summaryrefslogtreecommitdiff
path: root/compilerplugins/clang/unusedfields.py
blob: 763a11743acb4b24f28d53a83c32c67e30325b99 (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
#!/usr/bin/python

import sys
import re
import io

definitionSet = set()
definitionToSourceLocationMap = dict()
definitionToTypeMap = dict()
callSet = set()
readFromSet = set()
sourceLocationSet = set()
# things we need to exclude for reasons like :
# - it's a weird template thingy that confuses the plugin
exclusionSet = set([
    ])

# clang does not always use exactly the same numbers in the type-parameter vars it generates
# so I need to substitute them to ensure we can match correctly.
normalizeTypeParamsRegex = re.compile(r"type-parameter-\d+-\d+")
def normalizeTypeParams( line ):
    return normalizeTypeParamsRegex.sub("type-parameter-?-?", line)

# The parsing here is designed to avoid grabbing stuff which is mixed in from gbuild.
# I have not yet found a way of suppressing the gbuild output.
with io.open(sys.argv[1], "rb", buffering=1024*1024) as txt:
    for line in txt:
        if line.startswith("definition:\t"):
            idx1 = line.find("\t",12)
            idx2 = line.find("\t",idx1+1)
            idx3 = line.find("\t",idx2+1)
            funcInfo = (normalizeTypeParams(line[12:idx1]), line[idx1+1:idx2])
            definitionSet.add(funcInfo)
            definitionToTypeMap[funcInfo] = line[idx2+1:idx3].strip()
            definitionToSourceLocationMap[funcInfo] = line[idx3+1:].strip()
        elif line.startswith("touch:\t"):
            idx1 = line.find("\t",7)
            callInfo = (normalizeTypeParams(line[7:idx1]), line[idx1+1:].strip())
            callSet.add(callInfo)
        elif line.startswith("read:\t"):
            idx1 = line.find("\t",6)
            readInfo = (normalizeTypeParams(line[6:idx1]), line[idx1+1:].strip())
            readFromSet.add(readInfo)

# Invert the definitionToSourceLocationMap
# If we see more than one method at the same sourceLocation, it's being autogenerated as part of a template
# and we should just ignore
sourceLocationToDefinitionMap = {}
for k, v in definitionToSourceLocationMap.iteritems():
    sourceLocationToDefinitionMap[v] = sourceLocationToDefinitionMap.get(v, [])
    sourceLocationToDefinitionMap[v].append(k)
for k, definitions in sourceLocationToDefinitionMap.iteritems():
    if len(definitions) > 1:
        for d in definitions:
            definitionSet.remove(d)
    
untouchedSet = set()
for d in definitionSet:
    clazz = d[0] + " " + d[1]
    if clazz in exclusionSet:
        continue
    if d in callSet:
        continue
    srcLoc = definitionToSourceLocationMap[d];
    # ignore external source code
    if (srcLoc.startswith("external/")):
        continue
    # ignore build folder
    if (srcLoc.startswith("workdir/")):
        continue
    # ignore our stable/URE/UNO api
    if (srcLoc.startswith("include/com/")
        or srcLoc.startswith("include/cppu/")
        or srcLoc.startswith("include/cppuhelper/")
        or srcLoc.startswith("include/osl/")
        or srcLoc.startswith("include/rtl/")
        or srcLoc.startswith("include/sal/")
        or srcLoc.startswith("include/salhelper/")
        or srcLoc.startswith("include/systools/")
        or srcLoc.startswith("include/typelib/")
        or srcLoc.startswith("include/uno/")):
        continue
    # this is all representations of on-disk data structures
    if (srcLoc.startswith("sc/source/filter/inc/scflt.hxx")
        or srcLoc.startswith("sw/source/filter/ww8/")
        or srcLoc.startswith("vcl/source/filter/sgvmain.hxx")
        or srcLoc.startswith("vcl/source/filter/sgfbram.hxx")
        or srcLoc.startswith("vcl/inc/unx/XIM.h")
        or srcLoc.startswith("vcl/inc/unx/gtk/gloactiongroup.h")
        or srcLoc.startswith("include/svl/svdde.hxx")
        or srcLoc.startswith("lotuswordpro/source/filter/lwpsdwdrawheader.hxx")
        or srcLoc.startswith("svtools/source/dialogs/insdlg.cxx")):
        continue
    untouchedSet.add((clazz + " " + definitionToTypeMap[d], srcLoc))

writeonlySet = set()
for d in definitionSet:
    clazz = d[0] + " " + d[1]
    if d in readFromSet:
        continue
    srcLoc = definitionToSourceLocationMap[d];
    # ignore external source code
    if (srcLoc.startswith("external/")):
        continue
    # ignore build folder
    if (srcLoc.startswith("workdir/")):
        continue
    # ignore our stable/URE/UNO api
    if (srcLoc.startswith("include/com/")
        or srcLoc.startswith("include/cppu/")
        or srcLoc.startswith("include/cppuhelper/")
        or srcLoc.startswith("include/osl/")
        or srcLoc.startswith("include/rtl/")
        or srcLoc.startswith("include/sal/")
        or srcLoc.startswith("include/salhelper/")
        or srcLoc.startswith("include/systools/")
        or srcLoc.startswith("include/typelib/")
        or srcLoc.startswith("include/uno/")):
        continue
    # this is all representations of on-disk data structures
    if (srcLoc.startswith("sc/source/filter/inc/scflt.hxx")
        or srcLoc.startswith("sw/source/filter/ww8/")
        or srcLoc.startswith("vcl/source/filter/sgvmain.hxx")
        or srcLoc.startswith("vcl/source/filter/sgfbram.hxx")
        or srcLoc.startswith("vcl/inc/unx/XIM.h")
        or srcLoc.startswith("vcl/inc/unx/gtk/gloactiongroup.h")
        or srcLoc.startswith("include/svl/svdde.hxx")
        or srcLoc.startswith("lotuswordpro/source/filter/lwpsdwdrawheader.hxx")
        or srcLoc.startswith("svtools/source/dialogs/insdlg.cxx")):
        continue

    writeonlySet.add((clazz + " " + definitionToTypeMap[d], srcLoc))

# sort the results using a "natural order" so sequences like [item1,item2,item10] sort nicely
def natural_sort_key(s, _nsre=re.compile('([0-9]+)')):
    return [int(text) if text.isdigit() else text.lower()
            for text in re.split(_nsre, s)]

# sort results by name and line number
tmp1list = sorted(untouchedSet, key=lambda v: natural_sort_key(v[1]))
tmp2list = sorted(writeonlySet, key=lambda v: natural_sort_key(v[1]))

# print out the results
with open("unusedfields.untouched", "wt") as f:
    for t in tmp1list:
        f.write( t[1] + "\n" )
        f.write( "    " + t[0] + "\n" )
with open("unusedfields.writeonly", "wt") as f:
    for t in tmp2list:
        f.write( t[1] + "\n" )
        f.write( "    " + t[0] + "\n" )

    

# add an empty line at the end to make it easier for the unusedFieldsremove plugin to mmap() the output file 
print