Because the Git connector is loading whole files at once as resources, it means you cannot use parts of the JSON file as a resource. This might be an issue if your JSON file is pretty complex. Fear not, there are workarounds!

Another limitation is that some JSON files feature keys with $ signs in them. This is considered a special notation for reserved keywords in JSON and this is forbidden in MongoDB.

Reserved keywords in MongoDB¶

If you import a file with a key like $schema in Prancer, it will automatically be escaped like this:

\$schema

So when using a test in Prancer for such keys (You probably shouldn't be testing these fields) then they should use:

"\\\$schema"

as a reference to properly escape the value.

Use an external tool to chunk the file¶

By using a tool like jq, you can extract portions of your JSON file into smaller files. If you integrate this into your build system, you can easily generate chunks of your file and then use those temporary files as snapshots.

Take the following JSON file into account and note that this file is almost 3000 lines long normally:

{
    "AD": {
        "name": "Andorra",
        "native": "Andorra",
        "phone": "376",
        "continent": "EU",
        "capital": "Andorra la Vella",
        "currency": "EUR",
        "languages": [
            "ca"
        ]
    },
    "AE": {
        "name": "United Arab Emirates",
        "native": "دولة الإمارات العربية المتحدة",
        "phone": "971",
        "continent": "AS",
        "capital": "Abu Dhabi",
        "currency": "AED",
        "languages": [
            "ar"
        ]
    },
    "AF": {
        "name": "Afghanistan",
        "native": "افغانستان",
        "phone": "93",
        "continent": "AS",
        "capital": "Kabul",
        "currency": "AFN",
        "languages": [
            "ps",
            "uz",
            "tk"
        ]
    },
    "AG": {
        "name": "Antigua and Barbuda",
        "native": "Antigua and Barbuda",
        "phone": "1268",
        "continent": "NA",
        "capital": "Saint John's",
        "currency": "XCD",
        "languages": [
            "en"
        ]
    },
    "AI": {
        "name": "Anguilla",
        "native": "Anguilla",
        "phone": "1264",
        "continent": "NA",
        "capital": "The Valley",
        "currency": "XCD",
        "languages": [
            "en"
        ]
    }
}

Doing tests over this file isn't too complex because there aren't many layers of objects but all tests would still look like this:

{countries}.AE.languages[] == ['ca']

In a much more complex file, the repetition of {countries}.AE would be a real pain. Therefore, you might need a way to simplify your huge file. To do this, you can use the following workaround.

A quick and dirty example would be:

cat "countries.json" | jq '.AE.languages' > "output-ae.json"
cat "countries.json" | jq '.AI.languages' > "output-ai.json"

Then, you should just commit those files to Git and use them in your Prancer test suite just like a file that would normally exist.