def get_features(ltable, rtable, attr_types_ltable, attr_types_rtable, attr_corres, tok, sim_funcs):
if check_table_order(ltable, rtable, attr_types_ltable, attr_types_rtable, attr_corres) is False:
raise AssertionError('Table order seems to be different than what is mentioned in other params')
dict_list = []
autogen_lkp_tbl = get_autogen_lkp_tbl()
for attrs in attr_corres['corres']:
attr1_type = attr_types_ltable[attrs[0]]
attr2_type = attr_types_rtable[attrs[1]]
if attr1_type != attr2_type:
logger.warning('%s type and %s type are different' %attrs)
if attr1_type is 'str_eq_1w':
rec_fns = autogen_lkp_tbl['STR_EQ_1W']
elif attr1_type is 'str_bt_1w_5w':
rec_fns = autogen_lkp_tbl['STR_BT_1W_5W']
elif attr1_type is 'str_bt_5w_10w':
rec_fns = autogen_lkp_tbl['STR_BT_5W_10W']
elif attr1_type is 'str_gt_10w':
rec_fns = autogen_lkp_tbl['STR_GT_10W']
elif attr1_type is 'numeric':
rec_fns = autogen_lkp_tbl['NUM']
elif attr1_type is 'boolean':
rec_fns = autogen_lkp_tbl['BOOL']
else:
raise TypeError('Unknown type')
fn_obj = get_fn_objs(attrs, rec_fns, tok, sim_funcs)
dict_list.append(fn_obj)
df = pd.DataFrame(dict_list)
df = df[['function_name', 'left_attribute', 'right_attribute', 'left_attr_tokenizer', 'right_attr_tokenizer',
'simfunction', 'function']]
return df
def label(tbl, col_name):
from magellan.gui.mtable_gui import edit
table = tbl.copy()
table.properties = tbl.properties
table._set_key(table.get_key())
if col_name in table.columns:
logger.warning('Input table already contains table with name %s' %col_name)
else:
table[col_name] = 0
edit(table)
return table
def init_jvm(jvmpath=None):
if jpype.isJVMStarted():
logger.warning("JVM is already running")
return
# taken from https://github.com/konlpy/konlpy/blob/master/konlpy/jvm.py
folder_suffix = [
"{0}{2}",
"{0}{1}secondstring-20120620.jar{2}",
"{0}{1}simfunction.jar{2}",
"{0}{1}simmetrics_jar_v1_6_2_d07_02_07.jar{2}",
"{0}{1}*",
]
java_dir = "%s%sinst%sjava" % (get_install_path(), os.sep, os.sep)
if os.name == "nt":
args = [java_dir, os.sep, ";"]
else:
args = [java_dir, os.sep, ":"]
classpath = os.sep.join(f.format(*args) for f in folder_suffix)
jvmpath = jvmpath or jpype.getDefaultJVMPath()
if jvmpath:
jpype.startJVM(jvmpath, "-Djava.class.path=%s" % classpath)
logger.info("JVM started")
else:
raise ValueError("Please specify the JVM path")
