compilerplugins/clang/constantparam.py

   1 #!/usr/bin/python
   2
   3 import sys
   4 import re
   5 import io
   6
   7 callDict = dict() # callInfo tuple -> callValue
   8
   9 # clang does not always use exactly the same numbers in the type-parameter vars it generates
  10 # so I need to substitute them to ensure we can match correctly.
  11 normalizeTypeParamsRegex = re.compile(r"type-parameter-\d+-\d+")
  12 def normalizeTypeParams( line ):
  13     return normalizeTypeParamsRegex.sub("type-parameter-?-?", line)
  14
  15 # reading as binary (since we known it is pure ascii) is much faster than reading as unicode
  16 with io.open("workdir/loplugin.constantparam.log", "rb", buffering=1024*1024) as txt:
  17     for line in txt:
  18         try:
  19             tokens = line.strip().split("\t")
  20             returnType = normalizeTypeParams(tokens[0])
  21             nameAndParams = normalizeTypeParams(tokens[1])
  22             sourceLocation = tokens[2]
  23             paramName = tokens[3]
  24             paramType = normalizeTypeParams(tokens[4])
  25             callValue = tokens[5]
  26             callInfo = (returnType, nameAndParams, paramName, paramType, sourceLocation)
  27             if not callInfo in callDict:
  28                 callDict[callInfo] = set()
  29             callDict[callInfo].add(callValue)
  30         except IndexError:
  31             print "problem with line " + line.strip()
  32             raise
  33
  34 def RepresentsInt(s):
  35     try:
  36         int(s)
  37         return True
  38     except ValueError:
  39         return False
  40
  41 constructor_regex = re.compile("^\w+\(\)$")
  42
  43 tmp1list = list()
  44 tmp2list = list()
  45 tmp3list = list()
  46 tmp4list = list()
  47 for callInfo, callValues in callDict.iteritems():
  48     nameAndParams = callInfo[1]
  49     if len(callValues) != 1:
  50         continue
  51     callValue = next(iter(callValues))
  52     if "unknown" in callValue:
  53         continue
  54     sourceLoc = callInfo[4]
  55     functionSig = callInfo[0] + " " + callInfo[1]
  56
  57     # try to ignore setter methods
  58     if ("," not in nameAndParams) and (("::set" in nameAndParams) or ("::Set" in nameAndParams)):
  59         continue
  60     # ignore code that follows a common pattern
  61     if sourceLoc.startswith("sw/inc/swatrset.hxx"): continue
  62     if sourceLoc.startswith("sw/inc/format.hxx"): continue
  63     # template generated code
  64     if sourceLoc.startswith("include/sax/fshelper.hxx"): continue
  65     # debug code
  66     if sourceLoc.startswith("include/oox/dump"): continue
  67     # part of our binary API
  68     if sourceLoc.startswith("include/LibreOfficeKit"): continue
  69
  70     # ignore methods generated by SFX macros
  71     if "RegisterInterface(class SfxModule *)" in nameAndParams: continue
  72     if "RegisterChildWindow(_Bool,class SfxModule *,enum SfxChildWindowFlags)" in nameAndParams: continue
  73     if "RegisterControl(unsigned short,class SfxModule *)" in nameAndParams: continue
  74
  75     if RepresentsInt(callValue):
  76         if callValue == "0" or callValue == "1":
  77             tmp1list.append((sourceLoc, functionSig, callInfo[3] + " " + callInfo[2], callValue))
  78         else:
  79             tmp2list.append((sourceLoc, functionSig, callInfo[3] + " " + callInfo[2], callValue))
  80     # look for places where the callsite is always a constructor invocation
  81     elif constructor_regex.match(callValue) or callValue == "\"\"":
  82         if callValue.startswith("Get"): continue
  83         if callValue.startswith("get"): continue
  84         if "operator=" in functionSig: continue
  85         if "&&" in functionSig: continue
  86         if callInfo[2] == "###0" and callValue == "InitData()": continue
  87         if callInfo[2] == "###0" and callValue == "InitAggregate()": continue
  88         if callValue == "shared_from_this()": continue
  89         tmp3list.append((sourceLoc, functionSig, callInfo[3] + " " + callInfo[2], callValue))
  90     else:
  91         tmp4list.append((sourceLoc, functionSig, callInfo[3] + " " + callInfo[2], callValue))
  92
  93
  94 # sort results by filename:lineno
  95 def natural_sort_key(s, _nsre=re.compile('([0-9]+)')):
  96     return [int(text) if text.isdigit() else text.lower()
  97             for text in re.split(_nsre, s)]
  98 tmp1list.sort(key=lambda v: natural_sort_key(v[0]))
  99 tmp2list.sort(key=lambda v: natural_sort_key(v[0]))
 100 tmp3list.sort(key=lambda v: natural_sort_key(v[0]))
 101 tmp4list.sort(key=lambda v: natural_sort_key(v[0]))
 102
 103 # print out the results
 104 with open("compilerplugins/clang/constantparam.booleans.results", "wt") as f:
 105     for v in tmp1list:
 106         f.write(v[0] + "\n")
 107         f.write("    " + v[1] + "\n")
 108         f.write("    " + v[2] + "\n")
 109         f.write("    " + v[3] + "\n")
 110 with open("compilerplugins/clang/constantparam.numbers.results", "wt") as f:
 111     for v in tmp2list:
 112         f.write(v[0] + "\n")
 113         f.write("    " + v[1] + "\n")
 114         f.write("    " + v[2] + "\n")
 115         f.write("    " + v[3] + "\n")
 116 with open("compilerplugins/clang/constantparam.constructors.results", "wt") as f:
 117     for v in tmp3list:
 118         f.write(v[0] + "\n")
 119         f.write("    " + v[1] + "\n")
 120         f.write("    " + v[2] + "\n")
 121         f.write("    " + v[3] + "\n")
 122 with open("compilerplugins/clang/constantparam.others.results", "wt") as f:
 123     for v in tmp4list:
 124         f.write(v[0] + "\n")
 125         f.write("    " + v[1] + "\n")
 126         f.write("    " + v[2] + "\n")
 127         f.write("    " + v[3] + "\n")
 128
 129 # -------------------------------------------------------------
 130 # Now a fun set of heuristics to look for methods that
 131 # take bitmask parameters where one or more of the bits in the
 132 # bitmask is always one or always zero
 133
 134 # integer to hex str
 135 def hex(i):
 136     return "0x%x" % i
 137 # I can't use python's ~ operator, because that produces negative numbers
 138 def negate(i):
 139     return (1 << 32) - 1 - i
 140
 141 tmp2list = list()
 142 for callInfo, callValues in callDict.iteritems():
 143     nameAndParams = callInfo[1]
 144     if len(callValues) < 2:
 145         continue
 146     # we are only interested in enum parameters
 147     if not "enum" in callInfo[3]: continue
 148     if not "Flag" in callInfo[3] and not "flag" in callInfo[3] and not "Bit" in callInfo[3] and not "State" in callInfo[3]: continue
 149     # try to ignore setter methods
 150     if ("," not in nameAndParams) and (("::set" in nameAndParams) or ("::Set" in nameAndParams)):
 151         continue
 152
 153     setBits = 0
 154     clearBits = 0
 155     continue_flag = False
 156     first = True
 157     for callValue in callValues:
 158         if "unknown" == callValue or not callValue.isdigit():
 159             continue_flag = True
 160             break
 161         if first:
 162             setBits = int(callValue)
 163             clearBits = negate(int(callValue))
 164             first = False
 165         else:
 166             setBits = setBits & int(callValue)
 167             clearBits = clearBits & negate(int(callValue))
 168
 169     # estimate allBits by using the highest bit we have seen
 170     # TODO dump more precise information about the allBits values of enums
 171     allBits = (1 << setBits.bit_length()) - 1
 172     clearBits = clearBits & allBits
 173     if continue_flag or (setBits == 0 and clearBits == 0): continue
 174
 175     sourceLoc = callInfo[4]
 176     functionSig = callInfo[0] + " " + callInfo[1]
 177
 178     v2 = callInfo[3] + " " + callInfo[2]
 179     if setBits != 0: v2 += " setBits=" + hex(setBits)
 180     if clearBits != 0: v2 += " clearBits=" + hex(clearBits)
 181     tmp2list.append((sourceLoc, functionSig, v2))
 182
 183
 184 # sort results by filename:lineno
 185 tmp2list.sort(key=lambda v: natural_sort_key(v[0]))
 186
 187 # print out the results
 188 with open("compilerplugins/clang/constantparam.bitmask.results", "wt") as f:
 189     for v in tmp2list:
 190         f.write(v[0] + "\n")
 191         f.write("    " + v[1] + "\n")
 192         f.write("    " + v[2] + "\n")