vnegi10

function create_fruit_table(conn_master, db_name::String, table_name::String)
    
    create_db(conn_master, db_name)

    conn = connect_db(db_name)

    df_tables = DBInterface.execute(conn, 
	                           "SELECT * FROM information_schema.tables") |> DataFrame

    if table_name ∉ df_tables[!, :TABLE_NAME]

vnegi10

function create_db(conn, db_name::String)
	
    df_db = list_db(conn)

    if db_name ∉ df_db.name
        try
	    DBInterface.execute(conn, "CREATE DATABASE $(db_name)")
	    @info "Created a new database $(db_name)"
	catch e
	    error("Unable to create a new database $(db_name)")

vnegi10

function plot_failed_scatter(location::String, 
	                     num_files::Int64;
                             s1::String,
                             s2::String)

    df_all_nok = get_failed_drives(location, num_files)
    start_date, end_date = df_all_nok[!, :date][1], df_all_nok[!, :date][end]

    # Calculate Pearson's correlation coefficient
    df_corr     = select(df_all_nok, [Symbol(s1), Symbol(s2)]) |> dropmissing

vnegi10

function get_parameter_split(location::String, 
	                     num_files::Int64;
                             smart_stat::String)

    files = read_filepaths(location)
    files_to_use = files[1:num_files]

    dates = Date[]
    nok_match, ok_match = [Float64[] for i = 1:2]	

vnegi10

function get_stat_match(df::DataFrame, smart_stat::String)

    num_total   = nrow(df)

    df_stat     = select(df, Symbol(smart_stat) => Symbol("to_check"))
    num_stat    = count(row -> ~ismissing(row.to_check) && 
                                          row.to_check > 0, eachrow(df_stat))

    num_match   = (num_stat / num_total) * 100

vnegi10

function plot_failed_model_pert(location::String, 
	                      	num_files::Int64)

    files  = read_filepaths(location)

    # Read first file to get a population sample
    df_hdd = csv_to_df(files[1])
    df_hdd_model = get_model_count(df_hdd)

    # Get total failed drives

vnegi10

function get_model_count(df_hdd::DataFrame)
	
    # Get count of each model
    all_models = df_hdd[!, :model] |> unique
    all_counts = Int64[]

    for model in all_models
	num_counts = count(x -> x == model, df_hdd[!, :model])
	push!(all_counts, num_counts)
    end

vnegi10

function plot_failed_capacity_dist(location::String,
                                   num_files::Int64)

    df_all_nok = get_failed_drives(location, num_files)

    start_date, end_date = df_all_nok[!, :date][1], df_all_nok[!, :date][end]

    df_hdd_capacity = select(df_all_nok, 
	                     :model, 
	                     :capacity_bytes => (x -> x / 1e12) => :capacity_TB)

vnegi10

function get_failed_drives(location::String, 
	                   num_files::Int64)
                           
    files = read_filepaths(location)
    @assert num_files ≤ length(files) "Reduce the number of files below $(length(files))"
	
    files_to_use = files[1:num_files]
    all_nok = DataFrame[]

    for file in files_to_use

vnegi10

function csv_to_df(file)
    df_hdd = CSV.File(file, header = 1) |> DataFrame
    return df_hdd
end