def validate_article_xml(file_xml_path, print_error=True):
result = {}
logger.debug(file_xml_path)
try:
xmlvalidator = XMLValidator.parse(file_xml_path)
if config.get("VALIDATE_ALL") == "TRUE":
is_valid, errors = xmlvalidator.validate_all()
else:
is_valid, errors = xmlvalidator.validate()
except (exceptions.XMLSPSVersionError, etree.LxmlError) as e:
result[str(e)] = {
"count": 1,
"lineno": [1],
"message": [str(e)],
"filename": {file_xml_path},
}
return result
if not is_valid:
for error in errors:
if print_error:
logger.error("%s - %s - %s", error.level, error.line,
error.message)
message = error.message[:80]
data = {
"count": 1,
"lineno": [error.line],
"message": [error.message],
"filename": {file_xml_path},
}
dicts.merge(result, message, data)
return result
def validate_article_ALLxml(move_to_processed_source=False,
move_to_valid_xml=False):
logger.info("Iniciando Validação dos xmls")
list_files_xmls = files.xml_files_list(config.get("CONVERSION_PATH"))
success_path = config.get("VALID_XML_PATH")
errors_path = config.get("XML_ERRORS_PATH")
func = shutil.move if move_to_valid_xml else shutil.copyfile
result = {}
for file_xml in tqdm(list_files_xmls):
filename, _ = files.extract_filename_ext_by_path(file_xml)
converted_file = os.path.join(config.get("CONVERSION_PATH"), file_xml)
try:
errors = validate_article_xml(converted_file, False)
for k_error, v_error in errors.items():
dicts.merge(result, k_error, v_error)
if errors_path:
manage_error_file(
errors,
os.path.join(errors_path, "%s.err" % filename),
converted_file,
)
if not errors:
if success_path:
func(converted_file, os.path.join(success_path, file_xml))
if move_to_processed_source:
files.move_xml_to(
"%s.xml" % filename,
config.get("SOURCE_PATH"),
config.get("PROCESSED_SOURCE_PATH"),
)
except Exception as ex:
logger.exception(ex)
raise
analase = sorted(result.items(), key=lambda x: x[1]["count"], reverse=True)
for k_result, v_result in analase:
logger.error("%s - %s", k_result, v_result["count"])
def test_merge(self):
result = {"1": {"count": 2, "files": ("a", "b")}}
data = {"count": 1, "files": ("c", "d")}
dicts.merge(result, "1", data)
self.assertEqual(result["1"]["count"], 3)