compilerplugins/clang/unusedenumvalues.py

   1 #!/usr/bin/python
   2
   3 import sys
   4 import re
   5 import io
   6
   7 definitionSet = set()
   8 definitionToSourceLocationMap = dict()
   9 touchSet = set()
  10 # exclude some stuff, mostly because they are some kind of definition of external file formats
  11 excludedSourceFiles = set([
  12     "include/svx/msdffdef.hxx",
  13     "sw/source/filter/ww8/fields.hxx",
  14     "sw/source/filter/inc/wwstyles.hxx",
  15     "sw/inc/toxe.hxx",
  16     "sw/inc/poolfmt.hxx",
  17     "sw/inc/hintids.hxx",
  18     "vcl/inc/unx/XIM.h",
  19     ])
  20 excludedTypes = set([
  21     "SwVarFormat", "RES_FIELDS", "SwFillOrder", "SwIoDetect", "SwDocumentSettingsPropertyHandles",
  22     "SalGenericDataType", "SwDateSubFormat", "XclFutureRecType", "ds_status", "MediaCommand",
  23     "EmfPlusHatchStyle"
  24     ])
  25
  26 # clang does not always use exactly the same numbers in the type-parameter vars it generates
  27 # so I need to substitute them to ensure we can match correctly.
  28 normalizeTypeParamsRegex = re.compile(r"type-parameter-\d+-\d+")
  29 def normalizeTypeParams( line ):
  30     return normalizeTypeParamsRegex.sub("type-parameter-?-?", line)
  31
  32 # The parsing here is designed to avoid grabbing stuff which is mixed in from gbuild.
  33 # I have not yet found a way of suppressing the gbuild output.
  34 with io.open("loplugin.unusedenumvalues.log", "rb", buffering=1024*1024) as txt:
  35     for line in txt:
  36         tokens = line.strip().split("\t")
  37         if tokens[0] == "definition:":
  38             funcInfo = (normalizeTypeParams(tokens[1]), tokens[2])
  39             definitionSet.add(funcInfo)
  40             definitionToSourceLocationMap[funcInfo] = tokens[3]
  41         elif tokens[0] == "touch:":
  42             callInfo = (normalizeTypeParams(tokens[1]), tokens[2])
  43             touchSet.add(callInfo)
  44
  45 # Invert the definitionToSourceLocationMap
  46 # If we see more than one method at the same sourceLocation, it's being autogenerated as part of a template
  47 # and we should just ignore
  48 sourceLocationToDefinitionMap = {}
  49 for k, v in definitionToSourceLocationMap.iteritems():
  50     sourceLocationToDefinitionMap[v] = sourceLocationToDefinitionMap.get(v, [])
  51     sourceLocationToDefinitionMap[v].append(k)
  52 for k, definitions in sourceLocationToDefinitionMap.iteritems():
  53     if len(definitions) > 1:
  54         for d in definitions:
  55             definitionSet.remove(d)
  56
  57 untouchedSet = set()
  58 for d in definitionSet:
  59     clazz = d[0] + " " + d[1]
  60     if d in touchSet:
  61         continue
  62     srcLoc = definitionToSourceLocationMap[d]
  63     srcLocWithoutLineNo = srcLoc.split(":")[0]
  64     # ignore external source code
  65     if (srcLoc.startswith("external/")):
  66         continue
  67     # ignore build folder
  68     if (srcLoc.startswith("workdir/")):
  69         continue
  70     # ignore our stable/URE/UNO api
  71     if (srcLoc.startswith("include/com/")
  72         or srcLoc.startswith("include/cppu/")
  73         or srcLoc.startswith("include/cppuhelper/")
  74         or srcLoc.startswith("include/osl/")
  75         or srcLoc.startswith("include/rtl/")
  76         or srcLoc.startswith("include/sal/")
  77         or srcLoc.startswith("include/salhelper/")
  78         or srcLoc.startswith("include/systools/")
  79         or srcLoc.startswith("include/typelib/")
  80         or srcLoc.startswith("include/uno/")):
  81         continue
  82     if srcLocWithoutLineNo in excludedSourceFiles or d[0] in excludedTypes:
  83         continue
  84     # structure definitions
  85     if srcLoc.startswith("lotuswordpro/"):
  86         continue
  87
  88     # used in templates to find the last member of an enum
  89     if d[1] == "LAST" or d[1].endswith("_END"):
  90         continue
  91     # used to aid in alignment of enum values
  92     if d[1].endswith("FORCE_EQUAL_SIZE"):
  93         continue
  94
  95     untouchedSet.add((clazz, srcLoc))
  96
  97 # sort the results using a "natural order" so sequences like [item1,item2,item10] sort nicely
  98 def natural_sort_key(s, _nsre=re.compile('([0-9]+)')):
  99     return [int(text) if text.isdigit() else text.lower()
 100             for text in re.split(_nsre, s)]
 101
 102 # sort results by name and line number
 103 tmp1list = sorted(untouchedSet, key=lambda v: natural_sort_key(v[1]))
 104
 105 # print out the results
 106 with open("loplugin.unusedenumvalues.report-untouched", "wt") as f:
 107     for t in tmp1list:
 108         f.write( t[1] + "\n" )
 109         f.write( "    " + t[0] + "\n" )
 110