compilerplugins/clang/singlevalfields.py

   1 #!/usr/bin/python
   2
   3 import sys
   4 import re
   5 import io
   6
   7 definitionToSourceLocationMap = dict() # dict of tuple(parentClass, fieldName) to sourceLocation
   8 definitionToTypeMap = dict() # dict of tuple(parentClass, fieldName) to field type
   9 fieldAssignDict = dict() # dict of tuple(parentClass, fieldName) to (set of values)
  10
  11 # clang does not always use exactly the same numbers in the type-parameter vars it generates
  12 # so I need to substitute them to ensure we can match correctly.
  13 normalizeTypeParamsRegex = re.compile(r"type-parameter-\d+-\d+")
  14 def normalizeTypeParams( line ):
  15     return normalizeTypeParamsRegex.sub("type-parameter-?-?", line)
  16
  17 # reading as binary (since we known it is pure ascii) is much faster than reading as unicode
  18 with io.open("workdir/loplugin.singlevalfields.log", "rb", buffering=1024*1024) as txt:
  19     for line in txt:
  20         tokens = line.strip().split("\t")
  21         if tokens[0] == "defn:":
  22             parentClass = normalizeTypeParams(tokens[1])
  23             fieldName = normalizeTypeParams(tokens[2])
  24             fieldType = normalizeTypeParams(tokens[3])
  25             sourceLocation = tokens[4]
  26             fieldInfo = (parentClass, fieldName)
  27             definitionToSourceLocationMap[fieldInfo] = sourceLocation
  28             definitionToTypeMap[fieldInfo] = fieldType
  29         elif tokens[0] == "asgn:":
  30             parentClass = normalizeTypeParams(tokens[1])
  31             fieldName = normalizeTypeParams(tokens[2])
  32             if len(tokens) > 3:
  33                 assignValue = tokens[3]
  34             else:
  35                 assignValue = ""
  36             fieldInfo = (parentClass, fieldName)
  37             if not fieldInfo in fieldAssignDict:
  38                 fieldAssignDict[fieldInfo] = set()
  39             fieldAssignDict[fieldInfo].add(assignValue)
  40         else:
  41             print( "unknown line: " + line)
  42
  43 # look for stuff also has a single value
  44 tmp1list = list()
  45 # look for things which have two values - zero and one
  46 tmp2list = list()
  47 for fieldInfo, assignValues in fieldAssignDict.iteritems():
  48     v0 = fieldInfo[0] + " " + fieldInfo[1]
  49     v1 = (",".join(assignValues))
  50     v2 = ""
  51     if fieldInfo not in definitionToSourceLocationMap:
  52         continue
  53     v2 = definitionToSourceLocationMap[fieldInfo]
  54     if len(assignValues) > 2:
  55         continue
  56     if "?" in assignValues:
  57         continue
  58     #if len(assignValues - set(["0", "1", "-1", "nullptr"])) > 0:
  59     #    continue
  60     # ignore things which are locally declared but are actually redeclarations of things from 3rd party code
  61     containingClass = fieldInfo[0]
  62     if containingClass == "_mwmhints":
  63         continue
  64     # ignore things which are representations of on-disk structures
  65     if containingClass in ["SEPr", "WW8Dop", "BmpInfoHeader", "BmpFileHeader", "Exif::ExifIFD",
  66             "sw::WW8FFData", "FFDataHeader", "INetURLHistory_Impl::head_entry", "ImplPPTParaPropSet", "SvxSwAutoFormatFlags",
  67             "T602ImportFilter::T602ImportFilter::format602struct", "DataNode"]:
  68         continue
  69     if v2.startswith("hwpfilter/source"):
  70         continue
  71     # ignore things which are representations of structures from external code
  72     if v2.startswith("desktop/unx/source/splashx.c"):
  73         continue
  74     # Windows-only
  75     if containingClass in ["SfxAppData_Impl", "sfx2::ImplDdeItem", "SvFileStream",
  76             "DdeService", "DdeTopic", "DdeItem", "DdeConnection", "connectivity::sdbcx::OUser", "connectivity::sdbcx::OGroup", "connectivity::sdbcx::OCatalog",
  77             "cairocanvas::SpriteHelper"]:
  78         continue
  79     if v2.startswith("include/svl/svdde.hxx") or v2.startswith("embeddedobj/source/inc/oleembobj.hxx"):
  80         continue
  81     # Some of our supported compilers don't do constexpr, which means o3tl::typed_flags can't be 'static const'
  82     if containingClass in ["WaitWindow_Impl"]:
  83         continue
  84     if len(assignValues) == 2:
  85         if "0" in assignValues and "1" in assignValues:
  86             fieldType = definitionToTypeMap[fieldInfo]
  87             if not "_Bool" in fieldType and not "enum " in fieldType and not "boolean" in fieldType:
  88                 tmp2list.append((v0,v1,v2,fieldType))
  89     else:
  90         tmp1list.append((v0,v1,v2))
  91
  92 # sort results by filename:lineno
  93 def natural_sort_key(s, _nsre=re.compile('([0-9]+)')):
  94     return [int(text) if text.isdigit() else text.lower()
  95             for text in re.split(_nsre, s)]
  96 tmp1list.sort(key=lambda v: natural_sort_key(v[2]))
  97 tmp2list.sort(key=lambda v: natural_sort_key(v[2]))
  98
  99 # print out the results
 100 with open("compilerplugins/clang/singlevalfields.results", "wt") as f:
 101     for v in tmp1list:
 102         f.write(v[2] + "\n")
 103         f.write("    " + v[0] + "\n")
 104         f.write("    " + v[1] + "\n")
 105 with open("compilerplugins/clang/singlevalfields.could-be-bool.results", "wt") as f:
 106     for v in tmp2list:
 107         f.write(v[2] + "\n")
 108         f.write("    " + v[0] + "\n")
 109         f.write("    " + v[3] + "\n")
 110
 111