Datasets:

ArXiv:
File size: 6,768 Bytes
18e5876
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
import re


def split_words(s):
    # Split PascalCase or camelCase
    s = re.sub("([A-Z][a-z]+)", r" \1", re.sub("([A-Z]+)", r" \1", s)).strip()
    # Split snake_case or kebab-case
    s = re.sub("[_-]", " ", s)
    # Split numbers attached to strings
    s = re.sub("([a-zA-Z])(\d)", r"\1 \2", s)
    s = re.sub("(\d)([a-zA-Z])", r"\1 \2", s)
    # Split the string into words based on spaces
    words = s.split()
    return words


def is_camel_case(s):
    # The string must start with an uppercase letter, followed by zero or more sequences of an uppercase letter followed by zero or more lowercase letters.
    return re.match(r"^[A-Z]+([a-z0-9]*[A-Z]*[a-z0-9]*)*$", s) is not None


def is_snake_case(s):
    # The string must start with a lowercase letter, followed by zero or more sequences of an underscore followed by one or more lowercase letters.
    return re.match(r"^[a-z0-9]+(_[a-z0-9]+)*$", s) is not None


def camel_to_snake_case(s):
    # Add an underscore before every uppercase letter that is followed by a lowercase letter or digit and not preceded by an underscore, a hyphen or an uppercase letter
    s = re.sub("(?<=[^A-Z_-])([A-Z])", r"_\1", s)

    # Ensure there's an underscore before any uppercase letter that's followed by a lowercase letter or digit and comes after a sequence of uppercase letters
    s = re.sub("([A-Z]+)([A-Z][a-z0-9])", r"\1_\2", s)

    s = s.lower()
    return s


import shutil


def print_dict(d, indent=0, indent_delta=4, max_chars=None):
    max_chars = max_chars or shutil.get_terminal_size()[0] - 10  # Get terminal size if max_chars not set
    indent_str = " " * indent
    indent_delta_str = " " * indent_delta

    for key, value in d.items():
        if isinstance(value, dict):
            print(f"{indent_str}{key}:")
            print_dict(value, indent=indent + indent_delta, max_chars=max_chars)
        else:
            # Value is not a dict, print as a string
            str_value = str(value)

            line_width = max_chars - indent
            # Split value by newline characters and handle each line separately
            lines = str_value.split("\n")
            print(f"{indent_str}{key}:")
            for line in lines:
                if len(line) + len(indent_str) + indent_delta > line_width:
                    # Split long lines into multiple lines
                    print(f"{indent_str}{indent_delta_str}{line[:line_width]}")
                    for i in range(line_width, len(line), line_width):
                        print(f"{indent_str}{indent_delta_str}{line[i:i+line_width]}")
                else:
                    print(f"{indent_str}{indent_delta_str}{line}")
                key = ""  # Empty the key for lines after the first one


def nested_tuple_to_string(nested_tuple: tuple) -> str:
    result = []
    for item in nested_tuple:
        if isinstance(item, tuple):
            result.append(nested_tuple_to_string(item))
        else:
            result.append(str(item))
    return "_".join(result)


if __name__ == "__main__":
    # Define test cases
    test_cases = [
        ("example1", ["example", "1"]),
        ("exampleOne", ["example", "One"]),
        ("123example456", ["123", "example", "456"]),
        ("happyDay", ["happy", "Day"]),
        ("thisIsATest", ["this", "Is", "A", "Test"]),
        ("TestAI2023", ["Test", "AI", "2023"]),
        ("stringWith1Number", ["string", "With", "1", "Number"]),
        ("camelCaseExample", ["camel", "Case", "Example"]),
        ("snake_case_example", ["snake", "case", "example"]),
        ("snake_case2example3", ["snake", "case", "2", "example", "3"]),
        ("kebab-case-example", ["kebab", "case", "example"]),
        ("kebab-case2example3", ["kebab", "case", "2", "example", "3"]),
        ("PascalCaseExample", ["Pascal", "Case", "Example"]),
        ("Title Case Example", ["Title", "Case", "Example"]),
        ("Mixed1Example_case", ["Mixed", "1", "Example", "case"]),
        ("Mixed2Example-case", ["Mixed", "2", "Example", "case"]),
        ("Mixed3_Example-case", ["Mixed", "3", "Example", "case"]),
        ("UPPERCASEEXAMPLE", ["UPPERCASEEXAMPLE"]),
        ("lowercaseexample", ["lowercaseexample"]),
        ("mixedUPanddown", ["mixed", "U", "Panddown"]),
    ]

    # Loop through test cases
    for i, (input_string, expected_output) in enumerate(test_cases, 1):
        # Apply function and check result
        if split_words(input_string) != expected_output:
            print(f"Failed on example {i}: {input_string}")
            print(f"Expected: {expected_output}, but got: {split_words(input_string)}\n")

    is_camel_case_test_cases = [
        ("isCamelCase", False),
        ("notCamelCase", False),
        ("camelCase", False),
        ("Notcamelcase", True),
        ("camel_Case", False),
        ("camelCase123", False),
        ("camelcase", False),
        ("CAMELCASE", True),
        ("camel-case", False),
        ("HFLoader", True),
    ]

    for input_string, expected_output in is_camel_case_test_cases:
        if is_camel_case(input_string) != expected_output:
            print(f"Failed on is_camel_case: {input_string}")
            print(f"Expected: {expected_output}, but got: {is_camel_case(input_string)}\n")

    is_snake_case_test_cases = [
        ("is_snake_case", True),
        ("Not_snake_case", False),
        ("snake_case", True),
        ("snake_Case", False),
        ("Snakecase", False),
        ("snake-case", False),
        ("snake_case123", True),
        ("123snake_case", True),
        ("snakecase", True),
    ]

    for input_string, expected_output in is_snake_case_test_cases:
        if is_snake_case(input_string) != expected_output:
            print(f"Failed on is_snake_case: {input_string}")
            print(f"Expected: {expected_output}, but got: {is_snake_case(input_string)}\n")

    camel_to_snake_case_test_cases = [
        ("camelToSnake", "camel_to_snake"),
        ("CamelToSnake", "camel_to_snake"),
        ("CamelToSnakeCase", "camel_to_snake_case"),
        ("camelToSnakeCase123", "camel_to_snake_case123"),
        ("123CamelToSnakeCase", "123_camel_to_snake_case"),
        ("camelTo_Snake_Case", "camel_to__snake__case"),
        ("camelTo-Snake-Case", "camel_to-_snake-_case"),
        ("camelToSnakeCASE", "camel_to_snake_case"),
        ("CAMELToSnakeCase", "camel_to_snake_case"),
        ("camelToSNAKECase", "camel_to_snake_case"),
        ("HFLoader", "hf_loader"),
    ]

    for input_string, expected_output in camel_to_snake_case_test_cases:
        if camel_to_snake_case(input_string) != expected_output:
            print(f"Failed on camel_to_snake_case: {input_string}")
            print(f"Expected: {expected_output}, but got: {camel_to_snake_case(input_string)}\n")