import json

fin = open('./data/in/ds_dedup_2022-02-16_13.03.17.csv')
fin.readline()
dic = {}
for line in fin:
    line = line.strip().split(';')
    last = len(line) -1 
    if line[0] not in dic:
        dic[line[0]] = set()
    id = line[last ]
    for i in range(len(line[last]), 12):
        id += "_"
    dic[line[0]].add(id.lower() + "::" + line[2])

fin.close()

fin = open('./data/out/registryGroups.txt')
intersectionGroups = {}
onlyRegistries = []
onlyDedup = []
completeOverlap = []
cp = set()


for line in fin:
    rg = set(json.loads(line))
    found = False
    for dg in dic:
        intersection = rg.intersection(dic[dg])
        intl = len(intersection)
        if intl > 0:
            found = True
            if intl == len(dic[dg]) and intl == len(rg):
                completeOverlap.append(intersection)
                cp.add(dg)
            else:
                if not dg in intersectionGroups:
                    intersectionGroups[dg] = []
                intersectionGroups[dg].append([dic[dg], rg, intersection])
    if not found:
        onlyRegistries.append(rg)

fin.close()

freport = open('./data/out/allDuplicateSets.txt','w')
for dg in dic:
    if not (dg in intersectionGroups or dg in cp):
        onlyDedup.append(dic[dg])

fout = open('./data/out/onlyRegistry.txt','w')
for e in onlyRegistries:
    r = fout.write("%s\n"%json.dumps(list(e)))
    r = freport.write("%s\n"%json.dumps(list(e)))

fout.close()

fout = open('./data/out/completeOverlap.txt','w')
for e in completeOverlap:
    r = fout.write("%s\n"%json.dumps(list(e)))
    r = freport.write("%s\n"%json.dumps(list(e)))

fout.close()

fout = open('./data/out/intersection.txt','w')
for did in intersectionGroups:
    ret = []
    for g in intersectionGroups[did]:
        lista = []
        for l in g:
            lista.append(list(l))
        ret.append(lista )
    r = fout.write("%s\n"%json.dumps(list(ret)))
    
fout.close()

fout = open('./data/out/onlyDedup.txt','w')
for e in onlyDedup:
    r = fout.write("%s\n"%json.dumps(list(e)))
    r = freport.write("%s\n"%json.dumps(list(e)))

fout.close()

dedupExtention = 0
registryExtention = 0
mutualExtention = 0
groupMerging = 0
newGroup = []
newGroupCount = 0
recount = 0
extentionKeys = {'mutual':[], 'dedup':[],'registry':[],'merging':[]}

for dg in intersectionGroups:
    re = False 
    ng = set()
    if len(intersectionGroups[dg]) > 1:
        groupMerging += 1
        extentionKeys['merging'].append(dg)
        for g in intersectionGroups[dg]:
            ng = ng.union(g[0])
            ng = ng.union(g[1])
    else:
        intersection = intersectionGroups[dg][0]
        dedup = intersection[0]
        registry = intersection[1]
        inte = intersection[2]
        if(len(inte)) < len(dedup) and len(inte) < len(registry):
            mutualExtention += 1
            extentionKeys['mutual'].append(dg)
        if(len(inte)) < len(dedup) and len(inte) == len(registry):
            dedupExtention += 1
            extentionKeys['dedup'].append(dg)
        if(len(inte)) == len(dedup) and len(inte) < len(registry):
            registryExtention += 1
            re = True
            extentionKeys['registry'].append(dg)
        ng = dedup.union(registry)
    duplicate = False
    for tmp in newGroup:
        l = len(tmp.intersection(ng))
        if l == len(tmp) and l == len(ng):
            duplicate = True
            break 
    if not duplicate:
        newGroup.append(ng)
        if not re :
            newGroupCount += 1
        else:
            recount += 1


fout = open('./data/out/newGroups.txt','w')
for e in newGroup:
    r = fout.write("%s\n"%json.dumps(list(e)))
    r = freport.write("%s\n"%json.dumps(list(e)))

fout.close()
freport.close()
fout = open('./data/out/report.txt','w')
fout.write("Merged Groups = %s \n"%str(groupMerging))
fout.write("Mutual Extention = %s \n"%str(mutualExtention))
fout.write("Extention to registry group via dedup = %s \n"%str(dedupExtention))
fout.write("Extention to dedup group via registry = %s \n"%str(registryExtention))
fout.write("Number of new groups without duplicates = %s \n"%str(newGroupCount))
fout.write("Number of registry extentions without duplicates = %s \n"%str(recount))
across4 = 0
temp = []
for i in newGroup:
    tmp = []
    for e in i:
        tmp .append(e[0:4])
    if 'fair' in tmp and 'roar' in tmp and 're3d' in tmp and 'open' in tmp:
        across4 += 1
        temp.append(list(i))
fout.write("Number Groups with at least one entry for each registry = %s \n"%str(across4))

fout.write("\n\n\n********** ENTRIES FROM ALL REGISTRY  ************** \n\n\n")
for e in temp:
    fout.write("%s\n"%json.dumps(e))

fout.write("\n\n\n********** MUTUAL EXTENTION GROUPS ************** \n\n\n")
for e in extentionKeys['mutual']:
    ret = []
    for g in intersectionGroups[e]:
        lista = []
        for l in g:
            lista.append(list(l))
        ret.append(lista )
    r = fout.write("%s\n"%json.dumps(ret))

fout.write("\n\n\n********** MERGING EXTENTION GROUPS ************** \n\n\n")
for e in extentionKeys['merging']:
    ret = []
    for g in intersectionGroups[e]:
        lista = []
        for l in g:
            lista.append(list(l))
        ret.append(lista )
    r = fout.write("%s\n"%json.dumps(ret))

fout.write("\n\n\n********** DEDUP EXTENTION GROUPS ************** \n\n\n")
for e in extentionKeys['dedup']:
    ret = []
    for g in intersectionGroups[e]:
        lista = []
        for l in g:
            lista.append(list(l))
        ret.append(lista )
    r = fout.write("%s\n"%json.dumps(ret))

fout.write("\n\n\n********** REGISTRY EXTENTION GROUPS ************** \n\n\n")
for e in extentionKeys['registry']:
    ret = []
    for g in intersectionGroups[e]:
        lista = []
        for l in g:
            lista.append(list(l))
        ret.append(lista )
    r = fout.write("%s\n"%json.dumps(ret))


fout.close()