Array import magic

[nealrichardson]

Array/MR variables coming in from SPSS or other lossy file formats generally need some work to reconstruct the "right" metadata because they have to shove both parent and subvariable metadata into the "varlabels" of the subvariables. This often follows a pattern of having varlabels with a prefix containing the parent question wording (description) and a suffix that is the actual response label. (We also do this when we "unbind" an array.)

These functions allow us to detect this metadata pattern and attempt to repair it.

To do:

• Detect the common string prefix in subvariables
• Put that prefix in the parent's description and remove it from the subvars (needs test)
• Confirm that our round trip of unbind/bind behaves as expected
• Add argument to makeArray() so that this detection happens when the array is created
• Add function to comb all arrays in the dataset?
• Document

Ultimately we may want this done server side, but this will help us now and will inform any backend implementation.

[codecov]

Codecov Report

Merging #365 into master will decrease coverage by 0.15%.
The diff coverage is 51.72%.

@@            Coverage Diff             @@
##           master     #365      +/-   ##
==========================================
- Coverage   90.37%   90.22%   -0.16%     
==========================================
  Files         115      116       +1     
  Lines        7325     7354      +29     
==========================================
+ Hits         6620     6635      +15     
- Misses        705      719      +14

Impacted Files	Coverage Δ
R/add-variable.R	84.61% <28.57%> (-5.53%)	⬇️
R/clean-array.R	59.09% <59.09%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 6feab4a...c640f4d. Read the comment docs.

[codecov]

Codecov Report

Merging #365 into master will decrease coverage by 0.15%.
The diff coverage is 51.72%.

@@            Coverage Diff             @@
##           master     #365      +/-   ##
==========================================
- Coverage   90.37%   90.22%   -0.16%     
==========================================
  Files         115      116       +1     
  Lines        7325     7354      +29     
==========================================
+ Hits         6620     6635      +15     
- Misses        705      719      +14

Impacted Files	Coverage Δ
R/add-variable.R	84.61% <28.57%> (-5.53%)	⬇️
R/clean-array.R	59.09% <59.09%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 6feab4a...f876457. Read the comment docs.