compilerplugins/clang/expandablemethods.py

   1 #!/usr/bin/python
   2
   3 import sys
   4 import re
   5 import io
   6
   7 # --------------------------------------------------------------------------------------------
   8 # globals
   9 # --------------------------------------------------------------------------------------------
  10
  11 definitionSet = set() # set of tuple(return_type, name_and_params)
  12 definitionToSourceLocationMap = dict()
  13 calledFromDict = dict()
  14 calledFromOutsideSet = set()
  15 largeFunctionSet = set() # set of tuple(return_type, name_and_params)
  16 addressOfSet = set() # set of tuple(return_type, name_and_params)
  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 # --------------------------------------------------------------------------------------------
  25 # primary input loop
  26 # --------------------------------------------------------------------------------------------
  27
  28 with io.open("workdir/loplugin.expandablemethods.log", "rb", buffering=1024*1024) as txt:
  29     for line in txt:
  30         tokens = line.strip().split("\t")
  31         if tokens[0] == "definition:":
  32             access = tokens[1]
  33             returnType = tokens[2]
  34             nameAndParams = tokens[3]
  35             sourceLocation = tokens[4]
  36             funcInfo = (normalizeTypeParams(returnType), normalizeTypeParams(nameAndParams))
  37             definitionSet.add(funcInfo)
  38             definitionToSourceLocationMap[funcInfo] = sourceLocation
  39         elif tokens[0] == "calledFrom:":
  40             calleeLocation = tokens[1]
  41             returnType = tokens[2]
  42             nameAndParams = tokens[3]
  43             funcInfo = (normalizeTypeParams(returnType), normalizeTypeParams(nameAndParams))
  44             if not funcInfo in calledFromDict:
  45                 calledFromDict[funcInfo] = set()
  46             calledFromDict[funcInfo].add(calleeLocation)
  47         elif tokens[0] == "outside:":
  48             returnType = tokens[1]
  49             nameAndParams = tokens[2]
  50             calledFromOutsideSet.add((normalizeTypeParams(returnType), normalizeTypeParams(nameAndParams)))
  51         elif tokens[0] == "large:":
  52             returnType = tokens[1]
  53             nameAndParams = tokens[2]
  54             largeFunctionSet.add((normalizeTypeParams(returnType), normalizeTypeParams(nameAndParams)))
  55         elif tokens[0] == "addrof:":
  56             returnType = tokens[1]
  57             nameAndParams = tokens[2]
  58             addressOfSet.add((normalizeTypeParams(returnType), normalizeTypeParams(nameAndParams)))
  59         else:
  60             print( "unknown line: " + line)
  61
  62 # Invert the definitionToSourceLocationMap.
  63 # If we see more than one method at the same sourceLocation, it's being autogenerated as part of a template
  64 # and we should just ignore it.
  65 sourceLocationToDefinitionMap = {}
  66 for k, v in definitionToSourceLocationMap.iteritems():
  67     sourceLocationToDefinitionMap[v] = sourceLocationToDefinitionMap.get(v, [])
  68     sourceLocationToDefinitionMap[v].append(k)
  69 for k, definitions in sourceLocationToDefinitionMap.iteritems():
  70     if len(definitions) > 1:
  71         for d in definitions:
  72             definitionSet.remove(d)
  73
  74 def isOtherConstness( d, callSet ):
  75     method = d[0] + " " + d[1]
  76     # if this method is const, and there is a non-const variant of it, and the non-const variant is in use, then leave it alone
  77     if d[0].startswith("const ") and d[1].endswith(" const"):
  78         if ((d[0][6:],d[1][:-6]) in callSet):
  79            return True
  80     elif method.endswith(" const"):
  81         method2 = method[:len(method)-6] # strip off " const"
  82         if ((d[0],method2) in callSet):
  83            return True
  84     if method.endswith(" const") and ("::iterator" in method):
  85         method2 = method[:len(method)-6] # strip off " const"
  86         method2 = method2.replace("::const_iterator", "::iterator")
  87         if ((d[0],method2) in callSet):
  88            return True
  89     # if this method is non-const, and there is a const variant of it, and the const variant is in use, then leave it alone
  90     if (not method.endswith(" const")) and ((d[0],"const " + method + " const") in callSet):
  91            return True
  92     if (not method.endswith(" const")) and ("::iterator" in method):
  93         method2 = method.replace("::iterator", "::const_iterator") + " const"
  94         if ((d[0],method2) in callSet):
  95            return True
  96     return False
  97
  98 # sort the results using a "natural order" so sequences like [item1,item2,item10] sort nicely
  99 def natural_sort_key(s, _nsre=re.compile('([0-9]+)')):
 100     return [int(text) if text.isdigit() else text.lower()
 101             for text in re.split(_nsre, s)]
 102 def sort_set_by_natural_key(s):
 103     return sorted(s, key=lambda v: natural_sort_key(v[1]))
 104
 105
 106 # --------------------------------------------------------------------------------------------
 107 # Methods that are only called from inside their own class, and are only called from one spot
 108 # --------------------------------------------------------------------------------------------
 109
 110 tmp4set = set()
 111 for d in definitionSet:
 112     if d in calledFromOutsideSet:
 113         continue
 114     if isOtherConstness(d, calledFromOutsideSet):
 115         continue
 116     if d not in definitionToSourceLocationMap:
 117         print("warning, method has no location: " + d[0] + " " + d[1])
 118         continue
 119     # ignore external code
 120     if definitionToSourceLocationMap[d].startswith("external/"):
 121        continue
 122     # ignore constructors, calledFromOutsideSet does not provide accurate info for them
 123     tokens = d[1].split("(")[0].split("::")
 124     if len(tokens)>1 and tokens[-2] == tokens[-1]:
 125         continue
 126     # ignore large methods, which make the code clearer by being out of line
 127     if d in largeFunctionSet:
 128         continue
 129     # ignore methods whose address we take
 130     if d in addressOfSet:
 131         continue
 132     # ignore unused methods, leave them to the dedicated analysis
 133     if d not in calledFromDict:
 134         continue
 135     # ignore methods called from more than one site
 136     if len(calledFromDict[d]) > 1:
 137         continue
 138
 139     method = d[0] + " " + d[1]
 140     tmp4set.add((method, definitionToSourceLocationMap[d]))
 141
 142 # print output, sorted by name and line number
 143 with open("loplugin.expandablemethods.report", "wt") as f:
 144     for t in sort_set_by_natural_key(tmp4set):
 145         f.write(t[1] + "\n")
 146         f.write("    " + t[0] + "\n")