Getting started with cURL

cURL is an incredibly powerful tool when working on the web, but getting started (and being useful) with it can take some time. I'm going to walk through what I do with cURL 95% of the time.

At its very basic, cURL makes requests to URL's. Normally you want to interact with those URL's in someway.

curl www.google.com

By default, cURL makes an HTTP GET request, so that will fetch the HTML
from www.google.com. Awesome!

Spitting it out in the terminal isn't so useful, so let's save it to a file:

curl -o google.html www.google.com

Looking at google.html will show you the same contents as before. This is especially useful when you want to save a JSON response.

Great! Well, interacting with HTML is pretty boring, most of what I do is interacting with API's. cURL is a great way for quickly testing your PHP script, your JSON endpoint, or what that API actually returns.

I worked at CloudMine and we had a decent JSON API, so I'll use that just as an
example. Just keep in mind that the URL isn't important here, but
rather the curl options.

By default, curl makes GET requests, so let's GET some data!

curl https://api.cloudmine.me/v1/app/928a78ffd73e4ff78383d1d4c06dd5a7/text?keys=all

Woh! What?

{ "errors": ["API Key invalid"] }

Oh, CloudMine expects an API Key to be sent as well. As an additional header? Sure, no problem:

curl https://api.cloudmine.me/v1/app/928a78ffd73e4ff78383d1d4c06dd5a7/text \
-H X-CloudMine-ApiKey:e90ef1aeaadd48de93b45038ed592a06

Response:

{"success":{},"errors":{}}`

Excellent! Sending the correct header worked. But the response is rather vague, I'm not sure if it worked and found no data, or it didn't work and didn't return an error. Let's examine the headers on the reply:

curl https://api.cloudmine.me/v1/app/928a78ffd73e4ff78383d1d4c06dd5a7/text \
-H X-CloudMine-ApiKey:e90ef1aeaadd48de93b45038ed592a06 -i

Response:

HTTP/1.1 200 OK
Date: Sat, 20 Dec 2014 17:31:10 GMT
Content-Type: application/json; charset=utf-8
Transfer-Encoding: chunked
Status: 200 OK
X-Request-Id: a9fbdaec-3b3c-4b11-92d8-5af2e9f01e8e
Cache-Control: max-age=0, private, must-revalidate
X-Runtime: 0.020247
X-Rack-Cache: miss
Access-Control-Allow-Origin: *
Access-Control-Expose-Headers: X-Request-Id

{"success":{},"errors":{}}

Adding -i will return all the headers on the response. If you only want the headers, you can use -I.

Excellent. But the URL I want to hit is actually an endpoint for consuming data. I want to send information. Let's make a POST request!

curl -X POST https://api.cloudmine.me/v1/app/928a78ffd73e4ff78383d1d4c06dd5a7/text \
-H X-CloudMine-ApiKey:e90ef1aeaadd48de93b45038ed592a06 -i

{"errors":[{"code":400,"message":"Invalid payload"}]}
```

Oops, we forgot to send data. To send information, we use `-d`.

```bash
curl -X POST https://api.cloudmine.me/v1/app/928a78ffd73e4ff78383d1d4c06dd5a7/text \
  -d '{"myrandomkey":{"name":"ethan"}}' \
  -H X-CloudMine-ApiKey:e90ef1aeaadd48de93b45038ed592a06 \
  -H "content-type:application/json"
```

Response:

```json
{ "success": { "myrandomkey": "created" }, "errors": {} }

CloudMine expects the Content-Type to be explicitly stated, so we add that as a header too. Other common ones are application/xml and application/x-www-form-urlencoded

Cool. Well, I don't like that object, so let's delete it.

curl -X DELETE "https://api.cloudmine.me/v1/app/928a78ffd73e4ff78383d1d4c06dd5a7/data?keys=myrandomkey" \
-H X-CloudMine-ApiKey:e90ef1aeaadd48de93b45038ed592a06
```

Response:

```json
{ "success": { "myrandomkey": "deleted" }, "errors": {} }

Welp, there it goes.

That's 95% of what I do with cURL. Some other useful options to know are:

-d @./path/to/file Will send the information in the file as the body of the request. Great when POSTing large amounts of data.

-L The request will follow any redirects it gets. Useful for when site's redirect you from url/ to url.

-u username:password Useful for going through HTTP Basic authentication.

-v Makes the request verbose, which you should always do when submitting a cURL request and response to a support person.

Lastly, sometimes your terminal will do funky things with the URL to escape it. If you don't like this, just put the text in quotes. cURL will respect it either way.