def __init__(self, query, params, allow_minute_resolution=False):
self.query = query.get("query", "")
self.raw_fields = raw_fields = query.getlist("field", [])
self.raw_groupby = raw_groupby = query.getlist("groupBy", [])
if len(raw_fields) == 0:
raise InvalidField('Request is missing a "field"')
self.fields = {}
for key in raw_fields:
if key not in COLUMN_MAP:
raise InvalidField(f'Invalid field: "{key}"')
self.fields[key] = COLUMN_MAP[key]
self.groupby = []
for key in raw_groupby:
if key not in GROUPBY_MAP:
raise InvalidField(f'Invalid groupBy: "{key}"')
self.groupby.append(GROUPBY_MAP[key])
start, end, rollup = get_constrained_date_range(
query, allow_minute_resolution)
self.rollup = rollup
self.start = start
self.end = end
self.params = params
query_columns = set()
for field in self.fields.values():
query_columns.update(field.get_snuba_columns(raw_groupby))
for groupby in self.groupby:
query_columns.update(groupby.get_snuba_columns())
self.query_columns = list(query_columns)
query_groupby = set()
for groupby in self.groupby:
query_groupby.update(groupby.get_snuba_groupby())
self.query_groupby = list(query_groupby)
# the `params` are:
# project_id, organization_id, environment;
# also: start, end; but we got those ourselves.
snuba_filter = get_filter(self.query, params)
# this makes sure that literals in complex queries are properly quoted,
# and unknown fields are raised as errors
conditions = [
resolve_condition(c, resolve_column)
for c in snuba_filter.conditions
]
self.aggregations = snuba_filter.aggregations
self.conditions = conditions
self.filter_keys = snuba_filter.filter_keys
def resolve_discover_aliases(snuba_filter, function_translations=None):
"""
Resolve the public schema aliases to the discover dataset.
Returns a copy of the input structure, and includes a
`translated_columns` key containing the selected fields that need to
be renamed in the result set.
"""
resolved = snuba_filter.clone()
translated_columns = {}
derived_columns = set()
if function_translations:
for snuba_name, sentry_name in six.iteritems(function_translations):
derived_columns.add(snuba_name)
translated_columns[snuba_name] = sentry_name
selected_columns = resolved.selected_columns
if selected_columns:
for (idx, col) in enumerate(selected_columns):
if isinstance(col, (list, tuple)):
resolve_complex_column(col)
else:
name = resolve_column(col)
selected_columns[idx] = name
translated_columns[name] = col
resolved.selected_columns = selected_columns
groupby = resolved.groupby
if groupby:
for (idx, col) in enumerate(groupby):
name = col
if isinstance(col, (list, tuple)):
if len(col) == 3:
name = col[2]
elif col not in derived_columns:
name = resolve_column(col)
groupby[idx] = name
resolved.groupby = groupby
aggregations = resolved.aggregations
for aggregation in aggregations or []:
derived_columns.add(aggregation[2])
if isinstance(aggregation[1], six.string_types):
aggregation[1] = resolve_column(aggregation[1])
elif isinstance(aggregation[1], (set, tuple, list)):
aggregation[1] = [resolve_column(col) for col in aggregation[1]]
resolved.aggregations = aggregations
conditions = resolved.conditions
if conditions:
for (i, condition) in enumerate(conditions):
replacement = resolve_condition(condition, resolve_column)
conditions[i] = replacement
resolved.conditions = [c for c in conditions if c]
orderby = resolved.orderby
if orderby:
orderby = orderby if isinstance(orderby, (list, tuple)) else [orderby]
resolved_orderby = []
for field_with_order in orderby:
field = field_with_order.lstrip("-")
resolved_orderby.append(u"{}{}".format(
"-" if field_with_order.startswith("-") else "",
field if field in derived_columns else resolve_column(field),
))
resolved.orderby = resolved_orderby
return resolved, translated_columns
def resolve_discover_aliases(snuba_args):
"""
Resolve the public schema aliases to the discover dataset.
Returns a copy of the input structure, and includes a
`translated_columns` key containing the selected fields that need to
be renamed in the result set.
"""
resolved = deepcopy(snuba_args)
translated_columns = {}
derived_columns = set()
selected_columns = resolved.get("selected_columns")
if selected_columns:
for (idx, col) in enumerate(selected_columns):
if isinstance(col, (list, tuple)):
raise ValueError(
"discover selected_columns should only be str. got %s" %
col)
name = resolve_column(col)
selected_columns[idx] = name
translated_columns[name] = col
resolved["selected_columns"] = selected_columns
groupby = resolved.get("groupby")
if groupby:
for (idx, col) in enumerate(groupby):
name = col
if col not in derived_columns:
name = resolve_column(col)
groupby[idx] = name
resolved["groupby"] = groupby
aggregations = resolved.get("aggregations")
for aggregation in aggregations or []:
derived_columns.add(aggregation[2])
if isinstance(aggregation[1], six.string_types):
aggregation[1] = resolve_column(aggregation[1])
elif isinstance(aggregation[1], (set, tuple, list)):
aggregation[1] = [resolve_column(col) for col in aggregation[1]]
resolved["aggregations"] = aggregations
conditions = resolved.get("conditions")
if conditions:
for (i, condition) in enumerate(conditions):
replacement = resolve_condition(condition, resolve_column)
conditions[i] = replacement
resolved["conditions"] = list(filter(None, conditions))
# TODO add support for extracting having conditions.
orderby = resolved.get("orderby")
if orderby:
orderby = orderby if isinstance(orderby, (list, tuple)) else [orderby]
resolved_orderby = []
for field_with_order in orderby:
field = field_with_order.lstrip("-")
resolved_orderby.append(u"{}{}".format(
"-" if field_with_order.startswith("-") else "",
field if field in derived_columns else resolve_column(field),
))
resolved["orderby"] = resolved_orderby
return resolved, translated_columns
