compilerplugins/clang/unusedfields.py

   1 #!/usr/bin/python
   2
   3 import sys
   4 import re
   5 import io
   6
   7 definitionSet = set()
   8 protectedAndPublicDefinitionSet = set() # set of tuple(type, name)
   9 definitionToSourceLocationMap = dict()
  10 definitionToTypeMap = dict()
  11 callSet = set()
  12 readFromSet = set()
  13 sourceLocationSet = set()
  14 touchedFromOutsideSet = set()
  15
  16 # clang does not always use exactly the same numbers in the type-parameter vars it generates
  17 # so I need to substitute them to ensure we can match correctly.
  18 normalizeTypeParamsRegex = re.compile(r"type-parameter-\d+-\d+")
  19 def normalizeTypeParams( line ):
  20     return normalizeTypeParamsRegex.sub("type-parameter-?-?", line)
  21
  22 def parseFieldInfo( tokens ):
  23     if len(tokens) == 3:
  24         return (normalizeTypeParams(tokens[1]), tokens[2])
  25     else:
  26         return (normalizeTypeParams(tokens[1]), "")
  27
  28 # The parsing here is designed to avoid grabbing stuff which is mixed in from gbuild.
  29 # I have not yet found a way of suppressing the gbuild output.
  30 with io.open("loplugin.unusedfields.log", "rb", buffering=1024*1024) as txt:
  31     for line in txt:
  32         tokens = line.strip().split("\t")
  33         if tokens[0] == "definition:":
  34             access = tokens[1]
  35             fieldInfo = (normalizeTypeParams(tokens[2]), tokens[3])
  36             srcLoc = tokens[5]
  37             # ignore external source code
  38             if (srcLoc.startswith("external/")):
  39                 continue
  40             # ignore build folder
  41             if (srcLoc.startswith("workdir/")):
  42                 continue
  43             definitionSet.add(fieldInfo)
  44             definitionToTypeMap[fieldInfo] = tokens[4]
  45             if access == "protected" or access == "public":
  46                 protectedAndPublicDefinitionSet.add(fieldInfo)
  47             definitionToSourceLocationMap[fieldInfo] = tokens[5]
  48         elif tokens[0] == "touch:":
  49             callSet.add(parseFieldInfo(tokens))
  50         elif tokens[0] == "read:":
  51             readFromSet.add(parseFieldInfo(tokens))
  52         elif tokens[0] == "read:":
  53             readFromSet.add(parseFieldInfo(tokens))
  54         elif tokens[0] == "outside:":
  55             touchedFromOutsideSet.add(parseFieldInfo(tokens))
  56         else:
  57             print( "unknown line: " + line)
  58
  59 # Invert the definitionToSourceLocationMap
  60 # If we see more than one method at the same sourceLocation, it's being autogenerated as part of a template
  61 # and we should just ignore
  62 sourceLocationToDefinitionMap = {}
  63 for k, v in definitionToSourceLocationMap.iteritems():
  64     sourceLocationToDefinitionMap[v] = sourceLocationToDefinitionMap.get(v, [])
  65     sourceLocationToDefinitionMap[v].append(k)
  66 for k, definitions in sourceLocationToDefinitionMap.iteritems():
  67     if len(definitions) > 1:
  68         for d in definitions:
  69             definitionSet.remove(d)
  70
  71 untouchedSet = set()
  72 for d in definitionSet:
  73     if d in callSet:
  74         continue
  75     srcLoc = definitionToSourceLocationMap[d];
  76     # this is all representations of on-disk data structures
  77     if (srcLoc.startswith("sc/source/filter/inc/scflt.hxx")
  78         or srcLoc.startswith("sw/source/filter/ww8/")
  79         or srcLoc.startswith("vcl/source/filter/sgvmain.hxx")
  80         or srcLoc.startswith("vcl/source/filter/sgfbram.hxx")
  81         or srcLoc.startswith("vcl/inc/unx/XIM.h")
  82         or srcLoc.startswith("vcl/inc/unx/gtk/gloactiongroup.h")
  83         or srcLoc.startswith("include/svl/svdde.hxx")
  84         or srcLoc.startswith("lotuswordpro/source/filter/lwpsdwdrawheader.hxx")
  85         or srcLoc.startswith("hwpfilter/")
  86         or srcLoc.startswith("embeddedobj/source/inc/")
  87         or srcLoc.startswith("svtools/source/dialogs/insdlg.cxx")
  88         or srcLoc.startswith("bridges/")):
  89         continue
  90     if d[0] in set([ "AtkObjectWrapperClass", "AtkObjectWrapper", "GLOMenu", "GLOAction", "_XRegion", "SalMenuButtonItem", "Vertex",
  91             "OOoMountOperationClass", "SwCSS1ItemIds", "ScCompiler::AddInMap", "MemoryByteGrabber", "textcat_t", "fp_t", "ngram_t",
  92             "ImplPPTParaPropSet", "DataNode"]):
  93         continue
  94     # unit testing code
  95     if srcLoc.startswith("cppu/source/uno/check.cxx"):
  96         continue
  97     fieldType = definitionToTypeMap[d]
  98     if fieldType in set([ "class rptui::OModuleClient" ]):
  99         continue
 100     untouchedSet.add((d[0] + " " + d[1] + " " + fieldType, srcLoc))
 101
 102 writeonlySet = set()
 103 for d in definitionSet:
 104     clazz = d[0] + " " + d[1]
 105     if d in readFromSet:
 106         continue
 107     srcLoc = definitionToSourceLocationMap[d];
 108     # this is all representations of on-disk data structures
 109     if (srcLoc.startswith("sc/source/filter/inc/scflt.hxx")
 110         or srcLoc.startswith("sw/source/filter/ww8/")
 111         or srcLoc.startswith("vcl/source/filter/sgvmain.hxx")
 112         or srcLoc.startswith("vcl/source/filter/sgfbram.hxx")
 113         or srcLoc.startswith("vcl/inc/unx/XIM.h")
 114         or srcLoc.startswith("vcl/inc/unx/gtk/gloactiongroup.h")
 115         or srcLoc.startswith("include/svl/svdde.hxx")
 116         or srcLoc.startswith("lotuswordpro/source/filter/lwpsdwdrawheader.hxx")
 117         or srcLoc.startswith("svtools/source/dialogs/insdlg.cxx")):
 118         continue
 119
 120     writeonlySet.add((clazz + " " + definitionToTypeMap[d], srcLoc))
 121
 122
 123 canBePrivateSet = set()
 124 for d in protectedAndPublicDefinitionSet:
 125     clazz = d[0] + " " + d[1]
 126     if d in touchedFromOutsideSet:
 127         continue
 128     srcLoc = definitionToSourceLocationMap[d];
 129
 130     canBePrivateSet.add((clazz + " " + definitionToTypeMap[d], srcLoc))
 131
 132
 133 # sort the results using a "natural order" so sequences like [item1,item2,item10] sort nicely
 134 def natural_sort_key(s, _nsre=re.compile('([0-9]+)')):
 135     return [int(text) if text.isdigit() else text.lower()
 136             for text in re.split(_nsre, s)]
 137
 138 # sort results by name and line number
 139 tmp1list = sorted(untouchedSet, key=lambda v: natural_sort_key(v[1]))
 140 tmp2list = sorted(writeonlySet, key=lambda v: natural_sort_key(v[1]))
 141 tmp3list = sorted(canBePrivateSet, key=lambda v: natural_sort_key(v[1]))
 142
 143 # print out the results
 144 with open("loplugin.unusedfields.report-untouched", "wt") as f:
 145     for t in tmp1list:
 146         f.write( t[1] + "\n" )
 147         f.write( "    " + t[0] + "\n" )
 148 with open("loplugin.unusedfields.report-writeonly", "wt") as f:
 149     for t in tmp2list:
 150         f.write( t[1] + "\n" )
 151         f.write( "    " + t[0] + "\n" )
 152 with open("loplugin.unusedfields.report-can-be-private", "wt") as f:
 153     for t in tmp3list:
 154         f.write( t[1] + "\n" )
 155         f.write( "    " + t[0] + "\n" )
 156
 157