Response handlers and helpers to parse JSON and deal with HTTP errors when using the browser fetch API
Install with yarn:
yarn add fetch-helpers
or if that's not your thing:
npm install fetch-helpers --save
See here for a browser polyfill if you are using the fetch API in a browser that doesn't support it yet. This module ships ES2015 code. This means that you will need to compile it before you can use it in the browser since no browser supports all of ES2015 yet. If you are using webpack/babel, in your webpack.config:
{
module: {
rules: [{
test: /\.jsx?$/,
use: {
loader: "babel-loader"
},
include: [
path.resolve("./src"), // assuming your source code lives in src
path.resolve("./node_modules/fetch-helpers") // compile this library, too
]
}]
}
}It is recommended to use babel-preset-env to only compile what you need to target the specific environment you want. Your .babelrc might look something like this:
{
"presets": [["env", {
"targets": {
"browsers": "last 2 versions"
},
"modules": false
}]]
}If using this library in node, make use of the node-fetch library to polyfill fetch:
global.fetch = require("node-fetch");Do that at the beginning of the entry point for your app and then you can use fetch-helpers as normal. The main target is precompiled for node v8 so as long as you are using that version or greater, you shouldn't have to worry about compilation of this library.
This is a HOC (higher order component) heavily inspired by react-refetch tweaked to work more efficiently with how we work with data @ CivicSource. Here is an example:
const UserProfile = ({ user }) => (
<span title={user.username}>{user.email}</span>;
);
export default UserProfile;This is a simple component that will display user information given a user object. We can use this in places where we already have the user object loaded and not have to worry about any unwanted network requests. However, there are a lot of places where we don't have the user object loaded (we just have a username or user ID) deep in the component tree and we also want to show user information. We can solve this by also including a "containerized" version of this component which wraps the component and takes a username to fetch the User and pass it down to the original component.
That might end up looking something like this:
// add some more props to show loading statuses
const UserProfile = ({ user, isFetching, isFetched, error }) => {
if (error) {
return <span>Error loading user: {error}</span>;
}
if (isFetching && !isFetched) {
return <span>Loading...</span>;
}
if (!isFetched) {
return null;
}
// we may still need to show a loading indicator if the user is loaded
// but we are loading more data for the user, e.g. "refreshing the user"
const loadingIndicator = isFetching ? <span>Loading...</span> : null;
return <span title={user.username}>{user.email} {loadingIndicator}</span>;
};
export default class UserContainer extends Component {
constructor(props) {
super(props);
this.state = {
isFetching: false,
isFetched: false,
error: null
};
}
componentDidMount() {
this.fetchUser();
}
fetchUser = async () => {
this.setState({
isFetching: true,
error: null
});
try {
let response = await fetch(`http://example.com/api/${this.props.username}`, {
method: "GET"
});
response = await checkStatus(response);
response = await parseJSON(response);
this.setState({
isFetched: true,
isFetching: false,
user: response
});
} catch (ex) {
this.setState({
isFetching: false,
error: ex.message
});
}
};
render() {
return (
<UserProfile {...this.props} {...this.state} />
);
}
};That is a lot of boilerplate to load the data once the component mounts and doesn't even handle the case where the username prop changes after the component is mounted. fetchOnUpdate can reduce all of that to this equivalent code:
import { connect } from "fetch-helpers";
const UserProfileFetcher = connect(({ username }) => ({
user: {
url: `http://example.com/api/${username}`, // the URL to make a fetch request to
method: "GET" // any other standard fetch options
}
}))(UserProfile);
export default UserProfileFetcher;The first argument is a function that takes the current props and should return a keyed object describing how to fetch the requested data. connect will generate the isFetching, isFetched, & error statuses and pass that down to the wrapped component under the user prop.
By default, it will only run the fetch when the component is initially mounted. Optionally, if your component takes more props and you want to fetch on certain prop updates (which is usually the case), you can pass a list of keys to connect:
const FetchingComponent = connect(({ username }) => ({
user: {
url: `http://example.com/api/${username}`,
keys: ["username", "someOtherProp"]
}
}))(UserProfile);
export default UserProfileFetcher;The keys are the props connect should monitor for changes and if any of those props change (via a shallow compare), it will run the fetch again. You can pass an arbitrary number of props to monitor for changes. Each prop can use object paths of arbitrary length: e.g. user.username; in which case the fetch will only run if the username field on user changes. If no keys are specified, the fetch will only be run once when the component is first mounted.
You can fetch multiple resources at once by passing more than one key to the connect HOC:
import { connect } from "fetch-helpers";
const FetchingComponent = connect(({ fruitType }) => ({
users: `http://example.com/api/users`,
fruits: {
url: `http://example.com/api/fruits/${fruitType}`,
keys: ["fruitType"]
}
}))(MyComponent);This will fetch users & fruits from the server as soon as the component is first mounted. After that, any time the fruitType prop changes, it will fetch the fruits again.
You can pass a data manipulation function, onData, to alter the data returned from the server before passing it to the component:
const UserProfileFetcher = connect(({ username }) => ({
user:{
url: `http://example.com/api/${username}`,
bearerToken: "poop",
keys: ["username"]
}
}))(UserProfile);You can pass a function to connect in order to bind that function as a prop to the component to be invoked later. You can use this for lazy loading data and/or responding with user input to POST data to the server for example:
connect(({ username }) => ({
user:{
url: `http://example.com/api/${username}`,
bearerToken: "poop",
keys: ["username"]
},
saveUser: (birthday) => ({
userSaveResult: {
url: `http://example.com/api/${username}`,
method: "POST",
body: JSON.stringify({ birthday })
}
})
}))(UserProfile);This will pass a saveUser function to the UserProfile component which can be invoked in response to a user action. The userSaveResult prop will also be passed down with the standard isFetching, isFetched, & error fields.
You can add a timer to reset the fetch status of an item. This is useful for clearing an item-saved indicator for example:
connect(({ username }) => ({
saveUser: (birthday) => ({
userSaveResult: {
url: `http://example.com/api/${username}`,
method: "POST",
body: JSON.stringify({ birthday }),
reset: 2000
}
})
}))(UserProfile);After saveUser has been invoked with a successful response, the userSaveResult prop will be cleared after 2 seconds.
As a convenience, you can pass a bearerToken to add an Authorization header to the outgoing request:
const UserProfileFetcher = connect(({ username }) => ({
users: {
url: `http://example.com/api/users`,
bearerToken: "mytoken"
}
}))(UserProfile);You can pass arbitrary fetch options that will be passed to fetch unaltered:
const UserProfileFetcher = connect(({ username }) => ({
users: {
url: `http://example.com/api/users`,
headers: { Accept: "text/plain" }
}
}))(UserProfile);This HOC performs a very similar to function as that of connect but instead of calculating statuses itself, it will run a specified function any time any of the given props change (via a shallow compare). This is useful, for example, if you just want to trigger a redux action and retrieve loading statuses & such from the store when props change:
const UserProfileFetcher = fetchOnUpdate(({ username, fetchUser }) => {
//this function will run on mount & whenever "username" changes
fetchUser(username);
}, "username")(UserProfile);
// connect is from redux
const UserProfileContainer = connect((state, props) =>({
user: state.users[props.username]
}), dispatch => ({
fetchUser: username => dispatch(fetchUser(username))
}))(UserProfileFetcher);
export default UserProfileContainer;The second argument to fetchOnUpdate is the prop to monitor for changes and if that prop changes, it will run the fetch again. You can pass an arbitrary number of props to monitor for changes. Each prop can use object paths of arbitrary length: e.g. user.username; in which case the fetch will only run if the username field on user changes.
Read here for the inspiration for this function. It will reject fetch requests on any non-2xx response. It differs from the example in that it will try to parse a JSON body from the non-200 response and will set any message field (if it exists) from the JSON body as the error message.
import { checkStatus } from "fetch-helpers";
//given a 400 Bad Request response with a JSON body of:
//{ "message": "Invalid arguments. Try again.", "someOtherThing": 42 }
fetch("/data", {
method: "GET",
headers: {
Accept: "application/json"
}
})
.then(checkStatus)
.catch(err => {
console.log(err.message); //Invalid Arguments. Try again.
console.log(err.response.statusText); //Bad Request
console.log(err.response.jsonBody); //{ "message": "Invalid arguments. Try again.", "someOtherThing": 42 }
});It will try to look for a message field first, and then an exceptionMessage falling back to the statusText if neither one exist or if the response body is not JSON.
A simple response handler that will simply parse the response body as JSON.
import { parseJSON } from "fetch-helpers";
//given a 400 Bad Request response with a JSON body of:
//{ "message": "Invalid arguments. Try again.", "someOtherThing": 42 }
fetch("/data", {
method: "GET",
headers: {
Accept: "application/json"
}
})
.then(parseJSON)
.then(json => console.log(json));A utility to allow easily batching fetch requests. Calling code calls the function as if it will make a single request while, internally, it will wait a predetermined amount of time before actually making the request.
import { batchFetch } from "fetch-helpers";
const getItem = batchFetch("itemId", chunk => fetch(`http://example.com/api/items/${chunk.join(",")}/`, {
method: "GET"
}));
for (let i = 1; i <= 10; i++) {
getItem(i).then(item => console.log(`item with id ${item.itemId} retrieved from the server`));
}The above example will make one request to the URL http://example.com/api/items/1,2,3,4,5,6,7,8,9,10/ but resolve all promises separately so that calling code is none-the-wiser that its requests have been batched into one. The keyName (in this case itemId) must be returned from the server in the results as that is how batchFetch matches what promises to resolve/reject.
The default batch size is 10 and the default timeout is 100ms. Both can be overridden:
const getItem = batchFetch("itemId", chunk => fetch(`http://example.com/api/items/${chunk.join(",")}/`, {
method: "GET"
}), {
maxBatchSize: 30,
timeout = 300
});Any extra parameters passed to the resulting function will be passed to the performFetch function:
const getItem = batchFetch("itemId", (chunk, method) => fetch(`http://example.com/api/items/${chunk.join(",")}/`, {
method: method
}));
getItem(42, "GET");
getItem(13, "GET");
getItem(69, "GET");
getItem(420, "GET"); // the last one wins (as far as the extra params passed to performFetch)After cloning this repo, run:
npm install
npm run compile
This will build the src into lib using babel.