Pigweed provides an NPM package with modules to build web apps for Pigweed devices.

Also included is a basic React app that demonstrates using the npm package.

Getting Started#


If you have a bundler set up, you can install pigweedjs in your web application by:

$ npm install --save pigweedjs

After installing, you can import modules from pigweedjs in this way:

import { pw_rpc, pw_tokenizer, Device, WebSerial } from 'pigweedjs';

Import Directly in HTML#

If you don’t want to set up a bundler, you can also load Pigweed directly in your HTML page by:

<script src=""></script>
  const { pw_rpc, pw_hdlc, Device, WebSerial } from Pigweed;

Getting Started#

Easiest way to get started is to build pw_system demo and run it on a STM32F429I Discovery board. Discovery board is Pigweed’s primary target for development. Refer to target documentation for instructions on how to build the demo and try things out.

pigweedjs provides a Device API which simplifies common tasks. Here is an example to connect to device and call EchoService.Echo RPC service.

<h1>Hello Pigweed</h1>
<button onclick="connect()">Connect</button>
<button onclick="echo()">Echo RPC</button>
<br /><br />
<script src=""></script>
<script src=""></script>
  const { Device } = Pigweed;
  const { ProtoCollection } = PigweedProtoCollection;

  const device = new Device(new ProtoCollection());

  async function connect(){
    await device.connect();

  async function echo(){
    const [status, response] = await"Hello");
    document.querySelector('code').innerText = "Response: " + response;

pw_system demo uses pw_log_rpc; an RPC-based logging solution. pw_system also uses pw_tokenizer to tokenize strings and save device space. Below is an example that streams logs using the Device API.

<h1>Hello Pigweed</h1>
<button onclick="connect()">Connect</button>
<br /><br />
<script src=""></script>
<script src=""></script>
  const { Device, pw_tokenizer } = Pigweed;
  const { ProtoCollection } = PigweedProtoCollection;
  const tokenDBCsv = `...` // Load token database here

  const device = new Device(new ProtoCollection());
  const detokenizer = new pw_tokenizer.Detokenizer(tokenDBCsv);

  async function connect(){
    await device.connect();
    const call = => {
      msg.getEntriesList().forEach((entry) => {
        const frame = entry.getMessage();
        const detokenized = detokenizer.detokenizeUint8Array(frame);
        document.querySelector('code').innerHTML += detokenized + "<br/>";

The above example requires a token database in CSV format. You can generate one from the pw_system’s .elf file by running:

$ pw_tokenizer/py/pw_tokenizer/ create \
--database db.csv out/stm32f429i_disc1_stm32cube.size_optimized/obj/pw_system/bin/system_example.elf

You can then load this CSV in JavaScript using fetch() or by just copying the contents into the tokenDBCsv variable in the above example.



Device class is a helper API to connect to a device over serial and call RPCs easily.

To initialize device, it needs a ProtoCollection instance. pigweedjs includes a default one which you can use to get started, you can also generate one from your own .proto files using pw_proto_compiler.

Device goes through all RPC methods in the provided ProtoCollection. For each RPC, it reads all the fields in Request proto and generates a JavaScript function that accepts all the fields as it’s arguments. It then makes this function available under rpcs.* namespaced by its package name.

Device has following public API:

  • constructor(ProtoCollection, WebSerialTransport <optional>, rpcAddress <optional>)

  • connect() - Shows browser’s WebSerial connection dialog and let’s user make device selection

  • rpcs.* - Device API enumerates all RPC services and methods present in the provided proto collection and makes them available as callable functions under rpcs. Example: If provided proto collection includes Pigweed’s Echo service ie. pw.rpc.EchoService.Echo, it can be triggered by calling"some message"). The functions return a Promise that resolves an array with status and response.


To help with connecting to WebSerial and listening for serial data, a helper class is also included under WebSerial.WebSerialTransport. Here is an example usage:

import { WebSerial, pw_hdlc } from 'pigweedjs';

const transport = new WebSerial.WebSerialTransport();
const decoder = new pw_hdlc.Decoder();

// Present device selection prompt to user
await transport.connect();

// Or connect to an existing `SerialPort`
// await transport.connectPort(port);

// Listen and decode HDLC frames
transport.chunks.subscribe((item) => {
  const decoded = decoder.process(item);
  for (const frame of decoded) {
    if (frame.address === 1) {
      const decodedLine = new TextDecoder().decode(;

// Later, close all streams and close the port.

Individual Modules#

Following Pigweed modules are included in the NPM package:

Web Console#

Pigweed includes a web console that demonstrates pigweedjs usage in a React-based web app. Web console includes a log viewer and a REPL that supports autocomplete. Here’s how to run the web console locally:

$ cd pw_web/webconsole
$ npm install
$ npm run dev

Log Viewer Component#

The NPM package also includes a log viewer component that can be embedded in any webapp. The component works with Pigweed’s RPC stack out-of-the-box but also supports defining your own log source.

The component is composed of the component itself and a log source. Here is a simple example app that uses a mock log source:

<div id="log-viewer-container"></div>
<script src=""></script>

  const { createLogViewer, MockLogSource } = PigweedLogging;
  const logSource = new MockLogSource();
  const containerEl = document.querySelector(

  let unsubscribe = createLogViewer(logSource, containerEl);
  logSource.start(); // Start producing mock logs


The code above will render a working log viewer that just streams mock log entries.

It also comes with an RPC log source with support for detokenization. Here is an example app using that:

<div id="log-viewer-container"></div>
<script src=""></script>
<script src=""></script>
<script src=""></script>

  const { Device, pw_tokenizer } = Pigweed;
  const { ProtoCollection } = PigweedProtoCollection;
  const { createLogViewer, PigweedRPCLogSource } = PigweedLogging;

  const device = new Device(new ProtoCollection());
  const logSource = new PigweedRPCLogSource(device, "CSV TOKEN DB HERE");
  const containerEl = document.querySelector(

  let unsubscribe = createLogViewer(logSource, containerEl);


Custom Log Source#

You can define a custom log source that works with the log viewer component by just extending the abstract LogSource class and emitting the logEntry events like this:

import { LogSource, LogEntry, Severity } from 'pigweedjs/logging';

export class MockLogSource extends LogSource {
    // Do any initializations here
    // ...
    // Then emit logs
    const log1: LogEntry = {

      severity: Severity.INFO,
      timestamp: new Date(),
      fields: [
        { key: 'severity', value: severity }
        { key: 'timestamp', value: new Date().toISOString() },
        { key: 'source', value: "LEFT SHOE" },
        { key: 'message', value: "Running mode activated." }

After this, you just need to pass your custom log source object to createLogViewer(). See implementation of PigweedRPCLogSource for reference.

Column Order#

Column Order can be defined on initialization with the optional columnOrder parameter. Only fields that exist in the Log Source will render as columns in the Log Viewer.

createLogViewer(logSource, root, state, logStore, columnOrder)

columnOrder accepts an string[] and defaults to [log_source, time, timestamp]

createLogViewer(logSource, root, state, logStore, ['log_source', 'time', 'timestamp'])

Note, columns will always start with severity and end with message, these fields do not need to be defined. Columns are ordered in the following format:

  1. severity

  2. columnOrder

  3. Fields that exist in Log Source but not listed will be added here.

  4. message

Color Scheme#

The log viewer web component provides the ability to set the color scheme manually, overriding any default or system preferences.

To set the color scheme, first obtain a reference to the log-viewer element in the DOM. A common way to do this is by using querySelector():

const logViewer = document.querySelector('log-viewer');

You can then set the color scheme dynamically by updating the component’s colorScheme property or by setting a value for the colorscheme HTML attribute.

logViewer.colorScheme = 'dark';
logViewer.setAttribute('colorscheme', 'dark');

The color scheme can be set to 'dark', 'light', or the default 'auto' which allows the component to adapt to the preferences in the operating system settings.

Material Icon Font (Subsetting)#

Log Viewer uses a subset of Material Icons Rounded hosted on GitHub with codepoints listed in the codepoints file

(It’s easiest to look up the codepoints at e.g. see the sidebar shows the Codepoint for “home” is e88a)

The following icons with codepoints are curently used:

  • delete_sweep e16c

  • error e000

  • warning f083

  • cancel e5c9

  • bug_report e868

  • view_column e8ec

  • brightness_alert f5cf

  • wrap_text e25b

  • more_vert e5d4

To save load time and bandwidth, we provide a pre-made subset of the font with just the codepoints we need, which reduces the font size from 3.74MB to 12KB.

We use fonttools ( to create the subset. To create your own subset, find the codepoints you want to add and:

  1. Start a python virtualenv and install fonttools

virtualenv env
source env/bin/activate
pip install fonttools brotli
  1. Download the the raw MaterialSybmolsRounded woff2 file

# line below for example, the url is not stable: e.g.
curl -L -o MaterialSymbolsRounded.woff2 \
  1. Run fonttools, passing in the unicode codepoints of the necessary glyphs. (The points for letters a-z, numbers 0-9 and underscore character are necessary for creating ligatures)


Ensure there are nono spaces in the list of codepoints.

fonttools subset MaterialSymbolsRounded.woff2 \
   --unicodes=5f-7a,30-39,e16c,e000,e002,e8b2,e5c9,e868,e8ec,f083,f5cf,e25b,e5d4 \
   --no-layout-closure \
   --output-file=material_symbols_rounded_subset.woff2 \
  1. Update material_symbols_rounded_subset.woff2 in log_viewer/src/assets with the new subset