def test_read_write_storage(started_cluster):
hdfs_api = HDFSApi("root")
node1.query("create table SimpleHDFSStorage (id UInt32, name String, weight Float64) ENGINE = HDFS('hdfs://hdfs1:9000/simple_storage', 'TSV')")
node1.query("insert into SimpleHDFSStorage values (1, 'Mark', 72.53)")
assert hdfs_api.read_data("/simple_storage") == "1\tMark\t72.53\n"
assert node1.query("select * from SimpleHDFSStorage") == "1\tMark\t72.53\n"
def test_read_write_storage_with_globs(started_cluster):
hdfs_api = HDFSApi("root")
for i in ["1", "2", "3"]:
hdfs_api.write_data("/storage" + i, i + "\tMark\t72.53\n")
assert hdfs_api.read_data("/storage" + i) == i + "\tMark\t72.53\n"
node1.query(
"create table HDFSStorageWithRange (id UInt32, name String, weight Float64) ENGINE = HDFS('hdfs://hdfs1:9000/storage{1..5}', 'TSV')"
)
node1.query(
"create table HDFSStorageWithEnum (id UInt32, name String, weight Float64) ENGINE = HDFS('hdfs://hdfs1:9000/storage{1,2,3,4,5}', 'TSV')"
)
node1.query(
"create table HDFSStorageWithQuestionMark (id UInt32, name String, weight Float64) ENGINE = HDFS('hdfs://hdfs1:9000/storage?', 'TSV')"
)
node1.query(
"create table HDFSStorageWithAsterisk (id UInt32, name String, weight Float64) ENGINE = HDFS('hdfs://hdfs1:9000/storage*', 'TSV')"
)
assert node1.query("select count(*) from HDFSStorageWithRange") == '3\n'
assert node1.query("select count(*) from HDFSStorageWithEnum") == '3\n'
assert node1.query(
"select count(*) from HDFSStorageWithQuestionMark") == '3\n'
assert node1.query("select count(*) from HDFSStorageWithAsterisk") == '3\n'
def test_redirect(start_cluster):
hdfs_api = HDFSApi("root")
hdfs_api.write_data("/simple_storage", "1\t\n")
assert hdfs_api.read_data("/simple_storage") == "1\t\n"
node7.query(
"CREATE TABLE table_test_7_1 (word String) ENGINE=URL('http://hdfs1:50070/webhdfs/v1/simple_storage?op=OPEN&namenoderpcaddress=hdfs1:9000&offset=0', CSV)")
assert "not allowed" in node7.query_and_get_error("SET max_http_get_redirects=1; SELECT * from table_test_7_1")
def test_read_write_table(started_cluster):
hdfs_api = HDFSApi("root")
data = "1\tSerialize\t555.222\n2\tData\t777.333\n"
hdfs_api.write_data("/simple_table_function", data)
assert hdfs_api.read_data("/simple_table_function") == data
assert node1.query("select * from hdfs('hdfs://hdfs1:9000/simple_table_function', 'TSV', 'id UInt64, text String, number Float64')") == data
def test_read_write_table_with_parameter_none(started_cluster):
hdfs_api = HDFSApi("root")
data = "1\tHello Jessica\t555.222\n2\tI rolled a joint\t777.333\n"
hdfs_api.write_data("/simple_table_function.gz", data)
assert hdfs_api.read_data("/simple_table_function.gz") == data
assert node1.query("select * from hdfs('hdfs://hdfs1:9000/simple_table_function.gz', 'TSV', 'id UInt64, text String, number Float64', 'none')") == data
def test_read_write_table(started_cluster):
hdfs_api = HDFSApi("root")
data = "1\tSerialize\t555.222\n2\tData\t777.333\n"
hdfs_api.write_data("/simple_table_function", data)
assert hdfs_api.read_data("/simple_table_function") == data
assert node1.query("select * from hdfs('hdfs://hdfs1:9000/simple_table_function', 'TSV', 'id UInt64, text String, number Float64')") == data
def test_write_table(started_cluster):
hdfs_api = HDFSApi("root")
node1.query("create table OtherHDFSStorage (id UInt32, name String, weight Float64) ENGINE = HDFS('hdfs://hdfs1:9000/other_storage', 'TSV')")
node1.query("insert into OtherHDFSStorage values (10, 'tomas', 55.55), (11, 'jack', 32.54)")
result = "10\ttomas\t55.55\n11\tjack\t32.54\n"
assert hdfs_api.read_data("/other_storage") == result
assert node1.query("select * from OtherHDFSStorage order by id") == result
def test_write_table(started_cluster):
hdfs_api = HDFSApi("root")
node1.query("create table OtherHDFSStorage (id UInt32, name String, weight Float64) ENGINE = HDFS('hdfs://hdfs1:9000/other_storage', 'TSV')")
node1.query("insert into OtherHDFSStorage values (10, 'tomas', 55.55), (11, 'jack', 32.54)")
result = "10\ttomas\t55.55\n11\tjack\t32.54\n"
assert hdfs_api.read_data("/other_storage") == result
assert node1.query("select * from OtherHDFSStorage order by id") == result
def test_read_write_storage(started_cluster):
hdfs_api = HDFSApi("root")
hdfs_api.write_data("/simple_storage", "1\tMark\t72.53\n")
assert hdfs_api.read_data("/simple_storage") == "1\tMark\t72.53\n"
node1.query("create table SimpleHDFSStorage (id UInt32, name String, weight Float64) ENGINE = HDFS('hdfs://hdfs1:9000/simple_storage', 'TSV')")
assert node1.query("select * from SimpleHDFSStorage") == "1\tMark\t72.53\n"
def test_url_without_redirect(started_cluster):
hdfs_api = HDFSApi("root")
hdfs_api.write_data("/simple_storage", "1\tMark\t72.53\n")
assert hdfs_api.read_data("/simple_storage") == "1\tMark\t72.53\n"
# access datanode port directly
node1.query(
"create table WebHDFSStorage (id UInt32, name String, weight Float64) ENGINE = URL('http://hdfs1:50075/webhdfs/v1/simple_storage?op=OPEN&namenoderpcaddress=hdfs1:9000&offset=0', 'TSV')"
)
assert node1.query("select * from WebHDFSStorage") == "1\tMark\t72.53\n"
def test_url_with_redirect_not_allowed(started_cluster):
hdfs_api = HDFSApi("root")
hdfs_api.write_data("/simple_storage", "1\tMark\t72.53\n")
assert hdfs_api.read_data("/simple_storage") == "1\tMark\t72.53\n"
# access proxy port without allowing redirects
node1.query(
"create table WebHDFSStorageWithoutRedirect (id UInt32, name String, weight Float64) ENGINE = URL('http://hdfs1:50070/webhdfs/v1/simple_storage?op=OPEN&namenoderpcaddress=hdfs1:9000&offset=0', 'TSV')"
)
with pytest.raises(Exception):
assert node1.query("select * from WebHDFSStorageWithoutRedirect"
) == "1\tMark\t72.53\n"
def test_url_with_redirect_allowed(started_cluster):
hdfs_api = HDFSApi("root")
hdfs_api.write_data("/simple_storage", "1\tMark\t72.53\n")
assert hdfs_api.read_data("/simple_storage") == "1\tMark\t72.53\n"
# access proxy port with allowing redirects
# http://localhost:50070/webhdfs/v1/b?op=OPEN&namenoderpcaddress=hdfs1:9000&offset=0
node1.query(
"create table WebHDFSStorageWithRedirect (id UInt32, name String, weight Float64) ENGINE = URL('http://hdfs1:50070/webhdfs/v1/simple_storage?op=OPEN&namenoderpcaddress=hdfs1:9000&offset=0', 'TSV')"
)
assert node1.query(
"SET max_http_get_redirects=1; select * from WebHDFSStorageWithRedirect"
) == "1\tMark\t72.53\n"
def test_read_write_storage_with_globs(started_cluster):
hdfs_api = HDFSApi("root")
node1.query(
"create table HDFSStorageWithRange (id UInt32, name String, weight Float64) ENGINE = HDFS('hdfs://hdfs1:9000/storage{1..5}', 'TSV')"
)
node1.query(
"create table HDFSStorageWithEnum (id UInt32, name String, weight Float64) ENGINE = HDFS('hdfs://hdfs1:9000/storage{1,2,3,4,5}', 'TSV')"
)
node1.query(
"create table HDFSStorageWithQuestionMark (id UInt32, name String, weight Float64) ENGINE = HDFS('hdfs://hdfs1:9000/storage?', 'TSV')"
)
node1.query(
"create table HDFSStorageWithAsterisk (id UInt32, name String, weight Float64) ENGINE = HDFS('hdfs://hdfs1:9000/storage*', 'TSV')"
)
for i in ["1", "2", "3"]:
hdfs_api.write_data("/storage" + i, i + "\tMark\t72.53\n")
assert hdfs_api.read_data("/storage" + i) == i + "\tMark\t72.53\n"
assert node1.query("select count(*) from HDFSStorageWithRange") == "3\n"
assert node1.query("select count(*) from HDFSStorageWithEnum") == "3\n"
assert node1.query(
"select count(*) from HDFSStorageWithQuestionMark") == "3\n"
assert node1.query("select count(*) from HDFSStorageWithAsterisk") == "3\n"
try:
node1.query("insert into HDFSStorageWithEnum values (1, 'NEW', 4.2)")
assert False, "Exception have to be thrown"
except Exception as ex:
print ex
assert "in readonly mode" in str(ex)
try:
node1.query(
"insert into HDFSStorageWithQuestionMark values (1, 'NEW', 4.2)")
assert False, "Exception have to be thrown"
except Exception as ex:
print ex
assert "in readonly mode" in str(ex)
try:
node1.query(
"insert into HDFSStorageWithAsterisk values (1, 'NEW', 4.2)")
assert False, "Exception have to be thrown"
except Exception as ex:
print ex
assert "in readonly mode" in str(ex)