OPMExternalData.jl

# Rebase an amount of money in a given year to December 2016 dollars
# its bad form but I assume base_year and amt are singular
function opm_inflation_adjust(year, base_year, amt) 
    #base <- opm_get_inflation_index(base_year)
    #div <- sapply(year, function(x) { opm_get_inflation_index(x) })
    #amt * (base / div)
    b = opm_get_inflation_index(base_year)[1]
    mult = opm_get_inflation_index(year)
    amt .* (b ./ mult)
end

# We store the index in a hash for fast access, so breaking this out into a function
function opm_get_inflation_index(year) 
    #tmp <- as.character(year)
    #sapply(tmp, function(x) { ifelse(is.na(x), NA, get(x, envir = opm_inflation_hash)) })
    [(ismissing(y) ? (missing) : opm_inflation_hash[y]) for y in year]
end

# Takes a vector of years, gives percentage change between an element and the one before
function opm_percent_change_inflation(years) 
    #inflation <- opm_get_inflation_index(years)
    #opm_percent_change(inflation)
    inflation = opm_get_inflation_index(years)
    opm_percent_change(inflation)
end

# december of given year inflation index numbers.
opm_inflation_data_dec = DataFrame(
#    ~year, ~ index,
    [1970  39.8;
    1971  41.1;
    1972  42.5; 
    1973  46.2; 
    1974  51.9; 
    1975  55.5; 
    1976  58.2; 
    1977  62.1; 
    1978  67.7; 
    1979  76.7; 
    1980  86.3; 
    1981  94.0; 
    1982  97.6; 
    1983  101.3; 
    1984  105.3;
    1985  109.3; 
    1986  110.5; 
    1987  115.4; 
    1988  120.5; 
    1989  126.1; 
    1990  133.8; 
    1991  137.9; 
    1992  141.9; 
    1993  145.8; 
    1994  149.7; 
    1995  153.5; 
    1996  158.6; 
    1997  161.3; 
    1998  163.9; 
    1999  168.3; 
    2000  174.0; 
    2001  176.7; 
    2002  180.9; 
    2003  184.3; 
    2004  190.3; 
    2005  196.8; 
    2006  201.8; 
    2007  210.036; 
    2008  210.228; 
    2009  215.949; 
    2010  219.179; 
    2011  225.672; 
    2012  229.601; 
    2013  233.049; 
    2014  234.812; 
    2015  236.525; 
    2016  241.432]
)
# julia fixes for weird data frame behavior
names!(opm_inflation_data_dec, [:year, :index])
opm_inflation_data_dec[:year] = convert.(Int, opm_inflation_data_dec[:year])

# Build a hash of our inflation data becuase we do a lot of lookups into it and table filtering
# is really slow for that many iterations.
#opm_inflation_hash <- new.env(hash = TRUE)
#invisible(apply(opm_inflation_data_dec, 1, function(x) { assign(as.character(x['year']), x['index'], envir = opm_inflation_hash) }))
#opm_inflation_data_dec <- opm_inflation_data_dec %>% mutate(pct_change = opm_percent_change(index))
opm_inflation_hash  = Dict{Int, Float64}()
for row in eachrow(opm_inflation_data_dec)
    opm_inflation_hash[row[:year]] = row[:index]
end


# For lack of a better place, putting this here.
opm_presidental_transition_years = vcat(1974, 1977, 1981, 1989, 1993, 2001, 2009)
opm_presidential_names = vcat("Ford", "Carter", "Reagan", "Bush", "Clinton", "Bush", "Obama")
opm_presidents_tibble = DataFrame(years = opm_presidental_transition_years, names = opm_presidential_names)