1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
|
#!/usr/bin/python
# Simple script that will output the release where a given bug was fixed
# searching the NEWS file
import optparse
import re
import sys
class NewsParser(object):
paren_exp_re = re.compile('\(([^)]+)\)')
release_re = re.compile("bzr[ -]")
release_prefix_length = len('bzr ')
bugs_re = re.compile('#([0-9]+)')
def __init__(self, news):
self.news = news
# Temporary attributes used by the parser
self.release = None
self.date = None
self.may_be_release = None
self.release_markup = None
self.entry = ''
self.line = None
self.lrs = None
def set_line(self, line):
self.line = line
self.lrs = line.rstrip()
def try_release(self):
if self.release_re.match(self.lrs) is not None:
# May be a new release
self.may_be_release = self.lrs
# We know the markup will have the same length as the release
self.release_markup = '#' * len(self.may_be_release)
return True
return False
def confirm_release(self):
if self.may_be_release is not None and self.lrs == self.release_markup:
# The release is followed by the right markup
self.release = self.may_be_release[self.release_prefix_length:]
# Wait for the associated date
self.date = None
return True
return False
def try_date(self):
if self.release is None:
return False
date_re = re.compile(':%s: (NOT RELEASED YET|\d{4}-\d{2}-\d{2})'
% (self.release,))
match = date_re.match(self.lrs)
if match is not None:
self.date = match.group(1)
return True
# The old fashion way
released_re = re.compile(':Released:\s+(\d{4}-\d{2}-\d{2})')
match = released_re.match(self.lrs)
if match is not None:
self.date = match.group(1)
return True
return False
def add_line_to_entry(self):
if self.lrs == '':
return False
self.entry += self.line
return True
def extract_bugs_from_entry(self):
"""Possibly extract bugs from a NEWS entry and yield them.
Not all entries will contain bugs and some entries are even garbage and
we don't try to parse them (yet). The trigger is a '#' and what looks
like a bug number inside parens to start with. From that we extract
authors (when present) and multiple bugs if needed.
"""
# FIXME: Malone entries are different
# Join all entry lines to simplify multiple line matching
flat_entry = ' '.join(self.entry.splitlines())
# Fixed bugs are always inside parens
for par in self.paren_exp_re.findall(flat_entry):
sharp = par.find('#')
if sharp is not None:
# We have at least one bug inside parens.
bugs = list(self.bugs_re.finditer(par))
if bugs:
# See where the first bug is mentioned
start = bugs[0].start()
end = bugs[-1].end()
if start == 0:
# (bugs/authors)
authors = par[end:]
else:
# (authors/bugs)
authors = par[:start]
for bug_match in bugs:
bug_number = bug_match.group(0)
yield (bug_number, authors,
self.release, self.date, self.entry)
# We've consumed the entry
self.entry = ''
def parse_bugs(self):
for line in self.news:
self.set_line(line)
if self.try_release():
continue # line may a be release
try:
if self.confirm_release():
continue # previous line was indeed a release
finally:
self.may_be_release = None
if self.try_date():
continue # The release date has been seen
if self.add_line_to_entry():
continue # accumulate in self.enrty
for b in self.extract_bugs_from_entry():
yield b # all bugs in the news entry
def main():
opt_parser = optparse.OptionParser(
usage="""Usage: %prog [options] BUG_NUMBER
""")
opt_parser.add_option(
'-f', '--file', type='str', dest='news_file',
help='NEWS file (defaults to ./NEWS)')
opt_parser.add_option(
'-m', '--message', type='str', dest='msg_re',
help='A regexp to search for in the news entry '
'(BUG_NUMBER should not be specified in this case)')
opt_parser.set_defaults(news_file='./NEWS')
(opts, args) = opt_parser.parse_args(sys.argv[1:])
if opts.msg_re is not None:
if len(args) != 0:
opt_parser.error('BUG_NUMBER and -m are mutually exclusive')
bug = None
msg_re = re.compile(opts.msg_re)
elif len(args) != 1:
opt_parser.error('Expected a single bug number, got %r' % args)
else:
bug = args[0]
news = open(opts.news_file)
parser = NewsParser(news)
try:
seen = 0
for b in parser.parse_bugs():
(number, authors, release, date, entry,) = b
# indent entry
entry = '\n'.join([' ' + l for l in entry.splitlines()])
found = False
if bug is not None:
if number[1:] == bug: # Strip the leading '#'
found = True
elif msg_re.search(entry) is not None:
found = True
if found:
print 'Bug %s was fixed in bzr-%s/%s by %s:' % (
number, release, date, authors)
print entry
seen += 1
finally:
print '%s bugs seen' % (seen,)
news.close()
main()
|