curl Converter
Turn One Command Into Code

What A curl Converter Does

A curl command writes a request as a command line. This curl converter turns that one line into code you can drop into a project. Therefore the header you copied from devtools keeps its exact value, and you stop retyping it.

Four things come back from one paste. First, the request itself: method, address, headers and body, read back so you can check them. Then the code in the language you picked. After that, the switches curl would have applied, such as redirect following or a timeout. Finally, a list of anything the reader had to decide on its own.

That last list matters more than it looks. A silent guess inside generated code is a bug you find in production. So the page prints every decision next to the code instead.

You install nothing and you send nothing. So you paste a command, read the code, and close the tab when it suits you.

How A curl Converter Reads A Command

The work splits into three passes. First, the page splits the command into words the way a shell does. Then it reads each word as a flag, a value or the address. Finally it hands the request to a writer for the language you chose.

The splitting follows the shell rules for quoting, because a command pasted out of devtools leans on them. A body inside single quotes is one word even when it holds spaces. A backslash at the end of a line joins it to the next one.

Reading flags is where a converter earns its keep. Because curl lists over two hundred options, only some are worth carrying. The ones that matter change what a server sees. Those are the method, the headers, the body, the cookies, the redirects and the timeout. The page reports the rest rather than guessing at them.

The writers never touch the network. So the page works offline. It also works on a machine with no curl installed. A command can point at an address your browser cannot reach, and the page still converts it.

Why Quoting Is The Hard Part

A converter that splits on spaces alone gets most commands wrong. Because a header value often holds a space, a naive split cuts it in two. Then the code carries half a token and the request fails with a 401.

So the reader tracks the quote state as it walks the line. A double quote groups a run of words into one, and a single quote does the same without any escaping inside it. Only a quote that opens and closes at the right places counts.

A backslash is the third case. Outside quotes it escapes the next character, and before a newline it removes the newline itself. That is why a command pasted across twenty lines still reads as one command.

Which Languages It Writes

Thirteen targets cover the work most people bring here. For PHP there is plain cURL and the Guzzle client. For Python there is requests. For JavaScript there is fetch, axios and the Node.js http module. Go, Ruby, Java and C# each get one. Then come raw HTTP, a cleaned-up curl line, and JSON.

Each writer is a small function over the same request object. Because of that, every target sees identical bytes. The body sent by the PHP sample is the body sent by the Go sample. Nothing trims a header per language or quietly reformats the payload.

The PHP and Python writers aim at code you can run as-is. So the PHP one opens with the tag, sets the options and closes the handle. The Python one imports requests, builds the headers and prints both the status and the body.

JavaScript splits three ways on purpose. Fetch ships in the browser. Axios brings its own conveniences, and the Node.js http module suits a script with no dependencies. So all three earn a place.

Two targets are for reading rather than running. Raw HTTP prints the message as it would go over the wire. That is the fastest way to see whether a header is really there. JSON prints the same request as data, ready for a test fixture.

What The Read-Back Panel Shows

Before any code appears, the page prints the request it understood. Nine facts each get a line. They cover the method, the address, the header count, the body size and the content type. The authorization, the cookies, the switches and the timeout follow.

The addresses are not the interesting part. Rather, the point is the check. A header that went missing shows up here before you paste anything. So does a body that arrived as a form instead of JSON. A password in the wrong place shows up too.

The page prints the header list in full underneath, one header per line. Therefore you compare what curl would send against what your server expects. Then the difference is visible rather than assumed.

The panel names switches in plain words. A reader of the code should not have to know that -L means follow redirects. The panel says follow, insecure, compressed or silent instead.

From Devtools To A Script In One Paste

The usual way in is the browser. Every large browser can copy a request as curl. So open the network panel, right-click the call that failed, copy it, and paste it here. The command arrives with the real headers, the real body and the real cookies.

That is also the usual way to debug a private call. So you reproduce a failed request as code. Then you add a print and watch what the server says, instead of guessing from a network panel.

Line breaks do not matter. Therefore the page reads a command copied across twenty lines the same way it reads one line. Indentation does not matter either.