|
10 | 10 |
|
11 | 11 |
|
12 | 12 | KNOWN = {
|
13 |
| - 'a': {'aria-label', 'class', 'href', 'title'}, |
| 13 | + 'a': { |
| 14 | + 'aria-label': None, |
| 15 | + 'class': {'doi', 'dropdown-item'}, |
| 16 | + 'href': None, |
| 17 | + 'title': None |
| 18 | + }, |
14 | 19 | 'blockquote': {},
|
15 |
| - 'body': {'class'}, |
| 20 | + 'body': { |
| 21 | + 'class': {'draft'} |
| 22 | + }, |
16 | 23 | 'br': {},
|
17 | 24 | 'caption': {},
|
18 | 25 | 'cite': {},
|
19 |
| - 'code': {'class'}, |
20 |
| - 'dd': {'class'}, |
21 |
| - 'div': {'class', 'id'}, |
22 |
| - 'dl': {'class'}, |
23 |
| - 'dt': {'class', 'id'}, |
| 26 | + 'code': { |
| 27 | + 'class': { |
| 28 | + 'language-html', |
| 29 | + 'language-js', |
| 30 | + 'language-make', |
| 31 | + 'language-out', |
| 32 | + 'language-py', |
| 33 | + 'language-sh', |
| 34 | + 'language-txt', |
| 35 | + 'language-yml' |
| 36 | + } |
| 37 | + }, |
| 38 | + 'dd': { |
| 39 | + 'class': None |
| 40 | + }, |
| 41 | + 'div': { |
| 42 | + 'class': { |
| 43 | + 'glossary', |
| 44 | + 'callout', |
| 45 | + 'centered', |
| 46 | + 'container', |
| 47 | + 'content', |
| 48 | + 'dropdown', |
| 49 | + 'dropdown-content', |
| 50 | + 'file-link', |
| 51 | + 'terms' |
| 52 | + }, |
| 53 | + 'id': None |
| 54 | + }, |
| 55 | + 'dl': { |
| 56 | + 'class': { |
| 57 | + 'bibliography', |
| 58 | + 'glossary' |
| 59 | + } |
| 60 | + }, |
| 61 | + 'dt': { |
| 62 | + 'class': { |
| 63 | + 'bibliography', |
| 64 | + 'glossary' |
| 65 | + }, |
| 66 | + 'id': None |
| 67 | + }, |
24 | 68 | 'em': {},
|
25 | 69 | 'figcaption': {},
|
26 |
| - 'figure': {'class', 'id', 'slug'}, |
| 70 | + 'figure': { |
| 71 | + 'class': None, |
| 72 | + 'id': None, |
| 73 | + 'slug': None |
| 74 | + }, |
27 | 75 | 'footer': {},
|
28 |
| - 'h1': {'class', 'slug'}, |
29 |
| - 'h2': {'class', 'id'}, |
30 |
| - 'h3': {'id'}, |
| 76 | + 'h1': { |
| 77 | + 'class': { |
| 78 | + 'landing-title', |
| 79 | + 'nochaptertitle', |
| 80 | + 'page-title' |
| 81 | + }, |
| 82 | + 'slug': None |
| 83 | + }, |
| 84 | + 'h2': { |
| 85 | + 'class': { |
| 86 | + 'landing-subtitle' |
| 87 | + }, |
| 88 | + 'id': None |
| 89 | + }, |
| 90 | + 'h3': { |
| 91 | + 'id': None |
| 92 | + }, |
31 | 93 | 'head': {},
|
32 | 94 | 'header': {},
|
33 | 95 | 'hr': {},
|
34 |
| - 'html': {'lang'}, |
35 |
| - 'i': {'aria-hidden', 'class'}, |
36 |
| - 'img': {'alt', 'class', 'src'}, |
| 96 | + 'html': { |
| 97 | + 'lang': { |
| 98 | + 'en' |
| 99 | + } |
| 100 | + }, |
| 101 | + 'i': { |
| 102 | + 'aria-hidden': None, |
| 103 | + 'class': None |
| 104 | + }, |
| 105 | + 'img': { |
| 106 | + 'alt': None, |
| 107 | + 'class': { |
| 108 | + 'iconlink' |
| 109 | + }, |
| 110 | + 'src': None |
| 111 | + }, |
37 | 112 | 'li': {},
|
38 |
| - 'link': {'href', 'rel', 'type'}, |
| 113 | + 'link': { |
| 114 | + 'href': None, |
| 115 | + 'rel': { |
| 116 | + 'icon', |
| 117 | + 'shortcut', |
| 118 | + 'stylesheet', |
| 119 | + 'image/x-icon', |
| 120 | + 'text/css' |
| 121 | + }, |
| 122 | + 'type': None |
| 123 | + }, |
39 | 124 | 'main': {},
|
40 |
| - 'meta': {'charset', 'content', 'name'}, |
41 |
| - 'nav': {'class'}, |
| 125 | + 'meta': { |
| 126 | + 'charset': None, |
| 127 | + 'content': None, |
| 128 | + 'name': None |
| 129 | + }, |
| 130 | + 'nav': { |
| 131 | + 'class': { |
| 132 | + 'nav-main' |
| 133 | + } |
| 134 | + }, |
42 | 135 | 'ol': {},
|
43 |
| - 'p': {'class'}, |
44 |
| - 'pre': {'class', 'title'}, |
45 |
| - 'script': {'async', 'crossorigin', 'src', 'type'}, |
| 136 | + 'p': { |
| 137 | + 'class': { |
| 138 | + 'continue' |
| 139 | + } |
| 140 | + }, |
| 141 | + 'pre': { |
| 142 | + 'class': None, |
| 143 | + 'title': None |
| 144 | + }, |
| 145 | + 'script': { |
| 146 | + 'async': None, |
| 147 | + 'crossorigin': None, |
| 148 | + 'src': None, |
| 149 | + 'type': None |
| 150 | + }, |
46 | 151 | 'small': {},
|
47 |
| - 'span': {'class', 'f', 'g', 'i', 't', 'x'}, |
| 152 | + 'span': { |
| 153 | + 'class': { |
| 154 | + 'copyright', |
| 155 | + 'navtitle', |
| 156 | + 'nowrap' |
| 157 | + }, |
| 158 | + 'f': None, |
| 159 | + 'g': None, |
| 160 | + 'i': None, |
| 161 | + 't': None, |
| 162 | + 'x': None |
| 163 | + }, |
48 | 164 | 'strong': {},
|
49 |
| - 'table': {'class', 'id'}, |
| 165 | + 'table': { |
| 166 | + 'class': { |
| 167 | + 'links' |
| 168 | + }, |
| 169 | + 'id': None |
| 170 | + }, |
50 | 171 | 'tbody': {},
|
51 |
| - 'td': {'style'}, |
52 |
| - 'th': {'style'}, |
| 172 | + 'td': { |
| 173 | + 'style': None |
| 174 | + }, |
| 175 | + 'th': { |
| 176 | + 'style': None |
| 177 | + }, |
53 | 178 | 'thead': {},
|
54 |
| - 'time': {'datetime'}, |
| 179 | + 'time': { |
| 180 | + 'datetime': None |
| 181 | + }, |
55 | 182 | 'title': {},
|
56 | 183 | 'tr': {},
|
57 |
| - 'ul': {'class'} |
| 184 | + 'ul': { |
| 185 | + 'class': { |
| 186 | + 'toc' |
| 187 | + } |
| 188 | + } |
58 | 189 | }
|
59 | 190 |
|
60 | 191 |
|
61 | 192 | def check_dom(options):
|
62 | 193 | '''Main driver.'''
|
63 |
| - elements = set() |
64 |
| - attributes = {} |
| 194 | + problems = {} |
65 | 195 | for filename in options.sources:
|
66 | 196 | with open(filename, 'r') as reader:
|
67 | 197 | doc = bs4.BeautifulSoup(reader, features='lxml')
|
68 | 198 | for node in doc.descendants:
|
69 | 199 | if isinstance(node, bs4.element.Tag):
|
70 |
| - check(elements, attributes, node) |
71 |
| - report(elements, attributes) |
| 200 | + check(problems, filename, node) |
| 201 | + report(problems) |
72 | 202 |
|
73 | 203 |
|
74 |
| -def check(elements, attributes, node): |
| 204 | +def check(problems, filename, node): |
75 | 205 | '''Check individual element.'''
|
76 | 206 | if node.name not in KNOWN:
|
77 |
| - elements.add(node.name) |
| 207 | + problems[node.name] = f'unknown ({filename})' |
78 | 208 | return
|
79 | 209 | for attr in node.attrs:
|
80 | 210 | if attr not in KNOWN[node.name]:
|
81 |
| - if node.name not in attributes: |
82 |
| - attributes[node.name] = set() |
83 |
| - attributes[node.name].add(attr) |
| 211 | + if node.name not in problems: |
| 212 | + problems[node.name] = [] |
| 213 | + problems[node.name].append(f'unknown attribute "{attr}" in {filename}') |
| 214 | + elif KNOWN[node.name][attr] is not None: |
| 215 | + values = node[attr] if isinstance(node[attr], list) else [node[attr]] |
| 216 | + for value in values: |
| 217 | + if value not in KNOWN[node.name][attr]: |
| 218 | + if node.name not in problems: |
| 219 | + problems[node.name] = [] |
| 220 | + problems[node.name].append(f'"{attr}" has unknown value "{value}" in {filename}') |
84 | 221 |
|
85 | 222 |
|
86 |
| -def report(elements, attributes): |
| 223 | +def report(problems): |
87 | 224 | '''Report results.'''
|
88 |
| - if (not elements) and (not attributes): |
| 225 | + if not problems: |
89 | 226 | return
|
90 |
| - print('- unknown') |
91 |
| - if elements: |
92 |
| - print(' - elements') |
93 |
| - for name in sorted(elements): |
94 |
| - print(f' - {name}') |
95 |
| - if attributes: |
96 |
| - print(' - attributes') |
97 |
| - for name in sorted(attributes.keys()): |
98 |
| - print(f' - {name}') |
99 |
| - for attr in sorted(attributes[name]): |
100 |
| - print(f' - {attr}') |
| 227 | + print('- DOM') |
| 228 | + for name in sorted(problems.keys()): |
| 229 | + print(f' - {name}') |
| 230 | + for problem in problems[name]: |
| 231 | + print(f' - {problem}') |
101 | 232 |
|
102 | 233 |
|
103 | 234 | if __name__ == '__main__':
|
|
0 commit comments