Python code coverage for Doc/tools/extensions/suspicious.py

#	count	content
1	n/a	"""
2	n/a	Try to detect suspicious constructs, resembling markup
3	n/a	that has leaked into the final output.
4	n/a
5	n/a	Suspicious lines are reported in a comma-separated-file,
6	n/a	``suspicious.csv``, located in the output directory.
7	n/a
8	n/a	The file is utf-8 encoded, and each line contains four fields:
9	n/a
10	n/a	* document name (normalized)
11	n/a	* line number in the source document
12	n/a	* problematic text
13	n/a	* complete line showing the problematic text in context
14	n/a
15	n/a	It is common to find many false positives. To avoid reporting them
16	n/a	again and again, they may be added to the ``ignored.csv`` file
17	n/a	(located in the configuration directory). The file has the same
18	n/a	format as ``suspicious.csv`` with a few differences:
19	n/a
20	n/a	- each line defines a rule; if the rule matches, the issue
21	n/a	is ignored.
22	n/a	- line number may be empty (that is, nothing between the
23	n/a	commas: ",,"). In this case, line numbers are ignored (the
24	n/a	rule matches anywhere in the file).
25	n/a	- the last field does not have to be a complete line; some
26	n/a	surrounding text (never more than a line) is enough for
27	n/a	context.
28	n/a
29	n/a	Rules are processed sequentially. A rule matches when:
30	n/a
31	n/a	* document names are the same
32	n/a	* problematic texts are the same
33	n/a	* line numbers are close to each other (5 lines up or down)
34	n/a	* the rule text is completely contained into the source line
35	n/a
36	n/a	The simplest way to create the ignored.csv file is by copying
37	n/a	undesired entries from suspicious.csv (possibly trimming the last
38	n/a	field.)
39	n/a
40	n/a	Copyright 2009 Gabriel A. Genellina
41	n/a
42	n/a	"""
43	n/a
44	n/a	import os
45	n/a	import re
46	n/a	import csv
47	n/a	import sys
48	n/a
49	n/a	from docutils import nodes
50	n/a	from sphinx.builders import Builder
51	n/a
52	n/a	detect_all = re.compile(r'''
53	n/a	::(?=[^=])\| # two :: (but NOT ::=)
54	n/a	:[a-zA-Z][a-zA-Z0-9]+\| # :foo
55	n/a	`\| # ` (seldom used by itself)
56	n/a	(?<!\.)\.\.[ \t]*\w+: # .. foo: (but NOT ... else:)
57	n/a	''', re.UNICODE \| re.VERBOSE).finditer
58	n/a
59	n/a	py3 = sys.version_info >= (3, 0)
60	n/a
61	n/a
62	n/a	class Rule:
63	n/a	def __init__(self, docname, lineno, issue, line):
64	n/a	"""A rule for ignoring issues"""
65	n/a	self.docname = docname # document to which this rule applies
66	n/a	self.lineno = lineno # line number in the original source;
67	n/a	# this rule matches only near that.
68	n/a	# None -> don't care
69	n/a	self.issue = issue # the markup fragment that triggered this rule
70	n/a	self.line = line # text of the container element (single line only)
71	n/a	self.used = False
72	n/a
73	n/a	def __repr__(self):
74	n/a	return '{0.docname},,{0.issue},{0.line}'.format(self)
75	n/a
76	n/a
77	n/a
78	n/a	class dialect(csv.excel):
79	n/a	"""Our dialect: uses only linefeed as newline."""
80	n/a	lineterminator = '\n'
81	n/a
82	n/a
83	n/a	class CheckSuspiciousMarkupBuilder(Builder):
84	n/a	"""
85	n/a	Checks for possibly invalid markup that may leak into the output.
86	n/a	"""
87	n/a	name = 'suspicious'
88	n/a
89	n/a	def init(self):
90	n/a	# create output file
91	n/a	self.log_file_name = os.path.join(self.outdir, 'suspicious.csv')
92	n/a	open(self.log_file_name, 'w').close()
93	n/a	# load database of previously ignored issues
94	n/a	self.load_rules(os.path.join(os.path.dirname(__file__), '..',
95	n/a	'susp-ignored.csv'))
96	n/a
97	n/a	def get_outdated_docs(self):
98	n/a	return self.env.found_docs
99	n/a
100	n/a	def get_target_uri(self, docname, typ=None):
101	n/a	return ''
102	n/a
103	n/a	def prepare_writing(self, docnames):
104	n/a	pass
105	n/a
106	n/a	def write_doc(self, docname, doctree):
107	n/a	# set when any issue is encountered in this document
108	n/a	self.any_issue = False
109	n/a	self.docname = docname
110	n/a	visitor = SuspiciousVisitor(doctree, self)
111	n/a	doctree.walk(visitor)
112	n/a
113	n/a	def finish(self):
114	n/a	unused_rules = [rule for rule in self.rules if not rule.used]
115	n/a	if unused_rules:
116	n/a	self.warn('Found %s/%s unused rules:' %
117	n/a	(len(unused_rules), len(self.rules)))
118	n/a	for rule in unused_rules:
119	n/a	self.info(repr(rule))
120	n/a	return
121	n/a
122	n/a	def check_issue(self, line, lineno, issue):
123	n/a	if not self.is_ignored(line, lineno, issue):
124	n/a	self.report_issue(line, lineno, issue)
125	n/a
126	n/a	def is_ignored(self, line, lineno, issue):
127	n/a	"""Determine whether this issue should be ignored."""
128	n/a	docname = self.docname
129	n/a	for rule in self.rules:
130	n/a	if rule.docname != docname: continue
131	n/a	if rule.issue != issue: continue
132	n/a	# Both lines must match exactly. This is rather strict,
133	n/a	# and probably should be improved.
134	n/a	# Doing fuzzy matches with levenshtein distance could work,
135	n/a	# but that means bringing other libraries...
136	n/a	# Ok, relax that requirement: just check if the rule fragment
137	n/a	# is contained in the document line
138	n/a	if rule.line not in line: continue
139	n/a	# Check both line numbers. If they're "near"
140	n/a	# this rule matches. (lineno=None means "don't care")
141	n/a	if (rule.lineno is not None) and \
142	n/a	abs(rule.lineno - lineno) > 5: continue
143	n/a	# if it came this far, the rule matched
144	n/a	rule.used = True
145	n/a	return True
146	n/a	return False
147	n/a
148	n/a	def report_issue(self, text, lineno, issue):
149	n/a	if not self.any_issue: self.info()
150	n/a	self.any_issue = True
151	n/a	self.write_log_entry(lineno, issue, text)
152	n/a	if py3:
153	n/a	self.warn('[%s:%d] "%s" found in "%-.120s"' %
154	n/a	(self.docname, lineno, issue, text))
155	n/a	else:
156	n/a	self.warn('[%s:%d] "%s" found in "%-.120s"' % (
157	n/a	self.docname.encode(sys.getdefaultencoding(),'replace'),
158	n/a	lineno,
159	n/a	issue.encode(sys.getdefaultencoding(),'replace'),
160	n/a	text.strip().encode(sys.getdefaultencoding(),'replace')))
161	n/a	self.app.statuscode = 1
162	n/a
163	n/a	def write_log_entry(self, lineno, issue, text):
164	n/a	if py3:
165	n/a	f = open(self.log_file_name, 'a')
166	n/a	writer = csv.writer(f, dialect)
167	n/a	writer.writerow([self.docname, lineno, issue, text.strip()])
168	n/a	f.close()
169	n/a	else:
170	n/a	f = open(self.log_file_name, 'ab')
171	n/a	writer = csv.writer(f, dialect)
172	n/a	writer.writerow([self.docname.encode('utf-8'),
173	n/a	lineno,
174	n/a	issue.encode('utf-8'),
175	n/a	text.strip().encode('utf-8')])
176	n/a	f.close()
177	n/a
178	n/a	def load_rules(self, filename):
179	n/a	"""Load database of previously ignored issues.
180	n/a
181	n/a	A csv file, with exactly the same format as suspicious.csv
182	n/a	Fields: document name (normalized), line number, issue, surrounding text
183	n/a	"""
184	n/a	self.info("loading ignore rules... ", nonl=1)
185	n/a	self.rules = rules = []
186	n/a	try:
187	n/a	if py3:
188	n/a	f = open(filename, 'r')
189	n/a	else:
190	n/a	f = open(filename, 'rb')
191	n/a	except IOError:
192	n/a	return
193	n/a	for i, row in enumerate(csv.reader(f)):
194	n/a	if len(row) != 4:
195	n/a	raise ValueError(
196	n/a	"wrong format in %s, line %d: %s" % (filename, i+1, row))
197	n/a	docname, lineno, issue, text = row
198	n/a	if lineno:
199	n/a	lineno = int(lineno)
200	n/a	else:
201	n/a	lineno = None
202	n/a	if not py3:
203	n/a	docname = docname.decode('utf-8')
204	n/a	issue = issue.decode('utf-8')
205	n/a	text = text.decode('utf-8')
206	n/a	rule = Rule(docname, lineno, issue, text)
207	n/a	rules.append(rule)
208	n/a	f.close()
209	n/a	self.info('done, %d rules loaded' % len(self.rules))
210	n/a
211	n/a
212	n/a	def get_lineno(node):
213	n/a	"""Obtain line number information for a node."""
214	n/a	lineno = None
215	n/a	while lineno is None and node:
216	n/a	node = node.parent
217	n/a	lineno = node.line
218	n/a	return lineno
219	n/a
220	n/a
221	n/a	def extract_line(text, index):
222	n/a	"""text may be a multiline string; extract
223	n/a	only the line containing the given character index.
224	n/a
225	n/a	>>> extract_line("abc\ndefgh\ni", 6)
226	n/a	>>> 'defgh'
227	n/a	>>> for i in (0, 2, 3, 4, 10):
228	n/a	... print extract_line("abc\ndefgh\ni", i)
229	n/a	abc
230	n/a	abc
231	n/a	abc
232	n/a	defgh
233	n/a	defgh
234	n/a	i
235	n/a	"""
236	n/a	p = text.rfind('\n', 0, index) + 1
237	n/a	q = text.find('\n', index)
238	n/a	if q < 0:
239	n/a	q = len(text)
240	n/a	return text[p:q]
241	n/a
242	n/a
243	n/a	class SuspiciousVisitor(nodes.GenericNodeVisitor):
244	n/a
245	n/a	lastlineno = 0
246	n/a
247	n/a	def __init__(self, document, builder):
248	n/a	nodes.GenericNodeVisitor.__init__(self, document)
249	n/a	self.builder = builder
250	n/a
251	n/a	def default_visit(self, node):
252	n/a	if isinstance(node, (nodes.Text, nodes.image)): # direct text containers
253	n/a	text = node.astext()
254	n/a	# lineno seems to go backwards sometimes (?)
255	n/a	self.lastlineno = lineno = max(get_lineno(node) or 0, self.lastlineno)
256	n/a	seen = set() # don't report the same issue more than only once per line
257	n/a	for match in detect_all(text):
258	n/a	issue = match.group()
259	n/a	line = extract_line(text, match.start())
260	n/a	if (issue, line) not in seen:
261	n/a	self.builder.check_issue(line, lineno, issue)
262	n/a	seen.add((issue, line))
263	n/a
264	n/a	unknown_visit = default_visit
265	n/a
266	n/a	def visit_document(self, node):
267	n/a	self.lastlineno = 0
268	n/a
269	n/a	def visit_comment(self, node):
270	n/a	# ignore comments -- too much false positives.
271	n/a	# (although doing this could miss some errors;
272	n/a	# there were two sections "commented-out" by mistake
273	n/a	# in the Python docs that would not be caught)
274	n/a	raise nodes.SkipNode