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 touchedFromInsideSet = set()
  12 touchedFromOutsideSet = set()
  13 touchedFromOutsideConstructorSet = set()
  14 readFromSet = set()
  15 writeToSet = set()
  16 sourceLocationSet = set()
  17
  18 # clang does not always use exactly the same numbers in the type-parameter vars it generates
  19 # so I need to substitute them to ensure we can match correctly.
  20 normalizeTypeParamsRegex = re.compile(r"type-parameter-\d+-\d+")
  21 def normalizeTypeParams( line ):
  22     return normalizeTypeParamsRegex.sub("type-parameter-?-?", line)
  23
  24 def parseFieldInfo( tokens ):
  25     if len(tokens) == 3:
  26         return (normalizeTypeParams(tokens[1]), tokens[2])
  27     else:
  28         return (normalizeTypeParams(tokens[1]), "")
  29
  30 with io.open("workdir/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] == "inside:":
  49             touchedFromInsideSet.add(parseFieldInfo(tokens))
  50         elif tokens[0] == "outside:":
  51             touchedFromOutsideSet.add(parseFieldInfo(tokens))
  52         elif tokens[0] == "outside-constructor:":
  53             touchedFromOutsideConstructorSet.add(parseFieldInfo(tokens))
  54         elif tokens[0] == "read:":
  55             readFromSet.add(parseFieldInfo(tokens))
  56         elif tokens[0] == "write:":
  57             writeToSet.add(parseFieldInfo(tokens))
  58         else:
  59             print( "unknown line: " + line)
  60
  61 # Invert the definitionToSourceLocationMap
  62 # If we see more than one method at the same sourceLocation, it's being autogenerated as part of a template
  63 # and we should just ignore it
  64 sourceLocationToDefinitionMap = {}
  65 for k, v in definitionToSourceLocationMap.iteritems():
  66     sourceLocationToDefinitionMap[v] = sourceLocationToDefinitionMap.get(v, [])
  67     sourceLocationToDefinitionMap[v].append(k)
  68 for k, definitions in sourceLocationToDefinitionMap.iteritems():
  69     if len(definitions) > 1:
  70         for d in definitions:
  71             definitionSet.remove(d)
  72
  73 # Calculate untouched
  74 untouchedSet = set()
  75 untouchedSetD = set()
  76 for d in definitionSet:
  77     if d in touchedFromOutsideSet or d in touchedFromInsideSet:
  78         continue
  79     srcLoc = definitionToSourceLocationMap[d];
  80     # this is all representations of on-disk data structures
  81     if (srcLoc.startswith("sc/source/filter/inc/scflt.hxx")
  82         or srcLoc.startswith("sw/source/filter/ww8/")
  83         or srcLoc.startswith("vcl/source/filter/sgvmain.hxx")
  84         or srcLoc.startswith("vcl/source/filter/sgfbram.hxx")
  85         or srcLoc.startswith("vcl/inc/unx/XIM.h")
  86         or srcLoc.startswith("vcl/inc/unx/gtk/gloactiongroup.h")
  87         or srcLoc.startswith("include/svl/svdde.hxx")
  88         or srcLoc.startswith("lotuswordpro/source/filter/lwpsdwdrawheader.hxx")
  89         or srcLoc.startswith("hwpfilter/")
  90         or srcLoc.startswith("embeddedobj/source/inc/")
  91         or srcLoc.startswith("svtools/source/dialogs/insdlg.cxx")
  92         or srcLoc.startswith("bridges/")):
  93         continue
  94     if d[0] in set([ "AtkObjectWrapperClass", "AtkObjectWrapper", "GLOMenu", "GLOAction", "_XRegion", "SalMenuButtonItem", "Vertex",
  95             "OOoMountOperationClass", "SwCSS1ItemIds", "ScCompiler::AddInMap", "MemoryByteGrabber", "textcat_t", "fp_t", "ngram_t",
  96             "ImplPPTParaPropSet", "DataNode"]):
  97         continue
  98     # unit testing code
  99     if srcLoc.startswith("cppu/source/uno/check.cxx"):
 100         continue
 101     fieldType = definitionToTypeMap[d]
 102     if "ModuleClient" in fieldType:
 103         continue
 104     if "::sfx2::sidebar::ControllerItem" in fieldType:
 105         continue
 106     untouchedSet.add((d[0] + " " + d[1] + " " + fieldType, srcLoc))
 107     untouchedSetD.add(d)
 108
 109 # Calculate only-touched-in-constructor set
 110 onlyUsedInConstructorSet = set()
 111 for d in definitionSet:
 112     if d in touchedFromOutsideSet or d in touchedFromOutsideConstructorSet:
 113         continue
 114     srcLoc = definitionToSourceLocationMap[d];
 115     # this is all representations of on-disk data structures
 116     if (srcLoc.startswith("sc/source/filter/inc/scflt.hxx")
 117         or srcLoc.startswith("sw/source/filter/ww8/")
 118         or srcLoc.startswith("vcl/source/filter/sgvmain.hxx")
 119         or srcLoc.startswith("vcl/source/filter/sgfbram.hxx")
 120         or srcLoc.startswith("vcl/inc/unx/XIM.h")
 121         or srcLoc.startswith("vcl/inc/unx/gtk/gloactiongroup.h")
 122         or srcLoc.startswith("include/svl/svdde.hxx")
 123         or srcLoc.startswith("lotuswordpro/source/filter/lwpsdwdrawheader.hxx")
 124         or srcLoc.startswith("hwpfilter/")
 125         or srcLoc.startswith("embeddedobj/source/inc/")
 126         or srcLoc.startswith("svtools/source/dialogs/insdlg.cxx")
 127         or srcLoc.startswith("bridges/")):
 128         continue
 129     fieldType = definitionToTypeMap[d]
 130     if "std::unique_ptr" in fieldType:
 131         continue
 132     if "std::shared_ptr" in fieldType:
 133         continue
 134     if "Reference<" in fieldType:
 135         continue
 136     if "VclPtr<" in fieldType:
 137         continue
 138     if "osl::Mutex" in fieldType:
 139         continue
 140     if "::sfx2::sidebar::ControllerItem" in fieldType:
 141         continue
 142     onlyUsedInConstructorSet.add((d[0] + " " + d[1] + " " + fieldType, srcLoc))
 143
 144 writeonlySet = set()
 145 for d in definitionSet:
 146     parentClazz = d[0];
 147     if d in readFromSet or d in untouchedSetD:
 148         continue
 149     srcLoc = definitionToSourceLocationMap[d];
 150     # this is all representations of on-disk data structures
 151     if (srcLoc.startswith("sc/source/filter/inc/scflt.hxx")
 152         or srcLoc.startswith("sw/source/filter/ww8/")
 153         or srcLoc.startswith("vcl/source/filter/sgvmain.hxx")
 154         or srcLoc.startswith("vcl/source/filter/sgfbram.hxx")
 155         or srcLoc.startswith("vcl/inc/unx/XIM.h")
 156         or srcLoc.startswith("vcl/inc/unx/gtk/gloactiongroup.h")
 157         or srcLoc.startswith("include/svl/svdde.hxx")
 158         or srcLoc.startswith("lotuswordpro/source/filter/lwpsdwdrawheader.hxx")
 159         or srcLoc.startswith("svtools/source/dialogs/insdlg.cxx")):
 160         continue
 161     fieldType = definitionToTypeMap[d]
 162     if "ModuleClient" in fieldType:
 163         continue
 164     if "::sfx2::sidebar::ControllerItem" in fieldType:
 165         continue
 166     # ignore reference fields, because writing to them actually writes to another field somewhere else
 167     if fieldType.endswith("&"):
 168         continue
 169     # ignore the import/export data model stuff
 170     if srcLoc.startswith("sc/source/filter/inc/") and "Model" in fieldType:
 171         continue
 172     if srcLoc.startswith("sc/source/filter/inc/") and (parentClazz.startswith("Xcl") or parentClazz.startswith("oox::xls::")):
 173         continue
 174     # implement some kind of registration of errors
 175     if fieldType == "class SfxErrorHandler *":
 176         continue
 177     # mutex locking
 178     if "Guard" in fieldType:
 179         continue
 180     # these are just all model classes
 181     if (srcLoc.startswith("oox/")
 182         or srcLoc.startswith("lotuswordpro/")
 183         or srcLoc.startswith("include/oox/")
 184         or srcLoc.startswith("include/filter/")
 185         or srcLoc.startswith("hwpfilter/")
 186         or srcLoc.startswith("filter/")):
 187         continue
 188
 189     writeonlySet.add((d[0] + " " + d[1] + " " + definitionToTypeMap[d], srcLoc))
 190
 191
 192 readonlySet = set()
 193 for d in definitionSet:
 194     parentClazz = d[0];
 195     if d in writeToSet or d in untouchedSetD:
 196         continue
 197     fieldType = definitionToTypeMap[d]
 198     srcLoc = definitionToSourceLocationMap[d];
 199     if "ModuleClient" in fieldType:
 200         continue
 201     # this is all representations of on-disk data structures
 202     if (srcLoc.startswith("sc/source/filter/inc/scflt.hxx")
 203         or srcLoc.startswith("sw/source/filter/ww8/")
 204         or srcLoc.startswith("vcl/source/filter/sgvmain.hxx")
 205         or srcLoc.startswith("vcl/source/filter/sgfbram.hxx")
 206         or srcLoc.startswith("vcl/inc/unx/XIM.h")
 207         or srcLoc.startswith("vcl/inc/unx/gtk/gloactiongroup.h")
 208         or srcLoc.startswith("include/svl/svdde.hxx")):
 209         continue
 210     # I really don't care about these ancient file formats
 211     if (srcLoc.startswith("hwpfilter/")
 212         or srcLoc.startswith("lotuswordpro/")):
 213         continue
 214     readonlySet.add((d[0] + " " + d[1] + " " + definitionToTypeMap[d], srcLoc))
 215
 216
 217 canBePrivateSet = set()
 218 for d in protectedAndPublicDefinitionSet:
 219     clazz = d[0] + " " + d[1]
 220     if d in touchedFromOutsideSet:
 221         continue
 222     srcLoc = definitionToSourceLocationMap[d];
 223
 224     canBePrivateSet.add((clazz + " " + definitionToTypeMap[d], srcLoc))
 225
 226
 227 # sort the results using a "natural order" so sequences like [item1,item2,item10] sort nicely
 228 def natural_sort_key(s, _nsre=re.compile('([0-9]+)')):
 229     return [int(text) if text.isdigit() else text.lower()
 230             for text in re.split(_nsre, s)]
 231
 232 # sort results by name and line number
 233 tmp1list = sorted(untouchedSet, key=lambda v: natural_sort_key(v[1]))
 234 tmp2list = sorted(writeonlySet, key=lambda v: natural_sort_key(v[1]))
 235 tmp3list = sorted(canBePrivateSet, key=lambda v: natural_sort_key(v[1]))
 236 tmp4list = sorted(readonlySet, key=lambda v: natural_sort_key(v[1]))
 237 tmp5list = sorted(onlyUsedInConstructorSet, key=lambda v: natural_sort_key(v[1]))
 238
 239 # print out the results
 240 with open("compilerplugins/clang/unusedfields.untouched.results", "wt") as f:
 241     for t in tmp1list:
 242         f.write( t[1] + "\n" )
 243         f.write( "    " + t[0] + "\n" )
 244 with open("compilerplugins/clang/unusedfields.writeonly.results", "wt") as f:
 245     for t in tmp2list:
 246         f.write( t[1] + "\n" )
 247         f.write( "    " + t[0] + "\n" )
 248 # this one is not checked in yet because I haven't actually done anything with it
 249 with open("loplugin.unusedfields.report-can-be-private", "wt") as f:
 250     for t in tmp3list:
 251         f.write( t[1] + "\n" )
 252         f.write( "    " + t[0] + "\n" )
 253 with open("compilerplugins/clang/unusedfields.readonly.results", "wt") as f:
 254     for t in tmp4list:
 255         f.write( t[1] + "\n" )
 256         f.write( "    " + t[0] + "\n" )
 257 with open("compilerplugins/clang/unusedfields.only-used-in-constructor.results", "wt") as f:
 258     for t in tmp5list:
 259         f.write( t[1] + "\n" )
 260         f.write( "    " + t[0] + "\n" )
 261
 262