summaryrefslogtreecommitdiffstats
path: root/clang/utils/token-delta.py
blob: 327fa9221f0556dc59f6864710a26a39e953cc1f (plain)
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
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
#!/usr/bin/env python

import os
import re
import subprocess
import sys
import tempfile

###

class DeltaAlgorithm(object):
    def __init__(self):
        self.cache = set()

    def test(self, changes):
        abstract

    ###

    def getTestResult(self, changes):
        # There is no reason to cache successful tests because we will
        # always reduce the changeset when we see one.

        changeset = frozenset(changes)
        if changeset in self.cache:
            return False
        elif not self.test(changes):
            self.cache.add(changeset)
            return False
        else:
            return True

    def run(self, changes, force=False):
        # Make sure the initial test passes, if not then (a) either
        # the user doesn't expect monotonicity, and we may end up
        # doing O(N^2) tests, or (b) the test is wrong. Avoid the
        # O(N^2) case unless user requests it.
        if not force:
            if not self.getTestResult(changes):
                raise ValueError,'Initial test passed to delta fails.'

        # Check empty set first to quickly find poor test functions.
        if self.getTestResult(set()):
            return set()
        else:
            return self.delta(changes, self.split(changes))

    def split(self, S):
        """split(set) -> [sets]

        Partition a set into one or two pieces.
        """

        # There are many ways to split, we could do a better job with more
        # context information (but then the API becomes grosser).
        L = list(S)
        mid = len(L)//2
        if mid==0:
            return L,
        else:
            return L[:mid],L[mid:]
    
    def delta(self, c, sets):
        # assert(reduce(set.union, sets, set()) == c)

        # If there is nothing left we can remove, we are done.
        if len(sets) <= 1:
            return c
        
        # Look for a passing subset.
        res = self.search(c, sets)
        if res is not None:
            return res

        # Otherwise, partition sets if possible; if not we are done.
        refined = sum(map(list, map(self.split, sets)), [])
        if len(refined) == len(sets):
            return c
        
        return self.delta(c, refined)

    def search(self, c, sets):
        for i,S in enumerate(sets):
            # If test passes on this subset alone, recurse.
            if self.getTestResult(S):
                return self.delta(S, self.split(S))

            # Otherwise if we have more than two sets, see if test
            # pases without this subset.
            if len(sets) > 2:
                complement = sum(sets[:i] + sets[i+1:],[])
                if self.getTestResult(complement):
                    return self.delta(complement, sets[:i] + sets[i+1:])

###

class Token:
    def __init__(self, type, data, flags, file, line, column):
        self.type   = type
        self.data   = data
        self.flags  = flags
        self.file   = file
        self.line   = line
        self.column = column
        
kTokenRE = re.compile(r"""([a-z_]+) '(.*)'\t(.*)\tLoc=<(.*):(.*):(.*)>""",
                      re.DOTALL | re.MULTILINE)

def getTokens(path):
    p = subprocess.Popen(['clang','-dump-raw-tokens',path],
                         stdin=subprocess.PIPE,
                         stdout=subprocess.PIPE,
                         stderr=subprocess.PIPE)
    out,err = p.communicate()

    tokens = []
    collect = None
    for ln in err.split('\n'):
        # Silly programmers refuse to print in simple machine readable
        # formats. Whatever.
        if collect is None:
            collect = ln
        else:
            collect = collect + '\n' + ln
        if 'Loc=<' in ln and ln.endswith('>'):
            ln,collect = collect,None
            tokens.append(Token(*kTokenRE.match(ln).groups()))

    return tokens

###

class TMBDDelta(DeltaAlgorithm):
    def __init__(self, testProgram, tokenLists, log):
        def patchName(name, suffix):
            base,ext = os.path.splitext(name)
            return base + '.' + suffix + ext
        super(TMBDDelta, self).__init__()
        self.testProgram = testProgram
        self.tokenLists = tokenLists
        self.tempFiles = [patchName(f,'tmp')
                            for f,_ in self.tokenLists]
        self.targetFiles = [patchName(f,'ok')
                            for f,_ in self.tokenLists]
        self.log = log
        self.numTests = 0

    def writeFiles(self, changes, fileNames):
        assert len(fileNames) == len(self.tokenLists)
        byFile = [[] for i in self.tokenLists]
        for i,j in changes:
            byFile[i].append(j)

        for i,(file,tokens) in enumerate(self.tokenLists):
            f = open(fileNames[i],'w')
            for j in byFile[i]:
                f.write(tokens[j])
            f.close()

        return byFile

    def test(self, changes):
        self.numTests += 1

        byFile = self.writeFiles(changes, self.tempFiles)

        if self.log:
            print >>sys.stderr, 'TEST - ',
            if self.log > 1:
                for i,(file,_) in enumerate(self.tokenLists):
                    indices = byFile[i]
                    if i:
                        sys.stderr.write('\n      ')
                    sys.stderr.write('%s:%d tokens: [' % (file,len(byFile[i])))
                    prev = None
                    for j in byFile[i]:
                        if prev is None or j != prev + 1:
                            if prev:
                                sys.stderr.write('%d][' % prev)
                            sys.stderr.write(str(j))
                            sys.stderr.write(':')
                        prev = j
                    if byFile[i]:
                        sys.stderr.write(str(byFile[i][-1]))
                    sys.stderr.write('] ')
            else:
                print >>sys.stderr, ', '.join(['%s:%d tokens' % (file, len(byFile[i]))
                                               for i,(file,_) in enumerate(self.tokenLists)]),

        p = subprocess.Popen([self.testProgram] + self.tempFiles)
        res = p.wait() == 0

        if res:
            self.writeFiles(changes, self.targetFiles)

        if self.log:
            print >>sys.stderr, '=> %s' % res
        else:
            if res:
                print '\nSUCCESS (%d tokens)' % len(changes)
            else:                
                sys.stderr.write('.')

        return res

    def run(self):
        res = super(TMBDDelta, self).run([(i,j)
                                          for i,(file,tokens) in enumerate(self.tokenLists)
                                          for j in range(len(tokens))])
        self.writeFiles(res, self.targetFiles)
        if not self.log:
            print >>sys.stderr
        return res

def tokenBasedMultiDelta(program, files, log):            
    # Read in the lists of tokens.
    tokenLists = [(file, [t.data for t in getTokens(file)])
                  for file in files]

    numTokens = sum([len(tokens) for _,tokens in tokenLists])
    print "Delta on %s with %d tokens." % (', '.join(files), numTokens)
    
    tbmd = TMBDDelta(program, tokenLists, log)

    res = tbmd.run()

    print "Finished %s with %d tokens (in %d tests)." % (', '.join(tbmd.targetFiles),
                                                         len(res),
                                                         tbmd.numTests)
        
def main():
    from optparse import OptionParser, OptionGroup
    parser = OptionParser("%prog <test program> {files+}")
    parser.add_option("", "--debug", dest="debugLevel",
                     help="set debug level [default %default]",
                     action="store", type=int, default=0)
    (opts, args) = parser.parse_args()

    if len(args) <= 1:
        parser.error('Invalid number of arguments.')
        
    program,files = args[0],args[1:]

    md = tokenBasedMultiDelta(program, files, log=opts.debugLevel)
        
if __name__ == '__main__':
    try:
        main()
    except KeyboardInterrupt:
        print >>sys.stderr,'Interrupted.'
        os._exit(1) # Avoid freeing our giant cache.
OpenPOWER on IntegriCloud