def test_request(self):
# MAKE SOME DATA
data = {
"constant": "this is a test",
"random-data": convert.bytes2base64(Random.bytes(100))
}
client = Client(settings.url, unwrap(settings.hawk)) # unwrap() DUE TO BUG https://github.com/kumar303/mohawk/issues/21
link, id = client.send(data)
Log.note("Success! Located at {{link}} id={{id}}", link=link, id=id)
# FILL THE REST OF THE FILE
Log.note("Add ing {{num}} more...", num=99-id)
for i in range(id + 1, storage.BATCH_SIZE):
l, k = client.send(data)
if l != link:
Log.error("Expecting rest of data to have same link")
# TEST LINK HAS DATA
raw_content = requests.get(link).content
content = convert.zip2bytes(raw_content)
for line in convert.utf82unicode(content).split("\n"):
data = convert.json2value(line)
if data.etl.id == id:
Log.note("Data {{id}} found", id=id)
break
else:
Log.error("Expecting to find data at link")
def test_public_request(self):
# MAKE SOME DATA
data = {
"a": { # MATCHES SERVER PATTERN
"b": "good",
"c": [
{"k": "good", "m": 1},
{"k": 2, "m": 2}
]
},
"constant": "this is a test",
"random-data": convert.bytes2base64(Random.bytes(100))
}
content = json.dumps(data)
response = requests.post(
url=settings.url,
data=content,
headers={
'Content-Type': CONTENT_TYPE
}
)
self.assertEqual(response.status_code, 200, "Expecting 200")
about = json.loads(response.content)
return about['link'], about['etl']['id']
Log.note("Data located at {{link}} id={{id}}", link=link, id=id)
def test_public_request_too_big(self):
# MAKE SOME DATA
data = {
"a": { # MATCHES SERVER PATTERN
"b": "good",
"c": [
{"k": "good", "m": 1},
{"k": 2, "m": 2}
]
},
"constant": "this is a test",
"random-data": convert.bytes2base64(Random.bytes(500))
}
content = json.dumps(data)
def poster():
response = requests.post(
url=settings.url,
data=content,
headers={
'Content-Type': CONTENT_TYPE
}
)
self.assertEqual(response.status_code, 200, "Expecting 200")
self.assertRaises(Exception, poster)
def _deep_json_to_string(value, depth):
"""
:param value: SOME STRUCTURE
:param depth: THE MAX DEPTH OF PROPERTIES, DEEPER WILL BE STRING-IFIED
:return: FLATTER STRUCTURE
"""
if is_data(value):
if depth == 0:
return strings.limit(value2json(value), LOG_STRING_LENGTH)
return {
k: _deep_json_to_string(v, depth - 1)
for k, v in value.items()
}
elif is_sequence(value):
return strings.limit(value2json(value), LOG_STRING_LENGTH)
elif isinstance(value, number_types):
return value
elif is_text(value):
return strings.limit(value, LOG_STRING_LENGTH)
elif is_binary(value):
return strings.limit(bytes2base64(value), LOG_STRING_LENGTH)
elif isinstance(value, (date, datetime)):
return datetime2unix(value)
else:
return strings.limit(value2json(value), LOG_STRING_LENGTH)
def test_missing_auth(self):
# MAKE SOME DATA
data = {
"constant": "this is a test",
"random-data": convert.bytes2base64(Random.bytes(100))
}
response = requests.post(settings.bad_url, data=convert.unicode2utf8(convert.value2json(data)))
self.assertEqual(response.status_code, 403)
def save(self, query):
query.meta = None
json = convert.value2json(query)
hash = convert.unicode2utf8(json)
# TRY MANY HASHES AT ONCE
hashes = [None] * HASH_BLOCK_SIZE
for i in range(HASH_BLOCK_SIZE):
hash = hashlib.sha1(hash).digest()
hashes[i] = hash
short_hashes = [
convert.bytes2base64(h[0:6]).replace("/", "_") for h in hashes
]
available = {h: True for h in short_hashes}
existing = self.es.query({
"from": {
"type": "elasticsearch",
"settings": self.es.settings
},
"where": {
"terms": {
"hash": short_hashes
}
},
"meta": {
"timeout": "2second"
}
})
for e in Cube(select=existing.select,
edges=existing.edges,
data=existing.data).values():
if e.query == json:
return e.hash
available[e.hash] = False
# THIS WILL THROW AN ERROR IF THERE ARE NONE, HOW UNLUCKY!
best = [h for h in short_hashes if available[h]][0]
self.queue.add({
"id": best,
"value": {
"hash": best,
"create_time": Date.now(),
"last_used": Date.now(),
"query": json
}
})
Log.note("Saved query as {{hash}}", hash=best)
return best
def encrypt(text, _key, salt=None):
"""
RETURN JSON OF ENCRYPTED DATA {"salt":s, "length":l, "data":d}
"""
from pyLibrary.queries import jx
if not isinstance(text, unicode):
Log.error("only unicode is encrypted")
if _key is None:
Log.error("Expecting a key")
if isinstance(_key, str):
_key = bytearray(_key)
if salt is None:
salt = Random.bytes(16)
data = bytearray(text.encode("utf8"))
# Initialize encryption using key and iv
key_expander_256 = key_expander.KeyExpander(256)
expanded_key = key_expander_256.expand(_key)
aes_cipher_256 = aes_cipher.AESCipher(expanded_key)
aes_cbc_256 = cbc_mode.CBCMode(aes_cipher_256, 16)
aes_cbc_256.set_iv(salt)
output = Data()
output.type = "AES256"
output.salt = convert.bytes2base64(salt)
output.length = len(data)
encrypted = bytearray()
for _, d in jx.groupby(data, size=16):
encrypted.extend(aes_cbc_256.encrypt_block(d))
output.data = convert.bytes2base64(encrypted)
json = convert.value2json(output)
if DEBUG:
test = decrypt(json, _key)
if test != text:
Log.error("problem with encryption")
return json
def encrypt(text, _key, salt=None):
"""
RETURN JSON OF ENCRYPTED DATA {"salt":s, "length":l, "data":d}
"""
from pyLibrary.queries import jx
if not isinstance(text, unicode):
Log.error("only unicode is encrypted")
if _key is None:
Log.error("Expecting a key")
if isinstance(_key, str):
_key = bytearray(_key)
if salt is None:
salt = Random.bytes(16)
data = bytearray(text.encode("utf8"))
# Initialize encryption using key and iv
key_expander_256 = key_expander.KeyExpander(256)
expanded_key = key_expander_256.expand(_key)
aes_cipher_256 = aes_cipher.AESCipher(expanded_key)
aes_cbc_256 = cbc_mode.CBCMode(aes_cipher_256, 16)
aes_cbc_256.set_iv(salt)
output = Dict()
output.type = "AES256"
output.salt = convert.bytes2base64(salt)
output.length = len(data)
encrypted = bytearray()
for _, d in jx.groupby(data, size=16):
encrypted.extend(aes_cbc_256.encrypt_block(d))
output.data = convert.bytes2base64(encrypted)
json = convert.value2json(output)
if DEBUG:
test = decrypt(json, _key)
if test != text:
Log.error("problem with encryption")
return json
def test_save_then_load(self):
test = {
"data": [{
"a": "b"
}],
"query": {
"meta": {
"save": True
},
"from": TEST_TABLE,
"select": "a"
},
"expecting_list": {
"meta": {
"format": "list"
},
"data": ["b"]
}
}
settings = self.utils.fill_container(test)
bytes = unicode2utf8(
value2json({
"from": settings.index,
"select": "a",
"format": "list"
}))
expected_hash = convert.bytes2base64(
hashlib.sha1(bytes).digest()[0:6]).replace("/", "_")
wrap(test).expecting_list.meta.saved_as = expected_hash
self.utils.send_queries(test)
# ENSURE THE QUERY HAS BEEN INDEXED
Log.note("Flush saved query (with hash {{hash}})", hash=expected_hash)
container = elasticsearch.Index(index="saved_queries",
type=save_query.DATA_TYPE,
kwargs=settings)
container.flush(forced=True)
with Timer("wait for 5 seconds"):
Till(seconds=5).wait()
url = URL(self.utils.testing.query)
response = self.utils.try_till_response(url.scheme + "://" + url.host +
":" + text_type(url.port) +
"/find/" + expected_hash,
data=b'')
self.assertEqual(response.status_code, 200)
self.assertEqual(response.all_content, bytes)
def test_save_then_load(self):
test = {
"data": [
{"a": "b"}
],
"query": {
"meta": {"save": True},
"from": TEST_TABLE,
"select": "a"
},
"expecting_list": {
"meta": {
"format": "list"
},
"data": ["b"]
}
}
settings = self.utils.fill_container(test)
bytes = convert.unicode2utf8(convert.value2json({
"from": settings.index,
"select": "a",
"format": "list"
}))
expected_hash = convert.bytes2base64(hashlib.sha1(bytes).digest()[0:6]).replace("/", "_")
wrap(test).expecting_list.meta.saved_as = expected_hash
self.utils.send_queries(test)
#ENSURE THE QUERY HAS BEEN INDEXED
container = elasticsearch.Index(index="saved_queries", settings=settings)
container.flush()
Thread.sleep(seconds=5)
url = URL(self.utils.service_url)
response = self.utils.try_till_response(url.scheme+"://"+url.host+":"+unicode(url.port)+"/find/"+expected_hash, data=b'')
self.assertEqual(response.status_code, 200)
self.assertEqual(response.all_content, bytes)
def test_recovery_of_empty_string(self):
test = wrap({
"data": [
{"a": "bee"}
],
"query": {
"from": TEST_TABLE,
"select": "a",
"where": {"prefix": {"a": ""}},
"format": "list"
},
"expecting_list": {
"meta": {
"format": "list"
},
"data": ["bee"]
}
})
settings = self.utils.fill_container(test)
bytes = value2json(test.query).encode('utf8')
expected_hash = convert.bytes2base64(hashlib.sha1(bytes).digest()[0:6]).replace("/", "_")
test.expecting_list.meta.saved_as = expected_hash
test.query.meta = {"save": True}
self.utils.send_queries(test)
# ENSURE THE QUERY HAS BEEN INDEXED
Log.note("Flush saved query")
container = elasticsearch.Index(index="saved_queries", kwargs=settings)
container.flush(forced=True)
with Timer("wait for 5 seconds"):
Till(seconds=5).wait()
url = URL(self.utils.testing.query)
response = self.utils.try_till_response(url.scheme + "://" + url.host + ":" + text(url.port) + "/find/" + expected_hash, data=b'')
self.assertEqual(response.status_code, 200)
self.assertEqual(response.all_content, bytes)
def _deep_json_to_string(value, depth):
"""
:param value: SOME STRUCTURE
:param depth: THE MAX DEPTH OF PROPERTIES, DEEPER WILL BE STRING-IFIED
:return: FLATTER STRUCTURE
"""
if is_data(value):
if depth == 0:
return strings.limit(value2json(value), LOG_STRING_LENGTH)
return {k: _deep_json_to_string(v, depth - 1) for k, v in value.items()}
elif is_sequence(value):
return strings.limit(value2json(value), LOG_STRING_LENGTH)
elif isinstance(value, number_types):
return value
elif is_text(value):
return strings.limit(value, LOG_STRING_LENGTH)
elif is_binary(value):
return strings.limit(bytes2base64(value), LOG_STRING_LENGTH)
elif isinstance(value, (date, datetime)):
return datetime2unix(value)
else:
return strings.limit(value2json(value), LOG_STRING_LENGTH)
def _deep_json_to_string(value, depth):
"""
:param value: SOME STRUCTURE
:param depth: THE MAX DEPTH OF PROPERTIES, DEEPER WILL BE STRING-IFIED
:return: FLATTER STRUCTURE
"""
if isinstance(value, Mapping):
if depth == 0:
return strings.limit(value2json(value), LOG_STRING_LENGTH)
return {
k: _deep_json_to_string(v, depth - 1)
for k, v in value.items()
}
elif isinstance(value, (list, FlatList)):
return strings.limit(value2json(value), LOG_STRING_LENGTH)
elif isinstance(value, number_types):
return value
elif isinstance(value, text_type):
return strings.limit(value, LOG_STRING_LENGTH)
elif isinstance(value, binary_type):
return strings.limit(bytes2base64(value), LOG_STRING_LENGTH)
else:
return strings.limit(value2json(value), LOG_STRING_LENGTH)
import requests
from pyLibrary import convert, jsons
from pyLibrary.debugs.logs import Log
from pyLibrary.dot import unwrap
from pyLibrary.maths.randoms import Random
from modatasubmission import Client
settings = jsons.ref.get("file://~/MoDataSubmissionClient.json")
data={
"constant": "this is a test",
"random-data": convert.bytes2base64(Random.bytes(100))
}
link, id = Client(settings.url, unwrap(settings.hawk)).send(data)
Log.note("Success! Located at {{link}} id={{id}}", link=link, id=id)
data = convert.unicode2utf8(convert.value2json(settings.example))
response = requests.post(
settings.url,
data=data,
headers={
'Content-Type': b'application/json'
}
)
if response.status_code != 200:
def save(self, query):
"""
SAVE query TO ES FOR LATER RECOVERY
:param query:
:return: HAS TO USE FOR RECOVERY
"""
meta, query.meta = query.meta, None
json = convert.value2json(query)
query.meta = meta
hash = json.encode("utf8")
# TRY MANY HASHES AT ONCE
hashes = [None] * HASH_BLOCK_SIZE
for i in range(HASH_BLOCK_SIZE):
hash = hashlib.sha1(hash).digest()
hashes[i] = hash
short_hashes = [convert.bytes2base64(h[0:6]).replace("/", "_") for h in hashes]
available = {h: True for h in short_hashes}
existing = self.es.query(
{
"from": "saved_queries",
"where": {"terms": {"hash": short_hashes}},
"meta": {"timeout": "2second"},
}
)
for e in Cube(
select=existing.select, edges=existing.edges, data=existing.data
).values():
if e.query == json:
return e.hash
available[e.hash] = False
# THIS WILL THROW AN ERROR IF THERE ARE NONE, HOW UNLUCKY!
best = first(h for h in short_hashes if available[h])
self.queue.add(
{
"id": best,
"value": {
"hash": best,
"create_time": Date.now(),
"last_used": Date.now(),
"query": json,
},
}
)
if meta.testing:
while True:
verify = self.es.query(
{
"from": "saved_queries",
"where": {"terms": {"hash": best}},
"meta": {"timeout": "2second"},
"select": "query",
"format": "list",
}
)
if verify.data:
break
Log.alert("wait for saved query")
Till(seconds=1).wait()
Log.note("Saved {{json}} query as {{hash}}", json=json, hash=best)
return best
