compilerplugins/clang/constantparam.py

   1 #!/usr/bin/python3
   2
   3 import re
   4 import io
   5
   6 callDict = dict() # callInfo tuple -> callValue
   7
   8 # clang does not always use exactly the same numbers in the type-parameter vars it generates
   9 # so I need to substitute them to ensure we can match correctly.
  10 normalizeTypeParamsRegex = re.compile(r"type-parameter-\d+-\d+")
  11 def normalizeTypeParams( line ):
  12     return normalizeTypeParamsRegex.sub("type-parameter-?-?", line)
  13
  14 # reading as binary (since we known it is pure ascii) is much faster than reading as unicode
  15 with io.open("workdir/loplugin.constantparam.log", "r") as txt:
  16     line_no = 1;
  17     try:
  18         for line in txt:
  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             line_no += 1
  31     except (IndexError,UnicodeDecodeError):
  32         print("problem with line " + str(line_no))
  33         raise
  34
  35 def RepresentsInt(s):
  36     try:
  37         int(s)
  38         return True
  39     except ValueError:
  40         return False
  41
  42 constructor_regex = re.compile(r"^\w+\(\)$")
  43
  44 tmp1list = list()
  45 tmp2list = list()
  46 tmp3list = list()
  47 tmp4list = list()
  48 for callInfo, callValues in iter(callDict.items()):
  49     nameAndParams = callInfo[1]
  50     if len(callValues) != 1:
  51         continue
  52     callValue = next(iter(callValues))
  53     if "unknown" in callValue:
  54         continue
  55     sourceLoc = callInfo[4]
  56     functionSig = callInfo[0] + " " + callInfo[1]
  57
  58     # try to ignore setter methods
  59     if ("," not in nameAndParams) and (("::set" in nameAndParams) or ("::Set" in nameAndParams)):
  60         continue
  61     # ignore code that follows a common pattern
  62     if sourceLoc.startswith("sw/inc/swatrset.hxx"): continue
  63     if sourceLoc.startswith("sw/inc/format.hxx"): continue
  64     # template generated code
  65     if sourceLoc.startswith("include/sax/fshelper.hxx"): continue
  66     # debug code
  67     if sourceLoc.startswith("include/oox/dump"): continue
  68     # part of our binary API
  69     if sourceLoc.startswith("include/LibreOfficeKit"): continue
  70
  71     # ignore methods generated by SFX macros
  72     if "RegisterInterface(class SfxModule *)" in nameAndParams: continue
  73     if "RegisterChildWindow(_Bool,class SfxModule *,enum SfxChildWindowFlags)" in nameAndParams: continue
  74     if "RegisterControl(unsigned short,class SfxModule *)" in nameAndParams: continue
  75
  76     if RepresentsInt(callValue):
  77         if callValue == "0" or callValue == "1":
  78             tmp1list.append((sourceLoc, functionSig, callInfo[3] + " " + callInfo[2], callValue))
  79         else:
  80             tmp2list.append((sourceLoc, functionSig, callInfo[3] + " " + callInfo[2], callValue))
  81     # look for places where the callsite is always a constructor invocation
  82     elif constructor_regex.match(callValue) or callValue == "\"\"":
  83         if callValue.startswith("Get"): continue
  84         if callValue.startswith("get"): continue
  85         if "operator=" in functionSig: continue
  86         if "&&" in functionSig: continue
  87         if callInfo[2] == "###0" and callValue == "InitData()": continue
  88         if callInfo[2] == "###0" and callValue == "InitAggregate()": continue
  89         if callValue == "shared_from_this()": continue
  90         tmp3list.append((sourceLoc, functionSig, callInfo[3] + " " + callInfo[2], callValue))
  91     else:
  92         tmp4list.append((sourceLoc, functionSig, callInfo[3] + " " + callInfo[2], callValue))
  93
  94
  95 # sort results by filename:lineno
  96 def natural_sort_key(s, _nsre=re.compile('([0-9]+)')):
  97     return [int(text) if text.isdigit() else text.lower()
  98             for text in re.split(_nsre, s)]
  99 # sort by both the source-line and the datatype, so the output file ordering is stable
 100 # when we have multiple items on the same source line
 101 def v_sort_key(v):
 102     return natural_sort_key(v[0]) + [v[1]]
 103 tmp1list.sort(key=lambda v: v_sort_key(v))
 104 tmp2list.sort(key=lambda v: v_sort_key(v))
 105 tmp3list.sort(key=lambda v: v_sort_key(v))
 106 tmp4list.sort(key=lambda v: v_sort_key(v))
 107
 108 # print out the results
 109 with open("compilerplugins/clang/constantparam.booleans.results", "wt") as f:
 110     for v in tmp1list:
 111         f.write(v[0] + "\n")
 112         f.write("    " + v[1] + "\n")
 113         f.write("    " + v[2] + "\n")
 114         f.write("    " + v[3] + "\n")
 115 with open("compilerplugins/clang/constantparam.numbers.results", "wt") as f:
 116     for v in tmp2list:
 117         f.write(v[0] + "\n")
 118         f.write("    " + v[1] + "\n")
 119         f.write("    " + v[2] + "\n")
 120         f.write("    " + v[3] + "\n")
 121 with open("compilerplugins/clang/constantparam.constructors.results", "wt") as f:
 122     for v in tmp3list:
 123         f.write(v[0] + "\n")
 124         f.write("    " + v[1] + "\n")
 125         f.write("    " + v[2] + "\n")
 126         f.write("    " + v[3] + "\n")
 127 with open("compilerplugins/clang/constantparam.others.results", "wt") as f:
 128     for v in tmp4list:
 129         f.write(v[0] + "\n")
 130         f.write("    " + v[1] + "\n")
 131         f.write("    " + v[2] + "\n")
 132         f.write("    " + v[3] + "\n")
 133
 134 # -------------------------------------------------------------
 135 # Now a fun set of heuristics to look for methods that
 136 # take bitmask parameters where one or more of the bits in the
 137 # bitmask is always one or always zero
 138
 139 # integer to hex str
 140 def hex(i):
 141     return "0x%x" % i
 142 # I can't use python's ~ operator, because that produces negative numbers
 143 def negate(i):
 144     return (1 << 32) - 1 - i
 145
 146 tmp2list = list()
 147 for callInfo, callValues in iter(callDict.items()):
 148     nameAndParams = callInfo[1]
 149     if len(callValues) < 2:
 150         continue
 151     # we are only interested in enum parameters
 152     if not "enum" in callInfo[3]: continue
 153     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
 154     # try to ignore setter methods
 155     if ("," not in nameAndParams) and (("::set" in nameAndParams) or ("::Set" in nameAndParams)):
 156         continue
 157
 158     setBits = 0
 159     clearBits = 0
 160     continue_flag = False
 161     first = True
 162     for callValue in callValues:
 163         if "unknown" == callValue or not callValue.isdigit():
 164             continue_flag = True
 165             break
 166         if first:
 167             setBits = int(callValue)
 168             clearBits = negate(int(callValue))
 169             first = False
 170         else:
 171             setBits = setBits & int(callValue)
 172             clearBits = clearBits & negate(int(callValue))
 173
 174     # estimate allBits by using the highest bit we have seen
 175     # TODO dump more precise information about the allBits values of enums
 176     allBits = (1 << setBits.bit_length()) - 1
 177     clearBits = clearBits & allBits
 178     if continue_flag or (setBits == 0 and clearBits == 0): continue
 179
 180     sourceLoc = callInfo[4]
 181     functionSig = callInfo[0] + " " + callInfo[1]
 182
 183     v2 = callInfo[3] + " " + callInfo[2]
 184     if setBits != 0: v2 += " setBits=" + hex(setBits)
 185     if clearBits != 0: v2 += " clearBits=" + hex(clearBits)
 186     tmp2list.append((sourceLoc, functionSig, v2))
 187
 188
 189 # sort results by filename:lineno
 190 tmp2list.sort(key=lambda v: v_sort_key(v))
 191
 192 # print out the results
 193 with open("compilerplugins/clang/constantparam.bitmask.results", "wt") as f:
 194     for v in tmp2list:
 195         f.write(v[0] + "\n")
 196         f.write("    " + v[1] + "\n")
 197         f.write("    " + v[2] + "\n")