tvl-depot/web/bubblegum/default.nix

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

273 lines
7.2 KiB
Nix
Raw Normal View History

feat(web/bubblegum): nix CGI programming framework So here is what has been keeping me up at night: At some point I realized that nix actually made a somewhat passable language for CGI programming: * That `builtins.getEnv` exists as one of the impurities of Nix is perfect as environment variables are the main way of communication from the web server to the CGI application. * We can actually read from the filesystem via builtins.readDir and builtins.readFile with bearable overhead if we avoid importing the used paths into the nix store. * Templating and routing are convenient to implement via indented strings and attribute sets respectively. Of course there are obvious limitation: * The overhead of derivations is probably much to great for them to be useful via IfD. * Even without derivations, nix evaluation is very slow to the point were a trivial application takes between 100ms and 400ms to produce a response. * We can't really cause effects other than producing a response which makes it not viable for a lot of applications. There are some ways around this: * With a custom interpreter we could have streaming and multiplexed I/O (using lazy lists emulated via attrsets) to cause such effects, but it would probably perform terribly. * We can use builtins.fetchurl to call other HTTP-based microservices, but only in very limited constraints, i. e. only GET, no headers, and only if the tarball ttl is set to 0 in the global nix.conf. * Terrible error handling capabilities because builtins.tryEval actually doesn't catch a lot of errors. To prove that it actually works, there are some demo applications, which I invite you to run and potentially break horribly: nix-build -A web.bubblegum.examples && ./result # navigate to http://localhost:9000 The setup uses thttpd and executes the nix CGI scripts using users.sterni.nint which automatically passed `depot`, so they can import the cgi library. Change-Id: I3a22a749612211627e5f8301c31ec2e7a872812c Reviewed-on: https://cl.tvl.fyi/c/depot/+/2746 Tested-by: BuildkiteCI Reviewed-by: tazjin <mail@tazj.in>
2021-02-21 12:57:40 +01:00
{ depot, lib, pkgs, ... }:
let
inherit (depot.nix)
runExecline
getBins
utils
sparseTree
nint
feat(web/bubblegum): nix CGI programming framework So here is what has been keeping me up at night: At some point I realized that nix actually made a somewhat passable language for CGI programming: * That `builtins.getEnv` exists as one of the impurities of Nix is perfect as environment variables are the main way of communication from the web server to the CGI application. * We can actually read from the filesystem via builtins.readDir and builtins.readFile with bearable overhead if we avoid importing the used paths into the nix store. * Templating and routing are convenient to implement via indented strings and attribute sets respectively. Of course there are obvious limitation: * The overhead of derivations is probably much to great for them to be useful via IfD. * Even without derivations, nix evaluation is very slow to the point were a trivial application takes between 100ms and 400ms to produce a response. * We can't really cause effects other than producing a response which makes it not viable for a lot of applications. There are some ways around this: * With a custom interpreter we could have streaming and multiplexed I/O (using lazy lists emulated via attrsets) to cause such effects, but it would probably perform terribly. * We can use builtins.fetchurl to call other HTTP-based microservices, but only in very limited constraints, i. e. only GET, no headers, and only if the tarball ttl is set to 0 in the global nix.conf. * Terrible error handling capabilities because builtins.tryEval actually doesn't catch a lot of errors. To prove that it actually works, there are some demo applications, which I invite you to run and potentially break horribly: nix-build -A web.bubblegum.examples && ./result # navigate to http://localhost:9000 The setup uses thttpd and executes the nix CGI scripts using users.sterni.nint which automatically passed `depot`, so they can import the cgi library. Change-Id: I3a22a749612211627e5f8301c31ec2e7a872812c Reviewed-on: https://cl.tvl.fyi/c/depot/+/2746 Tested-by: BuildkiteCI Reviewed-by: tazjin <mail@tazj.in>
2021-02-21 12:57:40 +01:00
;
minimalDepot = sparseTree {
root = depot.path.origSrc;
name = "minimal-depot";
paths = [
# general depot things
"default.nix"
"nix/readTree"
# nixpkgs for lib and packages
"third_party/nixpkgs"
"third_party/overlays"
# bubblegum and its dependencies
"web/bubblegum"
"nix/runExecline"
"nix/utils"
"nix/sparseTree"
# tvix docs for svg demo
"tvix/docs"
# for blog.nix
"users/sterni/nix"
];
};
feat(web/bubblegum): nix CGI programming framework So here is what has been keeping me up at night: At some point I realized that nix actually made a somewhat passable language for CGI programming: * That `builtins.getEnv` exists as one of the impurities of Nix is perfect as environment variables are the main way of communication from the web server to the CGI application. * We can actually read from the filesystem via builtins.readDir and builtins.readFile with bearable overhead if we avoid importing the used paths into the nix store. * Templating and routing are convenient to implement via indented strings and attribute sets respectively. Of course there are obvious limitation: * The overhead of derivations is probably much to great for them to be useful via IfD. * Even without derivations, nix evaluation is very slow to the point were a trivial application takes between 100ms and 400ms to produce a response. * We can't really cause effects other than producing a response which makes it not viable for a lot of applications. There are some ways around this: * With a custom interpreter we could have streaming and multiplexed I/O (using lazy lists emulated via attrsets) to cause such effects, but it would probably perform terribly. * We can use builtins.fetchurl to call other HTTP-based microservices, but only in very limited constraints, i. e. only GET, no headers, and only if the tarball ttl is set to 0 in the global nix.conf. * Terrible error handling capabilities because builtins.tryEval actually doesn't catch a lot of errors. To prove that it actually works, there are some demo applications, which I invite you to run and potentially break horribly: nix-build -A web.bubblegum.examples && ./result # navigate to http://localhost:9000 The setup uses thttpd and executes the nix CGI scripts using users.sterni.nint which automatically passed `depot`, so they can import the cgi library. Change-Id: I3a22a749612211627e5f8301c31ec2e7a872812c Reviewed-on: https://cl.tvl.fyi/c/depot/+/2746 Tested-by: BuildkiteCI Reviewed-by: tazjin <mail@tazj.in>
2021-02-21 12:57:40 +01:00
statusCodes = {
# 1xx
"Continue" = 100;
"Switching Protocols" = 101;
"Processing" = 102;
"Early Hints" = 103;
# 2xx
"OK" = 200;
"Created" = 201;
"Accepted" = 202;
"Non-Authoritative Information" = 203;
"No Content" = 204;
"Reset Content" = 205;
"Partial Content" = 206;
"Multi Status" = 207;
"Already Reported" = 208;
"IM Used" = 226;
# 3xx
"Multiple Choices" = 300;
"Moved Permanently" = 301;
"Found" = 302;
"See Other" = 303;
"Not Modified" = 304;
"Use Proxy" = 305;
"Switch Proxy" = 306;
"Temporary Redirect" = 307;
"Permanent Redirect" = 308;
# 4xx
"Bad Request" = 400;
"Unauthorized" = 401;
"Payment Required" = 402;
"Forbidden" = 403;
"Not Found" = 404;
"Method Not Allowed" = 405;
"Not Acceptable" = 406;
"Proxy Authentication Required" = 407;
"Request Timeout" = 408;
"Conflict" = 409;
"Gone" = 410;
"Length Required" = 411;
"Precondition Failed" = 412;
"Payload Too Large" = 413;
"URI Too Long" = 414;
"Unsupported Media Type" = 415;
"Range Not Satisfiable" = 416;
"Expectation Failed" = 417;
"I'm a teapot" = 418;
"Misdirected Request" = 421;
"Unprocessable Entity" = 422;
"Locked" = 423;
"Failed Dependency" = 424;
"Too Early" = 425;
"Upgrade Required" = 426;
"Precondition Required" = 428;
"Too Many Requests" = 429;
"Request Header Fields Too Large" = 431;
"Unavailable For Legal Reasons" = 451;
# 5xx
"Internal Server Error" = 500;
"Not Implemented" = 501;
"Bad Gateway" = 502;
"Service Unavailable" = 503;
"Gateway Timeout" = 504;
"HTTP Version Not Supported" = 505;
"Variant Also Negotiates" = 506;
"Insufficient Storage" = 507;
"Loop Detected" = 508;
"Not Extended" = 510;
"Network Authentication Required" = 511;
};
/* Generate a CGI response. Takes three arguments:
1. Status of the response as a string which is
the descriptive name in the protocol, e. g.
`"OK"`, `"Not Found"` etc.
2. Attribute set describing extra headers to
send, keys and values should both be strings.
3. Response content as a string.
See the [README](./README.md) for an example.
Type: either int string -> attrs string -> string -> string
feat(web/bubblegum): nix CGI programming framework So here is what has been keeping me up at night: At some point I realized that nix actually made a somewhat passable language for CGI programming: * That `builtins.getEnv` exists as one of the impurities of Nix is perfect as environment variables are the main way of communication from the web server to the CGI application. * We can actually read from the filesystem via builtins.readDir and builtins.readFile with bearable overhead if we avoid importing the used paths into the nix store. * Templating and routing are convenient to implement via indented strings and attribute sets respectively. Of course there are obvious limitation: * The overhead of derivations is probably much to great for them to be useful via IfD. * Even without derivations, nix evaluation is very slow to the point were a trivial application takes between 100ms and 400ms to produce a response. * We can't really cause effects other than producing a response which makes it not viable for a lot of applications. There are some ways around this: * With a custom interpreter we could have streaming and multiplexed I/O (using lazy lists emulated via attrsets) to cause such effects, but it would probably perform terribly. * We can use builtins.fetchurl to call other HTTP-based microservices, but only in very limited constraints, i. e. only GET, no headers, and only if the tarball ttl is set to 0 in the global nix.conf. * Terrible error handling capabilities because builtins.tryEval actually doesn't catch a lot of errors. To prove that it actually works, there are some demo applications, which I invite you to run and potentially break horribly: nix-build -A web.bubblegum.examples && ./result # navigate to http://localhost:9000 The setup uses thttpd and executes the nix CGI scripts using users.sterni.nint which automatically passed `depot`, so they can import the cgi library. Change-Id: I3a22a749612211627e5f8301c31ec2e7a872812c Reviewed-on: https://cl.tvl.fyi/c/depot/+/2746 Tested-by: BuildkiteCI Reviewed-by: tazjin <mail@tazj.in>
2021-02-21 12:57:40 +01:00
*/
respond =
# response status as an integer (status code) or its
# textual representation in the HTTP protocol.
# See `statusCodes` for a list of valid options.
statusArg:
# headers as an attribute set of strings
headers:
# response body as a string
bodyArg:
let
status =
if builtins.isInt statusArg
then {
code = statusArg;
line = lib.findFirst
(line: statusCodes."${line}" == statusArg)
null
(builtins.attrNames statusCodes);
} else if builtins.isString statusArg then {
code = statusCodes."${statusArg}" or null;
line = statusArg;
} else {
code = null;
line = null;
};
renderedHeaders = lib.concatStrings
(lib.mapAttrsToList (n: v: "${n}: ${toString v}\r\n") headers);
internalError = msg: respond 500
{
Content-type = "text/plain";
} "bubblegum error: ${msg}";
body = builtins.tryEval bodyArg;
in
if status.code == null || status.line == null
then internalError "Invalid status ${lib.generators.toPretty {} statusArg}."
else if !body.success
then internalError "Unknown evaluation error in user code"
else
lib.concatStrings [
"Status: ${toString status.code} ${status.line}\r\n"
renderedHeaders
"\r\n"
body.value
];
feat(web/bubblegum): nix CGI programming framework So here is what has been keeping me up at night: At some point I realized that nix actually made a somewhat passable language for CGI programming: * That `builtins.getEnv` exists as one of the impurities of Nix is perfect as environment variables are the main way of communication from the web server to the CGI application. * We can actually read from the filesystem via builtins.readDir and builtins.readFile with bearable overhead if we avoid importing the used paths into the nix store. * Templating and routing are convenient to implement via indented strings and attribute sets respectively. Of course there are obvious limitation: * The overhead of derivations is probably much to great for them to be useful via IfD. * Even without derivations, nix evaluation is very slow to the point were a trivial application takes between 100ms and 400ms to produce a response. * We can't really cause effects other than producing a response which makes it not viable for a lot of applications. There are some ways around this: * With a custom interpreter we could have streaming and multiplexed I/O (using lazy lists emulated via attrsets) to cause such effects, but it would probably perform terribly. * We can use builtins.fetchurl to call other HTTP-based microservices, but only in very limited constraints, i. e. only GET, no headers, and only if the tarball ttl is set to 0 in the global nix.conf. * Terrible error handling capabilities because builtins.tryEval actually doesn't catch a lot of errors. To prove that it actually works, there are some demo applications, which I invite you to run and potentially break horribly: nix-build -A web.bubblegum.examples && ./result # navigate to http://localhost:9000 The setup uses thttpd and executes the nix CGI scripts using users.sterni.nint which automatically passed `depot`, so they can import the cgi library. Change-Id: I3a22a749612211627e5f8301c31ec2e7a872812c Reviewed-on: https://cl.tvl.fyi/c/depot/+/2746 Tested-by: BuildkiteCI Reviewed-by: tazjin <mail@tazj.in>
2021-02-21 12:57:40 +01:00
/* Returns the value of the `SCRIPT_NAME` environment
variable used by CGI.
*/
scriptName = builtins.getEnv "SCRIPT_NAME";
/* Returns the value of the `PATH_INFO` environment
variable used by CGI. All cases that could be
considered as the CGI script's root (i. e.
`PATH_INFO` is empty or `/`) is mapped to `"/"`
for convenience.
*/
pathInfo =
let
p = builtins.getEnv "PATH_INFO";
in
if builtins.stringLength p == 0
then "/"
else p;
/* Helper function which converts a path from the
root of the CGI script (i. e. something which
could be the content of `PATH_INFO`) to an
absolute path from the web root by also
utilizing `scriptName`.
Type: string -> string
*/
absolutePath = path:
if builtins.substring 0 1 path == "/"
then "${scriptName}${path}"
else "${scriptName}/${path}";
feat(web/bubblegum): nix CGI programming framework So here is what has been keeping me up at night: At some point I realized that nix actually made a somewhat passable language for CGI programming: * That `builtins.getEnv` exists as one of the impurities of Nix is perfect as environment variables are the main way of communication from the web server to the CGI application. * We can actually read from the filesystem via builtins.readDir and builtins.readFile with bearable overhead if we avoid importing the used paths into the nix store. * Templating and routing are convenient to implement via indented strings and attribute sets respectively. Of course there are obvious limitation: * The overhead of derivations is probably much to great for them to be useful via IfD. * Even without derivations, nix evaluation is very slow to the point were a trivial application takes between 100ms and 400ms to produce a response. * We can't really cause effects other than producing a response which makes it not viable for a lot of applications. There are some ways around this: * With a custom interpreter we could have streaming and multiplexed I/O (using lazy lists emulated via attrsets) to cause such effects, but it would probably perform terribly. * We can use builtins.fetchurl to call other HTTP-based microservices, but only in very limited constraints, i. e. only GET, no headers, and only if the tarball ttl is set to 0 in the global nix.conf. * Terrible error handling capabilities because builtins.tryEval actually doesn't catch a lot of errors. To prove that it actually works, there are some demo applications, which I invite you to run and potentially break horribly: nix-build -A web.bubblegum.examples && ./result # navigate to http://localhost:9000 The setup uses thttpd and executes the nix CGI scripts using users.sterni.nint which automatically passed `depot`, so they can import the cgi library. Change-Id: I3a22a749612211627e5f8301c31ec2e7a872812c Reviewed-on: https://cl.tvl.fyi/c/depot/+/2746 Tested-by: BuildkiteCI Reviewed-by: tazjin <mail@tazj.in>
2021-02-21 12:57:40 +01:00
bins = getBins pkgs.coreutils [ "env" "tee" "cat" "printf" "chmod" ]
// getBins nint [ "nint" ];
feat(web/bubblegum): nix CGI programming framework So here is what has been keeping me up at night: At some point I realized that nix actually made a somewhat passable language for CGI programming: * That `builtins.getEnv` exists as one of the impurities of Nix is perfect as environment variables are the main way of communication from the web server to the CGI application. * We can actually read from the filesystem via builtins.readDir and builtins.readFile with bearable overhead if we avoid importing the used paths into the nix store. * Templating and routing are convenient to implement via indented strings and attribute sets respectively. Of course there are obvious limitation: * The overhead of derivations is probably much to great for them to be useful via IfD. * Even without derivations, nix evaluation is very slow to the point were a trivial application takes between 100ms and 400ms to produce a response. * We can't really cause effects other than producing a response which makes it not viable for a lot of applications. There are some ways around this: * With a custom interpreter we could have streaming and multiplexed I/O (using lazy lists emulated via attrsets) to cause such effects, but it would probably perform terribly. * We can use builtins.fetchurl to call other HTTP-based microservices, but only in very limited constraints, i. e. only GET, no headers, and only if the tarball ttl is set to 0 in the global nix.conf. * Terrible error handling capabilities because builtins.tryEval actually doesn't catch a lot of errors. To prove that it actually works, there are some demo applications, which I invite you to run and potentially break horribly: nix-build -A web.bubblegum.examples && ./result # navigate to http://localhost:9000 The setup uses thttpd and executes the nix CGI scripts using users.sterni.nint which automatically passed `depot`, so they can import the cgi library. Change-Id: I3a22a749612211627e5f8301c31ec2e7a872812c Reviewed-on: https://cl.tvl.fyi/c/depot/+/2746 Tested-by: BuildkiteCI Reviewed-by: tazjin <mail@tazj.in>
2021-02-21 12:57:40 +01:00
/* Type: args -> either path derivation string -> derivation
*/
writeCGI =
{
# if given sets the `PATH` to search for `nix-instantiate`
# Useful when using for example thttpd which unsets `PATH`
# in the CGI environment.
binPath ? ""
# name of the resulting derivation. Defaults to `baseNameOf`
# the input path or name of the input derivation.
# Must be given if the input is a string.
, name ? null
, ...
}@args:
feat(web/bubblegum): nix CGI programming framework So here is what has been keeping me up at night: At some point I realized that nix actually made a somewhat passable language for CGI programming: * That `builtins.getEnv` exists as one of the impurities of Nix is perfect as environment variables are the main way of communication from the web server to the CGI application. * We can actually read from the filesystem via builtins.readDir and builtins.readFile with bearable overhead if we avoid importing the used paths into the nix store. * Templating and routing are convenient to implement via indented strings and attribute sets respectively. Of course there are obvious limitation: * The overhead of derivations is probably much to great for them to be useful via IfD. * Even without derivations, nix evaluation is very slow to the point were a trivial application takes between 100ms and 400ms to produce a response. * We can't really cause effects other than producing a response which makes it not viable for a lot of applications. There are some ways around this: * With a custom interpreter we could have streaming and multiplexed I/O (using lazy lists emulated via attrsets) to cause such effects, but it would probably perform terribly. * We can use builtins.fetchurl to call other HTTP-based microservices, but only in very limited constraints, i. e. only GET, no headers, and only if the tarball ttl is set to 0 in the global nix.conf. * Terrible error handling capabilities because builtins.tryEval actually doesn't catch a lot of errors. To prove that it actually works, there are some demo applications, which I invite you to run and potentially break horribly: nix-build -A web.bubblegum.examples && ./result # navigate to http://localhost:9000 The setup uses thttpd and executes the nix CGI scripts using users.sterni.nint which automatically passed `depot`, so they can import the cgi library. Change-Id: I3a22a749612211627e5f8301c31ec2e7a872812c Reviewed-on: https://cl.tvl.fyi/c/depot/+/2746 Tested-by: BuildkiteCI Reviewed-by: tazjin <mail@tazj.in>
2021-02-21 12:57:40 +01:00
input:
let
drvName =
if builtins.isString input || args ? name
then args.name
else utils.storePathName input;
feat(web/bubblegum): nix CGI programming framework So here is what has been keeping me up at night: At some point I realized that nix actually made a somewhat passable language for CGI programming: * That `builtins.getEnv` exists as one of the impurities of Nix is perfect as environment variables are the main way of communication from the web server to the CGI application. * We can actually read from the filesystem via builtins.readDir and builtins.readFile with bearable overhead if we avoid importing the used paths into the nix store. * Templating and routing are convenient to implement via indented strings and attribute sets respectively. Of course there are obvious limitation: * The overhead of derivations is probably much to great for them to be useful via IfD. * Even without derivations, nix evaluation is very slow to the point were a trivial application takes between 100ms and 400ms to produce a response. * We can't really cause effects other than producing a response which makes it not viable for a lot of applications. There are some ways around this: * With a custom interpreter we could have streaming and multiplexed I/O (using lazy lists emulated via attrsets) to cause such effects, but it would probably perform terribly. * We can use builtins.fetchurl to call other HTTP-based microservices, but only in very limited constraints, i. e. only GET, no headers, and only if the tarball ttl is set to 0 in the global nix.conf. * Terrible error handling capabilities because builtins.tryEval actually doesn't catch a lot of errors. To prove that it actually works, there are some demo applications, which I invite you to run and potentially break horribly: nix-build -A web.bubblegum.examples && ./result # navigate to http://localhost:9000 The setup uses thttpd and executes the nix CGI scripts using users.sterni.nint which automatically passed `depot`, so they can import the cgi library. Change-Id: I3a22a749612211627e5f8301c31ec2e7a872812c Reviewed-on: https://cl.tvl.fyi/c/depot/+/2746 Tested-by: BuildkiteCI Reviewed-by: tazjin <mail@tazj.in>
2021-02-21 12:57:40 +01:00
script =
if builtins.isPath input || lib.isDerivation input
then input
else if builtins.isString input
then pkgs.writeText "${drvName}-source" input
else builtins.throw "Unsupported input: ${lib.generators.toPretty {} input}";
shebang = lib.concatStringsSep " " ([
"#!${bins.env}"
# use the slightly cursed /usr/bin/env -S which allows us
# to pass any number of arguments to our interpreter
# instead of maximum one using plain shebang which considers
# everything after the first space as the second argument.
"-S"
] ++ lib.optionals (builtins.stringLength binPath > 0) [
"PATH=${binPath}"
] ++ [
"${bins.nint}"
# always pass depot so scripts can use this library
"--arg depot '(import ${minimalDepot} {})'"
feat(web/bubblegum): nix CGI programming framework So here is what has been keeping me up at night: At some point I realized that nix actually made a somewhat passable language for CGI programming: * That `builtins.getEnv` exists as one of the impurities of Nix is perfect as environment variables are the main way of communication from the web server to the CGI application. * We can actually read from the filesystem via builtins.readDir and builtins.readFile with bearable overhead if we avoid importing the used paths into the nix store. * Templating and routing are convenient to implement via indented strings and attribute sets respectively. Of course there are obvious limitation: * The overhead of derivations is probably much to great for them to be useful via IfD. * Even without derivations, nix evaluation is very slow to the point were a trivial application takes between 100ms and 400ms to produce a response. * We can't really cause effects other than producing a response which makes it not viable for a lot of applications. There are some ways around this: * With a custom interpreter we could have streaming and multiplexed I/O (using lazy lists emulated via attrsets) to cause such effects, but it would probably perform terribly. * We can use builtins.fetchurl to call other HTTP-based microservices, but only in very limited constraints, i. e. only GET, no headers, and only if the tarball ttl is set to 0 in the global nix.conf. * Terrible error handling capabilities because builtins.tryEval actually doesn't catch a lot of errors. To prove that it actually works, there are some demo applications, which I invite you to run and potentially break horribly: nix-build -A web.bubblegum.examples && ./result # navigate to http://localhost:9000 The setup uses thttpd and executes the nix CGI scripts using users.sterni.nint which automatically passed `depot`, so they can import the cgi library. Change-Id: I3a22a749612211627e5f8301c31ec2e7a872812c Reviewed-on: https://cl.tvl.fyi/c/depot/+/2746 Tested-by: BuildkiteCI Reviewed-by: tazjin <mail@tazj.in>
2021-02-21 12:57:40 +01:00
]);
in
runExecline.local drvName { } [
"importas"
"out"
"out"
"pipeline"
[
"foreground"
[
"if"
[ bins.printf "%s\n" shebang ]
]
"if"
[ bins.cat script ]
]
"if"
[ bins.tee "$out" ]
"if"
[ bins.chmod "+x" "$out" ]
"exit"
"0"
];
in
{
inherit
respond
pathInfo
scriptName
absolutePath
writeCGI
;
}