This commit is contained in:
Rashad Gover 2023-04-21 08:58:33 +00:00
parent 3220a29759
commit 05c490d5a4
2 changed files with 114 additions and 114 deletions

View File

@ -13,9 +13,9 @@ Okapi is a micro framework for implementing HTTP servers.
## Introduction
There are two ways to implement servers in Okapi.
There are two ways to implement *Servers* in Okapi.
The recommended way to implement a server in Okapi is via *Endpoints*:
The recommended way to implement a Server in Okapi is via *Endpoints*:
```haskell
-- | Define Endpoints using an Applicative eDSL
@ -40,7 +40,7 @@ myEndpoint = Endpoint
pure itsOk
```
An alternative, more concise way of defining a server in Okapi is via *Matchpoints*:
An alternative, more concise way of defining a Server in Okapi is via *Matchpoints*:
```haskell
-- | Define Matchpoint patterns using PatternSynonyms,
@ -85,18 +85,17 @@ main = Warp.run 3000 $ instantiate id myServer
```
The advantadge of using Endpoints over Matchpoints is that Okapi can
automatically generate specifications and clients for a server implemented
with Endpoints, but not a server implemented with Matchpoints.
automatically generate *Specifications* and *Clients* for a Server implemented
with Endpoints, but not a Server implemented with Matchpoints.
On the flip side, a server implemented with Matchpoints will be more
concise than the same server implemented with Endpoints.
On the flip side, a Server implemented with Matchpoints will be more
concise than the same Server implemented with Endpoints.
## Endpoint
An *Endpoint* is an *executable specification* that represents a single endpoint of
your API.
An Endpoint is an *executable specification* representing a single Operation that can be taken against your API.
An Endpoint has 6 fields:
An Endpoint has 6 fields.
```haskell
data Endpoint p q h b r = Endpoint
@ -109,16 +108,16 @@ data Endpoint p q h b r = Endpoint
}
```
The `method` field is a simple value, but the other fields point to a `Script` that represents their respective HTTP request parts.
The `method` field is a simple value, but the other fields point to what's called a *Script*. Scripts represent Okapi's DSL for extracting and parsing data from Requests. There's a specific type of Script for each part of a Request.
The type parameter of a Script represents the type of value it returns.
Therefore, the concrete type of an Endpoint is determined by the return types of
the Scripts that are used to construct the Endpoint.
All the different Script types are Applicatives, but not all are *lawful applicatives*.
All the different Script types are Applicatives, but not all are *lawful Applicatives*.
Since all Script types are Applicatives, we can use the Applicative typeclass methods to write our scripts. Here's an example of a query Script:
Since all Script types are Applicatives, we can use the Applicative typeclass methods to write our Scripts. Here's an example of a Query Script.
```haskell
data Filter = Filter
@ -136,7 +135,7 @@ myQueryScript = Filter
If you have the `-XApplicativeDo` language extension turned on, you can also write your Scripts using `do` syntax.
We recommend using `-XApplicativeDo` in conjuction with the `-XRecordWildCards` language extension if you're not comfortable with using the Applicative operators. Here's the same query script we defined above, but with these language extensions turned on:
We recommend using `-XApplicativeDo` in conjuction with the `-XRecordWildCards` language extension if you're not comfortable with using the Applicative operators. Here's the same Query Script we defined above, but with these language extensions turned on.
```haskell
{-# LANGUAGE ApplicativeDo #-}
@ -150,20 +149,20 @@ myQueryScript = do
pure Filter {..}
```
Each Script type has its' own operations suited to parsing its' respective part of the
request. These are defined in more detail below.
Each Script type has its own operations suited to parsing its respective part of the
Request. These operations are covered in more detail below.
1. #### Method
The `method` field represents the HTTP method that the Endpoint accepts.
The `method` field represents the HTTP Method that the Endpoint accepts.
Its' type is `StdMethod` from the `http-types` library.
Its type is `StdMethod` from the `http-types` library.
Only the standard methods mentioned in RFC-1234 are allowed.
2. #### Path Script
The `pathScript` field defines the request path that the Endpoint accepts, including *path parameters*.
The `pathScript` field defines the Request Path that the Endpoint accepts, including Path parameters.
Path Scripts have two operations: `static` and `param`.
@ -173,7 +172,7 @@ request. These are defined in more detail below.
3. #### Query Script
The `queryScript` field defines the query that the Endpoint accepts.
The `queryScript` field defines the Query that the Endpoint accepts.
There are two operations for Query Scripts: `param` and `flag`.
@ -193,7 +192,7 @@ request. These are defined in more detail below.
4. #### Body Script
The `bodyScript` field defines the request body and it's content type that the Endpoint accepts.
The `bodyScript` field defines the Request Body that the Endpoint accepts.
There are four operations for Body Scripts: `json`, `form`, `param`, `file`.
@ -206,7 +205,7 @@ request. These are defined in more detail below.
5. #### Headers Script
The `headersScript` field defines the request headers that the Endpoint accepts.
The `headersScript` field defines the Request Headers that the Endpoint accepts.
There are two operations for Headers Scripts: `param` and `cookie`.
@ -219,17 +218,17 @@ request. These are defined in more detail below.
6. #### Responder Script
The `responderScript` field defines the responses that the Endpoint's handler MUST return.
The `responderScript` field defines the Responses that the Endpoint's handler MUST return.
Responder Scripts have to be more complex than the other Script types in order for the Endpoint to have a contract with the handler that the handler will respond with the responses defined in the Responder Script.
Responder Scripts have to be more complex than the other Script types in order for the Endpoint to have a contract with its Handler. The contract ensures that the Handler will respond with the Responses defined in the Responder Script.
This is done using a combination of higher order functions, linear types, and smart constructors.
Responder Script operations have to take an *Add Header* Script as an argument to define what headers will be attached to the Response.
Responder Script operations have to take an *Add Header Script* as an argument to define what Headers will be attached to the Response.
For now, there is only one operation for Responder Scripts: `json`.
Add Header Scripts also only have one operation: `using`.
Add Header Scripts only have one operation as well: `using`.
```haskell
{-# LANGUAGE ApplicativeDo #-}
@ -264,13 +263,13 @@ request. These are defined in more detail below.
"Not Good!"
```
More information about *Responders* and *AddHeader* is available in the Handler section.
More information about *Responder* and *AddHeader* are available in the Handler section.
### Handler
Handlers are simple: they are contextful functions from the arguments provided by an Endpoint, to a Response.
The type synonym `Handler` represents the type of Handlers in Okapi:
The type synonym `Handler` represents the type of these contextful functions.
```haskell
type Handler m p q b h r = p -> q -> b -> h -> r -> m Response
@ -282,7 +281,7 @@ The type parameters `p`, `q`, `b`, `h` and `r` represent the types of the values
### Plan
A Plan is how your Endpoint and a compatible Handler come together.
A Plan is how your Endpoint and its designated Handler come together.
```haskell
data Plan m p q h b r = Plan
@ -292,7 +291,7 @@ data Plan m p q h b r = Plan
}
```
The `transformer` field represents a natural transformation from your handler's Monad `m` to `IO`. This is where you handle how you're custom effects are interpreted in an `IO` context.
The `transformer` field represents a *natural transformation* from your Handler's Monad `m`, to `IO`. This is where you decide how you're custom effects are interpreted in an `IO` context.
The `endpoint` field represents your Endpoint.
@ -347,14 +346,14 @@ data Server = Server
}
```
The `info` field represents your Server's metadata. It's used in the generation of the Server's specification. It's optional.
The `info` field represents your Server's metadata. It's used in the generation of the Server's Specification. It's optional.
The `artifacts` field is a list of Artifact. A single Artifact is generated from a single Plan, which contains a `transformer`, an Endpoint and a Handler. An Artifact contains two values that are generated from a Plan:
The `artifacts` field is a list of Artifact. A single Artifact is generated from a single Plan. An Artifact contains two values:
1. An IO action that returns a Response. It's only executed if the Endpoint used to generate the IO action matches the Request.
2. An OpenAPI PathItem value based on the structure of the Endpoint used to build the Artifact.
These two values, when combined with the Server's other Artifacts, are used to generate the final Application and OpenAPI Specification respectively.
These two values, when combined with the Server's other Artifacts, are used to generate the final Application and OpenAPI Specification.
```haskell
build ::
@ -383,7 +382,7 @@ myServer = Server
]
```
Notice, the types of the Plans don't have to be the same. The `build` function erases the types and gives us the end product we want. This allows us to mix and match various combinations of Endpoints, Handlers, and Monad transformations in the same Server definition. For example, you can have two handlers that run in two different Monads in the same Server.
Notice, the types of the Plans used to build your Server don't have to be the same. The `build` function erases the types and gives us the end products we need. This allows us to mix and match various combinations of Endpoints, Handlers, and Monad transformations in the same Server definition. For example, you can have two Handlers that operate in two different Monads in the same Server.
Now that you have you your Server, you can use it to:
@ -441,7 +440,7 @@ myServer = Server
#### DRY Endpoints
When implementing an API you will usually need the same path to have multiple methods, each with different parameters in the query, body and headers. Since Endpoints are Records, this is easy to deal with. Let's say we have a typical `/users/{userID : UserID}` route that accepts GET and PUT requests for fetching and updating a specific user respectively. The GET variant doesn't need a Body, but the PUT variant will.
When implementing an API you will usually need the same path to have multiple methods, each with different parameters in the query, body and headers. Since Endpoints are records this is easy to deal with. Let's say we have a typical `/users/{userID : UserID}` route that accepts GET and PUT requests for fetching and updating a specific user respectively. The GET variant doesn't need a Body, but the PUT variant will.
```haskell
getUser = Endpoint
@ -495,7 +494,7 @@ COMING SOON
## Matchpoint
A *Matchpoint* is a *pattern* that matches on `Request` values.
A Matchpoint is a *pattern* that matches on Request values.
```haskell
pattern Matchpoint :: Request -> Matchpoint
@ -520,7 +519,7 @@ The `GetUserByID` pattern defined above would match against any Request of the f
`PathParam` is a pattern synonym that you can use in your Matchpoints to match against path parameters of any type that are instances of both `ToHttpApiData` and `FromHttpApiData`. This is required since `PathParam` is a *bidirectional pattern synonym*. This property of `PathParam` makes it useful for generating URLs.
If our matching logic is more complicated, pattern synonyms alone may not be enough. For more complicated routes, we can use Okapi's DSL inside our Matchpoints by using `-XViewPatterns`. As an example, let's reimplement the first Endpoint on this page as a Matchpoint. Here's the Endpoint version first:
If your matching logic is more complicated, pattern synonyms alone may not be enough. For more complicated routes, we can use Okapi's DSL inside our Matchpoints by using `-XViewPatterns`. As an example, let's reimplement the first Endpoint on this page as a Matchpoint. Here's the Endpoint version first.
```haskell
-- | Define Endpoints using an Applicative eDSL
@ -545,7 +544,7 @@ myEndpoint = Endpoint
pure itsOk
```
Here's the equivalent Matchpoint version:
Here's the equivalent Matchpoint version.
```haskell
-- | Define Matchpoints using the same DSL
@ -562,7 +561,7 @@ xyQuery = do
pure (x, y)
```
We can simplify `MyMatchpoint` further by using more pattern synonyms:
We can simplify `MyMatchpoint` further by using more pattern synonyms.
```haskell
pattern MyMatchpoint n pair bar <- Matchpoint
@ -586,7 +585,7 @@ xyQuery = do
Pattern synonyms like `MagicNumber` or `XYQuery` in the example above come in handy when we need to use the same pattern inside multiple Matchpoints.
You would use the Matchpoint we defined above like so (using `-XLambdaCase`):
You can use the Matchpoint we defined above in a case statement with other Matchpoints to define a Server.
```haskell
type Server m = Request -> m Response
@ -611,13 +610,13 @@ Notice, the Server type for Matchpoints is much simpler than the Server type for
We recommend using Endpoints. Matchpoints are great if you're not worried about safety and just want to get something up and running quickly. Here are some downsides to using Matchpoints to implement your Server:
1. Can't do any static analysis. This means you can't generate OpenAPI specifications for Servers that use Matchpoints or perform any optimizations. You can perform static analysis on the Scripts that you use in your Matchpoints, if there are any.
1. We can't perform any static analysis on them. This means you can't generate OpenAPI specifications for Matchpoint Servers or perform any optimizations on them. You can perform static analysis on the Scripts that you use in your Matchpoints, if there are any.
2. All Handlers in a Matchpoint Server must operate within the same context. For Endpoints, this is not the case.
3. Endpoints are more modular. You can achieve some level of moduarity with your Matchpoints by using nested `-XPatternSynonyms` though.
4. Matchpoint Servers have no knowledge of what Responses you will return to the Client. Endpoint Servers know every possible Response you may return from your Handlers, besides ones caused by `IO` errors (the goal is for Endpoints to know about these as well).
4. Matchpoint Servers have no knowledge of what Responses you will return to the Client. Endpoint Servers know every possible Response you may return from your Handlers, besides the ones returned by `IO` errors (the goal is for Endpoints to know about these as well).
5. Requires knowledge of the `-XPatternSynonyms` and `-XViewPatterns` language extensions.

View File

@ -36,8 +36,8 @@ create a new Haskell project</li>
dependencies</li>
</ol>
<h2 id="introduction">Introduction</h2>
<p>There are two ways to implement servers in Okapi.</p>
<p>The recommended way to implement a server in Okapi is via
<p>There are two ways to implement <em>Servers</em> in Okapi.</p>
<p>The recommended way to implement a Server in Okapi is via
<em>Endpoints</em>:</p>
<div class="sourceCode" id="cb1"><pre
class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb1-1"><a href="#cb1-1" aria-hidden="true" tabindex="-1"></a><span class="co">-- | Define Endpoints using an Applicative eDSL</span></span>
@ -60,7 +60,7 @@ class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb1-1"><a
<span id="cb1-18"><a href="#cb1-18" aria-hidden="true" tabindex="-1"></a> addSecretNumber <span class="ot">&lt;-</span> AddHeader.using <span class="op">@</span><span class="dt">Int</span> <span class="st">&quot;X-SECRET&quot;</span></span>
<span id="cb1-19"><a href="#cb1-19" aria-hidden="true" tabindex="-1"></a> <span class="fu">pure</span> addSecretNumber</span>
<span id="cb1-20"><a href="#cb1-20" aria-hidden="true" tabindex="-1"></a> <span class="fu">pure</span> itsOk</span></code></pre></div>
<p>An alternative, more concise way of defining a server in Okapi is via
<p>An alternative, more concise way of defining a Server in Okapi is via
<em>Matchpoints</em>:</p>
<div class="sourceCode" id="cb2"><pre
class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb2-1"><a href="#cb2-1" aria-hidden="true" tabindex="-1"></a><span class="co">-- | Define Matchpoint patterns using PatternSynonyms,</span></span>
@ -103,15 +103,15 @@ class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb2-1"><a
<span id="cb2-38"><a href="#cb2-38" aria-hidden="true" tabindex="-1"></a><span class="co">-- | Run your Server using Warp</span></span>
<span id="cb2-39"><a href="#cb2-39" aria-hidden="true" tabindex="-1"></a>main <span class="ot">=</span> Warp.run <span class="dv">3000</span> <span class="op">$</span> instantiate <span class="fu">id</span> myServer</span></code></pre></div>
<p>The advantadge of using Endpoints over Matchpoints is that Okapi can
automatically generate specifications and clients for a server
implemented with Endpoints, but not a server implemented with
automatically generate <em>Specifications</em> and <em>Clients</em> for
a Server implemented with Endpoints, but not a Server implemented with
Matchpoints.</p>
<p>On the flip side, a server implemented with Matchpoints will be more
concise than the same server implemented with Endpoints.</p>
<p>On the flip side, a Server implemented with Matchpoints will be more
concise than the same Server implemented with Endpoints.</p>
<h2 id="endpoint">Endpoint</h2>
<p>An <em>Endpoint</em> is an <em>executable specification</em> that
represents a single endpoint of your API.</p>
<p>An Endpoint has 6 fields:</p>
<p>An Endpoint is an <em>executable specification</em> representing a
single Operation that can be taken against your API.</p>
<p>An Endpoint has 6 fields.</p>
<div class="sourceCode" id="cb3"><pre
class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb3-1"><a href="#cb3-1" aria-hidden="true" tabindex="-1"></a><span class="kw">data</span> <span class="dt">Endpoint</span> p q h b r <span class="ot">=</span> <span class="dt">Endpoint</span></span>
<span id="cb3-2"><a href="#cb3-2" aria-hidden="true" tabindex="-1"></a> {<span class="ot"> method ::</span> <span class="dt">StdMethod</span>, <span class="co">-- (1)</span></span>
@ -122,17 +122,18 @@ class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb3-1"><a
<span id="cb3-7"><a href="#cb3-7" aria-hidden="true" tabindex="-1"></a><span class="ot"> responderScript ::</span> <span class="dt">Responder.Script</span> r <span class="co">-- (6)</span></span>
<span id="cb3-8"><a href="#cb3-8" aria-hidden="true" tabindex="-1"></a> }</span></code></pre></div>
<p>The <code>method</code> field is a simple value, but the other fields
point to a <code>Script</code> that represents their respective HTTP
request parts.</p>
point to what's called a <em>Script</em>. Scripts represent Okapi's DSL
for extracting and parsing data from Requests. There's a specific type
of Script for each part of a Request.</p>
<p>The type parameter of a Script represents the type of value it
returns.</p>
<p>Therefore, the concrete type of an Endpoint is determined by the
return types of the Scripts that are used to construct the Endpoint.</p>
<p>All the different Script types are Applicatives, but not all are
<em>lawful applicatives</em>.</p>
<em>lawful Applicatives</em>.</p>
<p>Since all Script types are Applicatives, we can use the Applicative
typeclass methods to write our scripts. Here's an example of a query
Script:</p>
typeclass methods to write our Scripts. Here's an example of a Query
Script.</p>
<div class="sourceCode" id="cb4"><pre
class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb4-1"><a href="#cb4-1" aria-hidden="true" tabindex="-1"></a><span class="kw">data</span> <span class="dt">Filter</span> <span class="ot">=</span> <span class="dt">Filter</span></span>
<span id="cb4-2"><a href="#cb4-2" aria-hidden="true" tabindex="-1"></a> {<span class="ot"> color ::</span> <span class="dt">Text</span></span>
@ -150,9 +151,9 @@ turned on, you can also write your Scripts using <code>do</code>
syntax.</p>
<p>We recommend using <code>-XApplicativeDo</code> in conjuction with
the <code>-XRecordWildCards</code> language extension if you're not
comfortable with using the Applicative operators. Here's the same query
script we defined above, but with these language extensions turned
on:</p>
comfortable with using the Applicative operators. Here's the same Query
Script we defined above, but with these language extensions turned
on.</p>
<div class="sourceCode" id="cb5"><pre
class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb5-1"><a href="#cb5-1" aria-hidden="true" tabindex="-1"></a><span class="ot">{-# LANGUAGE ApplicativeDo #-}</span></span>
<span id="cb5-2"><a href="#cb5-2" aria-hidden="true" tabindex="-1"></a><span class="ot">{-# LANGUAGE RecordWildCards #-}</span></span>
@ -163,25 +164,25 @@ class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb5-1"><a
<span id="cb5-7"><a href="#cb5-7" aria-hidden="true" tabindex="-1"></a> categoryID <span class="ot">&lt;-</span> Query.param <span class="st">&quot;category&quot;</span></span>
<span id="cb5-8"><a href="#cb5-8" aria-hidden="true" tabindex="-1"></a> isOnSale <span class="ot">&lt;-</span> Query.optional <span class="op">$</span> Query.flag <span class="st">&quot;sale&quot;</span></span>
<span id="cb5-9"><a href="#cb5-9" aria-hidden="true" tabindex="-1"></a> <span class="fu">pure</span> <span class="dt">Filter</span> {<span class="op">..</span>}</span></code></pre></div>
<p>Each Script type has its' own operations suited to parsing its'
respective part of the request. These are defined in more detail
below.</p>
<p>Each Script type has its own operations suited to parsing its
respective part of the Request. These operations are covered in more
detail below.</p>
<ol type="1">
<li><h4>Method</h4>
<p>The <code>method</code> field represents the HTTP method that the
<p>The <code>method</code> field represents the HTTP Method that the
Endpoint accepts.</p>
<p>Its' type is <code>StdMethod</code> from the <code>http-types</code>
<p>Its type is <code>StdMethod</code> from the <code>http-types</code>
library.</p>
<p>Only the standard methods mentioned in RFC-1234 are allowed.</p></li>
<li><h4>Path Script</h4>
<p>The <code>pathScript</code> field defines the request path that the
Endpoint accepts, including <em>path parameters</em>.</p>
<p>The <code>pathScript</code> field defines the Request Path that the
Endpoint accepts, including Path parameters.</p>
<p>Path Scripts have two operations: <code>static</code> and
<code>param</code>.</p>
<div class="sourceCode" id="cb6"><pre
class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb6-1"><a href="#cb6-1" aria-hidden="true" tabindex="-1"></a>myPathScript <span class="ot">=</span> Path.static <span class="st">&quot;person&quot;</span> <span class="op">*&gt;</span> Path.param <span class="op">@</span><span class="dt">Int</span> <span class="st">&quot;personID&quot;</span></span></code></pre></div></li>
<li><h4>Query Script</h4>
<p>The <code>queryScript</code> field defines the query that the
<p>The <code>queryScript</code> field defines the Query that the
Endpoint accepts.</p>
<p>There are two operations for Query Scripts: <code>param</code> and
<code>flag</code>.</p>
@ -198,8 +199,8 @@ class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb7-1"><a
<span id="cb7-5"><a href="#cb7-5" aria-hidden="true" tabindex="-1"></a> minAge <span class="ot">&lt;-</span> Query.option <span class="dv">21</span> <span class="op">$</span> Query.param <span class="st">&quot;min_age&quot;</span></span>
<span id="cb7-6"><a href="#cb7-6" aria-hidden="true" tabindex="-1"></a> <span class="fu">pure</span> (lastName, optSalary, minAge)</span></code></pre></div></li>
<li><h4>Body Script</h4>
<p>The <code>bodyScript</code> field defines the request body and it's
content type that the Endpoint accepts.</p>
<p>The <code>bodyScript</code> field defines the Request Body that the
Endpoint accepts.</p>
<p>There are four operations for Body Scripts: <code>json</code>,
<code>form</code>, <code>param</code>, <code>file</code>.</p>
<p>There are two modifiers for Body Scripts: <code>optional</code> and
@ -208,7 +209,7 @@ content type that the Endpoint accepts.</p>
class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb8-1"><a href="#cb8-1" aria-hidden="true" tabindex="-1"></a><span class="ot">myBodyScript ::</span> <span class="dt">Body.Script</span> (<span class="dt">Maybe</span> <span class="dt">Value</span>)</span>
<span id="cb8-2"><a href="#cb8-2" aria-hidden="true" tabindex="-1"></a>myBodyScript <span class="ot">=</span> Body.optional <span class="op">$</span> Body.json <span class="op">@</span><span class="dt">Value</span></span></code></pre></div></li>
<li><h4>Headers Script</h4>
<p>The <code>headersScript</code> field defines the request headers that
<p>The <code>headersScript</code> field defines the Request Headers that
the Endpoint accepts.</p>
<p>There are two operations for Headers Scripts: <code>param</code> and
<code>cookie</code>.</p>
@ -218,20 +219,20 @@ and <code>option</code>.</p>
class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb9-1"><a href="#cb9-1" aria-hidden="true" tabindex="-1"></a><span class="ot">myHeadersScript ::</span> <span class="dt">Headers.Script</span> _</span>
<span id="cb9-2"><a href="#cb9-2" aria-hidden="true" tabindex="-1"></a>myHeadersScript <span class="ot">=</span> <span class="fu">undefined</span></span></code></pre></div></li>
<li><h4>Responder Script</h4>
<p>The <code>responderScript</code> field defines the responses that the
<p>The <code>responderScript</code> field defines the Responses that the
Endpoint's handler MUST return.</p>
<p>Responder Scripts have to be more complex than the other Script types
in order for the Endpoint to have a contract with the handler that the
handler will respond with the responses defined in the Responder
Script.</p>
in order for the Endpoint to have a contract with its Handler. The
contract ensures that the Handler will respond with the Responses
defined in the Responder Script.</p>
<p>This is done using a combination of higher order functions, linear
types, and smart constructors.</p>
<p>Responder Script operations have to take an <em>Add Header</em>
Script as an argument to define what headers will be attached to the
Response.</p>
<p>Responder Script operations have to take an <em>Add Header
Script</em> as an argument to define what Headers will be attached to
the Response.</p>
<p>For now, there is only one operation for Responder Scripts:
<code>json</code>.</p>
<p>Add Header Scripts also only have one operation:
<p>Add Header Scripts only have one operation as well:
<code>using</code>.</p>
<div class="sourceCode" id="cb10"><pre
class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb10-1"><a href="#cb10-1" aria-hidden="true" tabindex="-1"></a><span class="ot">{-# LANGUAGE ApplicativeDo #-}</span></span>
@ -264,14 +265,14 @@ class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb10-1"><a
<span id="cb10-28"><a href="#cb10-28" aria-hidden="true" tabindex="-1"></a> <span class="kw">else</span> <span class="fu">return</span> <span class="op">$</span> notGood</span>
<span id="cb10-29"><a href="#cb10-29" aria-hidden="true" tabindex="-1"></a> (\() response <span class="ot">-&gt;</span> response)</span>
<span id="cb10-30"><a href="#cb10-30" aria-hidden="true" tabindex="-1"></a> <span class="st">&quot;Not Good!&quot;</span></span></code></pre></div>
<p>More information about <em>Responders</em> and <em>AddHeader</em> is
<p>More information about <em>Responder</em> and <em>AddHeader</em> are
available in the Handler section.</p></li>
</ol>
<h3>Handler</h3>
<p>Handlers are simple: they are contextful functions from the arguments
provided by an Endpoint, to a Response.</p>
<p>The type synonym <code>Handler</code> represents the type of Handlers
in Okapi:</p>
<p>The type synonym <code>Handler</code> represents the type of these
contextful functions.</p>
<div class="sourceCode" id="cb11"><pre
class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb11-1"><a href="#cb11-1" aria-hidden="true" tabindex="-1"></a><span class="kw">type</span> <span class="dt">Handler</span> m p q b h r <span class="ot">=</span> p <span class="ot">-&gt;</span> q <span class="ot">-&gt;</span> b <span class="ot">-&gt;</span> h <span class="ot">-&gt;</span> r <span class="ot">-&gt;</span> m <span class="dt">Response</span></span></code></pre></div>
<p>The type parameter <code>m</code> represents the context in which the
@ -281,7 +282,7 @@ Handler creates the Response.</p>
returned by the Endpoint's Path, Query, Body, Headers and Responder
Scripts respectively.</p>
<h3>Plan</h3>
<p>A Plan is how your Endpoint and a compatible Handler come
<p>A Plan is how your Endpoint and its designated Handler come
together.</p>
<div class="sourceCode" id="cb12"><pre
class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb12-1"><a href="#cb12-1" aria-hidden="true" tabindex="-1"></a><span class="kw">data</span> <span class="dt">Plan</span> m p q h b r <span class="ot">=</span> <span class="dt">Plan</span></span>
@ -289,9 +290,9 @@ class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb12-1"><a
<span id="cb12-3"><a href="#cb12-3" aria-hidden="true" tabindex="-1"></a><span class="ot"> endpoint ::</span> <span class="dt">Endpoint</span> p q h b r,</span>
<span id="cb12-4"><a href="#cb12-4" aria-hidden="true" tabindex="-1"></a><span class="ot"> handler ::</span> <span class="dt">Monad</span> m <span class="ot">=&gt;</span> p <span class="ot">-&gt;</span> q <span class="ot">-&gt;</span> b <span class="ot">-&gt;</span> h <span class="ot">-&gt;</span> r <span class="ot">-&gt;</span> m <span class="dt">Response</span></span>
<span id="cb12-5"><a href="#cb12-5" aria-hidden="true" tabindex="-1"></a> }</span></code></pre></div>
<p>The <code>transformer</code> field represents a natural
transformation from your handler's Monad <code>m</code> to
<code>IO</code>. This is where you handle how you're custom effects are
<p>The <code>transformer</code> field represents a <em>natural
transformation</em> from your Handler's Monad <code>m</code>, to
<code>IO</code>. This is where you decide how you're custom effects are
interpreted in an <code>IO</code> context.</p>
<p>The <code>endpoint</code> field represents your Endpoint.</p>
<p>The <code>handler</code> field represents your Handler. The types
@ -340,11 +341,10 @@ class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb14-1"><a
<span id="cb14-4"><a href="#cb14-4" aria-hidden="true" tabindex="-1"></a><span class="ot"> artifacts ::</span> [<span class="dt">Artifact</span>]</span>
<span id="cb14-5"><a href="#cb14-5" aria-hidden="true" tabindex="-1"></a> }</span></code></pre></div>
<p>The <code>info</code> field represents your Server's metadata. It's
used in the generation of the Server's specification. It's optional.</p>
used in the generation of the Server's Specification. It's optional.</p>
<p>The <code>artifacts</code> field is a list of Artifact. A single
Artifact is generated from a single Plan, which contains a
<code>transformer</code>, an Endpoint and a Handler. An Artifact
contains two values that are generated from a Plan:</p>
Artifact is generated from a single Plan. An Artifact contains two
values:</p>
<ol type="1">
<li>An IO action that returns a Response. It's only executed if the
Endpoint used to generate the IO action matches the Request.</li>
@ -352,8 +352,8 @@ Endpoint used to generate the IO action matches the Request.</li>
used to build the Artifact.</li>
</ol>
<p>These two values, when combined with the Server's other Artifacts,
are used to generate the final Application and OpenAPI Specification
respectively.</p>
are used to generate the final Application and OpenAPI
Specification.</p>
<div class="sourceCode" id="cb15"><pre
class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb15-1"><a href="#cb15-1" aria-hidden="true" tabindex="-1"></a><span class="ot">build ::</span></span>
<span id="cb15-2"><a href="#cb15-2" aria-hidden="true" tabindex="-1"></a> <span class="kw">forall</span> m p q h b r<span class="op">.</span></span>
@ -379,12 +379,12 @@ class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb15-1"><a
<span id="cb15-22"><a href="#cb15-22" aria-hidden="true" tabindex="-1"></a> , build plan3</span>
<span id="cb15-23"><a href="#cb15-23" aria-hidden="true" tabindex="-1"></a> , <span class="op">...</span></span>
<span id="cb15-24"><a href="#cb15-24" aria-hidden="true" tabindex="-1"></a> ]</span></code></pre></div>
<p>Notice, the types of the Plans don't have to be the same. The
<code>build</code> function erases the types and gives us the end
product we want. This allows us to mix and match various combinations of
Endpoints, Handlers, and Monad transformations in the same Server
definition. For example, you can have two handlers that run in two
different Monads in the same Server.</p>
<p>Notice, the types of the Plans used to build your Server don't have
to be the same. The <code>build</code> function erases the types and
gives us the end products we need. This allows us to mix and match
various combinations of Endpoints, Handlers, and Monad transformations
in the same Server definition. For example, you can have two Handlers
that operate in two different Monads in the same Server.</p>
<p>Now that you have you your Server, you can use it to:</p>
<ol type="1">
<li>Generate a WAI Application</li>
@ -435,7 +435,7 @@ class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb18-1"><a
<h4>DRY Endpoints</h4>
<p>When implementing an API you will usually need the same path to have
multiple methods, each with different parameters in the query, body and
headers. Since Endpoints are Records, this is easy to deal with. Let's
headers. Since Endpoints are records this is easy to deal with. Let's
say we have a typical <code>/users/{userID : UserID}</code> route that
accepts GET and PUT requests for fetching and updating a specific user
respectively. The GET variant doesn't need a Body, but the PUT variant
@ -488,8 +488,8 @@ for those who aren't familiar with the
<h4>Idiom Brackets?</h4>
<p>COMING SOON</p>
<h2 id="matchpoint">Matchpoint</h2>
<p>A <em>Matchpoint</em> is a <em>pattern</em> that matches on
<code>Request</code> values.</p>
<p>A Matchpoint is a <em>pattern</em> that matches on Request
values.</p>
<div class="sourceCode" id="cb21"><pre
class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb21-1"><a href="#cb21-1" aria-hidden="true" tabindex="-1"></a><span class="kw">pattern</span> <span class="dt">Matchpoint</span><span class="ot"> ::</span> <span class="dt">Request</span> <span class="ot">-&gt;</span> <span class="dt">Matchpoint</span></span></code></pre></div>
<p>You can use the Matchpoint pattern synonym to create your own pattern
@ -519,11 +519,11 @@ instances of both <code>ToHttpApiData</code> and
<code>PathParam</code> is a <em>bidirectional pattern synonym</em>. This
property of <code>PathParam</code> makes it useful for generating
URLs.</p>
<p>If our matching logic is more complicated, pattern synonyms alone may
not be enough. For more complicated routes, we can use Okapi's DSL
<p>If your matching logic is more complicated, pattern synonyms alone
may not be enough. For more complicated routes, we can use Okapi's DSL
inside our Matchpoints by using <code>-XViewPatterns</code>. As an
example, let's reimplement the first Endpoint on this page as a
Matchpoint. Here's the Endpoint version first:</p>
Matchpoint. Here's the Endpoint version first.</p>
<div class="sourceCode" id="cb23"><pre
class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb23-1"><a href="#cb23-1" aria-hidden="true" tabindex="-1"></a><span class="co">-- | Define Endpoints using an Applicative eDSL</span></span>
<span id="cb23-2"><a href="#cb23-2" aria-hidden="true" tabindex="-1"></a>myEndpoint <span class="ot">=</span> <span class="dt">Endpoint</span></span>
@ -545,7 +545,7 @@ class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb23-1"><a
<span id="cb23-18"><a href="#cb23-18" aria-hidden="true" tabindex="-1"></a> addSecretNumber <span class="ot">&lt;-</span> AddHeader.using <span class="op">@</span><span class="dt">Int</span> <span class="st">&quot;X-SECRET&quot;</span></span>
<span id="cb23-19"><a href="#cb23-19" aria-hidden="true" tabindex="-1"></a> <span class="fu">pure</span> addSecretNumber</span>
<span id="cb23-20"><a href="#cb23-20" aria-hidden="true" tabindex="-1"></a> <span class="fu">pure</span> itsOk</span></code></pre></div>
<p>Here's the equivalent Matchpoint version:</p>
<p>Here's the equivalent Matchpoint version.</p>
<div class="sourceCode" id="cb24"><pre
class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb24-1"><a href="#cb24-1" aria-hidden="true" tabindex="-1"></a><span class="co">-- | Define Matchpoints using the same DSL</span></span>
<span id="cb24-2"><a href="#cb24-2" aria-hidden="true" tabindex="-1"></a><span class="kw">pattern</span> <span class="dt">MyMatchpoint</span> magicNumber pair foo <span class="ot">=</span> <span class="dt">Matchpoint</span></span>
@ -560,7 +560,7 @@ class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb24-1"><a
<span id="cb24-11"><a href="#cb24-11" aria-hidden="true" tabindex="-1"></a> y <span class="ot">&lt;-</span> Query.option <span class="dv">10</span> <span class="op">$</span> Query.param <span class="op">@</span><span class="dt">Int</span> <span class="st">&quot;y&quot;</span></span>
<span id="cb24-12"><a href="#cb24-12" aria-hidden="true" tabindex="-1"></a> <span class="fu">pure</span> (x, y)</span></code></pre></div>
<p>We can simplify <code>MyMatchpoint</code> further by using more
pattern synonyms:</p>
pattern synonyms.</p>
<div class="sourceCode" id="cb25"><pre
class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb25-1"><a href="#cb25-1" aria-hidden="true" tabindex="-1"></a><span class="kw">pattern</span> <span class="dt">MyMatchpoint</span> n pair bar <span class="ot">&lt;-</span> <span class="dt">Matchpoint</span></span>
<span id="cb25-2"><a href="#cb25-2" aria-hidden="true" tabindex="-1"></a> <span class="dt">GET</span></span>
@ -582,8 +582,8 @@ class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb25-1"><a
<p>Pattern synonyms like <code>MagicNumber</code> or
<code>XYQuery</code> in the example above come in handy when we need to
use the same pattern inside multiple Matchpoints.</p>
<p>You would use the Matchpoint we defined above like so (using
<code>-XLambdaCase</code>):</p>
<p>You can use the Matchpoint we defined above in a case statement with
other Matchpoints to define a Server.</p>
<div class="sourceCode" id="cb26"><pre
class="sourceCode haskell"><code class="sourceCode haskell"><span id="cb26-1"><a href="#cb26-1" aria-hidden="true" tabindex="-1"></a><span class="kw">type</span> <span class="dt">Server</span> m <span class="ot">=</span> <span class="dt">Request</span> <span class="ot">-&gt;</span> m <span class="dt">Response</span></span>
<span id="cb26-2"><a href="#cb26-2" aria-hidden="true" tabindex="-1"></a></span>
@ -607,10 +607,10 @@ worried about safety and just want to get something up and running
quickly. Here are some downsides to using Matchpoints to implement your
Server:</p>
<ol type="1">
<li><p>Can't do any static analysis. This means you can't generate
OpenAPI specifications for Servers that use Matchpoints or perform any
optimizations. You can perform static analysis on the Scripts that you
use in your Matchpoints, if there are any.</p></li>
<li><p>We can't perform any static analysis on them. This means you
can't generate OpenAPI specifications for Matchpoint Servers or perform
any optimizations on them. You can perform static analysis on the
Scripts that you use in your Matchpoints, if there are any.</p></li>
<li><p>All Handlers in a Matchpoint Server must operate within the same
context. For Endpoints, this is not the case.</p></li>
<li><p>Endpoints are more modular. You can achieve some level of
@ -618,8 +618,9 @@ moduarity with your Matchpoints by using nested
<code>-XPatternSynonyms</code> though.</p></li>
<li><p>Matchpoint Servers have no knowledge of what Responses you will
return to the Client. Endpoint Servers know every possible Response you
may return from your Handlers, besides ones caused by <code>IO</code>
errors (the goal is for Endpoints to know about these as well).</p></li>
may return from your Handlers, besides the ones returned by
<code>IO</code> errors (the goal is for Endpoints to know about these as
well).</p></li>
<li><p>Requires knowledge of the <code>-XPatternSynonyms</code> and
<code>-XViewPatterns</code> language extensions.</p></li>
</ol>