#!/usr/bin/python3

import sys
import pandas as pd
import numpy as np
import os

# infile = 'wa.csv'
infile = sys.argv[1]

f = open(infile)
l = f.readlines()
h = l[0]
hsub = l[1]  # TheExpertTA keep headers in 2 lines (mixed with max possible points)
maxPossible = l[1]
f.close()

# clean up of headers
h = h.strip()
h = h.replace('"', "")
hsub = hsub.strip()
hsub = hsub.replace('"', "")
hsub = hsub.split(",")
headers = h.split(",")

# we should fail hard if this column names are not present
headers[hsub.index("Last")] = "LastName"
headers[hsub.index("First")] = "FirstName"
headers[hsub.index("Email")] = "UserName"
headers[hsub.index("Student No")] = "SID"

# headers[0]='FullName'
# headers[1]='UserName'
# headers[2]='SID'
# headers[3]='TotalPcnt'
# headers[4]='TotalScore'

d = pd.read_csv(infile, skiprows=[0], header=None, names=headers)

# First row contains max points information, so let's reassign it
d.loc[0, "UserName"] = "_Max_Points_"
d.loc[0, "LastName"] = "MaxScore"
d.loc[0, "FirstName"] = "MaxScore"
d.loc[0, "SID"] = pd.NA

specialUsers = []
specialUsers.append("_Max_Points_")
# lets add row which will be in charge of the column type
d = pd.concat(
    [d, pd.DataFrame({"UserName": ["_Col_Category_"]}, columns=d.columns)],
    ignore_index=True,
)
specialUsers.append("_Col_Category_")

# cleanup
# c = d.columns
# c=c.drop(['FullName', 'UserName', 'SID'])
# index = d[c] == 'ND'
# d[index] = np.nan
# index = d[c] == 'NS'
# d[index] = np.nan

# TheExperTA last column contains 'Averages' per student which we do not need
d.drop(columns=["Averages"], inplace=True)
# TheExperTA last row contains Averages per assignment which we do not need
row = d[
    (d["LastName"] == "Averages") & (d["FirstName"].isna()) & (d["UserName"].isna())
]
d.drop(row.index, inplace=True)

# hand tuned fixes
d["UserName"].replace("@email.wm.edu$", "@wm.edu", regex=True, inplace=True)
# d['UserName'].replace('phanng@hotmail.com@tj.va$', 'kphan@wm.edu', regex=True, inplace=True)

# Now let's convert percentage which TheExperTA reports to points as GradeTable expects
for c in d.columns:
    if c in ["LastName", "FirstName", "UserName", "SID"]:
        continue
    maxP = d.loc[(d["UserName"] == "_Max_Points_")][c].values[0]
    index = ~d["UserName"].isin(specialUsers)
    d.loc[index, c] *= maxP / 100  # convert percentage to points

    # TheExpertTA due to their percentage system generates +/- 0.0000001 scores
    # which is useless and make very long numbers in tables.
    # So we are rounding here.
    d.loc[index, c] = d.loc[index, c].round(4)

    # now we are trying to guess column category
    if "hw" in c.lower():
        d.loc[(d["UserName"] == "_Col_Category_"), c] = "HomeWork"


d.to_csv("TheExpertTA.csv", index=False)

# now import to sqlite3

os.popen("rm -f TheExpertTA.db")
p = os.popen(
    'printf ".mode csv\n.import "TheExpertTA.csv" export_table\n.q" | sqlite3 TheExpertTA.db'
)
p.close()