| 1 | #!/usr/bin/env python3
|
| 2 | """fmt_check.py
|
| 3 |
|
| 4 | Check that the output HTML obeys the following rules:
|
| 5 |
|
| 6 | - No orphaned backticks '`' should be part of a `inline code block`
|
| 7 | (ie. any backticks not in a <code> block is treated as an error)
|
| 8 | - Lines in a <code> should be shorter than 70 chars (else they overflow)
|
| 9 | """
|
| 10 |
|
| 11 | import html.parser
|
| 12 | import sys
|
| 13 |
|
| 14 | from doctools.util import log
|
| 15 |
|
| 16 |
|
| 17 | class TagAwareHTMLParser(html.parser.HTMLParser):
|
| 18 |
|
| 19 | def __init__(self, file):
|
| 20 | super().__init__()
|
| 21 | self.tag_stack = []
|
| 22 | self.file = file
|
| 23 |
|
| 24 | def location_str(self):
|
| 25 | line, col = self.getpos()
|
| 26 | return '%s:%d:%d' % (self.file, line, col)
|
| 27 |
|
| 28 | def handle_starttag(self, tag, _attrs):
|
| 29 | # Skip self-closing elements
|
| 30 | if tag in ('meta', 'img'):
|
| 31 | return
|
| 32 |
|
| 33 | self.tag_stack.append(tag)
|
| 34 |
|
| 35 | def handle_endtag(self, tag):
|
| 36 | popped = self.tag_stack.pop()
|
| 37 | if tag != popped:
|
| 38 | print('%s [WARN] Mismatched tag!' % self.location_str(),
|
| 39 | 'Expected </%s> but got </%s>' % (popped, tag))
|
| 40 |
|
| 41 |
|
| 42 | class CheckBackticks(TagAwareHTMLParser):
|
| 43 |
|
| 44 | def __init__(self, file):
|
| 45 | super().__init__(file)
|
| 46 | self.has_error = False
|
| 47 |
|
| 48 | def handle_data(self, text):
|
| 49 | # Ignore eg, <code> tags
|
| 50 | if len(self.tag_stack) and (self.tag_stack[-1]
|
| 51 | not in ("p", "h1", "h2", "h3", "a")):
|
| 52 | return
|
| 53 |
|
| 54 | idx = text.find('`')
|
| 55 | if idx == -1:
|
| 56 | return
|
| 57 |
|
| 58 | print('%s [ERROR] Found stray backtick %r' %
|
| 59 | (self.location_str(), text))
|
| 60 |
|
| 61 | self.has_error = True
|
| 62 |
|
| 63 |
|
| 64 | class CheckCodeLines(TagAwareHTMLParser):
|
| 65 | # Found when the display is 801px in width
|
| 66 | MAX_LINE_LENGTH = 72
|
| 67 |
|
| 68 | def __init__(self, file):
|
| 69 | super().__init__(file)
|
| 70 | self.has_error = False
|
| 71 |
|
| 72 | def handle_data(self, text):
|
| 73 | # Ignore eg, <code> tags
|
| 74 | if len(self.tag_stack) and self.tag_stack[-1] != 'code':
|
| 75 | return
|
| 76 |
|
| 77 | for i, line in enumerate(text.splitlines()):
|
| 78 | if len(line) > self.MAX_LINE_LENGTH:
|
| 79 | print('%s [ERROR] Line %d of <code> is too long: %r' %
|
| 80 | (self.location_str(), i + 1, line))
|
| 81 | self.has_error = True
|
| 82 |
|
| 83 |
|
| 84 | def FormatCheck(filename):
|
| 85 | backticks = CheckBackticks(filename)
|
| 86 | with open(filename, "r") as f:
|
| 87 | backticks.feed(f.read())
|
| 88 |
|
| 89 | lines = CheckCodeLines(filename)
|
| 90 | with open(filename, "r") as f:
|
| 91 | lines.feed(f.read())
|
| 92 |
|
| 93 | return backticks.has_error or lines.has_error
|
| 94 |
|
| 95 |
|
| 96 | def main(argv):
|
| 97 | action = argv[1]
|
| 98 |
|
| 99 | any_error = False
|
| 100 | for path in argv[1:]:
|
| 101 | if not path.endswith('.html'):
|
| 102 | raise RuntimeError('Expected %r to be a .html file' % filename)
|
| 103 |
|
| 104 | this_error = FormatCheck(path)
|
| 105 | any_error = any_error or this_error
|
| 106 | log("%s %s" % ("ER" if this_error else "OK", path))
|
| 107 |
|
| 108 | if any_error:
|
| 109 | raise RuntimeError("Formatting errors found")
|
| 110 |
|
| 111 |
|
| 112 | if __name__ == '__main__':
|
| 113 | try:
|
| 114 | main(sys.argv)
|
| 115 | except RuntimeError as e:
|
| 116 | print('FATAL: %s' % e, file=sys.stderr)
|
| 117 | sys.exit(1)
|