1717import threading
1818from collections import OrderedDict
1919from collections .abc import MutableMapping
20- from typing import MutableSequence , Optional , Sequence
20+ from typing import MutableSequence , Optional , Sequence , Union
2121
2222from opentelemetry .util import types
2323
24- _VALID_ATTR_VALUE_TYPES = (bool , str , int , float )
24+ # bytes are accepted as a user supplied value for attributes but
25+ # decoded to strings internally.
26+ _VALID_ATTR_VALUE_TYPES = (bool , str , bytes , int , float )
2527
2628
2729_logger = logging .getLogger (__name__ )
2830
2931
30- def _is_valid_attribute_value (value : types .AttributeValue ) -> bool :
31- """Checks if attribute value is valid.
32+ def _clean_attribute (
33+ key : str , value : types .AttributeValue , max_len : Optional [int ]
34+ ) -> Optional [types .AttributeValue ]:
35+ """Checks if attribute value is valid and cleans it if required.
3236
3337 An attribute value is valid if it is either:
3438 - A primitive type: string, boolean, double precision floating
3539 point (IEEE 754-1985) or integer.
3640 - An array of primitive type values. The array MUST be homogeneous,
3741 i.e. it MUST NOT contain values of different types.
38- """
3942
40- if isinstance (value , _VALID_ATTR_VALUE_TYPES ):
41- return True
43+ An attribute needs cleansing if:
44+ - Its length is greater than the maximum allowed length.
45+ - It needs to be encoded/decoded e.g, bytes to strings.
46+ """
4247
43- if isinstance (value , Sequence ):
48+ if key is None or key is "" :
49+ _logger .warning ("invalid key `%s` (empty or null)" , key )
50+ return None
4451
52+ if isinstance (value , _VALID_ATTR_VALUE_TYPES ):
53+ return _clean_attribute_value (value , max_len )
54+ elif isinstance (value , Sequence ):
4555 sequence_first_valid_type = None
56+ cleaned_seq = []
57+
4658 for element in value :
59+ # None is considered valid in any sequence
60+ if element is None :
61+ cleaned_seq .append (element )
62+
63+ element = _clean_attribute_value (element , max_len )
4764 if element is None :
4865 continue
66+
4967 element_type = type (element )
68+ # Reject attribute value if sequence contains a value with an incompatible type.
5069 if element_type not in _VALID_ATTR_VALUE_TYPES :
5170 _logger .warning (
5271 "Invalid type %s in attribute value sequence. Expected one of "
@@ -57,56 +76,51 @@ def _is_valid_attribute_value(value: types.AttributeValue) -> bool:
5776 for valid_type in _VALID_ATTR_VALUE_TYPES
5877 ],
5978 )
60- return False
79+ return None
80+
6181 # The type of the sequence must be homogeneous. The first non-None
6282 # element determines the type of the sequence
6383 if sequence_first_valid_type is None :
6484 sequence_first_valid_type = element_type
65- elif not isinstance (element , sequence_first_valid_type ):
85+ # use equality instead of isinstance as isinstance(True, int) evaluates to True
86+ elif element_type != sequence_first_valid_type :
6687 _logger .warning (
6788 "Mixed types %s and %s in attribute value sequence" ,
6889 sequence_first_valid_type .__name__ ,
6990 type (element ).__name__ ,
7091 )
71- return False
72- return True
92+ return None
93+
94+ if element is not None :
95+ cleaned_seq .append (element )
96+
97+ # Freeze mutable sequences defensively
98+ return tuple (cleaned_seq )
7399
74100 _logger .warning (
75101 "Invalid type %s for attribute value. Expected one of %s or a "
76102 "sequence of those types" ,
77103 type (value ).__name__ ,
78104 [valid_type .__name__ for valid_type in _VALID_ATTR_VALUE_TYPES ],
79105 )
80- return False
81-
82106
83- def _filter_attributes (attributes : types .Attributes ) -> None :
84- """Applies attribute validation rules and drops (key, value) pairs
85- that doesn't adhere to attributes specification.
86107
87- https://github.com/open-telemetry/opentelemetry-specification/blob/main/specification/common/common.md#attributes.
88- """
89- if attributes :
90- for attr_key , attr_value in list (attributes .items ()):
91- if not attr_key :
92- _logger .warning ("invalid key `%s` (empty or null)" , attr_key )
93- attributes .pop (attr_key )
94- continue
108+ def _clean_attribute_value (
109+ value : types .AttributeValue , limit : Optional [int ]
110+ ) -> Union [types .AttributeValue , None ]:
111+ if value is None :
112+ return
95113
96- if _is_valid_attribute_value (attr_value ):
97- if isinstance (attr_value , MutableSequence ):
98- attributes [attr_key ] = tuple (attr_value )
99- if isinstance (attr_value , bytes ):
100- try :
101- attributes [attr_key ] = attr_value .decode ()
102- except ValueError :
103- attributes .pop (attr_key )
104- _logger .warning ("Byte attribute could not be decoded." )
105- else :
106- attributes .pop (attr_key )
114+ if isinstance (value , bytes ):
115+ try :
116+ value = value .decode ()
117+ except ValueError :
118+ _logger .warning ("Byte attribute could not be decoded." )
119+ return None
107120
108-
109- _DEFAULT_LIMIT = 128
121+ if limit is not None and isinstance (value , str ):
122+ value = value [:limit ]
123+ return value
110124
111125
112126class BoundedAttributes (MutableMapping ):
@@ -118,9 +132,10 @@ class BoundedAttributes(MutableMapping):
118132
119133 def __init__ (
120134 self ,
121- maxlen : Optional [int ] = _DEFAULT_LIMIT ,
135+ maxlen : Optional [int ] = None ,
122136 attributes : types .Attributes = None ,
123137 immutable : bool = True ,
138+ max_value_len : Optional [int ] = None ,
124139 ):
125140 if maxlen is not None :
126141 if not isinstance (maxlen , int ) or maxlen < 0 :
@@ -129,10 +144,10 @@ def __init__(
129144 )
130145 self .maxlen = maxlen
131146 self .dropped = 0
147+ self .max_value_len = max_value_len
132148 self ._dict = OrderedDict () # type: OrderedDict
133149 self ._lock = threading .Lock () # type: threading.Lock
134150 if attributes :
135- _filter_attributes (attributes )
136151 for key , value in attributes .items ():
137152 self [key ] = value
138153 self ._immutable = immutable
@@ -158,7 +173,10 @@ def __setitem__(self, key, value):
158173 elif self .maxlen is not None and len (self ._dict ) == self .maxlen :
159174 del self ._dict [next (iter (self ._dict .keys ()))]
160175 self .dropped += 1
161- self ._dict [key ] = value
176+
177+ value = _clean_attribute (key , value , self .max_value_len )
178+ if value is not None :
179+ self ._dict [key ] = value
162180
163181 def __delitem__ (self , key ):
164182 if getattr (self , "_immutable" , False ):
0 commit comments