compilerplugins/clang/virtualdead.py

   1 #!/usr/bin/python3
   2
   3 import re
   4 import io
   5
   6 callDict = dict() # callInfo tuple -> callValue
   7 definitionToSourceLocationMap = dict()
   8 paramSet = set() # paraminfo tuple
   9
  10 # clang does not always use exactly the same numbers in the type-parameter vars it generates
  11 # so I need to substitute them to ensure we can match correctly.
  12 normalizeTypeParamsRegex = re.compile(r"type-parameter-\d+-\d+")
  13 def normalizeTypeParams( line ):
  14     return normalizeTypeParamsRegex.sub("type-parameter-?-?", line)
  15
  16 # reading as binary (since we known it is pure ascii) is much faster than reading as unicode
  17 with io.open("workdir/loplugin.virtualdead.log", "r", encoding="ascii", errors="ignore", buffering=1024*1024) as txt:
  18     for line in txt:
  19         try:
  20             tokens = line.strip().split("\t")
  21             if tokens[0] == "virtual:":
  22                 nameAndParams = normalizeTypeParams(tokens[1])
  23                 sourceLocation = tokens[2]
  24                 returnValue = tokens[3]
  25                 callInfo = (nameAndParams, sourceLocation)
  26                 if callInfo not in callDict:
  27                     callDict[callInfo] = set()
  28                 callDict[callInfo].add(returnValue)
  29                 definitionToSourceLocationMap[nameAndParams] = sourceLocation
  30             elif tokens[0] == "param:":
  31                 name = normalizeTypeParams(tokens[1])
  32                 if len(tokens)>2:
  33                     bitfield = tokens[2]
  34                     paramSet.add((name,bitfield))
  35             else:
  36                 print( "unknown line: " + line)
  37         except IndexError:
  38             print("problem with line " + line.strip())
  39             raise
  40
  41 tmp1list = list()
  42 for callInfo, callValues in iter(callDict.items()):
  43     nameAndParams = callInfo[1]
  44     if len(callValues) != 1:
  45         continue
  46     callValue = next(iter(callValues))
  47     if "unknown-stmt" in callValue:
  48         continue
  49     if "unknown2" in callValue:
  50         continue
  51     if "unknown3" in callValue:
  52         continue
  53     if "unknown4" in callValue:
  54         continue
  55     if "pure" in callValue:
  56         continue
  57     srcloc = callInfo[1]
  58     if srcloc.startswith("workdir/"): continue
  59     # ignore Qt stuff
  60     if srcloc.startswith("Gui/"): continue
  61     if srcloc.startswith("Widgets/"): continue
  62     if srcloc.startswith("Core/"): continue
  63     if srcloc.startswith("/Qt"): continue
  64     if srcloc.startswith("Qt"): continue
  65     if srcloc.startswith("64-"): continue
  66     functionSig = callInfo[0]
  67     tmp1list.append((srcloc, functionSig, callValue))
  68
  69 def merge_bitfield(a, b):
  70     if len(a) == 0: return b
  71     ret = ""
  72     for i, c in enumerate(b):
  73         if c == "1" or a[i] == "1":
  74             ret += "1"
  75         else:
  76             ret += "0"
  77     return ret
  78 tmp2dict = dict()
  79 tmp2list = list()
  80 for paramInfo in paramSet:
  81     name = paramInfo[0]
  82     bitfield = paramInfo[1]
  83     if re.match( r"\w+ com::", name): continue
  84     if re.match( r"\w+ ooo::vba::", name): continue
  85     if re.match( r"\w+ orcus::", name): continue
  86     if re.match( r"\w+ std::", name): continue
  87     if name not in tmp2dict:
  88         tmp2dict[name] = bitfield
  89     else:
  90         tmp2dict[name] = merge_bitfield(tmp2dict[name], bitfield)
  91 for name, bitfield in iter(tmp2dict.items()):
  92     srcloc = definitionToSourceLocationMap[name]
  93     # ignore Qt stuff
  94     if srcloc.startswith("Gui/"): continue
  95     if srcloc.startswith("Widgets/"): continue
  96     if srcloc.startswith("Core/"): continue
  97     if srcloc.startswith("/Qt"): continue
  98     if srcloc.startswith("Qt"): continue
  99     if srcloc.startswith("64-"): continue
 100     # ignore external stuff
 101     if srcloc.startswith("workdir/"): continue
 102     # referenced by generated code in workdir/
 103     if srcloc.startswith("sw/source/writerfilter/ooxml/OOXMLFactory.hxx"): continue
 104     if "0" in bitfield:
 105         tmp2list.append((srcloc, name, bitfield))
 106
 107 # sort results by filename:lineno
 108 def natural_sort_key(s, _nsre=re.compile('([0-9]+)')):
 109     return [int(text) if text.isdigit() else text.lower()
 110             for text in re.split(_nsre, s)]
 111 # sort by both the source-line and the datatype, so the output file ordering is stable
 112 # when we have multiple items on the same source line
 113 def v_sort_key(v):
 114     return natural_sort_key(v[0]) + [v[1]]
 115 tmp1list.sort(key=lambda v: v_sort_key(v))
 116 tmp2list.sort(key=lambda v: v_sort_key(v))
 117
 118 # print out the results
 119 with open("compilerplugins/clang/virtualdead.results", "wt") as f:
 120     for v in tmp1list:
 121         f.write(v[0] + "\n")
 122         f.write("    " + v[1] + "\n")
 123         f.write("    " + v[2] + "\n")
 124 with open("compilerplugins/clang/virtualdead.unusedparams.results", "wt") as f:
 125     for v in tmp2list:
 126         f.write(v[0] + "\n")
 127         f.write("    " + v[1] + "\n")
 128         f.write("    " + v[2] + "\n")