Several conventions apply to CSV files across all domains, typically the meaning of certain columns (or headers). Headers are in general optional, it is up to the implementor to provide or not the fields that are required to create or edit the OpenMRS objects. Mandatory headers will be specifically indicated.
Each CSV line is meant to provide enough data to create, edit, or void/retire an OpenMRS object of the domain.
- If the value under this header is missing the OpenMRS object will be created with a new random UUID.
However if the domain supports secondary identifiers, those will be used to fetch the object if the UUID is missing. - If the value under this header is provided the initializer will attempt to retrieve the OpenMRS object that may already exist with this UUID.
- If the OpenMRS object already exists it will be modified and saved again according to the CSV line.
- If the OpenMRS object doesn't exist yet it will be created with the UUID specified on the CSV line.
Set this to true to indicate that the OpenMRS object with the provided UUID should be voided or retired.
When Void/Retire is set to true, the parsing of the remaining part of the CSV line is interrupted since the only objective is to retire the concept. And to this end, only the UUID and the retire flag are needed.
This is a locale-specific header that should not be used as such; it should be used with a locale ISO code appended to it, eg: Display:en or Display:fr or Display:es ... etc. The values under those display headers for each locale are used to localize the display string of the OpenMRS metadata entity being defined on the CSV line. Under the hood, this sets up Java localization messages that comply with the so-called UI Framework convention (see more on this here).
For example for an OpenMRS metadata entity display string to be localized in English (locale 'en') and French (locale 'fr'), simply name the headers Display:en and Display:fr and give them the corresponding localized values respectively in English and French making sure the 'uuid' field is filled and not left blank:
| UUID | ... | Display:en | Display:fr | ... | Void/Retired | ... |
|---|---|---|---|---|---|---|
| eb96c0d4-6248-476b-a079-0aaabfa2f614 | ... | Emergency Room | Salle d'urgences | ... | false | ... |
If the above CSV line was for an org.openmrs.Location, then the following messages would be created:
In locale 'en':
ui.i18n.Location.name.eb96c0d4-6248-476b-a079-0aaabfa2f614=Emergency Room
org.openmrs.Location.eb96c0d4-6248-476b-a079-0aaabfa2f614=Emergency Room
In locale 'fr':
ui.i18n.Location.name.eb96c0d4-6248-476b-a079-0aaabfa2f614=Salle d'urgences
org.openmrs.Location.eb96c0d4-6248-476b-a079-0aaabfa2f614=Salle d'urgences
In general, for each locale, a pair of messages is created as such
ui.i18n.<OpenMRS entity short class name>.name.<OpenMRS entity UUID>=<message in locale>
org.openmrs.<OpenMRS entity short class name>.<OpenMRS entity UUID>=<message in locale>
Special headers are used to provide metadata information about the CSV file itself.
All metadata headers start with an underscore, eg. _version:1, _order:1000, ... etc.
This metadata header specifies the order of loading of the CSV file within the domain. In many cases, the creation of OpenMRS objects relies on the existence of other OpenMRS objects that are referred to. This use case is covered by the order header that allows control of the order of loading of files in a given domain.
For example _order:1000 indicates that all CSV files with an order smaller than 1,000 will be processed before this file within the domain.
If the order metadata cannot be parsed or is missing, then the file will be processed last after all the ordered CSV files of the domain. However, if several CSV files have no order defined, then the loading order between them is undefined.
Versions are primarily introduced to allow for evolutions in the implementation of CSV parsers. Those evolutions may require to modify the CSV headers which will likely lead to backward compatibility issues. Using versions works around this by ensuring that specific parsers are used based on the version specified via the CSV file header.
When no version header is provided the initializer will fall back on a default one or may complain in its log messages that no version was specified.