Coverage for hiperta_stream/utils/substring_substitution_utils.py: 88%
49 statements
« prev ^ index » next coverage.py v7.4.3, created at 2024-07-16 10:16 +0000
1import copy
2import re
3import warnings
6def dict_defined_substring_substitution(d, external_variables=None, variable_identifier="@"):
7 """Replaces @{variable} by variable's value, in all values of `d` if variable is a unique key in `d` or `external_variables`.
9 This is typically used to perform sub-string substitution in a loaded configuration.
10 Keys can not be modified, only values.
12 Notes
13 -----
14 - Substitution only happens inside variable of type string.
15 - Only variables of type string, int or float can be substituted.
16 - Variables which are defined more than once in the dictionary are ignored.
18 Parameters
19 ----------
20 d : dict
21 The bash variable syntax in the string values of `d` will be replaced.
22 external_variables : dict, optional
23 Dictionary of variables not defined in `d`, to use for replacement as well.
24 variable_identifier : str
25 Character marking a sub-string to replace.
26 Examples: '@' to replace @{v} by v's value.
28 Returns
29 -------
30 dict
31 The modified dictionary with bash variable syntax replaced.
33 Examples
34 --------
35 >>> d = {
36 >>> "a": "b",
37 >>> "c": {"d": "@{a}"},
38 >>> "e": "@{a}",
39 >>> "f": "@{t}",
40 >>> }
41 >>> dict_defined_substring_substitution(d, {"t": "tt"})
42 {"a": "b", "c": {"d": "b"}, "e": "b", "f": "tt"}
43 """
45 # helper function to do the substitution
46 # This is an adaptation of os.path.expandvars, but replaces values of keys present in `variables`
47 # and instead of replacing environment variables.
48 # Also doesn't allow @qdf to be replaced by qdf's value. {} are mandatory.
49 # https://github.com/python/cpython/blob/3.11/Lib/posixpath.py
50 if len(variable_identifier) > 1 or not variable_identifier:
51 raise ValueError(f"variable identifier {variable_identifier} is not valid. It must be a single character.")
53 search = re.compile(rf"\{variable_identifier}" r"(\{[^}]*\})", re.ASCII).search
55 def substitute_in_str(string, variables):
56 i = 0
57 while True:
58 m = search(string, i)
59 if not m:
60 break
61 i, j = m.span(0)
62 name = m.group(1)
63 warning_name = name
64 name = name[1:-1] # remove starting {, ending }
65 try:
66 value = variables[name]
67 except KeyError:
68 warnings.warn(
69 "Value {}{} did not match any defined variables, ignoring...".format(
70 variable_identifier, warning_name
71 )
72 )
73 i = j
74 else:
75 tail = string[j:]
76 string = string[:i] + str(value)
77 i = len(string)
78 string += tail
79 return string
81 # Go through the dictionary to get all the variables that are defined only once.
82 def get_all_unique_keys_as_if_global(d, variables, variables_already_defined):
83 for key in d.keys():
84 if isinstance(d[key], dict):
85 get_all_unique_keys_as_if_global(d[key], variables, variables_already_defined)
86 elif isinstance(d[key], (str, int, float)):
87 # variable is defined more than 1 time, remove it from vars list
88 if key in variables_already_defined:
89 if key in variables.keys():
90 variables.pop(key)
91 else:
92 variables[key] = d[key]
93 variables_already_defined.add(key)
95 # Go through the dict to perform the substitution.
96 def recursive_dict_vars_substitution(d, variables):
97 for key in d.keys():
98 if isinstance(d[key], dict):
99 recursive_dict_vars_substitution(d[key], variables)
100 elif isinstance(d[key], str):
101 new_value = substitute_in_str(d[key], variables)
102 # perform substitution until the string remains unchanged,
103 # in case several level of substitution needs doing.
104 while new_value != d[key]:
105 d[key] = new_value
106 new_value = substitute_in_str(d[key], variables)
108 variables = {} if external_variables is None else copy.deepcopy(external_variables)
110 get_all_unique_keys_as_if_global(d, variables, set())
111 recursive_dict_vars_substitution(d, variables)
113 return d