2 # -*- coding: utf-8 -*-
5 htmldocck.py is a custom checker script for Rustdoc HTML outputs.
9 The principle is simple: This script receives a path to generated HTML
10 documentation and a "template" script, which has a series of check
11 commands like `@has` or `@matches`. Each command is used to check if
12 some pattern is present or not present in the particular file or in
13 a particular node of the HTML tree. In many cases, the template script
14 happens to be the source code given to rustdoc.
16 While it indeed is possible to test in smaller portions, it has been
17 hard to construct tests in this fashion and major rendering errors were
18 discovered much later. This script is designed to make black-box and
19 regression testing of Rustdoc easy. This does not preclude the needs for
20 unit testing, but can be used to complement related tests by quickly
21 showing the expected renderings.
23 In order to avoid one-off dependencies for this task, this script uses
24 a reasonably working HTML parser and the existing XPath implementation
25 from Python's standard library. Hopefully, we won't render
30 Commands start with an `@` followed by a command name (letters and
31 hyphens), and zero or more arguments separated by one or more whitespace
32 characters and optionally delimited with single or double quotes. The `@`
33 mark cannot be preceded by a non-whitespace character. Other lines
34 (including every text up to the first `@`) are ignored, but it is
35 recommended to avoid the use of `@` in the template file.
37 There are a number of supported commands:
39 * `@has PATH` checks for the existence of the given file.
41 `PATH` is relative to the output directory. It can be given as `-`
42 which repeats the most recently used `PATH`.
44 * `@has PATH PATTERN` and `@matches PATH PATTERN` checks for
45 the occurrence of the given pattern `PATTERN` in the specified file.
46 Only one occurrence of the pattern is enough.
48 For `@has`, `PATTERN` is a whitespace-normalized (every consecutive
49 whitespace being replaced by one single space character) string.
50 The entire file is also whitespace-normalized including newlines.
52 For `@matches`, `PATTERN` is a Python-supported regular expression.
53 The file remains intact but the regexp is matched without the `MULTILINE`
54 and `IGNORECASE` options. You can still use a prefix `(?m)` or `(?i)`
55 to override them, and `\A` and `\Z` for definitely matching
56 the beginning and end of the file.
58 (The same distinction goes to other variants of these commands.)
60 * `@has PATH XPATH PATTERN` and `@matches PATH XPATH PATTERN` checks for
61 the presence of the given XPath `XPATH` in the specified HTML file,
62 and also the occurrence of the given pattern `PATTERN` in the matching
63 node or attribute. Only one occurrence of the pattern in the match
66 `PATH` should be a valid and well-formed HTML file. It does *not*
67 accept arbitrary HTML5; it should have matching open and close tags
68 and correct entity references at least.
70 `XPATH` is an XPath expression to match. The XPath is fairly limited:
71 `tag`, `*`, `.`, `//`, `..`, `[@attr]`, `[@attr='value']`, `[tag]`,
72 `[POS]` (element located in given `POS`), `[last()-POS]`, `text()`
73 and `@attr` (both as the last segment) are supported. Some examples:
75 - `//pre` or `.//pre` matches any element with a name `pre`.
76 - `//a[@href]` matches any element with an `href` attribute.
77 - `//*[@class="impl"]//code` matches any element with a name `code`,
78 which is an ancestor of some element which `class` attr is `impl`.
79 - `//h1[@class="fqn"]/span[1]/a[last()]/@class` matches a value of
80 `class` attribute in the last `a` element (can be followed by more
81 elements that are not `a`) inside the first `span` in the `h1` with
82 a class of `fqn`. Note that there cannot be any additional elements
83 between them due to the use of `/` instead of `//`.
85 Do not try to use non-absolute paths, it won't work due to the flawed
86 ElementTree implementation. The script rejects them.
88 For the text matches (i.e. paths not ending with `@attr`), any
89 subelements are flattened into one string; this is handy for ignoring
90 highlights for example. If you want to simply check for the presence of
91 a given node or attribute, use an empty string (`""`) as a `PATTERN`.
93 * `@count PATH XPATH COUNT` checks for the occurrence of the given XPath
94 in the specified file. The number of occurrences must match the given
97 * `@snapshot NAME PATH XPATH` creates a snapshot test named NAME.
98 A snapshot test captures a subtree of the DOM, at the location
99 determined by the XPath, and compares it to a pre-recorded value
100 in a file. The file's name is the test's name with the `.rs` extension
101 replaced with `.NAME.html`, where NAME is the snapshot's name.
103 htmldocck supports the `--bless` option to accept the current subtree
104 as expected, saving it to the file determined by the snapshot's name.
105 compiletest's `--bless` flag is forwarded to htmldocck.
107 * `@has-dir PATH` checks for the existence of the given directory.
109 All conditions can be negated with `!`. `@!has foo/type.NoSuch.html`
110 checks if the given file does not exist, for example.
114 from __future__ import absolute_import, print_function, unicode_literals
122 from collections import namedtuple
124 from html.parser import HTMLParser
126 from HTMLParser import HTMLParser
128 from xml.etree import cElementTree as ET
130 from xml.etree import ElementTree as ET
133 from html.entities import name2codepoint
135 from htmlentitydefs import name2codepoint
137 # "void elements" (no closing tag) from the HTML Standard section 12.1.2
138 VOID_ELEMENTS = {'area', 'base', 'br', 'col', 'embed', 'hr', 'img', 'input', 'keygen',
139 'link', 'menuitem', 'meta', 'param', 'source', 'track', 'wbr'}
141 # Python 2 -> 3 compatibility
148 channel = os.environ["DOC_RUST_LANG_ORG_CHANNEL"]
150 # Initialized in main
151 rust_test_path = None
154 class CustomHTMLParser(HTMLParser):
155 """simplified HTML parser.
157 this is possible because we are dealing with very regular HTML from
158 rustdoc; we only have to deal with i) void elements and ii) empty
160 def __init__(self, target=None):
161 HTMLParser.__init__(self)
162 self.__builder = target or ET.TreeBuilder()
164 def handle_starttag(self, tag, attrs):
165 attrs = {k: v or '' for k, v in attrs}
166 self.__builder.start(tag, attrs)
167 if tag in VOID_ELEMENTS:
168 self.__builder.end(tag)
170 def handle_endtag(self, tag):
171 self.__builder.end(tag)
173 def handle_startendtag(self, tag, attrs):
174 attrs = {k: v or '' for k, v in attrs}
175 self.__builder.start(tag, attrs)
176 self.__builder.end(tag)
178 def handle_data(self, data):
179 self.__builder.data(data)
181 def handle_entityref(self, name):
182 self.__builder.data(unichr(name2codepoint[name]))
184 def handle_charref(self, name):
185 code = int(name[1:], 16) if name.startswith(('x', 'X')) else int(name, 10)
186 self.__builder.data(unichr(code))
189 HTMLParser.close(self)
190 return self.__builder.close()
193 Command = namedtuple('Command', 'negated cmd args lineno context')
196 class FailedCheck(Exception):
200 class InvalidCheck(Exception):
204 def concat_multi_lines(f):
205 """returns a generator out of the file object, which
206 - removes `\\` then `\n` then a shared prefix with the previous line then
208 - keeps a line number (starting from 0) of the first line being
210 lastline = None # set to the last line when the last line has a backslash
213 for lineno, line in enumerate(f):
214 line = line.rstrip('\r\n')
216 # strip the common prefix from the current line if needed
217 if lastline is not None:
218 common_prefix = os.path.commonprefix([line, lastline])
219 line = line[len(common_prefix):].lstrip()
221 firstlineno = firstlineno or lineno
222 if line.endswith('\\'):
225 catenated += line[:-1]
227 yield firstlineno, catenated + line
232 if lastline is not None:
233 print_err(lineno, line, 'Trailing backslash at the end of the file')
236 LINE_PATTERN = re.compile(r'''
237 (?<=(?<!\S))(?P<invalid>!?)@(?P<negated>!?)
238 (?P<cmd>[A-Za-z]+(?:-[A-Za-z]+)*)
240 ''', re.X | re.UNICODE)
243 def get_commands(template):
244 with io.open(template, encoding='utf-8') as f:
245 for lineno, line in concat_multi_lines(f):
246 m = LINE_PATTERN.search(line)
250 negated = (m.group('negated') == '!')
252 if m.group('invalid') == '!':
256 'Invalid command: `!@{0}{1}`, (help: try with `@!{1}`)'.format(
257 '!' if negated else '',
262 args = m.group('args')
263 if args and not args[:1].isspace():
264 print_err(lineno, line, 'Invalid template syntax')
267 args = shlex.split(args)
268 except UnicodeEncodeError:
269 args = [arg.decode('utf-8') for arg in shlex.split(args.encode('utf-8'))]
270 yield Command(negated=negated, cmd=cmd, args=args, lineno=lineno+1, context=line)
273 def _flatten(node, acc):
275 acc.append(node.text)
288 def normalize_xpath(path):
289 path = path.replace("{{channel}}", channel)
290 if path.startswith('//'):
291 return '.' + path # avoid warnings
292 elif path.startswith('.//'):
295 raise InvalidCheck('Non-absolute XPath is not supported due to implementation issues')
298 class CachedFiles(object):
299 def __init__(self, root):
303 self.last_path = None
305 def resolve_path(self, path):
307 path = os.path.normpath(path)
308 self.last_path = path
310 elif self.last_path is None:
311 raise InvalidCheck('Tried to use the previous path in the first command')
313 return self.last_path
315 def get_file(self, path):
316 path = self.resolve_path(path)
317 if path in self.files:
318 return self.files[path]
320 abspath = os.path.join(self.root, path)
321 if not(os.path.exists(abspath) and os.path.isfile(abspath)):
322 raise FailedCheck('File does not exist {!r}'.format(path))
324 with io.open(abspath, encoding='utf-8') as f:
326 self.files[path] = data
329 def get_tree(self, path):
330 path = self.resolve_path(path)
331 if path in self.trees:
332 return self.trees[path]
334 abspath = os.path.join(self.root, path)
335 if not(os.path.exists(abspath) and os.path.isfile(abspath)):
336 raise FailedCheck('File does not exist {!r}'.format(path))
338 with io.open(abspath, encoding='utf-8') as f:
340 tree = ET.fromstringlist(f.readlines(), CustomHTMLParser())
341 except Exception as e:
342 raise RuntimeError('Cannot parse an HTML file {!r}: {}'.format(path, e))
343 self.trees[path] = tree
344 return self.trees[path]
346 def get_dir(self, path):
347 path = self.resolve_path(path)
348 abspath = os.path.join(self.root, path)
349 if not(os.path.exists(abspath) and os.path.isdir(abspath)):
350 raise FailedCheck('Directory does not exist {!r}'.format(path))
353 def check_string(data, pat, regexp):
354 pat = pat.replace("{{channel}}", channel)
356 return True # special case a presence testing
358 return re.search(pat, data, flags=re.UNICODE) is not None
360 data = ' '.join(data.split())
361 pat = ' '.join(pat.split())
365 def check_tree_attr(tree, path, attr, pat, regexp):
366 path = normalize_xpath(path)
368 for e in tree.findall(path):
370 value = e.attrib[attr]
374 ret = check_string(value, pat, regexp)
380 def check_tree_text(tree, path, pat, regexp):
381 path = normalize_xpath(path)
384 for e in tree.findall(path):
390 ret = check_string(value, pat, regexp)
394 print('Failed to get path "{}"'.format(path))
399 def get_tree_count(tree, path):
400 path = normalize_xpath(path)
401 return len(tree.findall(path))
404 def check_snapshot(snapshot_name, tree, normalize_to_text):
405 assert rust_test_path.endswith('.rs')
406 snapshot_path = '{}.{}.{}'.format(rust_test_path[:-3], snapshot_name, 'html')
408 with open(snapshot_path, 'r') as snapshot_file:
409 expected_str = snapshot_file.read()
410 except FileNotFoundError:
414 raise FailedCheck('No saved snapshot value')
416 if not normalize_to_text:
417 actual_str = ET.tostring(tree).decode('utf-8')
419 actual_str = flatten(tree)
421 if expected_str != actual_str:
423 with open(snapshot_path, 'w') as snapshot_file:
424 snapshot_file.write(actual_str)
426 print('--- expected ---\n')
428 print('\n\n--- actual ---\n')
431 raise FailedCheck('Actual snapshot value is different than expected')
434 if sys.version_info.major < 3:
435 file = codecs.getwriter('utf-8')(sys.stderr)
439 print(*args, file=file)
442 def print_err(lineno, context, err, message=None):
445 stderr("{}: {}".format(lineno, message or err))
447 stderr("\t{}".format(err))
450 stderr("\t{}".format(context))
456 def check_command(c, cache):
459 if c.cmd == 'has' or c.cmd == 'matches': # string test
460 regexp = (c.cmd == 'matches')
461 if len(c.args) == 1 and not regexp: # @has <path> = file existence
463 cache.get_file(c.args[0])
465 except FailedCheck as err:
468 elif len(c.args) == 2: # @has/matches <path> <pat> = string test
469 cerr = "`PATTERN` did not match"
470 ret = check_string(cache.get_file(c.args[0]), c.args[1], regexp)
471 elif len(c.args) == 3: # @has/matches <path> <pat> <match> = XML tree test
472 cerr = "`XPATH PATTERN` did not match"
473 tree = cache.get_tree(c.args[0])
474 pat, sep, attr = c.args[1].partition('/@')
476 tree = cache.get_tree(c.args[0])
477 ret = check_tree_attr(tree, pat, attr, c.args[2], regexp)
478 else: # normalized text
480 if pat.endswith('/text()'):
482 ret = check_tree_text(cache.get_tree(c.args[0]), pat, c.args[2], regexp)
484 raise InvalidCheck('Invalid number of @{} arguments'.format(c.cmd))
486 elif c.cmd == 'count': # count test
487 if len(c.args) == 3: # @count <path> <pat> <count> = count test
488 expected = int(c.args[2])
489 found = get_tree_count(cache.get_tree(c.args[0]), c.args[1])
490 cerr = "Expected {} occurrences but found {}".format(expected, found)
491 ret = expected == found
493 raise InvalidCheck('Invalid number of @{} arguments'.format(c.cmd))
495 elif c.cmd == 'snapshot': # snapshot test
496 if len(c.args) == 3: # @snapshot <snapshot-name> <html-path> <xpath>
497 [snapshot_name, html_path, pattern] = c.args
498 tree = cache.get_tree(html_path)
499 xpath = normalize_xpath(pattern)
500 normalize_to_text = False
501 if xpath.endswith('/text()'):
503 normalize_to_text = True
505 subtrees = tree.findall(xpath)
506 if len(subtrees) == 1:
509 check_snapshot(snapshot_name, subtree, normalize_to_text)
511 except FailedCheck as err:
514 elif len(subtrees) == 0:
515 raise FailedCheck('XPATH did not match')
517 raise FailedCheck('Expected 1 match, but found {}'.format(len(subtrees)))
519 raise InvalidCheck('Invalid number of @{} arguments'.format(c.cmd))
521 elif c.cmd == 'has-dir': # has-dir test
522 if len(c.args) == 1: # @has-dir <path> = has-dir test
524 cache.get_dir(c.args[0])
526 except FailedCheck as err:
530 raise InvalidCheck('Invalid number of @{} arguments'.format(c.cmd))
532 elif c.cmd == 'valid-html':
533 raise InvalidCheck('Unimplemented @valid-html')
535 elif c.cmd == 'valid-links':
536 raise InvalidCheck('Unimplemented @valid-links')
539 raise InvalidCheck('Unrecognized @{}'.format(c.cmd))
542 raise FailedCheck(cerr)
544 except FailedCheck as err:
545 message = '@{}{} check failed'.format('!' if c.negated else '', c.cmd)
546 print_err(c.lineno, c.context, str(err), message)
547 except InvalidCheck as err:
548 print_err(c.lineno, c.context, str(err))
551 def check(target, commands):
552 cache = CachedFiles(target)
554 check_command(c, cache)
557 if __name__ == '__main__':
558 if len(sys.argv) not in [3, 4]:
559 stderr('Usage: {} <doc dir> <template> [--bless]'.format(sys.argv[0]))
562 rust_test_path = sys.argv[2]
563 if len(sys.argv) > 3 and sys.argv[3] == '--bless':
566 # We only support `--bless` at the end of the arguments.
567 # This assert is to prevent silent failures.
568 assert '--bless' not in sys.argv
570 check(sys.argv[1], get_commands(rust_test_path))
572 stderr("\nEncountered {} errors".format(ERR_COUNT))