Python code coverage for Lib/tabnanny.py

#	count	content
1	n/a	#! /usr/bin/env python3
2	n/a
3	n/a	"""The Tab Nanny despises ambiguous indentation. She knows no mercy.
4	n/a
5	n/a	tabnanny -- Detection of ambiguous indentation
6	n/a
7	n/a	For the time being this module is intended to be called as a script.
8	n/a	However it is possible to import it into an IDE and use the function
9	n/a	check() described below.
10	n/a
11	n/a	Warning: The API provided by this module is likely to change in future
12	n/a	releases; such changes may not be backward compatible.
13	n/a	"""
14	n/a
15	n/a	# Released to the public domain, by Tim Peters, 15 April 1998.
16	n/a
17	n/a	# XXX Note: this is now a standard library module.
18	n/a	# XXX The API needs to undergo changes however; the current code is too
19	n/a	# XXX script-like. This will be addressed later.
20	n/a
21	n/a	__version__ = "6"
22	n/a
23	n/a	import os
24	n/a	import sys
25	n/a	import getopt
26	n/a	import tokenize
27	n/a	if not hasattr(tokenize, 'NL'):
28	n/a	raise ValueError("tokenize.NL doesn't exist -- tokenize module too old")
29	n/a
30	n/a	__all__ = ["check", "NannyNag", "process_tokens"]
31	n/a
32	n/a	verbose = 0
33	n/a	filename_only = 0
34	n/a
35	n/a	def errprint(*args):
36	n/a	sep = ""
37	n/a	for arg in args:
38	n/a	sys.stderr.write(sep + str(arg))
39	n/a	sep = " "
40	n/a	sys.stderr.write("\n")
41	n/a
42	n/a	def main():
43	n/a	global verbose, filename_only
44	n/a	try:
45	n/a	opts, args = getopt.getopt(sys.argv[1:], "qv")
46	n/a	except getopt.error as msg:
47	n/a	errprint(msg)
48	n/a	return
49	n/a	for o, a in opts:
50	n/a	if o == '-q':
51	n/a	filename_only = filename_only + 1
52	n/a	if o == '-v':
53	n/a	verbose = verbose + 1
54	n/a	if not args:
55	n/a	errprint("Usage:", sys.argv[0], "[-v] file_or_directory ...")
56	n/a	return
57	n/a	for arg in args:
58	n/a	check(arg)
59	n/a
60	n/a	class NannyNag(Exception):
61	n/a	"""
62	n/a	Raised by tokeneater() if detecting an ambiguous indent.
63	n/a	Captured and handled in check().
64	n/a	"""
65	n/a	def __init__(self, lineno, msg, line):
66	n/a	self.lineno, self.msg, self.line = lineno, msg, line
67	n/a	def get_lineno(self):
68	n/a	return self.lineno
69	n/a	def get_msg(self):
70	n/a	return self.msg
71	n/a	def get_line(self):
72	n/a	return self.line
73	n/a
74	n/a	def check(file):
75	n/a	"""check(file_or_dir)
76	n/a
77	n/a	If file_or_dir is a directory and not a symbolic link, then recursively
78	n/a	descend the directory tree named by file_or_dir, checking all .py files
79	n/a	along the way. If file_or_dir is an ordinary Python source file, it is
80	n/a	checked for whitespace related problems. The diagnostic messages are
81	n/a	written to standard output using the print statement.
82	n/a	"""
83	n/a
84	n/a	if os.path.isdir(file) and not os.path.islink(file):
85	n/a	if verbose:
86	n/a	print("%r: listing directory" % (file,))
87	n/a	names = os.listdir(file)
88	n/a	for name in names:
89	n/a	fullname = os.path.join(file, name)
90	n/a	if (os.path.isdir(fullname) and
91	n/a	not os.path.islink(fullname) or
92	n/a	os.path.normcase(name[-3:]) == ".py"):
93	n/a	check(fullname)
94	n/a	return
95	n/a
96	n/a	try:
97	n/a	f = tokenize.open(file)
98	n/a	except OSError as msg:
99	n/a	errprint("%r: I/O Error: %s" % (file, msg))
100	n/a	return
101	n/a
102	n/a	if verbose > 1:
103	n/a	print("checking %r ..." % file)
104	n/a
105	n/a	try:
106	n/a	process_tokens(tokenize.generate_tokens(f.readline))
107	n/a
108	n/a	except tokenize.TokenError as msg:
109	n/a	errprint("%r: Token Error: %s" % (file, msg))
110	n/a	return
111	n/a
112	n/a	except IndentationError as msg:
113	n/a	errprint("%r: Indentation Error: %s" % (file, msg))
114	n/a	return
115	n/a
116	n/a	except NannyNag as nag:
117	n/a	badline = nag.get_lineno()
118	n/a	line = nag.get_line()
119	n/a	if verbose:
120	n/a	print("%r: * Line %d: trouble in tab city! *" % (file, badline))
121	n/a	print("offending line: %r" % (line,))
122	n/a	print(nag.get_msg())
123	n/a	else:
124	n/a	if ' ' in file: file = '"' + file + '"'
125	n/a	if filename_only: print(file)
126	n/a	else: print(file, badline, repr(line))
127	n/a	return
128	n/a
129	n/a	finally:
130	n/a	f.close()
131	n/a
132	n/a	if verbose:
133	n/a	print("%r: Clean bill of health." % (file,))
134	n/a
135	n/a	class Whitespace:
136	n/a	# the characters used for space and tab
137	n/a	S, T = ' \t'
138	n/a
139	n/a	# members:
140	n/a	# raw
141	n/a	# the original string
142	n/a	# n
143	n/a	# the number of leading whitespace characters in raw
144	n/a	# nt
145	n/a	# the number of tabs in raw[:n]
146	n/a	# norm
147	n/a	# the normal form as a pair (count, trailing), where:
148	n/a	# count
149	n/a	# a tuple such that raw[:n] contains count[i]
150	n/a	# instances of S * i + T
151	n/a	# trailing
152	n/a	# the number of trailing spaces in raw[:n]
153	n/a	# It's A Theorem that m.indent_level(t) ==
154	n/a	# n.indent_level(t) for all t >= 1 iff m.norm == n.norm.
155	n/a	# is_simple
156	n/a	# true iff raw[:n] is of the form (T)(S)
157	n/a
158	n/a	def __init__(self, ws):
159	n/a	self.raw = ws
160	n/a	S, T = Whitespace.S, Whitespace.T
161	n/a	count = []
162	n/a	b = n = nt = 0
163	n/a	for ch in self.raw:
164	n/a	if ch == S:
165	n/a	n = n + 1
166	n/a	b = b + 1
167	n/a	elif ch == T:
168	n/a	n = n + 1
169	n/a	nt = nt + 1
170	n/a	if b >= len(count):
171	n/a	count = count + [0] * (b - len(count) + 1)
172	n/a	count[b] = count[b] + 1
173	n/a	b = 0
174	n/a	else:
175	n/a	break
176	n/a	self.n = n
177	n/a	self.nt = nt
178	n/a	self.norm = tuple(count), b
179	n/a	self.is_simple = len(count) <= 1
180	n/a
181	n/a	# return length of longest contiguous run of spaces (whether or not
182	n/a	# preceding a tab)
183	n/a	def longest_run_of_spaces(self):
184	n/a	count, trailing = self.norm
185	n/a	return max(len(count)-1, trailing)
186	n/a
187	n/a	def indent_level(self, tabsize):
188	n/a	# count, il = self.norm
189	n/a	# for i in range(len(count)):
190	n/a	# if count[i]:
191	n/a	# il = il + (i//tabsize + 1)tabsize count[i]
192	n/a	# return il
193	n/a
194	n/a	# quicker:
195	n/a	# il = trailing + sum (i//ts + 1)tscount[i] =
196	n/a	# trailing + ts * sum (i//ts + 1)*count[i] =
197	n/a	# trailing + ts * sum i//ts*count[i] + count[i] =
198	n/a	# trailing + ts * [(sum i//ts*count[i]) + (sum count[i])] =
199	n/a	# trailing + ts * [(sum i//ts*count[i]) + num_tabs]
200	n/a	# and note that i//ts*count[i] is 0 when i < ts
201	n/a
202	n/a	count, trailing = self.norm
203	n/a	il = 0
204	n/a	for i in range(tabsize, len(count)):
205	n/a	il = il + i//tabsize * count[i]
206	n/a	return trailing + tabsize * (il + self.nt)
207	n/a
208	n/a	# return true iff self.indent_level(t) == other.indent_level(t)
209	n/a	# for all t >= 1
210	n/a	def equal(self, other):
211	n/a	return self.norm == other.norm
212	n/a
213	n/a	# return a list of tuples (ts, i1, i2) such that
214	n/a	# i1 == self.indent_level(ts) != other.indent_level(ts) == i2.
215	n/a	# Intended to be used after not self.equal(other) is known, in which
216	n/a	# case it will return at least one witnessing tab size.
217	n/a	def not_equal_witness(self, other):
218	n/a	n = max(self.longest_run_of_spaces(),
219	n/a	other.longest_run_of_spaces()) + 1
220	n/a	a = []
221	n/a	for ts in range(1, n+1):
222	n/a	if self.indent_level(ts) != other.indent_level(ts):
223	n/a	a.append( (ts,
224	n/a	self.indent_level(ts),
225	n/a	other.indent_level(ts)) )
226	n/a	return a
227	n/a
228	n/a	# Return True iff self.indent_level(t) < other.indent_level(t)
229	n/a	# for all t >= 1.
230	n/a	# The algorithm is due to Vincent Broman.
231	n/a	# Easy to prove it's correct.
232	n/a	# XXXpost that.
233	n/a	# Trivial to prove n is sharp (consider T vs ST).
234	n/a	# Unknown whether there's a faster general way. I suspected so at
235	n/a	# first, but no longer.
236	n/a	# For the special (but common!) case where M and N are both of the
237	n/a	# form (T)(S), M.less(N) iff M.len() < N.len() and
238	n/a	# M.num_tabs() <= N.num_tabs(). Proof is easy but kinda long-winded.
239	n/a	# XXXwrite that up.
240	n/a	# Note that M is of the form (T)(S) iff len(M.norm[0]) <= 1.
241	n/a	def less(self, other):
242	n/a	if self.n >= other.n:
243	n/a	return False
244	n/a	if self.is_simple and other.is_simple:
245	n/a	return self.nt <= other.nt
246	n/a	n = max(self.longest_run_of_spaces(),
247	n/a	other.longest_run_of_spaces()) + 1
248	n/a	# the self.n >= other.n test already did it for ts=1
249	n/a	for ts in range(2, n+1):
250	n/a	if self.indent_level(ts) >= other.indent_level(ts):
251	n/a	return False
252	n/a	return True
253	n/a
254	n/a	# return a list of tuples (ts, i1, i2) such that
255	n/a	# i1 == self.indent_level(ts) >= other.indent_level(ts) == i2.
256	n/a	# Intended to be used after not self.less(other) is known, in which
257	n/a	# case it will return at least one witnessing tab size.
258	n/a	def not_less_witness(self, other):
259	n/a	n = max(self.longest_run_of_spaces(),
260	n/a	other.longest_run_of_spaces()) + 1
261	n/a	a = []
262	n/a	for ts in range(1, n+1):
263	n/a	if self.indent_level(ts) >= other.indent_level(ts):
264	n/a	a.append( (ts,
265	n/a	self.indent_level(ts),
266	n/a	other.indent_level(ts)) )
267	n/a	return a
268	n/a
269	n/a	def format_witnesses(w):
270	n/a	firsts = (str(tup[0]) for tup in w)
271	n/a	prefix = "at tab size"
272	n/a	if len(w) > 1:
273	n/a	prefix = prefix + "s"
274	n/a	return prefix + " " + ', '.join(firsts)
275	n/a
276	n/a	def process_tokens(tokens):
277	n/a	INDENT = tokenize.INDENT
278	n/a	DEDENT = tokenize.DEDENT
279	n/a	NEWLINE = tokenize.NEWLINE
280	n/a	JUNK = tokenize.COMMENT, tokenize.NL
281	n/a	indents = [Whitespace("")]
282	n/a	check_equal = 0
283	n/a
284	n/a	for (type, token, start, end, line) in tokens:
285	n/a	if type == NEWLINE:
286	n/a	# a program statement, or ENDMARKER, will eventually follow,
287	n/a	# after some (possibly empty) run of tokens of the form
288	n/a	# (NL \| COMMENT)* (INDENT \| DEDENT+)?
289	n/a	# If an INDENT appears, setting check_equal is wrong, and will
290	n/a	# be undone when we see the INDENT.
291	n/a	check_equal = 1
292	n/a
293	n/a	elif type == INDENT:
294	n/a	check_equal = 0
295	n/a	thisguy = Whitespace(token)
296	n/a	if not indents[-1].less(thisguy):
297	n/a	witness = indents[-1].not_less_witness(thisguy)
298	n/a	msg = "indent not greater e.g. " + format_witnesses(witness)
299	n/a	raise NannyNag(start[0], msg, line)
300	n/a	indents.append(thisguy)
301	n/a
302	n/a	elif type == DEDENT:
303	n/a	# there's nothing we need to check here! what's important is
304	n/a	# that when the run of DEDENTs ends, the indentation of the
305	n/a	# program statement (or ENDMARKER) that triggered the run is
306	n/a	# equal to what's left at the top of the indents stack
307	n/a
308	n/a	# Ouch! This assert triggers if the last line of the source
309	n/a	# is indented and lacks a newline -- then DEDENTs pop out
310	n/a	# of thin air.
311	n/a	# assert check_equal # else no earlier NEWLINE, or an earlier INDENT
312	n/a	check_equal = 1
313	n/a
314	n/a	del indents[-1]
315	n/a
316	n/a	elif check_equal and type not in JUNK:
317	n/a	# this is the first "real token" following a NEWLINE, so it
318	n/a	# must be the first token of the next program statement, or an
319	n/a	# ENDMARKER; the "line" argument exposes the leading whitespace
320	n/a	# for this statement; in the case of ENDMARKER, line is an empty
321	n/a	# string, so will properly match the empty string with which the
322	n/a	# "indents" stack was seeded
323	n/a	check_equal = 0
324	n/a	thisguy = Whitespace(line)
325	n/a	if not indents[-1].equal(thisguy):
326	n/a	witness = indents[-1].not_equal_witness(thisguy)
327	n/a	msg = "indent not equal e.g. " + format_witnesses(witness)
328	n/a	raise NannyNag(start[0], msg, line)
329	n/a
330	n/a
331	n/a	if __name__ == '__main__':
332	n/a	main()