def _extract_java_data(self):
"""Extracts Java node data."""
root = self.oozie_node
props = self.props
if "mapred.child.java.opts" in props.merged:
self.java_opts.extend(
props.merged["mapred.child.java.opts"].split(" "))
if "mapreduce.map.java.opts" in props.merged:
self.java_opts.extend(
props.merged["mapreduce.map.java.opts"].split(" "))
self.main_class = xml_utils.get_tag_el_text(root=root,
tag=TAG_MAIN_CLASS,
props=props)
java_opts_string = xml_utils.get_tag_el_text(root=root,
tag=TAG_JAVA_OPTS,
props=props)
if java_opts_string:
self.java_opts.extend(java_opts_string.split(" "))
else:
self.java_opts.extend(
get_tags_el_array_from_text(root=root,
tag=TAG_JAVA_OPT,
props=props))
self.args = get_tags_el_array_from_text(root=root,
tag=TAG_ARG,
props=props)
def get_command(self) -> str:
cmd_txt = xml_utils.get_tag_el_text(self.oozie_node, TAG_CMD, self.props)
args = xml_utils.get_tags_el_array_from_text(self.oozie_node, TAG_ARG, self.props)
if not cmd_txt:
raise Exception("Missing or empty command node in SSH action {}".format(self.oozie_node))
cmd = " ".join([cmd_txt] + [shlex.quote(x) for x in args])
cmd = el_utils.convert_el_to_jinja(cmd)
return cmd
def get_command(self) -> str:
cmd_txt = xml_utils.get_tag_el_text(self.oozie_node, TAG_CMD)
args = xml_utils.get_tags_el_array_from_text(self.oozie_node, TAG_ARG)
if not cmd_txt:
raise Exception(f"Missing or empty command node in SSH action {self.oozie_node}")
cmd = " ".join([cmd_txt] + [shlex.quote(x) for x in args])
cmd = el_parser.translate(cmd, quote=True)
return cmd
def _parse_oozie_node(self):
self.resource_manager = get_tag_el_text(self.oozie_node, TAG_RESOURCE)
self.name_node = get_tag_el_text(self.oozie_node, TAG_NAME)
cmd_txt = get_tag_el_text(self.oozie_node, TAG_CMD)
args = get_tags_el_array_from_text(self.oozie_node, TAG_ARG)
cmd = " ".join([cmd_txt] + [x for x in args])
self.bash_command = el_parser.translate(cmd, quote=False)
self.pig_command = f"sh {self.bash_command}"
def _parse_oozie_node(self):
self.resource_manager = get_tag_el_text(self.oozie_node, TAG_RESOURCE,
self.props)
self.name_node = get_tag_el_text(self.oozie_node, TAG_NAME, self.props)
cmd_txt = get_tag_el_text(self.oozie_node, TAG_CMD, self.props)
args = get_tags_el_array_from_text(self.oozie_node, TAG_ARG,
self.props)
cmd = " ".join([cmd_txt] + [x for x in args])
self.bash_command = convert_el_to_jinja(cmd, quote=False)
self.pig_command = f"sh {self.bash_command}"
def on_parse_node(self):
super().on_parse_node()
_, self.hdfs_files = self.file_extractor.parse_node()
_, self.hdfs_archives = self.archive_extractor.parse_node()
self.java_jar = get_tag_el_text(self.oozie_node, tag=SPARK_TAG_JAR)
self.java_class = get_tag_el_text(self.oozie_node, tag=SPARK_TAG_CLASS)
if self.java_class and self.java_jar:
self.dataproc_jars = [self.java_jar]
self.java_jar = None
self.job_name = get_tag_el_text(self.oozie_node, tag=SPARK_TAG_JOB_NAME)
spark_opts = xml_utils.find_nodes_by_tag(self.oozie_node, SPARK_TAG_OPTS)
if spark_opts:
self.spark_opts.update(self._parse_spark_opts(spark_opts[0]))
self.application_args = xml_utils.get_tags_el_array_from_text(self.oozie_node, tag=SPARK_TAG_ARG)