The simplest and functional HTTP requests in Go.
Ever find yourself going back and forth between net/http and io docs and your code while making HTTP calls? Requests takes care of that by abstracting several types on both ends to just Request
and Response
--just the way it should be. Find out how or jump right in to examples.
I need an HTTP request package that:
- is atomic--All HTTP request configurations should be set, intuitively, on the request only.
- wraps useful channels-ridden asynchronous patterns.
- has helper functions like marshaling and posting JSON.
- stays true to net/http APIs.
- is idiomatic Go.
the following are the core differences from the standard net/http
package.
requests employs functional options as optional parameters, this approach being idiomatic, clean, and makes a friendly, extensible API. This pattern is adopted after feedback from the Go community.
jsontype := func(r *requests.Request) {
r.Header.Add("content-type", "application/json")
}
res, err := requests.Get("http://example.com", jsontype)
requests uses custom Request and Response types to embed standard http.Request, http.Response, and http.Client in order to insert helper methods, make configuring options atomic, and handle asynchronous errors.
The principle is, the caller should be able to set all the configurations on the
"Request" instead of doing it on the client, transport, vice versa. For instance,
Timeout
can be set on the Request
.
timeout := func(r *requests.Request) {
// Set Timeout on *Request instead of *http.Client
r.Timeout = time.Duration(5) * time.Second
}
res, err := requests.Get("http://example.com", timeout)
if err != nil {
panic(err)
}
// Helper method
htmlStr := res.String()
Also, where http.Transport
was normally needed to set control over
proxies, TLS configuration, keep-alives, compression, and other settings,
now everything is handled by Request
.
tlsConfig := func(r *requests.Request) {
r.TLSClientConfig = &tls.Config{RootCAs: x509.NewCertPool()}
r.DisableCompression = true
}
res, _ := requests.Get("http://example.com", tlsConfig)
See Types and Methods for more information.
requests provides the following abstractions around sending HTTP requests in goroutines:
requests.GetAsync
requests.PostAsync
requests.Pool
All return a receive-only channel on which *requests.Response
can be waited on.
rc, err := requests.GetAsync("http://httpbin.org/get")
if err != nil {
panic(err)
}
res := <-rc
// Handle connection errors.
if res.Error != nil {
panic(res.Error)
}
// Helper method
content := res.Bytes()
See Async and Handling Async Errors for more usage information.
go get github.com/jochasinga/requests
requests uses Go standard testing
package. Run this in the project's directory:
go test -v -cover
Sending a basic GET request is straightforward.
res, err := requests.Get("http://httpbin.org/get")
if err != nil {
panic(err)
}
fmt.Println(res.StatusCode) // 200
To send additional data, such as a query parameter, or set basic authorization header or content type, use functional options.
// Add a query parameter.
addFoo := func(r *requests.Request) {
r.Params.Add("foo", "bar")
}
// Set basic username and password.
setAuth := func(r *requests.Request) {
r.SetBasicAuth("user", "pass")
}
// Set the Content-Type.
setMime := func(r *requests.Request) {
r.Header.Add("content-type", "application/json")
}
// Pass as parameters to the function.
res, err := requests.Get("http://httpbin.org/get", addFoo, setAuth, setMime)
Or configure everything in one function.
opts := func(r *requests.Request) {
r.Params.Add("foo", "bar")
r.SetBasicAuth("user", "pass")
r.Header.Add("content-type", "application/json")
}
res, err := requests.Get("http://httpbin.org/get", opts)
Send POST requests with specified bodyType
and body
.
res, err := requests.Post("https://httpbin.org/post", "image/jpeg", &buf)
It also accepts variadic number of functional options:
notimeout := func(r *requests.Request) {
r.Timeout = 0
}
res, err := requests.Post("https://httpbin.org/post", "application/json", &buf, notimeout)
Encode your map or struct data as JSON and set bodyType
to application/json
implicitly.
first := map[string][]string{
"foo": []string{"bar", "baz"},
}
second := struct {
Foo []string `json:"foo"`
}{[]string{"bar", "baz"}}
payload := map[string][]interface{}{
"twins": {first, second}
}
res, err := requests.PostJSON("https://httpbin.org/post", payload)
HEAD
, PUT
, PATCH
, DELETE
, and OPTIONS
are supported. See the doc for more info.
After parsing all the options, GetAsync
spawns a goroutine to send a GET request and return <-chan *Response
immediately on which *Response
can be waited.
timeout := func(r *requests.Request) {
r.Timeout = time.Duration(5) * time.Second
}
rc, err := requests.GetAsync("http://golang.org", timeout)
if err != nil {
panic(err)
}
// Do other things...
// Block and wait
res := <-rc
// Handle a "reject" with Error field.
if res.Error != nil {
panic(res.Error)
}
fmt.Println(res.StatusCode) // 200
select
can be used to poll many channels asynchronously like normal.
res1, _ := requests.GetAsync("http://google.com")
res2, _ := requests.GetAsync("http://facebook.com")
res3, _ := requests.GetAsync("http://docker.com")
for i := 0; i < 3; i++ {
select {
case r1 := <-res1:
fmt.Println(r1.StatusCode)
case r2 := <-res2:
fmt.Println(r2.StatusCode)
case r3 := <-res3:
fmt.Println(r3.StatusCode)
}
}
requests.Pool is recommended for collecting concurrent responses from multiple requests.
An asynchronous counterpart of requests.Post
.
query := bytes.NewBufferString(`{
"query" : {
"term" : { "user" : "poco" }
}
}`)
// Sending query to Elasticsearch server
rc, err := PostAsync("http://localhost:9200/users/_search", "application/json", query)
if err != nil {
panic(err)
}
resp := <-rc
if resp.Error != nil {
panic(resp.Error)
}
result := resp.JSON()
Contains a Responses
field of type chan *Response
with variable-sized buffer specified in the constructor. Pool
is used to collect in-bound responses sent from numbers of goroutines corresponding
to the number of URLs provided in the slice.
urls := []string{
"http://golang.org",
"http://google.com",
"http://docker.com",
"http://medium.com",
"http://example.com",
"http://httpbin.org/get",
"https://en.wikipedia.org",
}
// Create a pool with the maximum buffer size.
p := requests.NewPool(10)
opts := func(r *requests.Request) {
r.Header.Set("user-agent", "GoBot(http://example.org/"))
r.Timeout = time.Duration(10) * time.Second
}
results, err := p.Get(urls, opts)
// An error is returned when an attempt to construct a
// request fails, probably from a malformed URL.
if err != nil {
panic(err)
}
for res := range results {
if res.Error != nil {
panic(res.Error)
}
fmt.Println(res.StatusCode)
}
You may want to ignore errors from malformed URLs instead of handling each of them, for instance, when crawling mass URLs.
To suppress the errors from being returned, either thrown away the error with _
or
set the IgnoreBadURL
field to true
, which suppress all internal errors
from crashing the pool:
results, err := p.Get(urls, func(r *requests.Request) {
r.IgnoreBadURL = true
})
Pool.Responses
channel is closed internally when all the responses are sent.
It has embedded types *http.Request
and *http.Client
, making it an atomic
type to pass into a functional option.
It also contains field Params
, which has the type url.Values
. Use this field
to add query parameters to your URL. Currently, parameters in Params
will replace
all the existing query string in the URL.
addParams := func(r *requests.Request) {
r.Params = url.Values{
"name" : { "Ava", "Sanchez", "Poco" },
}
}
// "q=cats" will be replaced by the new query string
res, err := requests.Get("https://httpbin.org/get?q=cats", addParams)
It has embedded type *http.Response
and provides extra byte-like helper methods
such as:
Len() int
String() string
Bytes() []byte
JSON() []byte
These methods will return an equivalent of nil
for each return type if a
certain condition isn't met. For instance:
res, _ := requests.Get("http://somecoolsite.io")
fmt.Println(res.JSON())
If the response from the server does not specify Content-Type
as "application/json",
res.JSON()
will return an empty bytes slice. It does not panic if the content type
is empty.
These methods close the response's body automatically.
Another helper method, ContentType
, is used to get the media type in the
response's header, and can be used with the helper methods to determine the
type before reading the output.
mime, _, err := res.ContentType()
if err != nil {
panic(err)
}
switch mime {
case "application/json":
fmt.Println(res.JSON())
case "text/html", "text/plain":
fmt.Println(res.String())
default:
fmt.Println(res.Bytes())
}
requests.Response
also has an Error
field which will contain any error
caused in the goroutine within requests.GetAsync
and carries it downstream
for proper handling (Think reject
in Promise but more
straightforward in Go-style).
rc, err := requests.GetAsync("http://www.docker.io")
// This error is returned before the goroutine i.e. malformed URL.
if err != nil {
panic(err)
}
res := <-rc
// This connection error is "attached" to the response.
if res.Error != nil {
panic(res.Error)
}
fmt.Println(res.StatusCode)
Response.Error
is default to nil
when there is no error or when the response
is being retrieved from a synchronous function.
Check out my other project relay, useful test proxies and round-robin switchers for end-to-end HTTP tests.
Yes, please fork away.
To support my ends in NYC and help me push commits, please consider to fuel me with quality ๐ต or ๐ this repo for spiritual octane.
Reach me at @jochasinga.