def parse_mapreduce_yaml(contents):
"""Parses mapreduce.yaml file contents.
Args:
contents: mapreduce.yaml file contents.
Returns:
MapReduceYaml object with all the data from original file.
Raises:
errors.BadYamlError: when contents is not a valid mapreduce.yaml file.
"""
try:
builder = yaml_object.ObjectBuilder(MapReduceYaml)
handler = yaml_builder.BuilderHandler(builder)
listener = yaml_listener.EventListener(handler)
listener.Parse(contents)
mr_info = handler.GetResults()
except (ValueError, yaml_errors.EventError), e:
raise errors.BadYamlError(e)
Raises:
errors.BadYamlError: when contents is not a valid mapreduce.yaml file.
"""
try:
builder = yaml_object.ObjectBuilder(MapReduceYaml)
handler = yaml_builder.BuilderHandler(builder)
listener = yaml_listener.EventListener(handler)
listener.Parse(contents)
mr_info = handler.GetResults()
except (ValueError, yaml_errors.EventError), e:
raise errors.BadYamlError(e)
if len(mr_info) < 1:
raise errors.BadYamlError("No configs found in mapreduce.yaml")
if len(mr_info) > 1:
raise errors.MultipleDocumentsInMrYaml("Found %d YAML documents" %
len(mr_info))
jobs = mr_info[0]
job_names = set(j.name for j in jobs.mapreduce)
if len(jobs.mapreduce) != len(job_names):
raise errors.BadYamlError(
"Overlapping mapreduce names; names must be unique")
return jobs
def get_mapreduce_yaml(parse=parse_mapreduce_yaml):
"""Locates mapreduce.yaml, loads and parses its info.