VRC is a Vim plug-in to help send requests to and display responses from RESTful services in Vim. It's useful for working with REST services that use JSON to exchange information between server and client such as ElasticSearch.

VRC can also be used as a cURL client for simple needs such as getting a HTTP page response or posting to a form.

Requirements:

• cURL
• Vim 7.4 (might work with the older Vim versions)

• Execute REST request and display the response on a separate display buffer.
• Make changing/adjusting request body easy.
• Can have multiple REST request blocks per VRC buffer.
• Can have multiple VRC buffers where they all share the same output buffer or each can have its own output buffer.
• Particularly useful for working with REST services that require the request body to be sent in JSON such as ElasticSearch.
• Syntax highlighting.
• Supported verbs: GET, POST, PUT, HEAD, PATCH, OPTIONS, and TRACE.

VRC requires cURL. It's tested with Vim 7.4 but might work with the older versions.

To install using pathogen.vim

cd ~/.vim/bundle
git clone https://github.com/Aadniz/vim-rest-console.git

To install using Vundle

" Add this line to .vimrc
Plugin 'Aadniz/vim-rest-console'

Other methods should work as well.

For more examples, check out

https://raw.githubusercontent.com/Aadniz/vim-rest-console/master/sample.rest

there is also an alternative version using global settings:

https://raw.githubusercontent.com/Aadniz/vim-rest-console/master/sample_global.rest

The following examples assume that an ElasticSearch service is running at localhost. The pipe (|) indicates the current position of the cursor.

• From the command line, run a new Vim instance.

• Set the buffer filetype to rest by

  :set ft=rest
  

• Type in

  http://localhost:9200
  GET /_cat/nodes?v|
  

• Hit the trigger key (<C-j> by default).

• A new vertically split buffer will be shown to display the output.

• Change the request block to (or add another one)

  http://localhost:9200
  POST /testindex/testtype
  {
    "key": "new key",
    "value": "new value"|
  }
  

• Hit the trigger key with the cursor placed anywhere within this request block.

• The display buffer will be updated with the new response.

This example continues the previous one.

• Open a new VRC buffer in a new tab

  :tabe NewVrc.rest
  

• Since the new buffer has the extension rest, the VRC plug-in is active for this one.

• Set b:vrc_output_buffer_name of this buffer to __NEW_VRC__

  :let b:vrc_output_buffer_name = '__NEW_VRC__'
  

• Type in a request block such as

  http://localhost:9200
  GET /testindex/_search?pretty|
  

• Hit the trigger key.

• A new display buffer will be created showing the response.

• Go back to the VRC buffer of the previous example (previous tab).

• Try to execute an existing request block.

• The corresponding display buffer will be updated.

This plug-in is activated when Vim opens a buffer of type rest. This may be a file with the extension .rest or a buffer with filetype explicitly set to rest by

:set ft=rest

A VRC buffer can have one or many REST request blocks. A request block contains a host, optional cUrl options, optional headers, query, and an optional request body (usually used by POST). A block is defined as follows.

# host
http[s]://domain[:port]

[optional cUrl options]

[optional headers]

# query
POST /path/to/resource
[optional request body]

A comment starts with # or // and must be on its own line. The following is an example of a VRC buffer with multiple request blocks.

# GETting from resource.
http://example.com
GET /path/to/resource?key=value

# POSTing to an ElasticSearch service.
http://example.com/elasticsearch

// Specify optional headers.
Content-Type: application/json; charset=utf-8

POST /index/type?pretty
{
    "key": "a key",
    "value": "a value"
}

# Submitting a form.
https://example.net:8080

Accept: */*
Authorization: Basic QWxhZGRpbjpvcGVuIHNlc2FtZQ==
Cache-Control: no-cache
Connection: keep-alive
Content-Type: application/x-www-form-urlencoded
Cookie: userId=ac32:dfbe:8f1a:249c; sid=cfb48e3d98fcb1
User-Agent: VRC

POST /form
var1=value of var1&
var2=value of var2

When the trigger key is called (<C-j> by default), VRC processes the request block that the cursor stays within. The response is displayed in a new vertically split buffer. This output buffer is reused if it's already present.

By default, the display/output buffer is named __REST_response__. If there are multiple VRC buffers, they all share the same display buffer. To have a separate output display for each VRC buffer, b:vrc_output_buffer_name can be set in the buffer scope.

A recent addition to VRC is the ability to specify cUrl options. These may be specified by the VRC option vrc_curl_opts or declaring in the global section of a REST buffer and request blocks.

All specified cUrl options are merged together when a cUrl command is built. For the same keys (cUrl switch) specified at different scopes, the ones of the request blocks overwrite the ones in the global section then overwrite the ones defined by vrc_curl_opts.

For the deprecated VRC options, they can be replaced by cUrl options. For example, assuming they have been defined as follows.

let g:vrc_connect_timeout = 10
let g:vrc_cookie_jar = '/path/to/cookie'
let g:vrc_follow_redirects = 1
let g:vrc_include_response_header = 1
let g:vrc_max_time = 60
let g:vrc_resolve_to_ipv4 = 1
let g:vrc_ssl_secure = 1

Using cUrl options,

let g:vrc_curl_opts = {
  \ '--connect-timeout' : 10,
  \ '-b': '/path/to/cookie',
  \ '-c': '/path/to/cookie',
  \ '-L': '',
  \ '-i': '',
  \ '--max-time': 60,
  \ '--ipv4': '',
  \ '-k': '',
\}

The global section is separated from the rest with two dashes -- and may include a default host, optional default cUrl options (buffer scope) and optional default headers. These values are always included in each request.

Each request block has to start with either two dashes indicating it uses the default host from the global section or any host only used by this block. If a 'local host' is given, it's used instead of the one specified in the global section. Additionally, a request block can specify extra cUrl options and headers. Local headers are merged with and overwrite global headers.

# Global definitions.
// Default host.
https://domain[:port]/...

// Default (buffer scope) cUrl options.
-L
--connect-timeout 10

// Default headers.
Accept: application/json
X-Header: Custom Data
--

# Request block that uses default values from the global section.
--
GET /some/query

# Request block that specifies its own host and extra headers.
// Local host.
http://example.net:9200

// Local cUrl opts.
-k
--ipv4
// This cUrl option overwrites the one in the global section.
--connect-timeout 30
-b /path/to/cookie
-c /path/to/cookie

// Extra headers.
Xtra-Header: Some Extra.
// This header will overwrite the one in the global section.
X-Header: New Data

POST /service
var1=value

VRC now supports variable declarations in the global scope. These variables then can be used in the query paths, headers, and the body. Notice: values are not url-encoded. Variables can contain static text, or reference an environment variable:

# Global scope.
http://host

// Variable declarations (value passed as is).
foobar = LoremIpsum
city = Some%20City
zip = 12345
population = 42
restpassword = $SECRET_ENV_VAR
--
# End global scope.

--
GET /city/:city

--
GET /city/:city/zip/:zip

--
custom-header :foobar
POST /city/:city
{ "population": :population }

Since version 2.3.0, the request body can be specified on a line-by-line basis. It's useful for name-value pair services. Each line of the request body is passed to cURL using --data or --data-urlencode depending on the verb.

To enable,

let g:vrc_split_request_body = 1

or

let b:vrc_split_request_body = 1

Then the request body can be specified as

#
# The following params in the request body will be
# sent using `--data-urlencode`
#
http://localhost
Content-Type: text/html; charset=UTF-8
GET /service
var1=value1
var2=value2

This option won't take effect for GET request if the option vrc_allow_get_request_body is set.

A request block may have consecutive request verbs. The output of each request verb is appended to the output view.

http://localhost:9200
PUT /test
GET /test
DELETE /test

Response handlers are used to parse response and do one of following actions:

• substitute headers or global variables with values from response
• send parse result to quickfix list
• provide assertions which can abort execution of following requests in same block

Response handler is effectively a comment of following format:

#{prefix}{target}<<{pattern}

where {prefix} defines the scope of handler (see below), {target} defines what to do with result, and {pattern} defines how to parse response to get the result.

There are three types of handler depending on {target}:

1. when {target} is non-empty string it is considered a substitution handler. It replaces remaining part of the line with {pattern} output. e.g. when {target} is "Authorization: " then lines which defines Authorization header will be replaced with respective value extracted from response.

2. when {target} is empty string it is considered trace handler. It outputs all pattern results to quickfix list.

3. when {target} is ? string it is considered assertion handler. In this case output of {pattern} is ignored but response code is considered. When response code is false (i.e. non-0) then it terminates execution of current block. See examples below.

{pattern} is any shell command which consumes response and produces some output used by {target}.

Handler can have one of following scopes depending on {prefix} value:

1. > - local scope. This handler consumes output of current request and makes substitutions (when it is substitution handler) only in current block and in global section. These handlers executed automatically right after request where they are defined. For this handler if {target} is ? and {pattern} exit code is non-0 then it terminates current block execution.
2. % - global scope. This handler consumes whole content of VRC output buffer and makes substitutions globally. It is ignored during usual request launch. These handlers can be invoked with :call VrcHandleResponse() for current line or selected region.

Examples

Local-scoped subtitution handler used to set request header value:

http://somehost
Authorization: Bearer <TBD>
--
--
POST /token
#>Authorization: Bearer << jq -r '.access_token'
GET /protected_resource

here we have Authorization header defined in global scope. First request returns some json with access_token field which can be used as bearer token for call which requires auth. This handler executed right after first request, extracts access token and put it in defined header. Second request will be executed with respective Authorization header.

Local-scoped subtitution handler used to set global variable value:

http://somehost
userId=
--
--
GET /user
#>userId=<< jq -r '.id'
PUT /user/:userId
{
.....
}

in this example, first we GET "user" resource. Handler extracts it's id from response and place it in userId global variable. In second request we use this userId to update resource.

Local-scoped assertion handler:

--
-i
GET /user
#>?<< grep "HTTP/" | grep "404"
POST /user/
{
....
}

in this example second request will be executed ONLY when first responds with 404.

Global-scoped handler, used to extract response codes of all requests:

--
-i
GET /user/1
GET /user/2
GET /user/3
#%<< grep "HTTP/"

this handler won't be executed automatically. Put cursor on it (or visually select line) and :call VrcHandleResponse() . It will show in quickfix list something like:

|144| HTTP/2 200
|144| HTTP/2 200
|144| HTTP/2 404

i.e. first and second resource exist. Last one - not. This types of handlers can be used to generate some kind of report after execution of series of requests.

Note: this handler uses as input whole content of VRC output buffer.

You can automate it even more. For example you need to quickly ensure that all requests from your batch finished with success (e.g. 200), then you can do:

--
-i
GET /user/1
GET /user/2
GET /user/3
#%?<< [[ -z $(grep "HTTP/" | grep -v "200") ]]

it will produce in quickfix list:

|145| True

if all requests respond with 200, or False otherwise. You can visually select several handlers and :call VrcHandleResponse() for region. They will be executed line by line. Output will be appended to quickfix list.

https://github.com/Aadniz/vim-rest-console/blob/master/doc/vim-rest-console.txt

Since v3.0, VRC supports POSTing data in bulk using in-line data or an external data file. It's helpful for such APIs as Elasticsearch's Bulk API.

To use in-line data, first enable the Elasticsearch support flag.

let g:vrc_elasticsearch_support = 1

The request would look like this.

http://localhost:9200
POST /testindex/_bulk
{ "index": { "_index": "test", "_type": "product" } }
{ "sku": "SKU1", "name": "Product name 1" }
{ "index": { "_index": "test", "_type": "product" } }
{ "sku": "SKU2", "name": "Product name 2" }

Using external data files doesn't need the support flag.

http://localhost:9200
POST /testindex/_bulk
@data.sample.json

You can also PUT contents of a file using the same syntax. This is equivalent to passing --data-binary flag to cURL.

http://localhost:9200
PUT /testindex/_bulk
@data.sample.json

Though VRC supports output syntax highlighting, it's based on the response Content-Type. When Content-Type is not present, the output can still be syntax-highlighted if the appropriate ftplugin is installed. To force the output highlighting based on filetype, place this setting in .vimrc:

let g:vrc_output_buffer_name = '__VRC_OUTPUT.<filetype>'

filetype can also be set in the output buffer on an ad hoc basis.

# vim: set ft=json

Thanks to the contributors (in alphabetical order of GitHub account)

@dan-silva
@dflupu
@diepm
@emaniacs
@gagara
@iamFIREcracker
@jojoyuji
@korin
@minhajuddin
@mjakl
@nathanaelkane
@nilsboy
@p1otr
@rawaludin
@rlisowski
@sethtrain
@shanesmith
@tdroxler
@tonyskn
@torbjornvatn

MIT