def __init__(self, xslt_processor='lxml'):
'''
Initiates MODSMapper, with option of what XSLT processor to use.
- lxml: faster, but does not provide XSLT 2.0 support (though the included stylesheet does not require)
- pyjxslt: slower, but offers XSLT 2.0 support
Args:
xslt_processor (str)['lxml','pyjxslt']: Selects which XSLT processor to use.
'''
self.xslt_processor = xslt_processor
self.xslt_filepath = '/opt/combine/inc/xslt/MODS_extract.xsl'
if self.xslt_processor == 'lxml':
# set xslt transformer
xslt_tree = etree.parse(self.xslt_filepath)
self.xsl_transform = etree.XSLT(xslt_tree)
elif self.xslt_processor == 'pyjxslt':
# prepare pyjxslt gateway
self.gw = pyjxslt.Gateway(6767)
with open(self.xslt_filepath,'r') as f:
self.gw.add_transform('xslt_transform', f.read())
class XMLToJsonTestCase(unittest.TestCase):
# Just a quick test as the actual transform is tested elsewhere. Our job is just to make sure
# that we get what we expect through the gateway
gw = pyjxslt.Gateway()
if not gw.gateway_connected(reconnect=False):
print("Gateway must be running on port 25333")
def compare_jsons(self, json1, json2):
json1d = json.loads(json1)
try:
json2d = json.loads(json2)
except json.JSONDecodeError as e:
print(str(e))
return False
success, txt = dict_compare(json1d, json2d)
if not success:
print(txt)
return success
def test1(self):
self.assertTrue(
self.compare_jsons(expected_json, self.gw.to_json(xml1)))
self.assertEqual(expected_bad, self.gw.to_json(bad_xml))
self.assertTrue(
self.compare_jsons(
expected_pi, self.gw.to_json(xml_with_processing_instruction)))
def transform_xml_udf(job_id, row, xslt_string):
# attempt transformation and save out put to 'document'
try:
# transform with pyjxslt gateway
gw = pyjxslt.Gateway(6767)
gw.add_transform('xslt_transform', xslt_string)
result = gw.transform('xslt_transform', row.document)
gw.drop_transform('xslt_transform')
# set trans_result tuple
trans_result = (result, '', 1)
# catch transformation exception and save exception to 'error'
except Exception as e:
# set trans_result tuple
trans_result = ('', str(e), 0)
# return Row
return Row(record_id=row.record_id,
document=trans_result[0],
error=trans_result[1],
job_id=int(job_id),
oai_set=row.oai_set,
success=trans_result[2])
def _transform_xslt(self, row):
try:
# attempt to parse xslt prior to submitting to pyjxslt
try:
etree.fromstring(self.payload.encode('utf-8'))
except Exception as err:
return str(err)
# transform with pyjxslt gateway
gateway = pyjxslt.Gateway(6767)
gateway.add_transform('xslt_transform', self.payload)
result = gateway.transform('xslt_transform', row.document)
gateway.drop_transform('xslt_transform')
# return
return result
except Exception as err:
return str(err)
def test_gw_down(self):
gw = pyjxslt.Gateway(port=23456) # a non-existent port
self.assertIsNone(gw.to_json(xml1))
class TestGateway(unittest.TestCase):
gw = pyjxslt.Gateway()
def testSimple(self):
self.gw.add_transform('k1', xsl1)
self.assertEqual(
"""<?xml version="1.0" encoding="UTF-8"?>
ENTRY: 17:FOO
ENTRY: 42:BAR""", self.gw.transform('k1', xml1))
def testParms(self):
self.gw.add_transform('k2', xsl2)
self.assertEqual(
"""<?xml version="1.0" encoding="UTF-8"?>
Parm1: DEVEL
Parm2: 42
ENTRY: 17:FOO
ENTRY: 42:BAR""", self.gw.transform('k2', xml1, p2=42))
self.assertEqual(
"""<?xml version="1.0" encoding="UTF-8"?>
Parm1: PROD
Parm2: 42
ENTRY: 17:FOO
ENTRY: 42:BAR""", self.gw.transform('k2', xml1, p2=42, p1="PROD"))
def testFile(self):
self.gw.add_transform('k3', os.path.join('data', 'file3.xsl'))
self.assertEqual(
"""<?xml version="1.0" encoding="UTF-8"?>
ENTRY: 17:FOO
ENTRY: 42:BAR""", self.gw.transform('k3', xml1))
self.gw.add_transform('k3', os.path.join('data', 'file3a.xsl'))
self.assertEqual(
"""<?xml version="1.0" encoding="UTF-8"?>
entry: 17:FOO
entry: 42:BAR""", self.gw.transform('k3', xml1))
def testReplace(self):
self.gw.add_transform('k4', xsl1)
self.assertEqual(
"""<?xml version="1.0" encoding="UTF-8"?>
ENTRY: 17:FOO
ENTRY: 42:BAR""", self.gw.transform('k4', xml1))
self.gw.add_transform('k4', xsl2)
self.assertEqual(
"""<?xml version="1.0" encoding="UTF-8"?>
Parm1: DEVEL
Parm2: 17
ENTRY: 17:FOO
ENTRY: 42:BAR""", self.gw.transform('k4', xml1))
def testBadXSL(self):
self.assertRaises(ValueError, self.gw.add_transform, *('e1', xsl3))
self.assertIsNone(self.gw.add_transform('e2', xsl4))
self.gw.add_transform('e2', xsl2)
self.assertEqual(
"""<?xml version="1.0" encoding="UTF-8"?>
Parm1: DEVEL
Parm2: 17
ENTRY: 17:FOO
ENTRY: 42:BAR""", self.gw.transform('e2', xml1))
def testBadXML(self):
self.gw.add_transform('k1', xsl1)
self.assertTrue(self.gw.transform('k1', xml2).startswith('ERROR:'))