summaryrefslogtreecommitdiffstats
path: root/lib/python/qmk/c_parse.py
blob: 67e196f0eaa564e6fdf877170e56d2060509e52a (plain)
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
165
166
167
168
169
170
171
172
173
174
175
"""Functions for working with config.h files.
"""
from pathlib import Path
import re

from milc import cli

from qmk.comment_remover import comment_remover

default_key_entry = {'x': -1, 'y': 0, 'w': 1}
single_comment_regex = re.compile(r' */[/*].*$')
multi_comment_regex = re.compile(r'/\*(.|\n)*\*/', re.MULTILINE)


def strip_line_comment(string):
    """Removes comments from a single line string.
    """
    return single_comment_regex.sub('', string)


def strip_multiline_comment(string):
    """Removes comments from a single line string.
    """
    return multi_comment_regex.sub('', string)


def c_source_files(dir_names):
    """Returns a list of all *.c, *.h, and *.cpp files for a given list of directories

    Args:

        dir_names
            List of directories relative to `qmk_firmware`.
    """
    files = []
    for dir in dir_names:
        files.extend(file for file in Path(dir).glob('**/*') if file.suffix in ['.c', '.h', '.cpp'])
    return files


def find_layouts(file):
    """Returns list of parsed LAYOUT preprocessor macros found in the supplied include file.
    """
    file = Path(file)
    aliases = {}  # Populated with all `#define`s that aren't functions
    parsed_layouts = {}

    # Search the file for LAYOUT macros and aliases
    file_contents = file.read_text()
    file_contents = comment_remover(file_contents)
    file_contents = file_contents.replace('\\\n', '')

    for line in file_contents.split('\n'):
        if line.startswith('#define') and '(' in line and 'LAYOUT' in line:
            # We've found a LAYOUT macro
            macro_name, layout, matrix = _parse_layout_macro(line.strip())

            # Reject bad macro names
            if macro_name.startswith('LAYOUT_kc') or not macro_name.startswith('LAYOUT'):
                continue

            # Parse the matrix data
            matrix_locations = _parse_matrix_locations(matrix, file, macro_name)

            # Parse the layout entries into a basic structure
            default_key_entry['x'] = -1  # Set to -1 so _default_key(key) will increment it to 0
            layout = layout.strip()
            parsed_layout = [_default_key(key) for key in layout.split(',')]

            for key in parsed_layout:
                if key['label'] in matrix_locations:
                    key['matrix'] = matrix_locations[key['label']]

            parsed_layouts[macro_name] = {
                'key_count': len(parsed_layout),
                'layout': parsed_layout,
                'filename': str(file),
            }

        elif '#define' in line:
            # Attempt to extract a new layout alias
            try:
                _, pp_macro_name, pp_macro_text = line.strip().split(' ', 2)
                aliases[pp_macro_name] = pp_macro_text
            except ValueError:
                continue

    # Populate our aliases
    for alias, text in aliases.items():
        if text in parsed_layouts and 'KEYMAP' not in alias:
            parsed_layouts[alias] = parsed_layouts[text]

    return parsed_layouts


def parse_config_h_file(config_h_file, config_h=None):
    """Extract defines from a config.h file.
    """
    if not config_h:
        config_h = {}

    config_h_file = Path(config_h_file)

    if config_h_file.exists():
        config_h_text = config_h_file.read_text()
        config_h_text = config_h_text.replace('\\\n', '')
        config_h_text = strip_multiline_comment(config_h_text)

        for linenum, line in enumerate(config_h_text.split('\n')):
            line = strip_line_comment(line).strip()

            if not line:
                continue

            line = line.split()

            if line[0] == '#define':
                if len(line) == 1:
                    cli.log.error('%s: Incomplete #define! On or around line %s' % (config_h_file, linenum))
                elif len(line) == 2:
                    config_h[line[1]] = True
                else:
                    config_h[line[1]] = ' '.join(line[2:])

            elif line[0] == '#undef':
                if len(line) == 2:
                    if line[1] in config_h:
                        if config_h[line[1]] is True:
                            del config_h[line[1]]
                        else:
                            config_h[line[1]] = False
                else:
                    cli.log.error('%s: Incomplete #undef! On or around line %s' % (config_h_file, linenum))

    return config_h


def _default_key(label=None):
    """Increment x and return a copy of the default_key_entry.
    """
    default_key_entry['x'] += 1
    new_key = default_key_entry.copy()

    if label:
        new_key['label'] = label

    return new_key


def _parse_layout_macro(layout_macro):
    """Split the LAYOUT macro into its constituent parts
    """
    layout_macro = layout_macro.replace('\\', '').replace(' ', '').replace('\t', '').replace('#define', '')
    macro_name, layout = layout_macro.split('(', 1)
    layout, matrix = layout.split(')', 1)

    return macro_name, layout, matrix


def _parse_matrix_locations(matrix, file, macro_name):
    """Parse raw matrix data into a dictionary keyed by the LAYOUT identifier.
    """
    matrix_locations = {}

    for row_num, row in enumerate(matrix.split('},{')):
        if row.startswith('LAYOUT'):
            cli.log.error('%s: %s: Nested layout macro detected. Matrix data not available!', file, macro_name)
            break

        row = row.replace('{', '').replace('}', '')
        for col_num, identifier in enumerate(row.split(',')):
            if identifier != 'KC_NO':
                matrix_locations[identifier] = [row_num, col_num]

    return matrix_locations