#!/usr/bin/python

import sys
import re
import io

definitionSet = set()
definitionToSourceLocationMap = dict()
readSet = set()
writeSet = set()
sourceLocationSet = set()

# 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)

def parseFieldInfo( tokens ):
    if len(tokens) == 3:
        return (normalizeTypeParams(tokens[1]), tokens[2])
    else:
        return (normalizeTypeParams(tokens[1]), "")

# The parsing here is designed to avoid grabbing stuff which is mixed in from gbuild.
# I have not yet found a way of suppressing the gbuild output.
with io.open("loplugin.unusedenumconstants.log", "rb", buffering=1024*1024) as txt:
    for line in txt:
        tokens = line.strip().split("\t")
        if tokens[0] == "definition:":
            fieldInfo = (normalizeTypeParams(tokens[1]), tokens[2])
            srcLoc = tokens[3]
            # ignore external source code
            if (srcLoc.startswith("external/")):
                continue
            # ignore build folder
            if (srcLoc.startswith("workdir/")):
                continue
            definitionSet.add(fieldInfo)
            definitionToSourceLocationMap[fieldInfo] = srcLoc
        elif tokens[0] == "read:":
            readSet.add(parseFieldInfo(tokens))
        elif tokens[0] == "write:":
            writeSet.add(parseFieldInfo(tokens))
        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
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 startswith_one_of( srcLoc, fileSet ):
    for f in fileSet:
        if srcLoc.startswith(f):
            return True;
    return False;

untouchedSet = set()
for d in definitionSet:
    if d in readSet or d in writeSet:
        continue
    srcLoc = definitionToSourceLocationMap[d];
    if startswith_one_of(srcLoc,
        [
        # this is all representations of on-disk or external data structures
         "basic/source/inc/filefmt.hxx",
         "basic/source/sbx/sbxscan.cxx",
         "cppcanvas/source/mtfrenderer/emfpbrush.hxx",
         "filter/source/graphicfilter/ipcd/ipcd.cxx",
         "filter/source/t602/t602filter.hxx",
         "include/filter/msfilter/escherex.hxx",
         "include/filter/msfilter/svdfppt.hxx",
         "hwpfilter/",
         "include/registry/types.hxx",
         "lotuswordpro/",
         "include/sot/formats.hxx",
         "include/svx/msdffdef.hxx",
         "sc/source/filter/inc/xlconst.hxx",
         "include/unotools/saveopt.hxx",
         "sw/inc/fldbas.hxx",
         "sw/source/filter/inc/wwstyles.hxx",
         "sw/source/filter/ww8/fields.hxx",
         "vcl/source/fontsubset/cff.cxx",
         "include/vcl/settings.hxx", # stored in a setting, can't remove it without potentially triggering UBSAN
        # unit test code
         "cppu/source/uno/check.cxx",
        # general weird nonsense going on
         "framework/inc/helper/mischelper.hxx"
         "include/sfx2/shell.hxx",
        # Windows or OSX only
         "include/canvas/rendering/icolorbuffer.hxx",
         "include/vcl/commandevent.hxx",
         "vcl/inc/unx/gendata.hxx",
        # must match some other enum
         "include/editeng/bulletitem.hxx",
         "include/editeng/svxenum.hxx",
         "include/formula/opcode.hxx",
         "include/i18nutil/paper.hxx",
         "include/oox/drawingml/shapepropertymap.hxx",
         "include/svl/nfkeytab.hx",
         "include/svl/zforlist.hxx",
        # represents constants from an external API
         "opencl/inc/opencl_device_selection.h",
         "vcl/inc/sft.hxx",
         "vcl/inc/unx/XIM.h",
         "vcl/unx/gtk/xid_fullscreen_on_all_monitors.c",
         ]):
        continue

    if d[1] == "UNKNOWN" or d[1].endswith("NONE") or d[1].endswith("None") or d[1].endswith("EQUAL_SIZE"): continue

    untouchedSet.add((d[0] + " " + d[1], srcLoc))

writeonlySet = set()
for d in writeSet:
    if d in readSet:
        continue
    srcLoc = definitionToSourceLocationMap[d];
    writeonlySet.add((d[0] + " " + d[1], srcLoc))

readonlySet = set()
for d in readSet:
    if d in writeSet:
        continue
    srcLoc = definitionToSourceLocationMap[d];
    readonlySet.add((d[0] + " " + d[1], srcLoc))

# 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 results by name and line number
tmp1list = sorted(untouchedSet, key=lambda v: natural_sort_key(v[1]))
tmp2list = sorted(writeonlySet, key=lambda v: natural_sort_key(v[1]))
tmp3list = sorted(readonlySet, key=lambda v: natural_sort_key(v[1]))

# print out the results
with open("loplugin.unusedenumconstants.report-untouched", "wt") as f:
    for t in tmp1list:
        f.write( t[1] + "\n" )
        f.write( "    " + t[0] + "\n" )
with open("loplugin.unusedenumconstants.report-writeonly", "wt") as f:
    for t in tmp2list:
        f.write( t[1] + "\n" )
        f.write( "    " + t[0] + "\n" )
with open("loplugin.unusedenumconstants.report-readonly", "wt") as f:
    for t in tmp3list:
        f.write( t[1] + "\n" )
        f.write( "    " + t[0] + "\n" )