import math import json from jmespath import exceptions from jmespath.compat import string_type as STRING_TYPE from jmespath.compat import get_methods # python types -> jmespath types TYPES_MAP = { 'bool': 'boolean', 'list': 'array', 'dict': 'object', 'NoneType': 'null', 'unicode': 'string', 'str': 'string', 'float': 'number', 'int': 'number', 'long': 'number', 'OrderedDict': 'object', '_Projection': 'array', '_Expression': 'expref', } # jmespath types -> python types REVERSE_TYPES_MAP = { 'boolean': ('bool',), 'array': ('list', '_Projection'), 'object': ('dict', 'OrderedDict',), 'null': ('NoneType',), 'string': ('unicode', 'str'), 'number': ('float', 'int', 'long'), 'expref': ('_Expression',), } def signature(*arguments): def _record_signature(func): func.signature = arguments return func return _record_signature class FunctionRegistry(type): def __init__(cls, name, bases, attrs): cls._populate_function_table() super(FunctionRegistry, cls).__init__(name, bases, attrs) def _populate_function_table(cls): function_table = {} # Any method with a @signature decorator that also # starts with "_func_" is registered as a function. # _func_max_by -> max_by function. for name, method in get_methods(cls): if not name.startswith('_func_'): continue signature = getattr(method, 'signature', None) if signature is not None: function_table[name[6:]] = { 'function': method, 'signature': signature, } cls.FUNCTION_TABLE = function_table class Functions(metaclass=FunctionRegistry): FUNCTION_TABLE = { } def call_function(self, function_name, resolved_args): try: spec = self.FUNCTION_TABLE[function_name] except KeyError: raise exceptions.UnknownFunctionError( "Unknown function: %s()" % function_name) function = spec['function'] signature = spec['signature'] self._validate_arguments(resolved_args, signature, function_name) return function(self, *resolved_args) def _validate_arguments(self, args, signature, function_name): if signature and signature[-1].get('variadic'): if len(args) < len(signature): raise exceptions.VariadictArityError( len(signature), len(args), function_name) elif len(args) != len(signature): raise exceptions.ArityError( len(signature), len(args), function_name) return self._type_check(args, signature, function_name) def _type_check(self, actual, signature, function_name): for i in range(len(signature)): allowed_types = signature[i]['types'] if allowed_types: self._type_check_single(actual[i], allowed_types, function_name) def _type_check_single(self, current, types, function_name): # Type checking involves checking the top level type, # and in the case of arrays, potentially checking the types # of each element. allowed_types, allowed_subtypes = self._get_allowed_pytypes(types) # We're not using isinstance() on purpose. # The type model for jmespath does not map # 1-1 with python types (booleans are considered # integers in python for example). actual_typename = type(current).__name__ if actual_typename not in allowed_types: raise exceptions.JMESPathTypeError( function_name, current, self._convert_to_jmespath_type(actual_typename), types) # If we're dealing with a list type, we can have # additional restrictions on the type of the list # elements (for example a function can require a # list of numbers or a list of strings). # Arrays are the only types that can have subtypes. if allowed_subtypes: self._subtype_check(current, allowed_subtypes, types, function_name) def _get_allowed_pytypes(self, types): allowed_types = [] allowed_subtypes = [] for t in types: type_ = t.split('-', 1) if len(type_) == 2: type_, subtype = type_ allowed_subtypes.append(REVERSE_TYPES_MAP[subtype]) else: type_ = type_[0] allowed_types.extend(REVERSE_TYPES_MAP[type_]) return allowed_types, allowed_subtypes def _subtype_check(self, current, allowed_subtypes, types, function_name): if len(allowed_subtypes) == 1: # The easy case, we know up front what type # we need to validate. allowed_subtypes = allowed_subtypes[0] for element in current: actual_typename = type(element).__name__ if actual_typename not in allowed_subtypes: raise exceptions.JMESPathTypeError( function_name, element, actual_typename, types) elif len(allowed_subtypes) > 1 and current: # Dynamic type validation. Based on the first # type we see, we validate that the remaining types # match. first = type(current[0]).__name__ for subtypes in allowed_subtypes: if first in subtypes: allowed = subtypes break else: raise exceptions.JMESPathTypeError( function_name, current[0], first, types) for element in current: actual_typename = type(element).__name__ if actual_typename not in allowed: raise exceptions.JMESPathTypeError( function_name, element, actual_typename, types) @signature({'types': ['number']}) def _func_abs(self, arg): return abs(arg) @signature({'types': ['array-number']}) def _func_avg(self, arg): if arg: return sum(arg) / float(len(arg)) else: return None @signature({'types': [], 'variadic': True}) def _func_not_null(self, *arguments): for argument in arguments: if argument is not None: return argument @signature({'types': []}) def _func_to_array(self, arg): if isinstance(arg, list): return arg else: return [arg] @signature({'types': []}) def _func_to_string(self, arg): if isinstance(arg, STRING_TYPE): return arg else: return json.dumps(arg, separators=(',', ':'), default=str) @signature({'types': []}) def _func_to_number(self, arg): if isinstance(arg, (list, dict, bool)): return None elif arg is None: return None elif isinstance(arg, (int, float)): return arg else: try: return int(arg) except ValueError: try: return float(arg) except ValueError: return None @signature({'types': ['array', 'string']}, {'types': []}) def _func_contains(self, subject, search): return search in subject @signature({'types': ['string', 'array', 'object']}) def _func_length(self, arg): return len(arg) @signature({'types': ['string']}, {'types': ['string']}) def _func_ends_with(self, search, suffix): return search.endswith(suffix) @signature({'types': ['string']}, {'types': ['string']}) def _func_starts_with(self, search, suffix): return search.startswith(suffix) @signature({'types': ['array', 'string']}) def _func_reverse(self, arg): if isinstance(arg, STRING_TYPE): return arg[::-1] else: return list(reversed(arg)) @signature({"types": ['number']}) def _func_ceil(self, arg): return math.ceil(arg) @signature({"types": ['number']}) def _func_floor(self, arg): return math.floor(arg) @signature({"types": ['string']}, {"types": ['array-string']}) def _func_join(self, separator, array): return separator.join(array) @signature({'types': ['expref']}, {'types': ['array']}) def _func_map(self, expref, arg): result = [] for element in arg: result.append(expref.visit(expref.expression, element)) return result @signature({"types": ['array-number', 'array-string']}) def _func_max(self, arg): if arg: return max(arg) else: return None @signature({"types": ["object"], "variadic": True}) def _func_merge(self, *arguments): merged = {} for arg in arguments: merged.update(arg) return merged @signature({"types": ['array-number', 'array-string']}) def _func_min(self, arg): if arg: return min(arg) else: return None @signature({"types": ['array-string', 'array-number']}) def _func_sort(self, arg): return list(sorted(arg)) @signature({"types": ['array-number']}) def _func_sum(self, arg): return sum(arg) @signature({"types": ['object']}) def _func_keys(self, arg): # To be consistent with .values() # should we also return the indices of a list? return list(arg.keys()) @signature({"types": ['object']}) def _func_values(self, arg): return list(arg.values()) @signature({'types': []}) def _func_type(self, arg): if isinstance(arg, STRING_TYPE): return "string" elif isinstance(arg, bool): return "boolean" elif isinstance(arg, list): return "array" elif isinstance(arg, dict): return "object" elif isinstance(arg, (float, int)): return "number" elif arg is None: return "null" @signature({'types': ['array']}, {'types': ['expref']}) def _func_sort_by(self, array, expref): if not array: return array # sort_by allows for the expref to be either a number of # a string, so we have some special logic to handle this. # We evaluate the first array element and verify that it's # either a string of a number. We then create a key function # that validates that type, which requires that remaining array # elements resolve to the same type as the first element. required_type = self._convert_to_jmespath_type( type(expref.visit(expref.expression, array[0])).__name__) if required_type not in ['number', 'string']: raise exceptions.JMESPathTypeError( 'sort_by', array[0], required_type, ['string', 'number']) keyfunc = self._create_key_func(expref, [required_type], 'sort_by') return list(sorted(array, key=keyfunc)) @signature({'types': ['array']}, {'types': ['expref']}) def _func_min_by(self, array, expref): keyfunc = self._create_key_func(expref, ['number', 'string'], 'min_by') if array: return min(array, key=keyfunc) else: return None @signature({'types': ['array']}, {'types': ['expref']}) def _func_max_by(self, array, expref): keyfunc = self._create_key_func(expref, ['number', 'string'], 'max_by') if array: return max(array, key=keyfunc) else: return None def _create_key_func(self, expref, allowed_types, function_name): def keyfunc(x): result = expref.visit(expref.expression, x) actual_typename = type(result).__name__ jmespath_type = self._convert_to_jmespath_type(actual_typename) # allowed_types is in term of jmespath types, not python types. if jmespath_type not in allowed_types: raise exceptions.JMESPathTypeError( function_name, result, jmespath_type, allowed_types) return result return keyfunc def _convert_to_jmespath_type(self, pyobject): return TYPES_MAP.get(pyobject, 'unknown')