Helper for testing Lua sandboxes

This is a helper to do unit testing of Heka lua sandboxes: encoders and decoders.

It is in alpha stage: some features may be missing or working incorrectly.

Contributions are welcome. See Contributing for details.

Usage

Install heka_mock, busted (or other Lua test suite) and other Lua modules you need:

luarocks install heka_mock busted lua-cjson lpeg

Create spec/ subdir in the dir with your lua modules:

---+--- spec/ --- my_encoder_spec.lua
   ---- my_encoder.lua

Example of specs can be found at https://github.com/timurb/heka_mock/tree/master/test/spec. For detailed APIs see API section below.

Run tests:

$ busted
●
1 success / 0 failures / 0 errors / 0 pending : 0.004586 seconds

Running tests for heka_mock

luarocks install heka_mock busted lua-cjson lpeg
git clone [email protected]:timurb/heka_mock.git
cd heka_mock/test
busted

Expected output:

$ busted
●●●●●●●●●●●●●●●●●●●●●●●●
24 successes / 0 failures / 0 errors / 0 pending : 0.017647 seconds

Dependencies of lua-cjson and lpeg are required only for running tests for heka_mock, you may not need this in your setup.

API

(See also Heka docs for real behaviour of the functions in heka)

add_to_payload(val)

Add val to payload.

inject_payload(payload_type, payload_name, ...)

Inject payload to injection list and clear current payload.

inject_message(msg)

Converts msg to string and injects payload.

decode_message()

Processes message returned by read_message("raw") to table.

NOTE: This only works in this scenario and does not works with protobuf structures and other native Heka formats.

injected()

(NOT IMPLEMENTED IN HEKA UPSTREAM)

Returns a table with all injected messages since last reset_payload().

reset_payload()

(NOT IMPLEMENTED IN HEKA UPSTREAM)

Reset payload and injection list. Run it before every test.

write_message(),

Produces error. These function provided by Heka are not implemented as I don't quite understand what is their use case in lua code. Probably you should use Heka-provided test suite to test again these.

read_next_field()

Not implemented. I didn't have any use case for this so didn't create this one. May be one day I fix this or merge your pull request for this.

Mocks

mock_read_message(hash)

Set values for read_message() called by encoder. As in upstream read_message("raw") should return table with original unprocessed message for further handling with decode_message().

Note: no checking is done for non-standard Heka fields -- no errors are produced and they are not dropped while Heka will do either of these

Note: array values (read_message("Fields[foo]", 1, 2) -- 2+ params) are not supported as they are not widely used. Probably they'll be supported in a later release of this library.

mock_read_config(hash)

Set values for read_config() called by encoder. If no paramater passed only the default config params will be set (see default_config()).

set_default_config(hash)

If you need to set some config values shared by all tests define them here. Note: you still need to run mock_read_config() to set them.

Contributing

1. Fork repository
2. Do the fix
3. Write the tests
4. Create pull request

License

Copyright 2015-2016, Timur Batyrshin. License: MIT. See LICENSE for details.