codes.csv maps the codes publishers use to the DFO nodes they stand for. A data
pipeline reads it to turn a code in raw data, such as a DAC provider code, into the dcid
of the node it means, without parsing the MCF.
The compiler writes it next to dfo.mcf on every passing run that is not --check. It
has no comment line, so it loads as a plain CSV.
| column | holds |
|---|---|
list |
The value list or class the entity belongs to. |
code |
The code, verbatim. |
dcid |
The dcid of the node the code stands for, in the same form as the MCF. |
label |
The entity's label, for a person reading the file. |
property |
The property that states the code, which names the codelist it comes from. |
There is one row per code. An entity with codes in several codelists gets a row for each, all naming the same dcid. Items that refer to a Data Commons node, such as countries, get rows too, even though they emit nothing into the MCF.
Join on list, code and property together. A code means something only within its
codelist and list: the same code can stand for different things in different codelists,
or for a currency and an organisation within one codelist.
The Codes section of report.txt shows, for each list, how many of its members carry
a code and names the ones that do not.