def main():
global configuration_directory
global do_keep_gen_files
global tmp_folder
# obtaining command line arguments for path to config directory
parser = argparse.ArgumentParser()
parser.add_argument(
'-c', dest='configuration_directory_path',
default=default_configuration_directory,
required=False,
help='Specify the path to the configuration directory')
# read the optional argument `-k` for keeping the generated files
parser.add_argument(
'-k', '--keep',
default=default_do_keep_gen_files,
required=False,
help = 'Specify `yes` to preserve the files generated during execution')
args = vars(parser.parse_args())
configuration_directory = args['configuration_directory_path'] + '/'
do_keep_gen_files = False if args['keep'] is None else True
#setup_json = configuration_directory + 'setup.json'
global setup
setup = gsm_lib.read_config(configuration_directory, 'setup.json')
site_catalog_file = configuration_directory + setup['site_catalog']
system_log_file = setup['system_log_file']
# Configure logging
global gsmlogger
gsmlogger = GSMLogger()
gsmlogger.configure_logging(system_log_file)
# Initialize Redcap Interface
rt = redcap_transactions()
rt.configuration_directory = configuration_directory
properties = rt.init_redcap_interface(setup, gsmlogger.logger)
transform_xsl = configuration_directory + setup['xml_formatting_tranform_xsl']
#get data from the redcap for the fields listed in the source_data_schema.xml
response = rt.get_data_from_redcap(properties, gsmlogger.logger)
#XSL Transformation 1: This transformation removes junk data, rename elements and extracts site_id and adds new tag site_id
xml_tree = etree.fromstring(response)
xslt = etree.parse(transform_xsl)
transform = etree.XSLT(xslt)
xml_transformed = transform(xml_tree)
xml_str = etree.tostring(xml_transformed, method='xml', pretty_print=True)
#XSL Transformation 2: This transformation groups the data based on site_id
transform2_xsl = proj_root + 'bin/utils/groupby_siteid_transform.xsl'
xslt = etree.parse(transform2_xsl)
transform = etree.XSLT(xslt)
xml_transformed2 = transform(xml_transformed)
#XSL Transformation 3: This transformation removes all the nodes which are not set
transform3_xsl = proj_root + 'bin/utils/remove_junktags_transform.xsl'
xslt = etree.parse(transform3_xsl)
transform = etree.XSLT(xslt)
xml_transformed3 = transform(xml_transformed2)
#Prettifying the output generated by XSL Transformation
xml_str2 = etree.tostring(xml_transformed3, method='xml', pretty_print=True)
tree = etree.fromstring(xml_str2, etree.XMLParser(remove_blank_text=True))
# Loop through the start_date elements and update theur values
for k in tree.iter('start_date'):
d = datetime.datetime.strptime(k.text, "%Y-%m-%d").date()-timedelta(days=180)
k.text = str(d)
#writing data to smi+site_id.xml. This xml will be saved to sftp of the site as smi.xml
smi_filenames = []
smi_ids = []
tmp_folder = gsm_lib.get_temp_path(do_keep_gen_files)
for k in tree:
file_name = tmp_folder + 'smi' + k.attrib['id']+'.xml'
gsm_lib.write_element_tree_to_file(ET.ElementTree(k), file_name)
smi_filenames.append(file_name)
smi_ids.append(k.attrib['id'])
print 'Using smi_filenames: '
pprint.pprint(smi_filenames)
parse_site_details_and_send(site_catalog_file, smi_filenames, smi_ids, gsmlogger)
def main():
global configuration_directory
global do_keep_gen_files
# obtaining command line arguments for path to config directory
parser = argparse.ArgumentParser()
parser.add_argument(
"-c",
dest="configuration_directory_path",
default=default_configuration_directory,
required=False,
help="Specify the path to the configuration directory",
)
# read the optional argument `-k` for keeping the generated files
parser.add_argument(
"-k",
"--keep",
default=default_do_keep_gen_files,
required=False,
help="Specify `yes` to preserve the files generated during execution",
)
args = vars(parser.parse_args())
configuration_directory = args["configuration_directory_path"] + "/"
do_keep_gen_files = False if args["keep"] is None else True
# read setup options
global setup
setup = gsm_lib.read_config(configuration_directory, "setup.json")
site_catalog_file = configuration_directory + setup["site_catalog"]
system_log_file = setup["system_log_file"]
# Configure logging
global gsmlogger
gsmlogger = GSMLogger()
gsmlogger.configure_logging(system_log_file)
# Initialize Redcap Interface
rt = redcap_transactions()
rt.configuration_directory = configuration_directory
properties = rt.init_redcap_interface(setup, gsmlogger.logger)
# gets data from the person index for the fields listed in the source_data_schema.xml
response = rt.get_data_from_redcap(properties, gsmlogger.logger)
xml_tree = etree.fromstring(response)
# XSL Transformation : transforms the person_index data
transform_xsl = proj_root + "bin/utils/person_index_transform.xsl"
xslt = etree.parse(transform_xsl)
transform = etree.XSLT(xslt)
person_index_data = transform(xml_tree)
# # # retrieve smi.xml from the sftp server
smi_path = get_smi_and_parse(site_catalog_file)
if not os.path.exists(smi_path):
raise GSMLogger().LogException("Error: file " + smi_path + " not found")
else:
smi = open(smi_path, "r")
# Below code merges the 2 xmls
smi_data = etree.parse(smi_path)
# sorting both the xml files.
gsm_lib.sort_element_tree(smi_data)
gsm_lib.sort_element_tree(person_index_data)
# generating the person index dictionary
person_index_dict = {}
for item in person_index_data.iter("item"):
person_index_dict[item.findtext("research_subject_id")] = [
item.findtext("yob"),
item.findtext("mrn"),
item.findtext("facility_code"),
]
# iterate through the smi data and generate a
# new merged xml's for subject_map and subject_map_exceptions
subjectmap_root = etree.Element("subject_map_records")
subjectmap_exceptions_root = etree.Element("subject_map_exception_records")
exceptions = False
for item in smi_data.iter("item"):
if item.findtext("research_subject_id") in person_index_dict.keys():
gsmlogger.logger.debug("Processing research_subject_id %s", item.findtext("research_subject_id"))
if person_index_dict[item.findtext("research_subject_id")][0] == item.findtext("yob"):
gsmlogger.logger.debug("yob matched for research_subject_id %s", item.findtext("research_subject_id"))
mrn = etree.SubElement(item, "mrn")
mrn.text = person_index_dict[item.findtext("research_subject_id")][1]
facility_code = etree.SubElement(item, "facility_code")
facility_code.text = person_index_dict[item.findtext("research_subject_id")][2]
item.remove(item.find("yob"))
subjectmap_root.append(item)
else:
gsmlogger.logger.debug(
"yob not matched for research_subject_id %s", item.findtext("research_subject_id")
)
exception_item = etree.Element("item")
research_subject_id = etree.SubElement(exception_item, "research_subject_id")
research_subject_id.text = item.findtext("research_subject_id")
if research_subject_id.text is not None:
exceptions = True
pi_yob = etree.SubElement(exception_item, "Person_Index_YOB")
pi_yob.text = person_index_dict[item.findtext("research_subject_id")][0]
hcvt_yob = etree.SubElement(exception_item, "HCVTarget_YOB")
hcvt_yob.text = item.findtext("yob")
subjectmap_exceptions_root.append(exception_item)
# Below code transforms the xml files to csv files
transform_xsl = proj_root + "bin/utils/xml2csv.xsl"
xslt = etree.parse(transform_xsl)
transform = etree.XSLT(xslt)
tmp_folder = gsm_lib.get_temp_path(do_keep_gen_files)
subject_map_file = tmp_folder + "subject_map.csv"
gsmlogger.logger.info("Using path subject map file path: " + subject_map_file)
try:
subject_map_csv = open(subject_map_file, "w")
subject_map_csv.write("%s" % '"research_subject_id","start_date","end_date","mrn","facility_code"\n')
for item in subjectmap_root.iter("item"):
line = '"{0}","{1}","{2}","{3}","{4}"\n'.format(
gsm_lib.handle_blanks(item.findtext("research_subject_id")),
gsm_lib.handle_blanks(item.findtext("start_date")),
gsm_lib.handle_blanks(item.findtext("end_date")),
gsm_lib.handle_blanks(item.findtext("mrn")),
gsm_lib.handle_blanks(item.findtext("facility_code")),
)
subject_map_csv.write("%s" % line)
subject_map_csv.close()
except IOError:
raise GSMLogger().LogException("Could not open file %s for write", subject_map_file)
# remove the smi.xml from the folder because the XSLT process
# writes data to smi.xml
try:
os.remove(smi_path)
except OSError:
raise GSMLogger().LogException("Could not remove file %s ", smi_path)
# send the subject_map.csv to EMR team (sftp server)
parse_site_details_and_send(site_catalog_file, subject_map_file, "sftp")
if do_keep_gen_files:
print " * Keeping the temporary file: " + subject_map_file
else:
print " * Removing the temporary file: " + subject_map_file
os.remove(subject_map_file)
# send subject_map_exceptions.csv as email attachment
if exceptions:
subject_map_exceptions_file = tmp_folder + "subject_map_exceptions.csv"
try:
subject_map_exceptions_csv = open(subject_map_exceptions_file, "w")
except IOError:
raise GSMLogger().LogException("Could not open file %s for write", subject_map_exceptions_file)
subject_map_exceptions_csv.write("%s" % '"research_subject_id","person_index_yob","redcap_yob"\n')
for item in subjectmap_exceptions_root.iter("item"):
line = '"{0}","{1}","{2}"\n'.format(
gsm_lib.handle_blanks(item.find("research_subject_id").text),
gsm_lib.handle_blanks(item.find("Person_Index_YOB").text),
gsm_lib.handle_blanks(item.find("HCVTarget_YOB").text),
)
subject_map_exceptions_csv.write("%s" % line)
subject_map_exceptions_csv.close()
parse_site_details_and_send(site_catalog_file, subject_map_exceptions_file, "email")
if do_keep_gen_files:
print " * Keeping the temporary file: " + subject_map_exceptions_file
else:
print " * Removing the temporary file: " + subject_map_exceptions_file
os.remove(subject_map_exceptions_file)