parsecounts: Parse counting units (thousands, millions, etc) from strings

In JGCRI/iamrpt: Convert GCAM results to the format required by various IAM experiment databases

Description

Counting units tend to get treated as words rather than symbols, and thus have a lot of variant forms. Here, we parse these with regular expressions that should catch the most common ones. When we find a unit, we return the conversion factor to convert it to ones. For example, "thousands" -> 0.001. This allows us to construct the conversion from unit 'a' to unit 'b' by dividing val(b)/val(a). Note that an empty input string is treated as ones (i.e., if you don't specify a count, it's assumed to be ones).

Usage

parsecounts(unit)

Arguments

unit	The counting unit to convert.

Details

If we can't identify a unit, then we issue a warning and return NA. This will cause unitconv_counts to abort the unit conversion.

Right now we cover all of the units I've seen in the GCAM output, plus a few other obvious extensions, but there are many more possibilities. The units are matched by case-insensitive regular expression; the matches currently implemented are:

(blank string)

Ones

hundred[ -]?thous

Hundred thousands

hundred

Hundreds (the compound 'hundred thousands' is checked first so that this match will not capture it.)

thous

Thousands (including abbreviations like 'thous' or 'thousand')

mil

Millions, including abbreviations

bil

Billions, including abbreviations

tril

Trillions, including abbreviations. This is the largest unit we expect to see written out (instead of denoted with a metric prefix).

JGCRI/iamrpt documentation built on June 30, 2019, 3:19 a.m.