotest/results',
134                                 job_relative_path)):
135            gsuri = utils.get_offload_gsuri().split('gs://')[1]
136            return ['https', GOOGLE_STORAGE_PATTERN, gsuri + job_relative_path]
137
138
139def get_full_url(info, log_path):
140    if info is not None:
141        protocol, host, path = info
142        prefix = '%s://%s' % (protocol, host)
143    else:
144        prefix = ''
145        path = log_path
146
147    if _is_json_request:
148        return '%s/tko/jsonp_fetcher.cgi?%s' % (prefix,
149                                                os.environ['QUERY_STRING'])
150    else:
151        return prefix + path
152
153
154log_path = _get_requested_path()
155info = find_repository_host(log_path)
156site_retrieve_logs(log_path)
157print _PAGE % get_full_url(info, log_path)
158