[redcap] add functionality to parse decimal options

[adamdaudrich]

---

The REDcap "radio option" sometimes includes no index and, in thtis case, a decimal (float) causing redcap2linst show the following error:

PHP Fatal error: Uncaught DomainException: Could not parse radio option: '0.01 ' in /var/www/loris/project/modules/redcap/php/client/models/redcapdictionaryrecord.class.inc:316 Stack trace: #0 /var/www/loris/project/modules/redcap/php/client/models/redcapdictionaryrecord.class.inc(268): LORIS\redcap\client\models\RedcapDictionaryRecord->_optionsToLINST() #1 /var/www/loris/project/tools/redcap2linst.php(74): LORIS\redcap\client\models\RedcapDictionaryRecord->toLINST() #2 {main} thrown in /var/www/loris/project/modules/redcap/php/client/models/redcapdictionaryrecord.class.inc on line 316

the fetch was 0.01 | 0.05 | 0.10 | 0.15 without indices, But this code fixes it as shown in image 2 and LINST code below

selectmultiple{@}wppsi_47_critic_5{@}Critical Value Significance

Level{@}'1'=>'0.01'{-}'2'=>'0.05'{-}'3'=>'0.10'{-}'4'=>'0.15'`

• Have you updated related documentation?
  Relevant documentation is the REDcap importer docs, which could be clarified to address this and other bugs that sometimes occur when interfacing with REDcap instance variances.

Testing instructions (if applicable)

Testing requires a redcap instance with an instrument containing this particular setting.

Link(s) to related issue(s)

This change replicates CBIGR PR #546

• Resolves # (Reference the issue this fixes, if any.) None

[MaximeBICMTL]

Thanks for the PR Adam. This does not break anything for MPN and the code looks reasonable (although I wonder why Q1K selects are different).

Maybe @regisoc has an opinion, but if not that is ok from me.

[MaximeBICMTL]

Does this case ever happen?

[adamdaudrich]

it does because look
lorisadmin@adaudrich-dev:/var/www/loris/project/tools$ php redcap2linst.php -o /data/imports/testzone/q1k/redcap -i api -r ooooooooooooooo/ -p 00

-- Getting metadata from REDCap connection.

-- Parsing records...
Warning: Auto-indexing malformed option:
'0.01' as index 1
Warning: Auto-indexing malformed option:
'0.05' as index 2
Warning: Auto-indexing malformed option:
'0.10' as index 3
Warning: Auto-indexing malformed option:
'0.15' as index 4
Warning: Auto-indexing malformed option:
'm' as index 1
Warning: Auto-indexing malformed option:
'n' as index 2
Warning: Auto-indexing malformed option:
'b/p' as index 3
Warning: Auto-indexing malformed option:
'd/t' as index 4
Warning: Auto-indexing malformed option:
'g/k' as index 5
Warning: Auto-indexing malformed option:
'w' as index 6
Warning: Auto-indexing malformed option:
'h' as index 7
Warning: Auto-indexing malformed option:
'f' as index 8
Warning: Auto-indexing malformed option:
'v' as index 9
Warning: Auto-indexing malformed option:
'y' as index 10
Warning: Auto-indexing malformed option:
's/sh' as index 11

-- Writing LINST/META files.

-> writing 'eeget_session_log'
-> writing 'mri_safety_questionnaire'
-> writing 'waisiv_scoring'
-> writing 'mri_acquisition_checklist'
etc...

[MaximeBICMTL]

By "this case", I mean this if (empty($choice)) specifically. It is different than the code change below that prints the warning. Please either provide an example where it is required or remove this if IMO.

[regisoc]

@adamdaudrich @MaximeBICMTL Trying to fully understand this case.
Is this only one radio triggering the issue? If so: the idea to normalize on LORIS is good, but as far as I can tell, both metadata and data come from REDCap, so REDCap is the source of truth.
If all other radio contain indices in their declaration (is it the case?), this one must too, and this normalization must happen in REDCap.

[adamdaudrich]

I think what happened here was the index was 0.01 => 0.01 and redcap2linst couldn't handle the 0.01 as the index

[adamdaudrich]

I'll take a look at the various tests in which it happened