def to_python(cls, data, slim_frames=True, raw=False):
is_valid, errors = validate_and_default_interface(data, cls.path)
if not is_valid:
raise InterfaceValidationError("Invalid stack frame data.")
frame_list = [
# XXX(dcramer): handle PHP sending an empty array for a frame
Frame.to_python(f or {}, raw=raw) for f in data['frames']
]
kwargs = {
'frames': frame_list,
}
kwargs['registers'] = None
if data.get('registers') and isinstance(data['registers'], dict):
kwargs['registers'] = data.get('registers')
if data.get('frames_omitted'):
kwargs['frames_omitted'] = data['frames_omitted']
else:
kwargs['frames_omitted'] = None
instance = cls(**kwargs)
if slim_frames:
slim_frame_data(instance)
return instance
def to_python(cls, data, rust_renormalized=RUST_RENORMALIZED_DEFAULT):
# TODO(markus): semaphore does not validate security interfaces yet
is_valid, errors = validate_and_default_interface(data, cls.path)
if not is_valid:
raise InterfaceValidationError("Invalid interface data")
return cls(**data)
def to_python(cls, data, slim_frames=True, raw=False):
is_valid, errors = validate_and_default_interface(data, cls.path)
if not is_valid:
raise InterfaceValidationError("Invalid stack frame data.")
# Trim down the frame list to a hard limit. Leave the last frame in place in case
# it's useful for debugging.
frameiter = data.get('frames') or []
if len(frameiter) > settings.SENTRY_STACKTRACE_FRAMES_HARD_LIMIT:
frameiter = chain(
islice(data['frames'], settings.SENTRY_STACKTRACE_FRAMES_HARD_LIMIT - 1), (data['frames'][-1],))
frame_list = []
for f in frameiter:
if f is None:
continue
# XXX(dcramer): handle PHP sending an empty array for a frame
frame_list.append(Frame.to_python(f or {}, raw=raw))
kwargs = {
'frames': frame_list,
}
kwargs['registers'] = None
if data.get('registers') and isinstance(data['registers'], dict):
kwargs['registers'] = data.get('registers')
kwargs['frames_omitted'] = data.get('frames_omitted') or None
instance = cls(**kwargs)
if slim_frames:
slim_frame_data(instance)
return instance
def to_python(cls, data, rust_renormalized=RUST_RENORMALIZED_DEFAULT):
if rust_renormalized:
for key in (
'abs_path',
'filename',
'context_line',
'lineno',
'pre_context',
'post_context',
):
data.setdefault(key, None)
return cls(**data)
is_valid, errors = validate_and_default_interface(data, cls.path)
if not is_valid:
raise InterfaceValidationError("Invalid template")
kwargs = {
'abs_path': trim(data.get('abs_path', None), 256),
'filename': trim(data.get('filename', None), 256),
'context_line': trim(data.get('context_line', None), 256),
'lineno': int(data['lineno']) if data.get('lineno', None) is not None else None,
# TODO(dcramer): trim pre/post_context
'pre_context': data.get('pre_context'),
'post_context': data.get('post_context'),
}
return cls(**kwargs)
def to_python(cls, data, slim_frames=True, raw=False):
is_valid, errors = validate_and_default_interface(data, cls.path)
if not is_valid:
raise InterfaceValidationError("Invalid stack frame data.")
# Trim down the frame list to a hard limit. Leave the last frame in place in case
# it's useful for debugging.
frameiter = data.get('frames') or []
if len(frameiter) > settings.SENTRY_STACKTRACE_FRAMES_HARD_LIMIT:
frameiter = chain(
islice(data['frames'], settings.SENTRY_STACKTRACE_FRAMES_HARD_LIMIT - 1), (data['frames'][-1],))
frame_list = []
for f in frameiter:
if f is None:
continue
# XXX(dcramer): handle PHP sending an empty array for a frame
frame_list.append(Frame.to_python(f or {}, raw=raw))
kwargs = {
'frames': frame_list,
}
kwargs['registers'] = None
if data.get('registers') and isinstance(data['registers'], dict):
kwargs['registers'] = data.get('registers')
kwargs['frames_omitted'] = data.get('frames_omitted') or None
instance = cls(**kwargs)
if slim_frames:
slim_frame_data(instance)
return instance
def to_python(cls, data):
# TODO(markus): semaphore does not validate security interfaces yet
is_valid, errors = validate_and_default_interface(data, cls.path)
if not is_valid:
raise InterfaceValidationError("Invalid interface data")
return cls(**data)
def to_python(cls, data, rust_renormalized=RUST_RENORMALIZED_DEFAULT):
# TODO(markus): semaphore does not validate security interfaces yet
is_valid, errors = validate_and_default_interface(data, cls.path)
if not is_valid:
raise InterfaceValidationError("Invalid interface data")
return cls(**data)
def to_python(cls, data, slim_frames=True, rust_renormalized=RUST_RENORMALIZED_DEFAULT):
if not rust_renormalized:
is_valid, errors = validate_and_default_interface(data, cls.path)
if not is_valid:
raise InterfaceValidationError("Invalid exception")
if not (data.get('type') or data.get('value')):
raise InterfaceValidationError("No 'type' or 'value' present")
if get_path(data, 'stacktrace', 'frames', filter=True):
stacktrace = Stacktrace.to_python(
data['stacktrace'],
slim_frames=slim_frames,
rust_renormalized=rust_renormalized
)
else:
stacktrace = None
if get_path(data, 'raw_stacktrace', 'frames', filter=True):
raw_stacktrace = Stacktrace.to_python(
data['raw_stacktrace'], slim_frames=slim_frames, raw=True,
rust_renormalized=rust_renormalized
)
else:
raw_stacktrace = None
type = data.get('type')
value = data.get('value')
if not rust_renormalized:
if isinstance(value, six.string_types):
if type is None:
m = _type_value_re.match(value)
if m:
type = m.group(1)
value = m.group(2).strip()
elif value is not None:
value = json.dumps(value)
value = trim(value, 4096)
if data.get('mechanism'):
mechanism = Mechanism.to_python(data['mechanism'],
rust_renormalized=rust_renormalized)
else:
mechanism = None
kwargs = {
'type': trim(type, 128),
'value': value,
'module': trim(data.get('module'), 128),
'mechanism': mechanism,
'stacktrace': stacktrace,
'thread_id': trim(data.get('thread_id'), 40),
'raw_stacktrace': raw_stacktrace,
}
return cls(**kwargs)
def to_python(cls, data, slim_frames=True, rust_renormalized=RUST_RENORMALIZED_DEFAULT):
if not rust_renormalized:
is_valid, errors = validate_and_default_interface(data, cls.path)
if not is_valid:
raise InterfaceValidationError("Invalid exception")
if not (data.get('type') or data.get('value')):
raise InterfaceValidationError("No 'type' or 'value' present")
if get_path(data, 'stacktrace', 'frames', filter=True):
stacktrace = Stacktrace.to_python(
data['stacktrace'],
slim_frames=slim_frames,
rust_renormalized=rust_renormalized
)
else:
stacktrace = None
if get_path(data, 'raw_stacktrace', 'frames', filter=True):
raw_stacktrace = Stacktrace.to_python(
data['raw_stacktrace'], slim_frames=slim_frames, raw=True,
rust_renormalized=rust_renormalized
)
else:
raw_stacktrace = None
type = data.get('type')
value = data.get('value')
if not rust_renormalized:
if isinstance(value, six.string_types):
if type is None:
m = _type_value_re.match(value)
if m:
type = m.group(1)
value = m.group(2).strip()
elif value is not None:
value = json.dumps(value)
value = trim(value, 4096)
if data.get('mechanism'):
mechanism = Mechanism.to_python(data['mechanism'],
rust_renormalized=rust_renormalized)
else:
mechanism = None
kwargs = {
'type': trim(type, 128),
'value': value,
'module': trim(data.get('module'), 128),
'mechanism': mechanism,
'stacktrace': stacktrace,
'thread_id': trim(data.get('thread_id'), 40),
'raw_stacktrace': raw_stacktrace,
}
return cls(**kwargs)
def to_python(cls, data):
data = upgrade_legacy_mechanism(data)
is_valid, errors = validate_and_default_interface(data, cls.path)
if not is_valid:
raise InterfaceValidationError("Invalid mechanism")
if not data.get('type'):
raise InterfaceValidationError("No 'type' present")
mechanism_meta = data.get('meta') or {}
mach_exception = mechanism_meta.get('mach_exception')
if mach_exception is not None:
mach_exception = prune_empty_keys({
'exception':
mach_exception['exception'],
'code':
mach_exception['code'],
'subcode':
mach_exception['subcode'],
'name':
mach_exception.get('name'),
})
signal = mechanism_meta.get('signal')
if signal is not None:
signal = prune_empty_keys({
'number': signal['number'],
'code': signal.get('code'),
'name': signal.get('name'),
'code_name': signal.get('code_name'),
})
errno = mechanism_meta.get('errno')
if errno is not None:
errno = prune_empty_keys({
'number': errno['number'],
'name': errno.get('name'),
})
kwargs = {
'type': trim(data['type'], 128),
'synthetic': data.get('synthetic'),
'description': trim(data.get('description'), 1024),
'help_link': trim(data.get('help_link'), 1024),
'handled': data.get('handled'),
'data': trim(data.get('data'), 4096),
'meta': {
'errno': errno,
'mach_exception': mach_exception,
'signal': signal,
},
}
return cls(**kwargs)
def to_python(cls, data, slim_frames=True):
is_valid, errors = validate_and_default_interface(data, cls.path)
if not is_valid:
raise InterfaceValidationError("Invalid exception")
if not (data.get('type') or data.get('value')):
raise InterfaceValidationError("No 'type' or 'value' present")
if data.get('stacktrace') and data['stacktrace'].get('frames'):
stacktrace = Stacktrace.to_python(
data['stacktrace'],
slim_frames=slim_frames,
)
else:
stacktrace = None
if data.get('raw_stacktrace') and data['raw_stacktrace'].get('frames'):
raw_stacktrace = Stacktrace.to_python(data['raw_stacktrace'],
slim_frames=slim_frames,
raw=True)
else:
raw_stacktrace = None
type = data.get('type')
value = data.get('value')
if isinstance(value, six.string_types):
if type is None:
m = _type_value_re.match(value)
if m:
type = m.group(1)
value = m.group(2).strip()
elif value is not None:
value = json.dumps(value)
value = trim(value, 4096)
mechanism = data.get('mechanism')
if mechanism is not None:
mechanism = trim(data.get('mechanism'), 4096)
mechanism.setdefault('type', 'generic')
kwargs = {
'type': trim(type, 128),
'value': value,
'module': trim(data.get('module'), 128),
'mechanism': mechanism,
'stacktrace': stacktrace,
'thread_id': trim(data.get('thread_id'), 40),
'raw_stacktrace': raw_stacktrace,
}
return cls(**kwargs)
def to_python(cls, data):
data = upgrade_legacy_mechanism(data)
is_valid, errors = validate_and_default_interface(data, cls.path)
if not is_valid:
raise InterfaceValidationError("Invalid mechanism")
if not data.get('type'):
raise InterfaceValidationError("No 'type' present")
meta = data.get('meta', {})
mach_exception = meta.get('mach_exception')
if mach_exception is not None:
mach_exception = prune_empty_keys({
'exception': mach_exception['exception'],
'code': mach_exception['code'],
'subcode': mach_exception['subcode'],
'name': mach_exception.get('name'),
})
signal = meta.get('signal')
if signal is not None:
signal = prune_empty_keys({
'number': signal['number'],
'code': signal.get('code'),
'name': signal.get('name'),
'code_name': signal.get('code_name'),
})
errno = meta.get('errno')
if errno is not None:
errno = prune_empty_keys({
'number': errno['number'],
'name': errno.get('name'),
})
kwargs = {
'type': trim(data['type'], 128),
'description': trim(data.get('description'), 1024),
'help_link': trim(data.get('help_link'), 1024),
'handled': data.get('handled'),
'data': trim(data.get('data'), 4096),
'meta': {
'errno': errno,
'mach_exception': mach_exception,
'signal': signal,
},
}
return cls(**kwargs)
def to_python(cls, data):
is_valid, errors = validate_and_default_interface(data, cls.path)
if not is_valid:
raise InterfaceValidationError("Invalid template")
kwargs = {
'abs_path': trim(data.get('abs_path', None), 256),
'filename': trim(data.get('filename', None), 256),
'context_line': trim(data.get('context_line', None), 256),
'lineno': int(data['lineno']) if data.get('lineno', None) is not None else None,
# TODO(dcramer): trim pre/post_context
'pre_context': data.get('pre_context'),
'post_context': data.get('post_context'),
}
return cls(**kwargs)
def to_python(cls, data):
is_valid, errors = validate_and_default_interface(data, cls.path)
if not is_valid:
raise InterfaceValidationError("Invalid template")
kwargs = {
'abs_path': trim(data.get('abs_path', None), 256),
'filename': trim(data.get('filename', None), 256),
'context_line': trim(data.get('context_line', None), 256),
'lineno': int(data['lineno']) if data.get('lineno', None) is not None else None,
# TODO(dcramer): trim pre/post_context
'pre_context': data.get('pre_context'),
'post_context': data.get('post_context'),
}
return cls(**kwargs)
def to_python(cls, data):
is_valid, errors = validate_and_default_interface(data, cls.path)
if not is_valid:
raise InterfaceValidationError("Invalid device")
data = data.copy()
extra_data = data.pop('data', data)
name = trim(data.pop('name'), 64)
version = trim(data.pop('version'), 64)
build = trim(data.pop('build', None), 64)
kwargs = {
'name': name,
'version': version,
'build': build,
'data': trim_dict(extra_data),
}
return cls(**kwargs)
def to_python(cls, data, raw=False):
is_valid, errors = validate_and_default_interface(data, cls.path)
if not is_valid:
raise InterfaceValidationError("Invalid stack frame data.")
abs_path = data.get('abs_path')
filename = data.get('filename')
symbol = data.get('symbol')
function = data.get('function')
module = data.get('module')
package = data.get('package')
# For legacy reasons
if function in ('?', ''):
function = None
# For consistency reasons
if symbol in ('?', ''):
symbol = None
# Some of this processing should only be done for non raw frames
if not raw:
# absolute path takes priority over filename
# (in the end both will get set)
if not abs_path:
abs_path = filename
filename = None
if not filename and abs_path:
if is_url(abs_path):
urlparts = urlparse(abs_path)
if urlparts.path:
filename = urlparts.path
else:
filename = abs_path
else:
filename = abs_path
platform = data.get('platform')
context_locals = data.get('vars') or {}
if isinstance(context_locals, (list, tuple)):
context_locals = dict(enumerate(context_locals))
elif not isinstance(context_locals, dict):
context_locals = {}
context_locals = trim_dict(context_locals, object_hook=handle_nan)
# extra data is used purely by internal systems,
# so we dont trim it
extra_data = data.get('data') or {}
if isinstance(extra_data, (list, tuple)):
extra_data = dict(enumerate(extra_data))
# XXX: handle lines which were sent as 'null'
context_line = trim(data.get('context_line'), 256)
pre_context = data.get('pre_context', None)
if isinstance(pre_context, list) and pre_context:
pre_context = [c or '' for c in pre_context]
else:
pre_context = None
post_context = data.get('post_context', None)
if isinstance(post_context, list) and post_context:
post_context = [c or '' for c in post_context]
else:
post_context = None
if not context_line and (pre_context or post_context):
context_line = ''
in_app = validate_bool(data.get('in_app'), False)
kwargs = {
'abs_path': trim(abs_path, 2048),
'filename': trim(filename, 256),
'platform': platform,
'module': trim(module, 256),
'function': trim(function, 256),
'package': package,
'image_addr': to_hex_addr(data.get('image_addr')),
'symbol': trim(symbol, 256),
'symbol_addr': to_hex_addr(data.get('symbol_addr')),
'instruction_addr': to_hex_addr(data.get('instruction_addr')),
'trust': trim(data.get('trust'), 16),
'in_app': in_app,
'context_line': context_line,
# TODO(dcramer): trim pre/post_context
'pre_context': pre_context,
'post_context': post_context,
'vars': context_locals or None,
'data': extra_data or None,
'errors': data.get('errors'),
}
if data.get('lineno') is not None:
lineno = int(data['lineno'])
if lineno < 0:
lineno = None
kwargs['lineno'] = lineno
else:
kwargs['lineno'] = None
if data.get('colno') is not None:
colno = int(data['colno'])
if colno < 0:
colno = None
kwargs['colno'] = colno
else:
kwargs['colno'] = None
return cls(**kwargs)
def normalize(self, request_env=None):
request_env = request_env or {}
data = self.data
errors = data['errors'] = []
# Ignore event meta data for now.
data.pop('_meta', None)
# Before validating with a schema, attempt to cast values to their desired types
# so that the schema doesn't have to take every type variation into account.
text = six.text_type
fp_types = six.string_types + six.integer_types + (float, )
def to_values(v):
return {'values': v} if v and isinstance(v, (tuple, list)) else v
def stringify(f):
if isinstance(f, float):
return text(int(f)) if abs(f) < (1 << 53) else None
return text(f)
casts = {
'environment': lambda v: text(v) if v is not None else v,
'fingerprint': lambda v: list(x for x in map(stringify, v) if x is not None) if isinstance(v, list) and all(isinstance(f, fp_types) for f in v) else v,
'release': lambda v: text(v) if v is not None else v,
'dist': lambda v: text(v).strip() if v is not None else v,
'time_spent': lambda v: int(v) if v is not None else v,
'tags': lambda v: [(text(v_k).replace(' ', '-').strip(), text(v_v).strip()) for (v_k, v_v) in dict(v).items()],
'timestamp': lambda v: process_timestamp(v),
'platform': lambda v: v if v in VALID_PLATFORMS else 'other',
'logentry': lambda v: v if isinstance(v, dict) else {'message': v},
# These can be sent as lists and need to be converted to {'values': [...]}
'exception': to_values,
'breadcrumbs': to_values,
'threads': to_values,
}
for c in casts:
if c in data:
try:
data[c] = casts[c](data[c])
except InvalidTimestamp as it:
errors.append({'type': it.args[0], 'name': c, 'value': data[c]})
del data[c]
except Exception as e:
errors.append({'type': EventError.INVALID_DATA, 'name': c, 'value': data[c]})
del data[c]
# raw 'message' is coerced to the Message interface, as its used for pure index of
# searchable strings. If both a raw 'message' and a Message interface exist, try and
# add the former as the 'formatted' attribute of the latter.
# See GH-3248
msg_str = data.pop('message', None)
if msg_str:
msg_if = data.get('logentry')
msg_meta = data.get('_meta', {}).get('message')
if not msg_if:
msg_if = data['logentry'] = {'message': msg_str}
if msg_meta:
data.setdefault('_meta', {}).setdefault('logentry', {})['message'] = msg_meta
if msg_if.get('message') != msg_str:
if not msg_if.get('formatted'):
msg_if['formatted'] = msg_str
if msg_meta:
data.setdefault('_meta', {}).setdefault(
'logentry', {})['formatted'] = msg_meta
# Fill in ip addresses marked as {{auto}}
client_ip = request_env.get('client_ip')
if client_ip:
if get_path(data, ['sentry.interfaces.Http', 'env', 'REMOTE_ADDR']) == '{{auto}}':
data['sentry.interfaces.Http']['env']['REMOTE_ADDR'] = client_ip
if get_path(data, ['request', 'env', 'REMOTE_ADDR']) == '{{auto}}':
data['request']['env']['REMOTE_ADDR'] = client_ip
if get_path(data, ['sentry.interfaces.User', 'ip_address']) == '{{auto}}':
data['sentry.interfaces.User']['ip_address'] = client_ip
if get_path(data, ['user', 'ip_address']) == '{{auto}}':
data['user']['ip_address'] = client_ip
# Validate main event body and tags against schema.
# XXX(ja): jsonschema does not like CanonicalKeyDict, so we need to pass
# in the inner data dict.
is_valid, event_errors = validate_and_default_interface(data.data, 'event')
errors.extend(event_errors)
if 'tags' in data:
is_valid, tag_errors = validate_and_default_interface(data['tags'], 'tags', name='tags')
errors.extend(tag_errors)
# Validate interfaces
for k in list(iter(data)):
if k in CLIENT_RESERVED_ATTRS:
continue
value = data.pop(k)
if not value:
self.logger.debug('Ignored empty interface value: %s', k)
continue
try:
interface = get_interface(k)
except ValueError:
self.logger.debug('Ignored unknown attribute: %s', k)
errors.append({'type': EventError.INVALID_ATTRIBUTE, 'name': k})
continue
try:
inst = interface.to_python(value)
data[inst.get_path()] = inst.to_json()
except Exception as e:
log = self.logger.debug if isinstance(
e, InterfaceValidationError) else self.logger.error
log('Discarded invalid value for interface: %s (%r)', k, value, exc_info=True)
errors.append({'type': EventError.INVALID_DATA, 'name': k, 'value': value})
# Additional data coercion and defaulting
level = data.get('level') or DEFAULT_LOG_LEVEL
if isinstance(level, int) or (isinstance(level, six.string_types) and level.isdigit()):
level = LOG_LEVELS.get(int(level), DEFAULT_LOG_LEVEL)
data['level'] = LOG_LEVELS_MAP.get(level, LOG_LEVELS_MAP[DEFAULT_LOG_LEVEL])
if data.get('dist') and not data.get('release'):
data['dist'] = None
timestamp = data.get('timestamp')
if not timestamp:
timestamp = timezone.now()
# TODO (alex) can this all be replaced by utcnow?
# it looks like the only time that this would even be hit is when timestamp
# is not defined, as the earlier process_timestamp already converts existing
# timestamps to floats.
if isinstance(timestamp, datetime):
# We must convert date to local time so Django doesn't mess it up
# based on TIME_ZONE
if settings.TIME_ZONE:
if not timezone.is_aware(timestamp):
timestamp = timestamp.replace(tzinfo=timezone.utc)
elif timezone.is_aware(timestamp):
timestamp = timestamp.replace(tzinfo=None)
timestamp = float(timestamp.strftime('%s'))
data['timestamp'] = timestamp
data['received'] = float(timezone.now().strftime('%s'))
data.setdefault('checksum', None)
data.setdefault('culprit', None)
data.setdefault('dist', None)
data.setdefault('environment', None)
data.setdefault('extra', {})
data.setdefault('fingerprint', None)
data.setdefault('logger', DEFAULT_LOGGER_NAME)
data.setdefault('platform', None)
data.setdefault('server_name', None)
data.setdefault('site', None)
data.setdefault('tags', [])
data.setdefault('transaction', None)
# Fix case where legacy apps pass 'environment' as a tag
# instead of a top level key.
# TODO (alex) save() just reinserts the environment into the tags
if not data.get('environment'):
tagsdict = dict(data['tags'])
if 'environment' in tagsdict:
data['environment'] = tagsdict['environment']
del tagsdict['environment']
data['tags'] = tagsdict.items()
# the SDKs currently do not describe event types, and we must infer
# them from available attributes
data['type'] = eventtypes.infer(data).key
data['version'] = self.version
exception = data.get('sentry.interfaces.Exception')
stacktrace = data.get('sentry.interfaces.Stacktrace')
if exception and len(exception['values']) == 1 and stacktrace:
exception['values'][0]['stacktrace'] = stacktrace
del data['sentry.interfaces.Stacktrace']
# Exception mechanism needs SDK information to resolve proper names in
# exception meta (such as signal names). "SDK Information" really means
# the operating system version the event was generated on. Some
# normalization still works without sdk_info, such as mach_exception
# names (they can only occur on macOS).
if exception:
sdk_info = get_sdk_from_event(data)
for ex in exception['values']:
if 'mechanism' in ex:
normalize_mechanism_meta(ex['mechanism'], sdk_info)
# If there is no User ip_addres, update it either from the Http interface
# or the client_ip of the request.
auth = request_env.get('auth')
is_public = auth and auth.is_public
add_ip_platforms = ('javascript', 'cocoa', 'objc')
http_ip = data.get('sentry.interfaces.Http', {}).get('env', {}).get('REMOTE_ADDR')
if http_ip:
data.setdefault('sentry.interfaces.User', {}).setdefault('ip_address', http_ip)
elif client_ip and (is_public or data.get('platform') in add_ip_platforms):
data.setdefault('sentry.interfaces.User', {}).setdefault('ip_address', client_ip)
# Trim values
data['logger'] = trim(data['logger'].strip(), 64)
trim_dict(data['extra'], max_size=settings.SENTRY_MAX_EXTRA_VARIABLE_SIZE)
if data['culprit']:
data['culprit'] = trim(data['culprit'], MAX_CULPRIT_LENGTH)
if data['transaction']:
data['transaction'] = trim(data['transaction'], MAX_CULPRIT_LENGTH)
return data
def normalize(self, request_env=None):
request_env = request_env or {}
data = self.data
errors = data['errors'] = []
# Before validating with a schema, attempt to cast values to their desired types
# so that the schema doesn't have to take every type variation into account.
text = six.text_type
fp_types = six.string_types + six.integer_types + (float, )
def to_values(v):
return {'values': v} if v and isinstance(v, (tuple, list)) else v
def convert_fingerprint(values):
rv = values[:]
bad_float = False
for idx, item in enumerate(rv):
if isinstance(item, float) and \
(abs(item) >= (1 << 53) or int(item) != item):
bad_float = True
rv[idx] = text(item)
if bad_float:
metrics.incr(
'events.bad_float_fingerprint',
skip_internal=True,
tags={
'project_id': data.get('project'),
},
)
return rv
casts = {
'environment':
lambda v: text(v) if v is not None else v,
'fingerprint':
lambda v: convert_fingerprint(v)
if isinstance(v, list) and all(isinstance(f, fp_types)
for f in v) else v,
'release':
lambda v: text(v) if v is not None else v,
'dist':
lambda v: text(v).strip() if v is not None else v,
'time_spent':
lambda v: int(v) if v is not None else v,
'tags':
lambda v: [(text(v_k).replace(' ', '-').strip(), text(v_v).strip())
for (v_k, v_v) in dict(v).items()],
'timestamp':
lambda v: process_timestamp(v),
'platform':
lambda v: v if v in VALID_PLATFORMS else 'other',
'sentry.interfaces.Message':
lambda v: v if isinstance(v, dict) else {
'message': v
},
# These can be sent as lists and need to be converted to {'values': [...]}
'exception':
to_values,
'sentry.interfaces.Exception':
to_values,
'breadcrumbs':
to_values,
'sentry.interfaces.Breadcrumbs':
to_values,
'threads':
to_values,
'sentry.interfaces.Threads':
to_values,
}
for c in casts:
if c in data:
try:
data[c] = casts[c](data[c])
except InvalidTimestamp as it:
errors.append({
'type': it.args[0],
'name': c,
'value': data[c]
})
del data[c]
except Exception as e:
errors.append({
'type': EventError.INVALID_DATA,
'name': c,
'value': data[c]
})
del data[c]
# raw 'message' is coerced to the Message interface, as its used for pure index of
# searchable strings. If both a raw 'message' and a Message interface exist, try and
# add the former as the 'formatted' attribute of the latter.
# See GH-3248
msg_str = data.pop('message', None)
if msg_str:
msg_if = data.setdefault('sentry.interfaces.Message',
{'message': msg_str})
if msg_if.get('message') != msg_str:
msg_if.setdefault('formatted', msg_str)
# Fill in ip addresses marked as {{auto}}
client_ip = request_env.get('client_ip')
if client_ip:
if get_path(data, ['sentry.interfaces.Http', 'env', 'REMOTE_ADDR'
]) == '{{auto}}':
data['sentry.interfaces.Http']['env'][
'REMOTE_ADDR'] = client_ip
if get_path(data, ['request', 'env', 'REMOTE_ADDR']) == '{{auto}}':
data['request']['env']['REMOTE_ADDR'] = client_ip
if get_path(
data,
['sentry.interfaces.User', 'ip_address']) == '{{auto}}':
data['sentry.interfaces.User']['ip_address'] = client_ip
if get_path(data, ['user', 'ip_address']) == '{{auto}}':
data['user']['ip_address'] = client_ip
# Validate main event body and tags against schema
is_valid, event_errors = validate_and_default_interface(data, 'event')
errors.extend(event_errors)
if 'tags' in data:
is_valid, tag_errors = validate_and_default_interface(data['tags'],
'tags',
name='tags')
errors.extend(tag_errors)
# Validate interfaces
for k in list(iter(data)):
if k in CLIENT_RESERVED_ATTRS:
continue
value = data.pop(k)
if not value:
self.logger.debug('Ignored empty interface value: %s', k)
continue
try:
interface = get_interface(k)
except ValueError:
self.logger.debug('Ignored unknown attribute: %s', k)
errors.append({
'type': EventError.INVALID_ATTRIBUTE,
'name': k
})
continue
try:
inst = interface.to_python(value)
data[inst.get_path()] = inst.to_json()
except Exception as e:
log = self.logger.debug if isinstance(
e, InterfaceValidationError) else self.logger.error
log('Discarded invalid value for interface: %s (%r)',
k,
value,
exc_info=True)
errors.append({
'type': EventError.INVALID_DATA,
'name': k,
'value': value
})
# Additional data coercion and defaulting
level = data.get('level') or DEFAULT_LOG_LEVEL
if isinstance(level, int) or (isinstance(level, six.string_types)
and level.isdigit()):
level = LOG_LEVELS.get(int(level), DEFAULT_LOG_LEVEL)
data['level'] = LOG_LEVELS_MAP.get(level,
LOG_LEVELS_MAP[DEFAULT_LOG_LEVEL])
if data.get('dist') and not data.get('release'):
data['dist'] = None
timestamp = data.get('timestamp')
if not timestamp:
timestamp = timezone.now()
# TODO (alex) can this all be replaced by utcnow?
# it looks like the only time that this would even be hit is when timestamp
# is not defined, as the earlier process_timestamp already converts existing
# timestamps to floats.
if isinstance(timestamp, datetime):
# We must convert date to local time so Django doesn't mess it up
# based on TIME_ZONE
if settings.TIME_ZONE:
if not timezone.is_aware(timestamp):
timestamp = timestamp.replace(tzinfo=timezone.utc)
elif timezone.is_aware(timestamp):
timestamp = timestamp.replace(tzinfo=None)
timestamp = float(timestamp.strftime('%s'))
data['timestamp'] = timestamp
data['received'] = float(timezone.now().strftime('%s'))
data.setdefault('checksum', None)
data.setdefault('culprit', None)
data.setdefault('dist', None)
data.setdefault('environment', None)
data.setdefault('extra', {})
data.setdefault('fingerprint', None)
data.setdefault('logger', DEFAULT_LOGGER_NAME)
data.setdefault('platform', None)
data.setdefault('server_name', None)
data.setdefault('site', None)
data.setdefault('tags', [])
data.setdefault('transaction', None)
# Fix case where legacy apps pass 'environment' as a tag
# instead of a top level key.
# TODO (alex) save() just reinserts the environment into the tags
if not data.get('environment'):
tagsdict = dict(data['tags'])
if 'environment' in tagsdict:
data['environment'] = tagsdict['environment']
del tagsdict['environment']
data['tags'] = tagsdict.items()
# the SDKs currently do not describe event types, and we must infer
# them from available attributes
data['type'] = eventtypes.infer(data).key
data['version'] = self.version
exception = data.get('sentry.interfaces.Exception')
stacktrace = data.get('sentry.interfaces.Stacktrace')
if exception and len(exception['values']) == 1 and stacktrace:
exception['values'][0]['stacktrace'] = stacktrace
del data['sentry.interfaces.Stacktrace']
# Exception mechanism needs SDK information to resolve proper names in
# exception meta (such as signal names). "SDK Information" really means
# the operating system version the event was generated on. Some
# normalization still works without sdk_info, such as mach_exception
# names (they can only occur on macOS).
if exception:
sdk_info = get_sdk_from_event(data)
for ex in exception['values']:
if 'mechanism' in ex:
normalize_mechanism_meta(ex['mechanism'], sdk_info)
# If there is no User ip_addres, update it either from the Http interface
# or the client_ip of the request.
auth = request_env.get('auth')
is_public = auth and auth.is_public
add_ip_platforms = ('javascript', 'cocoa', 'objc')
http_ip = data.get('sentry.interfaces.Http',
{}).get('env', {}).get('REMOTE_ADDR')
if http_ip:
data.setdefault('sentry.interfaces.User',
{}).setdefault('ip_address', http_ip)
elif client_ip and (is_public
or data.get('platform') in add_ip_platforms):
data.setdefault('sentry.interfaces.User',
{}).setdefault('ip_address', client_ip)
# Trim values
data['logger'] = trim(data['logger'].strip(), 64)
trim_dict(data['extra'],
max_size=settings.SENTRY_MAX_EXTRA_VARIABLE_SIZE)
if data['culprit']:
data['culprit'] = trim(data['culprit'], MAX_CULPRIT_LENGTH)
if data['transaction']:
data['transaction'] = trim(data['transaction'], MAX_CULPRIT_LENGTH)
return data
def to_python(cls, data):
is_valid, errors = validate_and_default_interface(data, cls.path)
if not is_valid:
raise InterfaceValidationError("Invalid interface data")
return cls(**data)
def to_python(cls, data):
is_valid, errors = validate_and_default_interface(data, cls.path)
if not is_valid:
raise InterfaceValidationError("Invalid interface data")
kwargs = {}
if data.get('method'):
method = data['method'].upper()
# Optimize for the common path here, where it's a GET/POST, falling
# back to a regular expresion test
if method not in ('GET',
'POST') and not http_method_re.match(method):
raise InterfaceValidationError("Invalid value for 'method'")
kwargs['method'] = method
else:
kwargs['method'] = None
scheme, netloc, path, query_bit, fragment_bit = urlsplit(data['url'])
query_string = data.get('query_string') or query_bit
if query_string:
# if querystring was a dict, convert it to a string
if isinstance(query_string, dict):
query_string = urlencode([(to_bytes(k), to_bytes(v))
for k, v in query_string.items()])
else:
if query_string[0] == '?':
# remove '?' prefix
query_string = query_string[1:]
kwargs['query_string'] = trim(query_string, 4096)
else:
kwargs['query_string'] = ''
fragment = data.get('fragment') or fragment_bit
cookies = data.get('cookies')
# if cookies were [also] included in headers we
# strip them out
headers = data.get('headers')
if headers:
headers, cookie_header = format_headers(headers)
if not cookies and cookie_header:
cookies = cookie_header
else:
headers = ()
# We prefer the body to be a string, since we can then attempt to parse it
# as JSON OR decode it as a URL encoded query string, without relying on
# the correct content type header being passed.
body = data.get('data')
content_type = next((v for k, v in headers if k == 'Content-Type'),
None)
# Remove content type parameters
if content_type is not None:
content_type = content_type.partition(';')[0].rstrip()
# We process request data once during ingestion and again when
# requesting the http interface over the API. Avoid overwriting
# decoding the body again.
inferred_content_type = data.get('inferred_content_type', content_type)
if 'inferred_content_type' not in data and not isinstance(body, dict):
body, inferred_content_type = heuristic_decode(body, content_type)
if body:
body = trim(body, settings.SENTRY_MAX_HTTP_BODY_SIZE)
env = data.get('env', {})
# TODO (alex) This could also be accomplished with schema (with formats)
if 'REMOTE_ADDR' in env:
try:
validate_ip(env['REMOTE_ADDR'], required=False)
except ValueError:
del env['REMOTE_ADDR']
kwargs['inferred_content_type'] = inferred_content_type
kwargs['cookies'] = trim_pairs(format_cookies(cookies))
kwargs['env'] = trim_dict(env)
kwargs['headers'] = trim_pairs(headers)
kwargs['data'] = fix_broken_encoding(body)
kwargs['url'] = urlunsplit((scheme, netloc, path, '', ''))
kwargs['fragment'] = trim(fragment, 1024)
return cls(**kwargs)
def to_python(cls, data, rust_renormalized=RUST_RENORMALIZED_DEFAULT):
if rust_renormalized:
data.setdefault('query_string', [])
for key in (
"method",
"url",
"fragment",
"cookies",
"headers",
"data",
"env",
"inferred_content_type",
):
data.setdefault(key, None)
return cls(**data)
is_valid, errors = validate_and_default_interface(data, cls.path)
if not is_valid:
raise InterfaceValidationError("Invalid interface data")
kwargs = {}
if data.get('method'):
method = data['method'].upper()
# Optimize for the common path here, where it's a GET/POST, falling
# back to a regular expresion test
if method not in ('GET', 'POST') and not http_method_re.match(method):
raise InterfaceValidationError("Invalid value for 'method'")
kwargs['method'] = method
else:
kwargs['method'] = None
if data.get('url', None):
url = to_unicode(data['url'])
# The JavaScript SDK used to send an ellipsis character for
# truncated URLs. Canonical URLs do not contain UTF-8 characters in
# either the path, query string or fragment, so we replace it with
# three dots (which is the behavior of other SDKs). This effectively
# makes the string two characters longer, but it will be trimmed
# again down below.
if url.endswith(u"\u2026"):
url = url[:-1] + "..."
scheme, netloc, path, query_bit, fragment_bit = urlsplit(url)
else:
scheme = netloc = path = query_bit = fragment_bit = None
query_string = data.get('query_string') or query_bit
if query_string:
if isinstance(query_string, six.string_types):
if query_string[0] == '?':
query_string = query_string[1:]
if query_string.endswith(u"\u2026"):
query_string = query_string[:-1] + "..."
query_string = [
(to_unicode(k), jsonify(v))
for k, v in parse_qsl(query_string, keep_blank_values=True)
]
elif isinstance(query_string, dict):
query_string = [(to_unicode(k), jsonify(v)) for k, v in six.iteritems(query_string)]
elif isinstance(query_string, list):
query_string = [
tuple(tup) for tup in query_string
if isinstance(tup, (tuple, list)) and len(tup) == 2
]
else:
query_string = []
kwargs['query_string'] = trim(query_string, 4096)
else:
kwargs['query_string'] = []
fragment = data.get('fragment') or fragment_bit
cookies = data.get('cookies')
# if cookies were [also] included in headers we
# strip them out
if data.get("headers"):
headers, cookie_header = format_headers(get_path(data, "headers", filter=True))
if not cookies and cookie_header:
cookies = cookie_header
else:
headers = ()
# We prefer the body to be a string, since we can then attempt to parse it
# as JSON OR decode it as a URL encoded query string, without relying on
# the correct content type header being passed.
body = data.get('data')
content_type = next((v for k, v in headers if k == 'Content-Type'), None)
# Remove content type parameters
if content_type is not None:
content_type = content_type.partition(';')[0].rstrip()
# We process request data once during ingestion and again when
# requesting the http interface over the API. Avoid overwriting
# decoding the body again.
inferred_content_type = data.get('inferred_content_type', content_type)
if 'inferred_content_type' not in data and not isinstance(body, dict):
body, inferred_content_type = heuristic_decode(body, content_type)
if body:
body = trim(body, settings.SENTRY_MAX_HTTP_BODY_SIZE)
env = data.get('env', {})
# TODO (alex) This could also be accomplished with schema (with formats)
if 'REMOTE_ADDR' in env:
try:
validate_ip(env['REMOTE_ADDR'], required=False)
except ValueError:
del env['REMOTE_ADDR']
kwargs['inferred_content_type'] = inferred_content_type
kwargs['cookies'] = trim_pairs(format_cookies(cookies))
kwargs['env'] = trim_dict(env)
kwargs['headers'] = trim_pairs(headers)
kwargs['data'] = fix_broken_encoding(body)
kwargs['url'] = urlunsplit((scheme, netloc, path, '', ''))
kwargs['fragment'] = trim(fragment, 1024)
return cls(**kwargs)
def to_python(cls, data, raw=False):
is_valid, errors = validate_and_default_interface(data, cls.path)
if not is_valid:
raise InterfaceValidationError("Invalid stack frame data.")
abs_path = data.get('abs_path')
filename = data.get('filename')
symbol = data.get('symbol')
function = data.get('function')
module = data.get('module')
package = data.get('package')
# For legacy reasons
if function in ('?', ''):
function = None
# For consistency reasons
if symbol in ('?', ''):
symbol = None
# Some of this processing should only be done for non raw frames
if not raw:
# absolute path takes priority over filename
# (in the end both will get set)
if not abs_path:
abs_path = filename
filename = None
if not filename and abs_path:
if is_url(abs_path):
urlparts = urlparse(abs_path)
if urlparts.path:
filename = urlparts.path
else:
filename = abs_path
else:
filename = abs_path
platform = data.get('platform')
context_locals = data.get('vars') or {}
if isinstance(context_locals, (list, tuple)):
context_locals = dict(enumerate(context_locals))
elif not isinstance(context_locals, dict):
context_locals = {}
context_locals = trim_dict(context_locals, object_hook=handle_nan)
# extra data is used purely by internal systems,
# so we dont trim it
extra_data = data.get('data') or {}
if isinstance(extra_data, (list, tuple)):
extra_data = dict(enumerate(extra_data))
# XXX: handle lines which were sent as 'null'
context_line = trim(data.get('context_line'), 256)
if context_line is not None:
pre_context = data.get('pre_context', None)
if pre_context:
pre_context = [c or '' for c in pre_context]
post_context = data.get('post_context', None)
if post_context:
post_context = [c or '' for c in post_context]
else:
pre_context, post_context = None, None
in_app = validate_bool(data.get('in_app'), False)
kwargs = {
'abs_path': trim(abs_path, 2048),
'filename': trim(filename, 256),
'platform': platform,
'module': trim(module, 256),
'function': trim(function, 256),
'package': package,
'image_addr': to_hex_addr(data.get('image_addr')),
'symbol': trim(symbol, 256),
'symbol_addr': to_hex_addr(data.get('symbol_addr')),
'instruction_addr': to_hex_addr(data.get('instruction_addr')),
'trust': trim(data.get('trust'), 16),
'in_app': in_app,
'context_line': context_line,
# TODO(dcramer): trim pre/post_context
'pre_context': pre_context,
'post_context': post_context,
'vars': context_locals or None,
'data': extra_data or None,
'errors': data.get('errors'),
}
if data.get('lineno') is not None:
lineno = int(data['lineno'])
if lineno < 0:
lineno = None
kwargs['lineno'] = lineno
else:
kwargs['lineno'] = None
if data.get('colno') is not None:
kwargs['colno'] = int(data['colno'])
else:
kwargs['colno'] = None
return cls(**kwargs)
def to_python(cls, data, rust_renormalized=RUST_RENORMALIZED_DEFAULT):
if rust_renormalized:
data.setdefault('query_string', [])
for key in (
"method",
"url",
"fragment",
"cookies",
"headers",
"data",
"env",
"inferred_content_type",
):
data.setdefault(key, None)
return cls(**data)
is_valid, errors = validate_and_default_interface(data, cls.path)
if not is_valid:
raise InterfaceValidationError("Invalid interface data")
kwargs = {}
if data.get('method'):
method = data['method'].upper()
# Optimize for the common path here, where it's a GET/POST, falling
# back to a regular expresion test
if method not in ('GET',
'POST') and not http_method_re.match(method):
raise InterfaceValidationError("Invalid value for 'method'")
kwargs['method'] = method
else:
kwargs['method'] = None
if data.get('url', None):
url = to_unicode(data['url'])
# The JavaScript SDK used to send an ellipsis character for
# truncated URLs. Canonical URLs do not contain UTF-8 characters in
# either the path, query string or fragment, so we replace it with
# three dots (which is the behavior of other SDKs). This effectively
# makes the string two characters longer, but it will be trimmed
# again down below.
if url.endswith(u"\u2026"):
url = url[:-1] + "..."
scheme, netloc, path, query_bit, fragment_bit = urlsplit(url)
else:
scheme = netloc = path = query_bit = fragment_bit = None
query_string = data.get('query_string') or query_bit
if query_string:
if isinstance(query_string, six.string_types):
if query_string[0] == '?':
query_string = query_string[1:]
if query_string.endswith(u"\u2026"):
query_string = query_string[:-1] + "..."
query_string = [
(to_unicode(k), jsonify(v))
for k, v in parse_qsl(query_string, keep_blank_values=True)
]
elif isinstance(query_string, dict):
query_string = [(to_unicode(k), jsonify(v))
for k, v in six.iteritems(query_string)]
elif isinstance(query_string, list):
query_string = [
tuple(tup) for tup in query_string
if isinstance(tup, (tuple, list)) and len(tup) == 2
]
else:
query_string = []
kwargs['query_string'] = trim(query_string, 4096)
else:
kwargs['query_string'] = []
fragment = data.get('fragment') or fragment_bit
cookies = data.get('cookies')
# if cookies were [also] included in headers we
# strip them out
if data.get("headers"):
headers, cookie_header = format_headers(
get_path(data, "headers", filter=True))
if not cookies and cookie_header:
cookies = cookie_header
else:
headers = ()
# We prefer the body to be a string, since we can then attempt to parse it
# as JSON OR decode it as a URL encoded query string, without relying on
# the correct content type header being passed.
body = data.get('data')
content_type = next((v for k, v in headers if k == 'Content-Type'),
None)
# Remove content type parameters
if content_type is not None:
content_type = content_type.partition(';')[0].rstrip()
# We process request data once during ingestion and again when
# requesting the http interface over the API. Avoid overwriting
# decoding the body again.
inferred_content_type = data.get('inferred_content_type', content_type)
if 'inferred_content_type' not in data and not isinstance(body, dict):
body, inferred_content_type = heuristic_decode(body, content_type)
if body:
body = trim(body, settings.SENTRY_MAX_HTTP_BODY_SIZE)
env = data.get('env', {})
# TODO (alex) This could also be accomplished with schema (with formats)
if 'REMOTE_ADDR' in env:
try:
validate_ip(env['REMOTE_ADDR'], required=False)
except ValueError:
del env['REMOTE_ADDR']
kwargs['inferred_content_type'] = inferred_content_type
kwargs['cookies'] = trim_pairs(format_cookies(cookies))
kwargs['env'] = trim_dict(env)
kwargs['headers'] = trim_pairs(headers)
kwargs['data'] = fix_broken_encoding(body)
kwargs['url'] = urlunsplit((scheme, netloc, path, '', ''))
kwargs['fragment'] = trim(fragment, 1024)
return cls(**kwargs)
def normalize(self):
data = self.data
errors = data.get('errors', [])
# Before validating with a schema, attempt to cast values to their desired types
# so that the schema doesn't have to take every type variation into account.
text = six.text_type
fp_types = six.string_types + six.integer_types + (float, )
def to_values(v):
return {'values': v} if v and isinstance(v, (tuple, list)) else v
casts = {
'environment':
lambda v: text(v) if v is not None else v,
'fingerprint':
lambda v: list(map(text, v))
if isinstance(v, list) and all(isinstance(f, fp_types)
for f in v) else v,
'release':
lambda v: text(v) if v is not None else v,
'dist':
lambda v: text(v).strip() if v is not None else v,
'time_spent':
lambda v: int(v) if v is not None else v,
'tags':
lambda v: [(text(v_k).replace(' ', '-').strip(), text(v_v).strip())
for (v_k, v_v) in dict(v).items()],
'timestamp':
lambda v: process_timestamp(v),
'platform':
lambda v: v if v in VALID_PLATFORMS else 'other',
# These can be sent as lists and need to be converted to {'values': [...]}
'exception':
to_values,
'sentry.interfaces.Exception':
to_values,
'breadcrumbs':
to_values,
'sentry.interfaces.Breadcrumbs':
to_values,
'threads':
to_values,
'sentry.interfaces.Threads':
to_values,
}
for c in casts:
if c in data:
try:
data[c] = casts[c](data[c])
except Exception as e:
errors.append({
'type': EventError.INVALID_DATA,
'name': c,
'value': data[c]
})
del data[c]
# raw 'message' is coerced to the Message interface, as its used for pure index of
# searchable strings. If both a raw 'message' and a Message interface exist, try and
# add the former as the 'formatted' attribute of the latter.
# See GH-3248
msg_str = data.pop('message', None)
if msg_str:
msg_if = data.setdefault('sentry.interfaces.Message',
{'message': msg_str})
if msg_if.get('message') != msg_str:
msg_if.setdefault('formatted', msg_str)
# Validate main event body and tags against schema
is_valid, event_errors = validate_and_default_interface(data, 'event')
errors.extend(event_errors)
if 'tags' in data:
is_valid, tag_errors = validate_and_default_interface(data['tags'],
'tags',
name='tags')
errors.extend(tag_errors)
# Validate interfaces
for k in list(iter(data)):
if k in CLIENT_RESERVED_ATTRS:
continue
value = data.pop(k)
if not value:
self.logger.debug('Ignored empty interface value: %s', k)
continue
try:
interface = get_interface(k)
except ValueError:
self.logger.debug('Ignored unknown attribute: %s', k)
errors.append({
'type': EventError.INVALID_ATTRIBUTE,
'name': k
})
continue
try:
inst = interface.to_python(value)
data[inst.get_path()] = inst.to_json()
except Exception as e:
log = self.logger.debug if isinstance(
e, InterfaceValidationError) else self.logger.error
log('Discarded invalid value for interface: %s (%r)',
k,
value,
exc_info=True)
errors.append({
'type': EventError.INVALID_DATA,
'name': k,
'value': value
})
level = data.get('level') or DEFAULT_LOG_LEVEL
if isinstance(level, int) or (isinstance(level, six.string_types)
and level.isdigit()):
level = LOG_LEVELS.get(int(level), DEFAULT_LOG_LEVEL)
data['level'] = LOG_LEVELS_MAP.get(level,
LOG_LEVELS_MAP[DEFAULT_LOG_LEVEL])
if data.get('dist') and not data.get('release'):
data['dist'] = None
timestamp = data.get('timestamp')
if not timestamp:
timestamp = timezone.now()
# TODO (alex) can this all be replaced by utcnow?
# it looks like the only time that this would even be hit is when timestamp
# is not defined, as the earlier process_timestamp already converts existing
# timestamps to floats.
if isinstance(timestamp, datetime):
# We must convert date to local time so Django doesn't mess it up
# based on TIME_ZONE
if settings.TIME_ZONE:
if not timezone.is_aware(timestamp):
timestamp = timestamp.replace(tzinfo=timezone.utc)
elif timezone.is_aware(timestamp):
timestamp = timestamp.replace(tzinfo=None)
timestamp = float(timestamp.strftime('%s'))
data['timestamp'] = timestamp
data['received'] = float(timezone.now().strftime('%s'))
data.setdefault('culprit', None)
data.setdefault('transaction', None)
data.setdefault('server_name', None)
data.setdefault('site', None)
data.setdefault('checksum', None)
data.setdefault('fingerprint', None)
data.setdefault('platform', None)
data.setdefault('dist', None)
data.setdefault('environment', None)
data.setdefault('extra', {})
data.setdefault('tags', [])
# Fix case where legacy apps pass 'environment' as a tag
# instead of a top level key.
# TODO (alex) save() just reinserts the environment into the tags
if not data.get('environment'):
tagsdict = dict(data['tags'])
if 'environment' in tagsdict:
data['environment'] = tagsdict['environment']
del tagsdict['environment']
data['tags'] = tagsdict.items()
# the SDKs currently do not describe event types, and we must infer
# them from available attributes
data['type'] = eventtypes.infer(data).key
data['version'] = self.version
exception = data.get('sentry.interfaces.Exception')
stacktrace = data.get('sentry.interfaces.Stacktrace')
if exception and len(exception['values']) == 1 and stacktrace:
exception['values'][0]['stacktrace'] = stacktrace
del data['sentry.interfaces.Stacktrace']
if 'sentry.interfaces.Http' in data:
try:
ip_address = validate_ip(
data['sentry.interfaces.Http'].get('env',
{}).get('REMOTE_ADDR'),
required=False,
)
if ip_address:
data.setdefault('sentry.interfaces.User',
{}).setdefault('ip_address', ip_address)
except ValueError:
pass
# Trim values
logger = data.get('logger', DEFAULT_LOGGER_NAME)
data['logger'] = trim(logger.strip(), 64)
trim_dict(data['extra'],
max_size=settings.SENTRY_MAX_EXTRA_VARIABLE_SIZE)
if data['culprit']:
data['culprit'] = trim(data['culprit'], MAX_CULPRIT_LENGTH)
if data['transaction']:
data['transaction'] = trim(data['transaction'], MAX_CULPRIT_LENGTH)
data['errors'] = errors
return data
def to_python(cls, data):
is_valid, errors = validate_and_default_interface(data, cls.path)
if not is_valid:
raise InterfaceValidationError("Invalid interface data")
kwargs = {}
if data.get('method'):
method = data['method'].upper()
# Optimize for the common path here, where it's a GET/POST, falling
# back to a regular expresion test
if method not in ('GET', 'POST') and not http_method_re.match(method):
raise InterfaceValidationError("Invalid value for 'method'")
kwargs['method'] = method
else:
kwargs['method'] = None
if data.get('url', None):
scheme, netloc, path, query_bit, fragment_bit = urlsplit(data['url'])
else:
scheme = netloc = path = query_bit = fragment_bit = None
query_string = data.get('query_string') or query_bit
if query_string:
# if querystring was a dict, convert it to a string
if isinstance(query_string, dict):
query_string = urlencode(
[(to_bytes(k), to_bytes(v)) for k, v in query_string.items()]
)
else:
if query_string[0] == '?':
# remove '?' prefix
query_string = query_string[1:]
kwargs['query_string'] = trim(query_string, 4096)
else:
kwargs['query_string'] = ''
fragment = data.get('fragment') or fragment_bit
cookies = data.get('cookies')
# if cookies were [also] included in headers we
# strip them out
headers = data.get('headers')
if headers:
headers, cookie_header = format_headers(headers)
if not cookies and cookie_header:
cookies = cookie_header
else:
headers = ()
# We prefer the body to be a string, since we can then attempt to parse it
# as JSON OR decode it as a URL encoded query string, without relying on
# the correct content type header being passed.
body = data.get('data')
content_type = next((v for k, v in headers if k == 'Content-Type'), None)
# Remove content type parameters
if content_type is not None:
content_type = content_type.partition(';')[0].rstrip()
# We process request data once during ingestion and again when
# requesting the http interface over the API. Avoid overwriting
# decoding the body again.
inferred_content_type = data.get('inferred_content_type', content_type)
if 'inferred_content_type' not in data and not isinstance(body, dict):
body, inferred_content_type = heuristic_decode(body, content_type)
if body:
body = trim(body, settings.SENTRY_MAX_HTTP_BODY_SIZE)
env = data.get('env', {})
# TODO (alex) This could also be accomplished with schema (with formats)
if 'REMOTE_ADDR' in env:
try:
validate_ip(env['REMOTE_ADDR'], required=False)
except ValueError:
del env['REMOTE_ADDR']
kwargs['inferred_content_type'] = inferred_content_type
kwargs['cookies'] = trim_pairs(format_cookies(cookies))
kwargs['env'] = trim_dict(env)
kwargs['headers'] = trim_pairs(headers)
kwargs['data'] = fix_broken_encoding(body)
kwargs['url'] = urlunsplit((scheme, netloc, path, '', ''))
kwargs['fragment'] = trim(fragment, 1024)
return cls(**kwargs)
def to_python(cls, data, rust_renormalized=RUST_RENORMALIZED_DEFAULT):
if rust_renormalized:
for key in (
'type',
'synthetic',
'description',
'help_link',
'handled',
'data',
'meta',
):
data.setdefault(key, None)
return cls(**data)
data = upgrade_legacy_mechanism(data)
is_valid, errors = validate_and_default_interface(data, cls.path)
if not is_valid:
raise InterfaceValidationError("Invalid mechanism")
if not data.get('type'):
raise InterfaceValidationError("No 'type' present")
mechanism_meta = data.get('meta') or {}
mach_exception = mechanism_meta.get('mach_exception')
if mach_exception is not None:
mach_exception = prune_empty_keys({
'exception': mach_exception['exception'],
'code': mach_exception['code'],
'subcode': mach_exception['subcode'],
'name': mach_exception.get('name'),
})
signal = mechanism_meta.get('signal')
if signal is not None:
signal = prune_empty_keys({
'number': signal['number'],
'code': signal.get('code'),
'name': signal.get('name'),
'code_name': signal.get('code_name'),
})
errno = mechanism_meta.get('errno')
if errno is not None:
errno = prune_empty_keys({
'number': errno['number'],
'name': errno.get('name'),
})
kwargs = {
'type': trim(data['type'], 128),
'synthetic': data.get('synthetic'),
'description': trim(data.get('description'), 1024),
'help_link': trim(data.get('help_link'), 1024),
'handled': data.get('handled'),
'data': trim(data.get('data'), 4096),
'meta': {
'errno': errno,
'mach_exception': mach_exception,
'signal': signal,
},
}
return cls(**kwargs)
def to_python(cls, data):
is_valid, errors = validate_and_default_interface(data, cls.path)
if not is_valid:
raise InterfaceValidationError("Invalid interface data")
return cls(**data)
