compilerplugins/clang/singlevalfields.py

   1 #!/usr/bin/python3
   2
   3 import re
   4 import io
   5
   6 definitionToSourceLocationMap = dict() # dict of tuple(parentClass, fieldName) to sourceLocation
   7 definitionToTypeMap = dict() # dict of tuple(parentClass, fieldName) to field type
   8 fieldAssignDict = dict() # dict of tuple(parentClass, fieldName) to (set of values)
   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.singlevalfields.log", "r", buffering=1024*1024) as txt:
  18     for line in txt:
  19         tokens = line.strip().split("\t")
  20         if tokens[0] == "defn:":
  21             parentClass = normalizeTypeParams(tokens[1])
  22             fieldName = normalizeTypeParams(tokens[2])
  23             fieldType = normalizeTypeParams(tokens[3])
  24             srcLoc = tokens[4]
  25             # ignore some external stuff that is somehow sneaking through
  26             if not(srcLoc.startswith("workdir/") or srcLoc.startswith("-3.0/") or srcLoc.startswith("_64-linux-gnu/")):
  27                 fieldInfo = (parentClass, fieldName)
  28                 definitionToSourceLocationMap[fieldInfo] = srcLoc
  29                 definitionToTypeMap[fieldInfo] = fieldType
  30         elif tokens[0] == "asgn:":
  31             parentClass = normalizeTypeParams(tokens[1])
  32             fieldName = normalizeTypeParams(tokens[2])
  33             if len(tokens) > 3:
  34                 assignValue = tokens[3]
  35             else:
  36                 assignValue = ""
  37             fieldInfo = (parentClass, fieldName)
  38             if fieldInfo not in fieldAssignDict:
  39                 fieldAssignDict[fieldInfo] = set()
  40             fieldAssignDict[fieldInfo].add(assignValue)
  41         else:
  42             print( "unknown line: " + line)
  43
  44 # look for stuff also has a single value
  45 tmp1list = list()
  46 # look for things which have two values - zero and one
  47 tmp2list = list()
  48 for fieldInfo, assignValues in fieldAssignDict.items():
  49     v0 = fieldInfo[0] + " " + fieldInfo[1]
  50     v1 = (",".join(assignValues))
  51     v2 = ""
  52     if fieldInfo not in definitionToSourceLocationMap:
  53         continue
  54     v2 = definitionToSourceLocationMap[fieldInfo]
  55     if len(assignValues) > 2:
  56         continue
  57     if "?" in assignValues:
  58         continue
  59     # ignore some random noise, no idea why this does not get filtered out by the normal checks in the C++
  60     if v2.startswith("ux-gnu/") or v2.startswith(":") or v2.startswith(".h:") or v2.startswith("bxml/"):
  61         continue
  62     if v2.startswith("ib.h:") or v2.startswith("freetype/") or v2.startswith("k/") or v2.startswith("n.h"):
  63         continue
  64     if v2.startswith("pango/") or v2.startswith("t.h") or v2.startswith("h:"):
  65         continue
  66     #if len(assignValues - set(["0", "1", "-1", "nullptr"])) > 0:
  67     #    continue
  68     # ignore things which are locally declared but are actually redeclarations of things from 3rd party code
  69     containingClass = fieldInfo[0]
  70     if containingClass == "_mwmhints":
  71         continue
  72     # ignore things which are representations of on-disk structures
  73     if containingClass in ["SEPr", "WW8Dop", "BmpInfoHeader", "BmpFileHeader", "Exif::ExifIFD",
  74             "sw::WW8FFData", "FFDataHeader", "INetURLHistory_Impl::head_entry", "ImplPPTParaPropSet", "SvxSwAutoFormatFlags",
  75             "T602ImportFilter::T602ImportFilter::format602struct", "DataNode"]:
  76         continue
  77     if v2.startswith("hwpfilter/source"):
  78         continue
  79     # ignore things which are representations of structures from external code
  80     if v2.startswith("desktop/unx/source/splashx.c"):
  81         continue
  82     # Windows-only
  83     if containingClass in ["SfxAppData_Impl", "sfx2::ImplDdeItem", "SvFileStream",
  84             "DdeService", "DdeTopic", "DdeItem", "DdeConnection", "connectivity::sdbcx::OUser", "connectivity::sdbcx::OGroup", "connectivity::sdbcx::OCatalog",
  85             "cairocanvas::SpriteHelper"]:
  86         continue
  87     if v2.startswith("include/svl/svdde.hxx") or v2.startswith("embeddedobj/source/inc/oleembobj.hxx"):
  88         continue
  89     # Some of our supported compilers don't do constexpr, which means o3tl::typed_flags can't be 'static const'
  90     if containingClass in ["WaitWindow_Impl"]:
  91         continue
  92     if len(assignValues) == 2:
  93         if "0" in assignValues and "1" in assignValues:
  94             fieldType = definitionToTypeMap[fieldInfo]
  95             if "_Bool" not in fieldType and "enum " not in fieldType and "boolean" not in fieldType:
  96                 tmp2list.append((v0,v1,v2,fieldType))
  97     elif len(assignValues) == 1:
  98         # ignore timers/idles
  99         if not("Idle" in v1 or "Timer" in v1):
 100             tmp1list.append((v0,v1,v2))
 101     else:
 102         tmp1list.append((v0,v1,v2))
 103
 104 # sort results by filename:lineno
 105 def natural_sort_key(s, _nsre=re.compile('([0-9]+)')):
 106     return [int(text) if text.isdigit() else text.lower()
 107             for text in re.split(_nsre, s)]
 108 tmp1list.sort(key=lambda v: natural_sort_key(v[2]))
 109 tmp2list.sort(key=lambda v: natural_sort_key(v[2]))
 110
 111 # print out the results
 112 with open("compilerplugins/clang/singlevalfields.results", "wt") as f:
 113     for v in tmp1list:
 114         f.write(v[2] + "\n")
 115         f.write("    " + v[0] + "\n")
 116         f.write("    " + v[1] + "\n")
 117 with open("compilerplugins/clang/singlevalfields.could-be-bool.results", "wt") as f:
 118     for v in tmp2list:
 119         f.write(v[2] + "\n")
 120         f.write("    " + v[0] + "\n")
 121         f.write("    " + v[3] + "\n")
 122
 123