Parse variables into tibble #81
Conversation
* attempt to parse text to make tibble like the parameter table on pangaea
|
Thanks for the suggestion @lukasjonkers Looks like the changes are failing the testthat tests. I'll look in more detail later, but perhaps it would be better to keep the function returning what it currently does but provide an That way your workflow would be read_meta() |>
as_tibble()leaving the original function most untouched? |
|
Hi @gavinsimpson I saw too that the tests are failing. Perhaps this is because the structure of the output changes with my suggestion? Your suggestion to simply convert the output of the original function to a |
|
@lukasjonkers Sorry that I wasn't clear. Let's assume we fix the meta data parsing problems (which seems like something we certainly should do), would an |
|
Hi @gavinsimpson I understand your concern not to break existing code, although I am not sure how many people have written code to parse the metadata list that is currently returned. Nevertheless, in that case it would make sense to return a more structured list (with the same elements as columns in the tibble?), rather than a The testthat tests fail exactly because I suggested changing the output from a list to a tibble, but I guess you saw that already. Below is copied from |
Description
Amended the read_meta function in an attempt to better parse the parameter description. Rather than a list of lists the parameters element in metadata now contains a more structured tibble with name, short name, unit, PI, method and comment like on the pangaea website. This should allow searching the metadata and facilitate using the metadata in a hopefully more meaningful way.
Related Issue
None. Just a personal wish that I thought others might find useful too.
Example
Based on https://doi.pangaea.de/10.1594/PANGAEA.115823
Old
New
Test
I ran this with for 10s of DOIs and think it works. No guarantee though, I'm no regexp expert.