def entries(self):
builder = None
for prefix, event, value in self._parser:
if (prefix, event) == ('entries.item', 'start_map'):
builder = ijson.ObjectBuilder()
if builder is not None:
builder.event(event, value)
if (prefix, event) == ('entries.item', 'end_map'):
yield self._parse_entry(builder.value)
builder = None
def streamJsonArrayItems(f):
# read items from a json array -- without loading the entire file into memory
level = 0
currentObject = ijson.ObjectBuilder()
parsed = ijson.parse(f)
# eat the initial start_array event
assertEqual('start_array', next(parsed)[1])
# construct objects. use level in order to support objects within objects
for _, event, value in parsed:
currentObject.event(event, value)
if event == 'start_map':
level += 1
elif event == 'end_map':
level -= 1
if level == 0:
yield currentObject.value
currentObject = ijson.ObjectBuilder()
def jsonObjectReader(filepath):
"""
Creates a generator that parses an array of json objects from a valid
json array file, yielding each top level json object in the array.
:param filepath: path to json file.
"""
top_level_array = False
array_stack = 0
top_level_object = False
object_stack = 0
parser = ijson.parse(open(filepath, 'r'))
for prefix, event, value in parser:
if event == 'start_array':
if not top_level_array:
top_level_array = True
continue
else:
array_stack += 1
if event == 'start_map':
if not top_level_object:
top_level_object = True
builder = ijson.ObjectBuilder()
else:
object_stack += 1
if event == 'end_map':
if not top_level_object:
raise Exception('end_map without a top level object')
else:
if object_stack == 0:
top_level_object = False
yield builder.value
else:
object_stack -= 1
if event == 'end_array':
if not top_level_array:
raise Exception('end_array without a top level array')
else:
if array_stack == 0:
top_level_array = False
else:
array_stack -= 1
# convert Decimal to float because mongo can't serialize Decimal
# TODO is this the right place to do this? Should it be done instead
# upon save?
if isinstance(value, decimal.Decimal):
# TODO this has different behavior on python 2.6 vs 2.7 due to
# different rounding behavior
value = float(value)
builder.event(event, value)
def _read_file_generator(self, f, skip):
try:
events = iter(ijson.parse(f))
while True:
current, event, value = next(events)
if current and '.' not in current:
if event in ('start_map', 'start_array'):
builder = ijson.ObjectBuilder()
end_event = event.replace('start', 'end')
while '.' in current or event != end_event:
builder.event(event, value)
current, event, value = next(events)
yield builder.value
except StopIteration:
pass
def _parse_defs(self):
builder = None
for prefix, event, value in self._parser:
# If we reach entries array, there are no more tracepoint definitions
if prefix == 'entries':
break
elif prefix == 'tsc_rate':
self._tsc_rate = value
continue
if (prefix, event) == ('tpoints', 'start_array'):
builder = ijson.ObjectBuilder()
if builder is not None:
builder.event(event, value)
if (prefix, event) == ('tpoints', 'end_array'):
self._parse_tpoints(builder.value)
builder = None
def ijson_top_level_items(file, local_streaming_backend):
ijson = get_ijson(local_streaming_backend)
parser = ijson.parse(file)
prefixed_events = iter(parser)
wanted = None
try:
while True:
current, event, value = next(prefixed_events)
if current != '':
wanted = current
if event in ('start_map', 'start_array'):
builder = ijson_mod.ObjectBuilder()
end_event = event.replace('start', 'end')
while (current, event) != (wanted, end_event):
builder.event(event, value)
current, event, value = next(prefixed_events)
yield current, builder.value
except StopIteration:
pass