office-gobmx/compilerplugins/clang/expandablemethods.py

#!/usr/bin/python

import sys
import re
import io

# --------------------------------------------------------------------------------------------
# globals
# --------------------------------------------------------------------------------------------

definitionSet = set() # set of tuple(return_type, name_and_params)
definitionToSourceLocationMap = dict()
calledFromDict = dict()
calledFromOutsideSet = set()
largeFunctionSet = set() # set of tuple(return_type, name_and_params)
addressOfSet = set() # set of tuple(return_type, name_and_params)

# clang does not always use exactly the same numbers in the type-parameter vars it generates
# so I need to substitute them to ensure we can match correctly.
normalizeTypeParamsRegex = re.compile(r"type-parameter-\d+-\d+")
def normalizeTypeParams( line ):
    return normalizeTypeParamsRegex.sub("type-parameter-?-?", line)

# --------------------------------------------------------------------------------------------
# primary input loop
# --------------------------------------------------------------------------------------------

with io.open("workdir/loplugin.expandablemethods.log", "rb", buffering=1024*1024) as txt:
    for line in txt:
        tokens = line.strip().split("\t")
        if tokens[0] == "definition:":
            access = tokens[1]
            returnType = tokens[2]
            nameAndParams = tokens[3]
            sourceLocation = tokens[4]
            funcInfo = (normalizeTypeParams(returnType), normalizeTypeParams(nameAndParams))
            definitionSet.add(funcInfo)
            definitionToSourceLocationMap[funcInfo] = sourceLocation
        elif tokens[0] == "calledFrom:":
            calleeLocation = tokens[1]
            returnType = tokens[2]
            nameAndParams = tokens[3]
            funcInfo = (normalizeTypeParams(returnType), normalizeTypeParams(nameAndParams))
            if not funcInfo in calledFromDict:
                calledFromDict[funcInfo] = set()
            calledFromDict[funcInfo].add(calleeLocation)
        elif tokens[0] == "outside:":
            returnType = tokens[1]
            nameAndParams = tokens[2]
            calledFromOutsideSet.add((normalizeTypeParams(returnType), normalizeTypeParams(nameAndParams)))
        elif tokens[0] == "large:":
            returnType = tokens[1]
            nameAndParams = tokens[2]
            largeFunctionSet.add((normalizeTypeParams(returnType), normalizeTypeParams(nameAndParams)))
        elif tokens[0] == "addrof:":
            returnType = tokens[1]
            nameAndParams = tokens[2]
            addressOfSet.add((normalizeTypeParams(returnType), normalizeTypeParams(nameAndParams)))
        else:
            print( "unknown line: " + line)

# Invert the definitionToSourceLocationMap.
# If we see more than one method at the same sourceLocation, it's being autogenerated as part of a template
# and we should just ignore it.
sourceLocationToDefinitionMap = {}
for k, v in definitionToSourceLocationMap.iteritems():
    sourceLocationToDefinitionMap[v] = sourceLocationToDefinitionMap.get(v, [])
    sourceLocationToDefinitionMap[v].append(k)
for k, definitions in sourceLocationToDefinitionMap.iteritems():
    if len(definitions) > 1:
        for d in definitions:
            definitionSet.remove(d)

def isOtherConstness( d, callSet ):
    method = d[0] + " " + d[1]
    # 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
    if d[0].startswith("const ") and d[1].endswith(" const"):
        if ((d[0][6:],d[1][:-6]) in callSet):
           return True
    elif method.endswith(" const"):
        method2 = method[:len(method)-6] # strip off " const"
        if ((d[0],method2) in callSet):
           return True
    if method.endswith(" const") and ("::iterator" in method):
        method2 = method[:len(method)-6] # strip off " const"
        method2 = method2.replace("::const_iterator", "::iterator")
        if ((d[0],method2) in callSet):
           return True
    # 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
    if (not method.endswith(" const")) and ((d[0],"const " + method + " const") in callSet):
           return True
    if (not method.endswith(" const")) and ("::iterator" in method):
        method2 = method.replace("::iterator", "::const_iterator") + " const"
        if ((d[0],method2) in callSet):
           return True
    return False

# sort the results using a "natural order" so sequences like [item1,item2,item10] sort nicely
def natural_sort_key(s, _nsre=re.compile('([0-9]+)')):
    return [int(text) if text.isdigit() else text.lower()
            for text in re.split(_nsre, s)]
# sort by both the source-line and the datatype, so the output file ordering is stable
# when we have multiple items on the same source line
def v_sort_key(v):
    return natural_sort_key(v[1]) + [v[0]]
def sort_set_by_natural_key(s):
    return sorted(s, key=lambda v: v_sort_key(v))


# --------------------------------------------------------------------------------------------
# Methods that are only called from inside their own class, and are only called from one spot
# --------------------------------------------------------------------------------------------

tmp4set = set()
for d in definitionSet:
    if d in calledFromOutsideSet:
        continue
    if isOtherConstness(d, calledFromOutsideSet):
        continue
    if d not in definitionToSourceLocationMap:
        print("warning, method has no location: " + d[0] + " " + d[1])
        continue
    # ignore external code
    if definitionToSourceLocationMap[d].startswith("external/"):
       continue
    # ignore constructors, calledFromOutsideSet does not provide accurate info for them
    tokens = d[1].split("(")[0].split("::")
    if len(tokens)>1 and tokens[-2] == tokens[-1]:
        continue
    # ignore large methods, which make the code clearer by being out of line
    if d in largeFunctionSet:
        continue
    # ignore methods whose address we take
    if d in addressOfSet:
        continue
    # ignore unused methods, leave them to the dedicated analysis
    if d not in calledFromDict:
        continue
    # ignore methods called from more than one site
    if len(calledFromDict[d]) > 1:
        continue

    method = d[0] + " " + d[1]
    tmp4set.add((method, definitionToSourceLocationMap[d]))

# print output, sorted by name and line number
with open("loplugin.expandablemethods.report", "wt") as f:
    for t in sort_set_by_natural_key(tmp4set):
        f.write(t[1] + "\n")
        f.write("    " + t[0] + "\n")