Serving structured data

Structured data (YAML and JSON files) is very typical for UI development to keep settings, texts and other structured information.

EVA ICS provides a feature, called “serve as”, which allows to convert any structured data file on-the-flow and load it into UI applications in the most convenient way.

This feature is supported by both public UI (“/ui” URLs) and SFA PVT and enabled automatically for all files with suffixes “.yml”, “.yaml” and “.json”.

Format conversion

To convert a structured file into another format, request it as:


where FMT:

  • yml (or yaml) - convert the file into YAML
  • json - convert the file into JSON
  • js - convert the file into JavaScript (requires either “var” or “func” as the additional parameter)

e.g. let’s convert YAML, which is usually more human-editable and preferred to keep configs, into JSON:


The file can also be converted on-the-flow to JavaScript variable, or, as copying JavaScript arrays and dicts is usually tricky, into the function, which returns the structured data on every call:




An additional argument “lang” can be used to apply the chosen locale on all string fields of structured data file. Multi-line strings are processed correctly, string formatting (left and right white spaces) is preserved:


(as seen from the above example, “lang” can be combined with “as”).

Firstly, a locale file should be created. Create a directory EVA_DIR/pvt/locales (used for both UI and SFA PVT). After, create inside that directory language directories with LC_MESSAGES subdirectory inside each one. So the tree should look like e.g. (“cs” is for Czech language):

└── locales
    └── cs
        └── LC_MESSAGES
            ├── messages.po
            ├── tests
            └────── test.po

“.po” files are standard Gettext files, which look like:

msgid "this is a test"
msgstr "je to test"

or e.g. for Japanese (UTF-8):

#, fuzzy
msgid ""
msgstr ""
"Content-Type: text/plain; charset=utf-8\n"

msgid "this is a test"
msgstr "これはテストです"

(note that if diacritic is used e.g. in Czech lang messages, the file should specify UTF-8 encoding as well)

The files can be compiled with “msgfmt” Linux command from “gettext” package (installed by default by majority Linux distributions):

msgfmt file.po -o

EVA ICS uses the following strategy to find locale files. E.g. if the document


is served, the message files are looked up in the following order:

  • EVA_DIR/pvt/locales/cs/LC_MESSAGES/tests/
  • EVA_DIR/pvt/locales/cs/LC_MESSAGES/
  • EVA_DIR/pvt/locales/cs/LC_MESSAGES/

(the last file is the standard common message file). If no message file is found, the strings are served as-is, without any conversion.


Altrenatively, locale files can be kept in EVA_DIR/ui/locales. SCADA Final Aggregator automatically searches for the locale files in “ui” if no locale files found in “pvt”.

The option “-o EVA_DIR/ui/locales” for gen-intl can be used to automatically generate / compile locale files in EVA_DIR/ui/locales.


To auto-generate / update “.po” files from JSON or YAML strings, a supplied tool “gen-intl” can be used (multiple languages can be specified at once):

/opt/eva/bin/gen-intl test.yml -l cs generate

The above command will auto-generate or update “test.po” file and put it to the corresponding locale path. E.g. if the file absolute path is /opt/eva/ui/tests/test.yml, the result “.po” file will be written to /opt/eva/pvt/locales/cs/LC_MESSAGES/tests/test.po.

After editing, compile “.po” file manually with “msgfmt”, or run

/opt/eva/bin/gen-intl test.yml -l cs compile

Locale cache

Message files are cached by EVA ICS gettext library, until the SCADA Final Aggregator server is restarted.

The cache can be turned off by setting development mode (development = yes) in [server] section of etc/sfa.ini configuration file.

On production, the API method clear_lang_cache can be used, either by calling it manually or during a deployment.