import numpy as np
import pandas as pd
import matplotlib.pyplot as plt

men = pd.read_csv("US Rank List Men.csv")
women = pd.read_csv("US Rank List Women.csv")
mixed = pd.read_csv("US Rank List Mixed.csv")
us_open = pd.read_csv("US Rank List Open.csv")
athlete_list = pd.concat([men, women, mixed, us_open], axis = 0)
athlete_list.drop(athlete_list.columns[[6,5,7,8,9,12,13,14,15,16,17,18]],
                  axis = 1, inplace = True)
athlete_list["Person/Team"] = athlete_list["Person/Team"].str.replace(r"\(.*\)","")
athlete_list["Discipline"] = athlete_list["Discipline"].str.strip()
athlete_list["DoB"] = pd.to_datetime(athlete_list["DoB"], errors = "coerce")

# Create nested dictionary for {Sport: [Dispcline, Risk]}
risk_dict = {
    "44": {
        "37": 1.00,
        "39": .986182,
        "36": .980809,
        "35": .963153,
        "38": .952406,
        "40": .943961
        },
    "1": {
        "55": .017437,
        "54": .017437,
        "61": .017437,
        },
    "12":{
        "7": .605391,
        "90": .570079,
        "52": .566241,
        "6": .565473,
        "8": .559332,
        "57": .558565,
        },
    "51":{
        "224": .525208,
        },
    "48" :{
        "42": .286292,
        "41": .053626,
        },
    "42":{
        "180": .27401,
        },
    "52":{
        "47": .177908,
        "46": .161019,
        },
    "38": {
        "85": .171079,
        "84": .151888,
        "86": .140373,
        "87": .136535,
        },
    "20":{
        "15": .152899,
        "16": .12987,
        "17": .129102,
        },
    "22":{
        "164": .136172,
        },
    "8":{
        "155": .132829,
        },
    "36":{
        "31": .125921,
        "135": .125921,
        "34": .125921,
        "136": .11978,
        },
    "27":{
        "174": .096624,
        },
    "85":{
        "200": .094119,
        "244": .094119,
        },
    "47":{
        "81": .081165,
        "80": .068883,
        },
    "90":{
        "122": .074478,
        "124": .065266,
        "121": .064499,
        "125": .062963,
        "123": .062963,
        },
    "18":{
        "162": .067781,
        },
    "30":{
        "22": .065472,
        "23": .065472,
        "24": .065472,
        "21": .065472,
        },
    "34":{
        "201": .062192,
        },
    "107":{
        "194": .062084,
        },
    "41":{
        "202": .061201,
        },
    "15":{
        "216": .059223,
        },
    "88":{
        "219": .056303,
        },
    "43":{
        "139": .055173,
        },
    "111":{
        "68": .047926,
        "69": .047926,
        },
    "7":{
        "154": .040582,
        },
    "19":{
        "176": .038383,
        },
    "21":{
        "163": .037286,
        },
    "94":{
        "82": .034222,
        "83": .033455,
        "110": .026546
        },
    "49":{
        "147": .033779,
        },
    "89":{
        "183": .032947,
        },
    "6":{
        "2": .032132,
        "3": .015572,
        },
    "33":{
        "64": .032132,
        "25": .028842,
        "27": .021494,
        "75": .01305,
        "77": .010747,
        "26": .008663,
        "29": .005922,
        "28": .005922,
        "76": .032132,
        "63": .028842,
        "62": .021494,
        "133": .01305,
        },
    "29":{
        "175": .027745,
        },
    "143":{
        "196": .02435,
        },
    "110":{
        "184": .022262,
        },
    "16":{
        "9": .021277,
        "10": .021277,
        "11": .021277,
        "102": .02051,
        },
    "11":{
        "158": .020946,
        },
    "57":{
        "49": .017656,
        "48": .0176556,
        },
    "108":{
        "198": .016781,
        },
    "23":{
        "167": .01645,
        },
    "86":{
        "191": .016121,
        },
    "35":{
        "177": .015463,
        },
    "84":{
        "189": .015353,
        },
    "50":{
        "45": .014585,
        "43": .014585,
        "44": .014585,
        "56": .014585,
        },
    "114":{
        "187": .014478,
        },
    "97":{
        "190": .01305,
        },
    "39":{
        "146": .011953,
        },
    "102":{
        "210": .011515,
        "209": .011515,
        },
    "103":{
        "118": .010747,
        "117": .004606,
        "119": .10747,
        },
    "25":{
        "173": .010747,
        },
    "9":{
        "4": .005812,
        "88": .005045,
        "89": .004277,
        "5": .00965,
        },
    "17":{
        "12": .007786,
        "13": .007786,
        "14": .007019,
        },
    "2":{
        "148": .007677,
        },
    "5":{
        "151": .007677,
        },
    "40":{
        "179": .007677,
        },
    "24":{
        "168": .007348,
        },
    "123":{
        "193": .006909,
        },
    "101":{
        "73": .006141,
        "72": .006141,
        "74": .006141,
        },
    "138":{
        "1199": .006141,
        },
    "37":{
        "178": .006141,
        },
    "99":{
        "192": .003071,
        },
    "153":{
        "186": .002303,
        },
    "93":{
        "114": .002303,
        "113": .002303
        },
    "104":{
        "197": .001535,},}

#TODO Take the athlete_list dataframe and convert the sport/discipline names
# to those conained in the risk dictionary.


# Replace command does not like matching k:v pairs, so those are commented out
sport_dict={
    "Athletics": "44",
    "Archery": "1",
    "Alpine Skiing": "33",
    "Badminton": "2",
    "Baseball": "3",
    "Basketball": "4",
    "Beach Volleyball": "48",
    "Biathlon": "5",
    "BMX": "12",
    "Bobsleigh": "6",
    "Boxing": "8",
    "Canoe Slalom": "9",
    "Canoe Sprint": "9",
    "Cross Country Skiing": "33",
    "Curling": "11",
    "Cycling - Road": "12",
    "Cycling - Track": "12",
    "Diving": "15",
    "Fencing": "17",
    "Figure Skating": "19",
    "Football": "34",
    "Freestyle Skiing": "33",
    "Golf": "142",
    "Gymnastics - Artistic": "20",
    "Gymnastics - Trampolining": "20",
    "Handball": "41",
    "Hockey": "18",
    "Ice Hockey": "21",
    "Judo": "22",
    "Karate": "57",
    "Luge": "23",
    "Modern Pentathlon": "24",
    "Mountain Bike": "12",
    "Nordic Combined": "33",
    "Open Water Swimming": "38",
    "Rowing": "27",
    "Rugby Sevens": "111",
    "Sailing": "29",
    "Shooting": "30",
    "Short Track": "36",
    "Skateboarding": "133",
    "Skeleton": "6",
    "Ski Jumping": "33",
    "Snowboard": "33",
    "Speed Skating": "36",
    "Sport Climbing": "121",
    "Surfing": "141",
    "Swimming": "38",
    "Table Tennis": "40",
    "Taekwondo": "43",
    "Tennis": "42",
    "Triathlon": "47",
    "Volleyball": "48",
    "Water Polo": "49",
    "Weightlifting": "51",
    "Wrestling - Freestyle": "52",
    "Wrestling - Greco-Roman": "52",
    "Gymnastics - Rhythmic": "20",
    "Softball": "35",
    "Synchronised Swimming": "39",
    "Equestrian - Dressage": "16",
    "Equestrian - Eventing": "16",
    "Equestrian - Jumping": "16"}

athlete_list = athlete_list.replace({"Sport": sport_dict})

discipline_change_dict = {
    "44": {
        "4 x 100m Relay": "37",
        "4 x 400m Relay": "37",
        "4 x 400m Relay Indoor": "37",
        "60m Hurdles Indoor": "37",
        "50km Walk" : "39",
        "60m Hurdles Indoor": "37",
        "60m Indoor": "37",
        "100m": "37",
        "110m Hurdles": "37",
        "200m":"37",
        "400m":"37",
        "400m Hurdles":"37",
        "400m Indoor": "37",
        "800m": "38",
        "800m Indoor": "38",
        "1500m": "38",
        "1500m Indoor": "38",
        "3000m Indoor": "39",
        "3000m Steeplechase": "39",
        "5000m": "39",
        "10000m": "39",
        "Decathlon": "40",
        "Heptathlon": "40",
        "Discus Throw": "35",
        "Hammer Throw": "35",
        "High Jump": "36",
        "Javelin Throw": "35",
        "Long Jump": "36",
        "Marathon": "39",
        "Pole Vault": "36",
        "Shot Put": "35",
        "Triple Jump": "36",
        "100m Hurdles": "37",
        "20km Walk": "39",
        },
    "2":{
        "Doubles": "148",
        "Singles": "148",
        },
    "5":{
        "4 x 7.5km Relay": "151",
        "10k Sprint": "151",
        "12.5km Pursuit": "151",
        "15km Mass Start": "151",
        "20km Individual": "151",
        "4 x 6km Relay": "151",
        "7.5km Sprint": "151",
        "10km Pursuit": "151",
        "10km Sprint": "151",
        "12.5km Mass Start": "151",
        "15km Individual": "151",
        "2 x 6 + 2 x 7.5km Relay": "151",
        "6 + 7.5km Relay": "151",
        },
    "6":{
        "2-Man Competition": "2",
        "Individual": "3",
        "4-Man Competition": "2",
        },
    "8":{
        "Bantamweight": "155",
        "Flyweight": "155",
        "Light Flyweight": "155",
        "Light Heavyweight": "155",
        "Light Welterweight": "155",
        "Lightweight": "155",
        "Middleweight": "155",
        "Super Heavyweight": "155",
        "Welterweight": "155",
        },
    "9":{
        "C1 200m": "4",
        "C1 1000m": "89",
        "K1 200m": "4",
        "K1 500m": "88",
        "K1 1000m": "89",
        "K1 5000m": "89",
        "K2 500m": "88",
        "K4 1000m": "89",
        "K4 500m": "88",
        "C1": "5",
        "C1 Team": "5",
        "C2": "5",
        "C2 Team": "5",
        "K1": "5",
        "K1 Cross": "5",
        "K1 Team": "5",
        },
    "11":{
        "Team": "158",
        "Doubles": "158",
        },
    "12":{
        "Individual": "52",
        "1km Time Trial": "8",
        "3km Individual Pursuit": "8",
        "4-Cross": "57",
        "4km Individual Pursuit": "8",
        "500m Time Trial": "8",
        "Cross Country": "6",
        "Cross Country Eliminator": "6",
        "Downhill": "6",
        "Individual time trial": "7",
        "Keirin": "90",
        "Madison": "90",
        "Marathon": "6",
        "Omnium": "90",
        "Points Race": "90",
        "Road race": "7",
        "Scratch": "90",
        "Sprint": "8",
        "Team Pursuit": "8",
        "Team Sprint": "8",
        },
    "15":{
        "10m Platform": "216",
        "10m Platform Synchro": "216",
        "1m Springboard": "216",
        "3m Springboard": "216",
        "3m Springboard Synchro": "216",
        },
    "16":{
        "Individual": "10",
        "Team": "10",# Nielson breaks on 3 levels, us on 2, so everything into 1 category
        },
    "17":{
        "Individual Epee": "12",
        "Individual Foil": "13",
        "Individual Sabre": "13",
        "Team Epee": "12",
        "Team Foil": "13",
        "Team Sabre": "13",
        },
    "18":{
        "Team": "162",
        },
    "19":{
        "Ice Dance": "176",
        "Pairs": "176",
        "Singles": "176",
        "Team": "176",
        },
    "142":{
        "Individual": "217",
        },
    "20":{
        "Ball": "16",
        "Beam": "15",
        "Clubs": "16",
        "Floor Exercise": "15",
        "Group All-Around": "16",
        "Group Competition - 3x Clubs + 2x Hoop": "16",
        "Group Competition - 5x Hoop": "16",
        "Group Competition - 5x Ribbon": "16",
        "Hoop": "16",
        "Horizontal Bar": "15",
        "Individual All-Around": "15",
        "Parallel Bars": "15",
        "Pommel Horse": "15",
        "Individual Trampoline": "17",
        "Ribbon": "16",
        "Rings": "15",
        "Team": "15",
        "Uneven Bars": "15",
        "Vault": "15",
        },
    "21":{
        "Team": "163",
        },
    "22":{
        "+100kg": "164",
        "+78kg": "164",
        "100kg": "164",
        "48kg": "164",
        "52kg": "164",
        "57kg": "164",
        "60kg": "164",
        "63kg": "164",
        "66kg": "164",
        "70kg": "164",
        "73kg": "164",
        "78kg": "164",
        "81kg": "164",
        "90kg": "164",
        },
    "57":{
        "+68kg Kumite": "48", #We have no history on Kumite, so cominbing all into karate
        "+84kg Kumite": "48",
        "-50kg Kumite": "48",
        "-55kg Kumite": "48",
        "-60kg Kumite": "48",
        "-61kg Kumite": "48",
        "-67kg Kumite": "48",
        "-68kg Kumite": "48",
        "-75kg Kumite": "48",
        "-84kg Kumite": "48",
        "Individual Kata": "49",
        "Team Kata": "49",
        "Team Kumite": "48",
        "Individual Kumite": "48",
        },
    "23":{
        "Double": "167",
        "Single": "167",
        "Sprint - Double": "167",
        "Sprint - Single": "167",
        "Team - Relay": "167",
        },
    "24":{
        "Individual": "168",
        },
    "27":{
        "Coxed Pair": "174",
        "Double Sculls": "174",
        "Eight": "174",
        "Four": "174",
        "Lightweight Double Sculls": "174",
        "Lightweight Eight": "174",
        "Lightweight Four": "174",
        "Lightweight Pair": "174",
        "Lightweight Quadruple Sculls": "174",
        "Lightweight Single Sculls": "174",
        "Pair": "174",
        "Quadruple Sculls": "174",
        "Single Sculls": "174",
        },
    "111":{
        "Team": "69",
        },
    "29":{
        "470": "175",
        "49er": "175",
        "49er FX": "175",
        "Finn": "175",
        "Laser": "175",
        "Laser Radial": "175",
        "Nacra 17": "175",
        "RS:X": "175",
        },
    "30":{
        "10m Air Pistol": "22",
        "10m Air Rifle": "23",
        "25m Pistol": "22",
        "50m Pistol": "22",
        "50m Rifle 3 Positions": "23",
        "50m Rifle Prone": "23",
        "Double Trap": "21",
        "Skeet": "21",
        "Trap": "21",
        "25m Rapid Fire Pistol": "22",
        },
    "133":{
        "Park": "222",
        "Street": "222",
        },
    "33":{
        "5km Classic Style": "26",
        "5km Freestyle": "26",
        "10km Classic Style": "26",
        "10km Freestyle": "26",
        "15km Classic Style": "26",
        "15km Freestyle": "26",
        "30km Classic Style": "26",
        "30km Freestyle": "26",
        "2 x 10km Skiathlon": "26",
        "4 x 10km Skiathlon": "26",
        "2 x 15km Skiathlon": "26",
        "2 x 5km Skiathlon": "26",
        "2 x 7.5km Skiathlon": "26",
        "4 x 5km Relay": "26",
        "4 x 7.5km Relay": "26",
        "50km Classic Style": "26",
        "50km Freestyle": "26",
        "5km Classic Style": "26",
        "5km Freestyle": "26",
        "Aerials": "27",
        "Alpine Combined": "25",
        "Big Air": "27",
        "City Event": "25",
        "Downhill": "25",
        "Dual Moguls": "27",
        "Giant Slalom": "25",
        "Gundersen Large Hill / 10km": "29",
        "Gundersen Normal Hill / 5km": "29",
        "Gundersen Normal Hill / 10km": "29",
        "Gundersen Normal Hill / 15km": "29",
        "Halfpipe": "75",
        "Individual - Large Hill": "28",
        "Individual - Normal Hill": "28",
        "Individual - Ski Flying": "28",
        "Moguls": "27",
        "Parallel Giant Slalom": "25",
        "Parallel Slalom": "25",
        "Pursuit": "26",
        "Ski Cross": "76",
        "Slalom": "25",
        "Slopestyle": "77",
        "Snowboard Cross": "64",
        "Snowboard Cross Team": "64",
        "Sprint Classic Style": "26",
        "Sprint Freestyle": "26",
        "Super G": "25",
        "Team - Large Hill": "28",
        "Team - Normal Hill": "28",
        "Team - Ski Flying": "28",
        "Team Normal Hill / 4 x 5km": "29",
        "Team Sprint Classic Style": "26",
        "Team Sprint Freestyle": "26",
        "Team Sprint Large Hill / 2 x 7.5km": "26",
        "10km Classic Style": "26",
        "4 x 10km Relay": "26",
        "Team": "25",
        },
    "34":{
        "Team": "201",
        },
    "36":{
        "10000m": "136",
        "1000m": "31",
        "1500m": "31",
        "3000m": "136",
        "3000m Relay": "136",
        "5000m": "136",
        "5000m Relay": "136",
        "500m": "31",
        "Mass Start 16 Laps": "136",
        "Overall": "31",
        "Small Combination": "135",
        "Sprint Combination": "135",
        "Team Pursuit 6 Laps": "31",
        "Team Pursuit 8 Laps": "31",
        },
    "121":{
        "Bouldering": "60g",
        "Combined": "59",
        "Lead": "59",
        "Speed": "58",
        },
    "141":{
        "Longboard": "79",
        "Shortboard": "79",
        "Team": "79",
        },
    "38":{
        "100m Backstroke": "84",
        "100m Backstroke SC": "84",
        "100m Breaststroke": "84",
        "100m Breaststroke SC": "84",
        "100m Butterfly": "84",
        "100m Butterfly SC": "84",
        "100m Freestyle": "84",
        "100m Freestyle SC": "84",
        "100m Individual Medley SC": "84",
        "10km": "87",
        "1500m Freestyle": "86",
        "1500m Freestyle SC": "86",
        "200m Backstroke": "85",
        "200m Backstroke SC": "85",
        "200m Breaststroke": "85",
        "200m Breaststroke SC": "85",
        "200m Butterfly": "85",
        "200m Butterfly SC": "85",
        "200m Freestyle": "85",
        "200m Freestyle SC": "85",
        "200m Individual Medley": "85",
        "200m Individual Medley SC": "85",
        "4 x 100m Freestyle Relay": "85",
        "4 x 100m Freestyle Relay SC": "85",
        "4 x 100m Medley Relay": "85",
        "4 x 100m Medley Relay SC": "85",
        "4 x 200m Freestyle Relay": "86",
        "4 x 200m Freestyle Relay SC": "86",
        "4 x 50m Freestyle Relay SC": "85",
        "4 x 50m Medley Relay SC": "85",
        "400m Freestyle": "85",
        "400m Freestyle SC": "85",
        "400m Individual Medley": "85",
        "400m Individual Medley SC": "85",
        "50m Backstroke": "84",
        "50m Backstroke SC": "84",
        "50m Breaststroke": "84",
        "50m Breaststroke SC": "84",
        "50m Butterfly": "84",
        "50m Butterfly SC": "84",
        "50m Freestyle": "84",
        "50m Freestyle SC": "84",
        "5km": "87",
        "800m Freestyle": " 86",
        "800m Freestyle SC": "86",
        },
    "39":{
        "Duet": "146",
        "Team": "146",
        "Team Free Routine": "146",
        "Team Technical Routine": "146",
        },
    "40":{
        "Doubles": "179",
        "Singles": "179",
        "Team": "179",
        },
    "43":{
        "Bantamweight 49-53kg": "139",
        "Bantamweight 58-63kg": "139",
        "Featherweight 53-57kg": "139",
        "Featherweight 63-68kg": "139",
        "Finweight -46kg": "139",
        "Finweight -54kg": "139",
        "Flyweight 46-49kg": "139",
        "Flyweight 54-58kg": "139",
        "Heavyweight +67kg": "139",
        "Heavyweight +73kg": "139",
        "Heavyweight +80kg": "139",
        "Heavyweight +87kg": "139",
        "Lightweight 57-62kg": "139",
        "Lightweight 68-74kg": "139",
        "Middleweight 67-73kg": "139",
        "Middleweight 80-87kg": "139",
        "Welterweight 57-67kg": "139",
        "Welterweight 62-67kg": "139",
        "Welterweight 68-80kg": "139",
        "Welterweight 74-80kg": "139",
        },
    "41":{
        "Team Handball": "202",
        "Team": "202",
        },
    "42":{
        "Doubles": "180",
        "Singles": "180",
        },
    "47":{
        "Olympic Distance": "80",
        "Sprint Distance": "80",
        },
    "48":{
        "Team": "42",
        },
    "51":{
        "+105kg": "224",
        "+105kg - Clean & Jerk": "224",
        "+105kg - Snatch": "224",
        "+75kg":"224",
        "+75kg Clean & Jerk":"224",
        "+75kg - Snatch": "224",
        "105kg": "224",
        "105kg - Clean & Jerk": "224",
        "105kg - Snatch": "224",
        "48kg": "224",
        "48kg - Clean & Jerk": "224",
        "48kg - Snatch": "224",
        "53kg": "224",
        "53kg - Clean & Jerk": "224",
        "53kg - Snatch": "224",
        "56kg": "224",
        "56kg - Clean & Jerk": "224",
        "56kg - Snatch": "224",
        "58kg": "224",
        "58kg - Clean & Jerk": "224",
        "58kg - Snatch": "224",
        "63kg": "224",
        "63kg - Clean & Jerk": "224",
        "63kg - Snatch": "224",
        "69kg": "224",
        "69kg - Clean & Jerk": "224",
        "69kg - Snatch": "224",
        "75kg": "224",
        "+75kg - Clean & Jerk": "224",
        "75kg - Clean & Jerk": "224",
        "75kg - Snatch": "224",
        "77kg": "224",
        "77kg - Clean & Jerk": "224",
        "77kg - Snatch": "224",
        "85kg": "224",
        "85kg - Clean & Jerk": "224",
        "85kg - Snatch": "224",
        "94kg": "224",
        "94kg - Clean & Jerk": "224",
        "94kg - Snatch": "224",
        },
    "52":{
        "125kg": "47",
        "130kg": "46",
        "48kg": "47",
        "53kg": "47",
        "57kg": "47",
        "58kg": "47",
        "59kg": "46",
        "63kg": "47",
        "65kg": "47",
        "66kg": "46",
        "69kg": "47",
        "74kg": "47",
        "75kg": "46",
        "85kg": "46",
        "86kg": "47",
        "97kg": "47",
        "98kg": "46",
        },
    "1":{
        "Compound Individual": "54",
        "Compound Team": "54",
        "Individual": "55",
        "Team": "55",
        },
    "3":{
        "Team": "149",
        },
    "4":{
        "Team": "150",
        },
    "35":{
        "Team": "177",
        },
    "49":{
        "Team": "147",
        },}

sport_name_change_df = pd.DataFrame(sport_dict).stack().to_frame("Renamed Discipline")
out_df = athlete_list.merge(sport_dict_df, how = "left", left_on=["Discipline", "Sport"], right_index = True)