Skip to content

Repository files navigation

Inspiration

With the rise in usage of microservices, Making http requests is essential job of most devs. For these, there are multiple options like curl, insomnia postman. My Ideal choice is to use curl but problems with it is, having no history and no easy way to save and rerun. Postman solves that problem, but once user logs in(earlier it used to be optional, now its mandatory), it stores all request metadata(body/headers/urls/credentails) to their servers although it helps solve using it across multiple devices, backup but it could be potential security loop hole if they can access it. (now a days, postman desktop is super slow also).

This project aims to solve save&reuse requests, maintaining history and no sync in proprietery servers (although you can achive sync via commiting http files into git).

GOAL

dothttp will provide simple, cleaner architecture for making http requests. It uses xtext (eclipse developed dsl) to build a custom dsl.

Documentation

More information or docs can be cound at https://docs.dothttp.dev


Go through this example for better understanding. for babysteps click here

# users.http

#!/usr/bin/env /home/prasanth/cedric05/dothttp/dist/dothttp-cli

var name=adam; // this is how variable is created;
var username='username'; 
var password='password';

# this is comment

// this is also a comment

/*
   this is multi line
   comment
*/

# http file can have multiple requests, name tag/annotation is used to identify
@name("fetch 100 users, skip first 50")

# makes are get request, with url `https://req.dothttp.dev/user`
GET https://req.dothttp.dev/user

# below is an header example
"Authorization": "Basic dXNlcm5hbWU6cGFzc3dvcmQ="

# below is how you set url params '?' --> signifies url quary param
? ("fetch", "100") #
? ("skip", "50")
? projection, name
? projection, org
? projection, location




# makes are post request, with url `https://req.dothttp.dev/user`
POST https://req.dothttp.dev/user

basicauth('username', 'password')
/*
   below defines payload for the post request.
   json --> signifies payload is json data
*/
json({
    "name": "{{name}}", # name is templated, if spcified via env or property, it will be replaced
    "org": "dothttp",
    "location": "Hyderabad",
    # "interests": ["exploring", "listening to music"],
})



# makes put request, with url `https://req.dothttp.dev/user/1`
PUT https://req.dothttp.dev/post

# define headers in .dothttp.json with env
basicauth("{{username}}, "{{password}}")

# posts with urlencoded
urlencoded({
    "name": "Adam A",
    "org": "dothttp",
    "location": "Hyderabad",
    "interests": ["exploring", "listening to music"],
})

// or use below one
// data('name=Adam+A&org=dothttp&location=Hyderabad&interests=%5B%27exploring%27%2C+%27listening+to+music%27%5D')

KICKSTART

From pypi

pip install dothttp-req==0.0.10

From source

git clone git@github.com:cedric05/dothttp.git
cd dothttp

python3 -m pip install pipenv
pipenv install

python3.9

python3 -m dothttp examples/dothttpazure.http

docker

docker build -t dothttp .
docker run -it --rm dothttp

whalebrew

docker run -it --rm dothttp

Features

  1. easy and cleaner http syntax
  2. variable substitution with property file
  3. generates curl from http for easy sharing
docker build -t dothttp .
whalebrew install dothttp
dothttp examples/dothttpazure.http

First DotHttpRequest and more

GET "http://localhost:8000/get"

dothttp get.http or python -m dothttp get.http

Run

dothttp simple.http

prints

{
  "args": {},
  "headers": {
    "Accept-Encoding": "identity",
    "Host": "httpbin.org",
    "User-Agent": "python-urllib3/1.26.3",
    "X-Amzn-Trace-Id": "Root=1-6022266a-20fb552e530ba3d90c75be6d"
  },
  "origin": "117.216.243.24",
  "url": "http://localhost:8000/get"
}

POST request

POST "http://localhost:8000/post"
{
  "args": {},
  "data": "",
  "files": {},
  "form": {},
  "headers": {
    "Accept-Encoding": "identity",
    "Content-Length": "0",
    "Host": "httpbin.org",
    "User-Agent": "python-urllib3/1.26.3",
    "X-Amzn-Trace-Id": "Root=1-602228fa-3c3ed5213b6d8c2d2a223148"
  },
  "json": null,
  "origin": "117.216.243.24",
  "url": "http://localhost:8000/post"
}

similarly, other methodsGET, POST, OPTIONS, DELETE, CONNECT, PUT, HEAD, TRACE support is available.

Query

query params can be added to request by specifying query ( "key", "value") ? "key", "value" ? "key": "value" ? "key"= "value" all four are accepted. going with query("hi", "hi2) is more readable. ?"key"= "value" is more concise

Payload

user can specify payload by mentioning below four forms (for various scenarios).

  • data("ram")

    user can also mention its content-type with data("ram", "text/plain")

  • urlencoded({"key": "value"}) for form input.

  • json({"key": "value"}) for json payload.

  • fileinput("path/to/file", "type") uploads file as payload (type is optional).

  • files(("photo", "path/to/file/photo.jpg", "image/jpeg"), ("photo details", '{"name":"prasanth"}', "application/json") )

    for multipart upload dothttp will figure out content type by going through file/data, when type is not mentioned.

Comments

dothttp will use # for commenting entire line.

  1. // line comment. follows java, javascript
  2. # line comment. follows python's comment style
  3. /* */ multi line comment. follows java/javascript style

Templating

POST 'http://localhost:8000/post'
? ("{{key}}", "{{value}}")
data('{"{{key}}" :"{{value}}"}', 'application/json')
  • specify variable values through property file (sample.json).
    • user can define environments and can activate multiple environments at a time

    • dothttp by default will read variables from "*" section

    • for example dothttp --property-file path/to/file.json --env ram chandra

      will activate * section properties, ram section properties and chandra section properties dothttp --env ram chandra will activate * section properties, ram section properties and chandra section properties from .dothttp.json in httpfile name space

  • through command line dothttp --property key=ram value=ranga will replace {{ram}} to ranga from the file
  • through file itself. (will be helpful for default properties)
POST 'https://{{host=httpbin.org}}/post'

Headers

User can define headers in below three formats

  1. header('content-type', 'application/json') readable
  2. 'content-type': 'application/json' concise
  3. property file headers section from property-file can also be used. in most scenarios, headers section will be common for a host. having them in property file would ease them.

Authentication

BasicAuth

basicauth('username','password')' --> will compute add respective headers.

DigestAuth

digestauth('username','password')' --> will compute add respective headers.

NtlmAuth

ntlmauth('username','password')' --> will compute add respective headers.

Property file

{
  "*": {
    "host": "httpbin.org"
  },
  "headers": {
    "content-type": "plain/text"
  },
  "preprod": {
    "host": "preprod.httpbin.org"
  }
}

Special sections in property file

  1. * section in property file will be activated once user specifies property file if user didn't specifiy file and .dothttp.json exists, it will be activated
  2. headers once a property file is activated. headers from property file will be added to request by default without user having to specify in .http file

Formatter (experimental phase)

dothttp can format a http file using below command dothttp -fmt examples/dothttpazure.http --experimental or dothttp --format examples/dothttpazure.http --experimental

to print to command line

dothttp --format examples/dothttpazure.http --experimental --stdout

Editor support

syntax highlighting for visual studio code is supported via dothttp-code

Command line options

usage: dothttp [-h] [--curl] [--property-file PROPERTY_FILE] [--no-cookie]
               [--env ENV [ENV ...]] [--debug] [--info] [--format]
               [--experimental] [--stdout]
               [--property PROPERTY [PROPERTY ...]] [--target TARGET]
               [file]

http requests for humans

options:
  -h, --help            show this help message and exit

general:
  --curl                generates curl script
  --no-cookie, -nc      cookie storage is disabled
  --debug, -d           debug will enable logs and exceptions
  --info, -i            more information
  file                  http file. use `-` or omit to read http content from
                        stdin
  --target, -t TARGET   targets a particular http definition

property:
  --property-file, -p PROPERTY_FILE
                        property file
  --env, -e ENV [ENV ...]
                        environment to select in property file. properties
                        will be enabled on FIFO
  --experimental, --b   enable experimental
  --property PROPERTY [PROPERTY ...]
                        list of property's

format:
  --format, -fmt        format http file
  --stdout              print to commandline

checkout examples


Using dothttp with AI agents / coding assistants

dothttp is stdin/stdout friendly, so an AI agent (Claude Code, Cursor, a custom LLM tool, a CI script, etc.) can fire a request without ever writing a .http file to disk or leaving temp files behind.

Pipe the .http content on stdin using - as the file argument (or just omit the file argument altogether):

echo 'GET "https://httpbin.org/get"' | dothttp -
# or, equivalently
echo 'GET "https://httpbin.org/get"' | dothttp

The response body is written to stdout, exactly like the file-based flow — no extra banners or separators clutter the output unless a test script (see Templating) actually produces something to report.

This composes with the regular flags, so an agent can still target a specific request, substitute properties, or ask for a curl script instead of executing the request:

# multiple requests in one payload, run the 2nd one
# a heredoc is the easiest way to hand over multi-line content
dothttp - --target 2 <<EOF
GET "https://httpbin.org/get"

POST "https://httpbin.org/post"
EOF

# substitute a property from the command line
echo 'GET "https://httpbin.org/get?key={{myprop}}"' \
  | dothttp - --property myprop=myvalue

# get a shareable curl command instead of executing the request
echo 'GET "https://httpbin.org/get"' | dothttp - --curl

Property-file auto-discovery (.dothttp.json/.yaml/.yml/.toml in the current directory) still works in stdin mode, since it's resolved relative to the working directory the command is run from — handy for agents that run from within a project checkout.

An explicit >> "path/to/file" output directive inside the piped content still writes to that file instead of stdout, same as file-based input.

Reusing existing request definitions (import + @name(...) : base)

An agent working inside a project usually doesn't need to reinvent auth, headers, or retry/timeout settings for every new request — a project's existing .http files already define them. dothttp supports import and named request inheritance (@name("child") : "base"), so an agent's stdin snippet can just import the file with the shared setup and extend it, inheriting everything from the base and overriding only what's different (the path, method, payload, etc).

For example, examples/inheritance_example.http already defines an "api base" request with a host, Authorization header, timeout, and retry configured. An agent can reuse all of it with a two-line snippet:

dothttp - --target "get repo" <<'EOF'
import "examples/inheritance_example.http";

@name("get repo") : "api base"
GET /repos/cedric05/dothttp
EOF

This resolves to GET https://api.github.com/repos/cedric05/dothttp with the Authorization, timeout, and retry settings from "api base" applied automatically — the agent never had to know or copy those details. import paths (like --property-file auto-discovery above) are resolved relative to the working directory the command is run from, so this works best when the agent runs from the project root.


Vscode alternatives


Non Vscode alternatives