cichy / rescript-logger

Logging implementation for ReScript

rescript-logger

Logging implementation for ReScript.

Features

• Zero runtime in production builds.
• Multiple logging levels.
• Customizable verbosity.
• [@log] helper.
• @rescript/react integration.
• Custom loggers.
• Logging in libraries.

Installation

Get the package:

# yarn
yarn add rescript-logger
# or npm
npm install --save rescript-logger

Then add it to bsconfig.json:

"bs-dependencies": [
  "rescript-logger"
],
"ppx-flags": ["rescript-logger/ppx"]

Usage

There are 5 log levels:

• trace
• debug
• info
• warn
• error

You can log message of specific level using appropriate macros:

// ReScript
%log.info("Info message")
%log.error("Error message")

// Reason
[%log.info "Info message"];
[%log.error "Error message"];

Additional data

You can add data to log entry like this:

// ReScript
%log.info(
  "Info message"
  ("Foo", 42)
)
%log.info(
  "Info message"
  ("Foo", {x: 42})
  ("Bar", [1, 2, 3])
)

// Reason
[%log.info "Info message"; ("Foo", 42)];
[%log.info
  "Info message";
  ("Foo", {x: 42});
  ("Bar", [1, 2, 3]);
];

Currently, logger can accept up to 7 additional entries.

Verbosity customization

Output verbosity can be customized by providing specific log level and/or code locations.

Log level

You can set maximum log level via environment variable RES_LOG.

Let's say you want to log only warnings and errors. To make it happen, run your build like this:

RES_LOG=warn rescript build

Available RES_LOG values:

• *: log everything
• trace: basically, the same as *
• debug: log everything except trace level messages
• info: log everything except trace & debug level messages
• warn: log warn & error messages only
• error: log error messages only
• off: don't log anything

If RES_LOG is set to off, nothing will be logged and none of the log entries will appear in your JS assets.

In case if RES_LOG environment variable is not set, log level warn will be used.

Also, see Usage in libraries.

Code location

If you want to focus on logging from specific part(s) of your code, you can use RES_LOG_ONLY environment variable.

For example, if you want to see logs only from module Test, run the build as following:

RES_LOG_ONLY=Test rescript build

You can pass submodules and functions to it as well. If you want to log from multiple locations, separate them by ,.

Consider the following source:

// Test.res
%log.warn("Top level message")

module Submodule1 = {
  %log.warn("Message from Submodule1")
}

module Submodule2 = {
  %log.warn("Message from Submodule2")

  let fn = () => %log.warn("Message from function within Submodule2")
  fn()
}

Here is what will be logged with different build configurations:

# build
RES_LOG_ONLY=Test rescript build

# output
WARNING  [Test] Top level message
WARNING  [Test.Submodule1] Message from Submodule1
WARNING  [Test.Submodule2] Message from Submodule2
WARNING  [Test.Submodule2.fn] Message from function within Submodule2

# build
RES_LOG_ONLY=Test.Submodule2 rescript build

# output
WARNING  [Test.Submodule2] Message from Submodule2
WARNING  [Test.Submodule2.fn] Message from function within Submodule2

# build
RES_LOG_ONLY=Test.Submodule1,Test.Submodule2.fn rescript build

# output
WARNING  [Test.Submodule1] Message from Submodule1
WARNING  [Test.Submodule2.fn] Message from function within Submodule2

[@log] helper

This helper can be placed in front of any switch expression with constructor patterns and it will inject debug expressions into each branch.

// ReScript
let _ =
  x =>
    @log
    switch x {
    | A => "A"
    | B(b) => b
    }

// Reason
let _ =
  x =>
    [@log]
    switch (x) {
    | A => "A"
    | B(b) => b
    }

Without a @log helper, an equivalent would be:

// ReScript
let _ =
  x =>
    switch (x) {
    | A =>
      %log.debug("A")
      "A"
    | B(b) =>
      %log.debug("B with payload" ("b", b))
      b
    }

// Reason
let _ =
  x =>
    switch (x) {
    | A =>
      [%log.debug "A"];
      "A";
    | B(b) =>
      [%log.debug "B with payload"; ("b", b)];
      b;
    }

You can pass optional custom namespace to helper like this: @log("MyNamespace").

[@log] helper works only for switch expressions with constructor patterns, for now. Let us know in the issues if you need to handle more cases.

@rescript/react integration

Using @log helper, you can log dispatched actions in your components.

Annotate reducer function like this:

// ReScript
let reducer =
  (state, action) =>
    @log
    switch action {
      ...
    }

// Reason
let reducer =
  (state, action) =>
    [@log]
    switch (action) {
      ...
    }

These entries are logged on the debug level so none of it will appear in your production builds.

Custom loggers

rescript-logger ships with 3 loggers:

• ReScriptLogger.Browser (default)
• ReScriptLogger.Node
• ReScriptLogger.Universal (picks either Browser or Node logger at runtime depending on an environment, useful for apps with SSR)

And you can easily plug your own.

For example, in development, you want to log everything to console using default logger, but in production, you want to disable console logging and send error level events to bug tracker.

To implement your own logger, you need to create a module (e.g. BugTracker.re) and set the following environment variables for production build.

RES_LOG=error
RES_LOGGER=BugTracker

Considering that you want to log only error level messages, you need to create functions only for errors logging.

// BugTracker.res

let error = (loc, msg) => BugTrackerSDK.notify(`${msg} in ${loc.rootModule}`)

let error1 =
  (
    loc,
    msg,
    (label, payload),
  ) =>
    BugTrackerSDK.notify(
      `${msg} in ${loc.rootModule}`,
      [|(label, payload)|],
    );

let error2 =
  (
    loc,
    msg,
    (label1, payload1),
    (label2, payload2),
  ) =>
    BugTrackerSDK.notify(
      `${msg} in ${loc.rootModule}`,
      [|
        (label1, payload1),
        (label2, payload2),
      |],
    );

// Up to 7

The first argument loc is a ReScriptLogger.Location.t record. It's passed by PPX and contains the location data.

type t = {
  rootModule: string,
  subModulePath: list<string>,
  value: option<string>,
  fullPath: string,
  filePath: string,
}

If Test.Submodule.fn gets called, logger would receive the following location:

// Test.res
module Submodule = {
  let fn = () => %log.warn("Warn!")
}

// Location
{
  rootModule: "Test",
  subModulePath: list{"Submodule"},
  value: Some("fn"),
  fullPath: "Test.Submodule.fn",
  filePath: "/absolute/path/to/project/src/Test.res",
}

---

Note, you don't have to re-implement all functions from the default logger, only the ones you actually use. Don't worry to forget to implement something. If later on, you will attempt to use unimplemented method it will be compile time error.

Usage in libraries

I you develop a library and want to use rescript-logger during development process, you can do so without spamming output of consumers of your library.

rescript-logger/ppx accepts --lib flag:

"ppx-flags": [
  ["rescript-logger/ppx", "--lib=my-lib"]
]

Once this flag is passed, you need to provide special value of RES_LOG to log your entries:

RES_LOG=my-lib=* rescript build

If consumers of your lib would like to see log output from your lib, they can do so too by extending a value of RES_LOG variable:

RES_LOG=*,my-lib=error rescript build

Few more examples to illustrate how it works:

# log everything from application code only
RES_LOG=* rescript build

# log everything from application code
# log errors from `my-lib`
RES_LOG=*,my-lib=error rescript build

# log everything from application code
# log errors from `my-lib-1`
# log warnings and errors from `my-lib-2`
RES_LOG=*,my-lib-1=error,my-lib-2=warn rescript build

Caveats

Logging is disabled after file save
If you run bsb via editor integration, make sure editor picked up RES_LOG variable. E.g. if you use Atom run it like this:

RES_LOG=info atom .

If your editor is telling you, variables used in ppx are unused, you can either:

1. prefix such variables with _
2. or open editor with RES_LOG variable set to appropriate level.

Changing value of RES_LOG/RES_LOGGER/RES_LOG_ONLY doesn't make any effect
When you change a value of environment variable, rescript clean before the next build.

Developing

Repo consists of 2 parts:

• ReScript lib: dependencies are managed by yarn
• OCaml PPX: dependencies are managed either by nix (in development) or esy (in development and/or on CI)

Nix flow

Clone repo and either enter the Nix shell:

nix-shell

Or use direnv and create .envrc file in the root directory of the project with the folowing content:

use nix

Then install deps:

yarn install

Build loggers and ppx:

dune build
cd lib && yarn run build
cd ../examples && yarn run build

Esy flow

Clone repo and install deps:

esy install
yarn install

Build loggers and ppx:

esy build
cd lib && yarn run build
cd ../examples && yarn run build

Auto-formatting

Note, this project doesn't use auto-formatting in OCaml files (*.ml), so if you're intended to contribute, please, turn off auto-formatting in the editor while editing such files.