compilerplugins/clang/constantparam.py

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