For the tree data collection form I have been creating in ODK with XLSForm, I wanted to allow users to quickly re-use species names from previous species identified within the plot to cut down on typing. It stands to reason that if a species is encountered within a plot, there are likely other trees of the same species within the plot. My mental model for the ODK form was to present the user with a repeating question, one iteration for each individual, which said: “from the choices listed below, choose the species of tree”, where the list of choices was populated by answers given by the user during previous iterations. If the list didn’t contain the correct species the user could enter it manually in a question below and the name would be added to the list of choices for future iterations of the repeat. Currently XLSForm+pyxform lack a method to accomplish this, so I posted a reply to a similar question on the ODK support forums
. I got an incredibly helpful response from a member of the core developer team which involved editing the raw .xml after the form had been converted. Here is a rundown of the method posted using an example where users are asked to record the names of animals encountered. First, the .xls file, approximated using a markdown table:
| type | name | label | choice_filter | required | appearance | relevance | calculation |
|---|---|---|---|---|---|---|---|
| begin_repeat | pet | Pet | |||||
| begin_group | pet_group | Pet | field-list | ||||
| calculate | pos | position(../..) | |||||
| select_one ${animal_type} | animal_select | Select the animal type. If not in the list, type below. | position() != current()/../pos and animal_type != '' | autocomplete | |||
| text | animal_type | Animal type | yes | ${animal_select} = '' | |||
| end_group | pet_group | ||||||
| end_repeat | pet |
Technically this file is invalid in its specification. It can be awkward to produce XForm .xml from an invalid .xls. I used xls2xform from pyxform:
xls2xform --pretty_print --skip_validate pets.xls pets.xml
pretty_print creates un-minified .xml that is easier to edit, --skip_validate allows the .xml file to be produced, where otherwise it would be discarded.
After the XML is created it is necessary to edit it. These are the offending lines:
<itemset nodeset="..[position() != ./pos and animal_type != '']">
<value ref="animal_type"/>
<label ref="animal_type"/>
</itemset>
Which must be amended to:
<itemset nodeset="/data/pet[position() != current()/../pos and pet_group/animal_type != '']">
<value ref="pet_group/animal_type"/>
<label ref="pet_group/animal_type"/>
</itemset>
This update basically changes the references to take into account the nested grouping within the repeat, and changes where the pos item updates from.