mirrors/cosmopolitan
1
0
Fork 0
mirror of https://github.com/jart/cosmopolitan.git synced 2025-02-01 03:53:33 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions
cosmopolitan/tool/net/help.txt
Paul Kulchenko 1b93066883
Add GetCookie to redbean Lua (#269)
2021-09-11 16:46:21 -07:00

1237 lines
57 KiB
Text
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

SYNOPSIS
redbean.com [-hvduzmbagf] [-p PORT] [-- SCRIPTARGS...]
DESCRIPTION
redbean - single-file distributable web server
OVERVIEW
redbean makes it possible to share web applications that run offline
as a single-file Actually Portable Executable PKZIP archive which
contains your assets. All you need to do is download the redbean.com
program below, change the filename to .zip, add your content in a zip
editing tool, and then change the extension back to .com.
redbean can serve 1 million+ gzip encoded responses per second on a
cheap personal computer. That performance is thanks to zip and gzip
using the same compression format, which enables kernelspace copies.
Another reason redbean goes fast is that it's a tiny static binary,
which makes fork memory paging nearly free.
redbean is also easy to modify to suit your own needs. The program
itself is written as a single .c file. It embeds the Lua programming
language and SQLite which let you write dynamic pages.
FLAGS
-h help
-d daemonize
-u uniprocess
-z print port
-m log messages
-b log message bodies
-a log resource usage
-g log handler latency
-j enable ssl client verify
-k disable ssl fetch verify
-B use stronger cryptography
-f log worker function calls
-s increase silence [repeatable]
-v increase verbosity [repeatable]
-V increase ssl verbosity [repeatable]
-H K:V sets http header globally [repeatable]
-D DIR overlay assets in local directory [repeatable]
-r /X=/Y redirect X to Y [repeatable]
-R /X=/Y rewrites X to Y [repeatable]
-K PATH tls private key path [repeatable]
-C PATH tls certificate(s) path [repeatable]
-M INT tunes max message payload size [def. 65536]
-t INT timeout ms or keepalive sec if <0 [def. 60000]
-p PORT listen port [def. 8080; repeatable]
-l ADDR listen addr [def. 0.0.0.0; repeatable]
-c SEC configures static cache-control
-L PATH log file location
-P PATH pid file location
-U INT daemon set user id
-G INT daemon set group id
FEATURES
- Lua v5.4
- SQLite 3.35.5
- TLS v1.2 / v1.1 / v1.0
- HTTP v1.1 / v1.0 / v0.9
- Chromium-Zlib Compression
- Statusz Monitoring Statistics
- Self-Modifying PKZIP Object Store
- Linux + Windows + Mac + FreeBSD + OpenBSD + NetBSD
USAGE
This executable is also a ZIP file that contains static assets.
You can run redbean interactively in your terminal as follows:
./redbean.com -vvvmbag # starts server verbosely
open http://127.0.0.1:8080/ # shows zip listing page
CTRL-C # 1x: graceful shutdown
CTRL-C # 2x: forceful shutdown
You can override the default listing page by adding:
zip redbean.com index.lua # lua server pages take priority
zip redbean.com index.html # default page for directory
The listing page only applies to the root directory. However the
default index page applies to subdirectories too. In order for it
to work, there needs to be an empty directory entry in the zip.
That should already be the default practice of your zip editor.
wget \
--mirror \
--convert-links \
--adjust-extension \
--page-requisites \
--no-parent \
--no-if-modified-since \
http://a.example/index.html
zip -r redbean.com a.example/ # default page for directory
redbean normalizes the trailing slash for you automatically:
$ printf 'GET /a.example HTTP/1.0\n\n' | nc 127.0.0.1 8080
HTTP/1.0 307 Temporary Redirect
Location: /a.example/
Virtual hosting is accomplished this way too. The Host is simply
prepended to the path, and if it doesn't exist, it gets removed.
$ printf 'GET / HTTP/1.1\nHost:a.example\n\n' | nc 127.0.0.1 8080
HTTP/1.1 200 OK
Link: <http://127.0.0.1/a.example/index.html>; rel="canonical"
If you mirror a lot of websites within your redbean then you can
actually tell your browser that redbean is your proxy server, in
which redbean will act as your private version of the Internet.
$ printf 'GET http://a.example HTTP/1.0\n\n' | nc 127.0.0.1 8080
HTTP/1.0 200 OK
Link: <http://127.0.0.1/a.example/index.html>; rel="canonical"
If you use a reverse proxy, then redbean recognizes the following
provided that the proxy forwards requests over the local network:
X-Forwarded-For: 203.0.113.42:31337
X-Forwarded-Host: foo.example:80
There's a text/plain statistics page called /statusz that makes
it easy to track and monitor the health of your redbean:
printf 'GET /statusz\n\n' | nc 127.0.0.1 8080
redbean will display an error page using the /redbean.png logo
by default, embedded as a bas64 data uri. You can override the
custom page for various errors by adding files to the zip root.
zip redbean.com 404.html # custom not found page
Audio video content should not be compressed in your ZIP files.
Uncompressed assets enable browsers to send Range HTTP request.
On the other hand compressed assets are best for gzip encoding.
zip redbean.com index.html # adds file
zip -0 redbean.com video.mp4 # adds without compression
You can have redbean run as a daemon by doing the following:
sudo ./redbean.com -vvdp80 -p443 -L redbean.log -P redbean.pid
kill -TERM $(cat redbean.pid) # 1x: graceful shutdown
kill -TERM $(cat redbean.pid) # 2x: forceful shutdown
redbean currently has a 32kb limit on request messages and 64kb
including the payload. redbean will grow to whatever the system
limits allow. Should fork() or accept() fail redbean will react
by going into "meltdown mode" which closes lingering workers.
You can trigger this at any time using:
kill -USR2 $(cat redbean.pid)
Another failure condition is running out of disk space in which
case redbean reacts by truncating the log file. Lastly, redbean
does the best job possible reporting on resource usage when the
logger is in debug mode noting that NetBSD is the best at this.
Your redbean is an actually portable executable, that's able to
run on six different operating systems. To do that, it needs to
overwrite its own MZ header at startup, with ELF or Mach-O, and
then puts the original back once the program loads. If you want
your redbean to follow the platform-local executable convention
then delete the /.ape file from zip.
redbean contains software licensed ISC, MIT, BSD-2, BSD-3, zlib
which makes it a permissively licensed gift to anyone who might
find it useful. The transitive closure of legalese can be found
inside the binary. redbean also respects your privacy and won't
phone home because your computer is its home.
SECURITY
redbean uses a protocol polyglot for serving HTTP and HTTPS on
the same port numbers. For example, both of these are valid:
http://127.0.0.1:8080/
https://127.0.0.1:8080/
The easiest way to use a self-signed certificate is to provide
redbean with a key-signing key:
openssl req -x509 -newkey rsa:2048 \
-keyout .ca.key -out .ca.crt -days 6570 -nodes \
-subj '/C=US/ST=CA/O=Jane Doe/CN=My Root CA 1' \
-addext 'keyUsage = critical,cRLSign,keyCertSign'
sudo ./redbean.com -C ca.crt -K .ca.key -p 80 -p 443
Let's say you're migrating away from NGINX and you use Let's Encrypt.
In that case you'll likely want something like the following:
certbot certonly --nginx --key-type ecdsa \
--cert-name redbean-ecdsa -d redbean.dev -d www.redbean.dev
certbot certonly --nginx --key-type rsa \
--cert-name redbean-rsa -d redbean.dev -d www.redbean.dev
You can then program /var/www/html/.init.lua as such:
ProgramPrivateKey(Slurp('/etc/letsencrypt/live/redbean-ecdsa/privkey.pem'))
ProgramCertificate(Slurp('/etc/letsencrypt/live/redbean-ecdsa/fullchain.pem'))
ProgramPrivateKey(Slurp('/etc/letsencrypt/live/redbean-rsa/privkey.pem'))
ProgramCertificate(Slurp('/etc/letsencrypt/live/redbean-rsa/fullchain.pem'))
if IsDaemon() then
ProgramUid(33) # see `vipw` to get appropriate number
ProgramGid(33) # see `vigr` to get appropriate number
ProgramPort(80)
ProgramPort(443)
ProgramLogPath('/var/log/redbean.log')
ProgramPidPath('/var/run/redbean.pid')
end
function OnHttpRequest()
path = GetPath()
if path == '/favicon.ico' or
path == '/site.webmanifest' or
path == '/favicon-16x16.png' or
path == '/favicon-32x32.png' or
path == '/apple-touch-icon' then
SetLogLevel(kLogWarn)
end
Route()
SetHeader('Content-Language', 'en-US')
end
You'd then run redbean as follows:
redbean.com -dD /var/www/html
You can load as many public and private keys as you want. They can be
specified as pem, der, concatenated ascii, bundles, or chains. If you
don't specify specific chains then redbean will automatically infer it
based on SUBJECT → ISSUER relationships. Your redbean won't serve the
self-signed root certificate at the end of the chain where self-signed
is defined as SUBJECT == ISSUER. Otherwise you can control when chains
terminate by setting the max length constraint to zero.
Your redbean supports SSL virtual hosting. 99.76% of TLS clients send
a Server Name Indicator (SNI), which is matched against DNS or IPs in
Subject Alternative Names (SAN) or the Common Name (CN) of subject if
SAN isn't used. This means you don't need to reveal your whole domain
portfolio to each client just to have ssl. You can just use different
certificates for each domain if you choose to do so. If redbean can't
find an appropriate match, then the first certificate will be chosen.
Your redbean has been secured with algorithms so strong that, until a
few decades ago, it was illegal to share them with with those outside
the United States. By default your redbean offers roughly 128 bits of
security with modern clients but will fall back to at minimum 112 bit
security depending on the preferences of legacy and iot clients. Both
are secure based on public knowledge until 2030 according to NIST. If
you'd rather restrict yourself to just 150+ bits of security but with
the tradeoff of dropping support for old Internet Explorer and making
embedded clients less happy, then pass the -B flag, which'll restrict
redbean to a very short list of protocols, algorithms, and parameters
that the NSA, NIST, and IANA all agree upon.
SSL verbosity is controlled as follows for troubleshooting:
-V log ssl errors
-VV log ssl state changes too
-VVV log ssl informational messages too
-VVVV log ssl verbose details too
That's in addition to existing flags like -vvvm.
LUA SERVER PAGES
Any files with the extension .lua will be dynamically served by redbean.
Here's the simplest possible example:
Write('<b>Hello World</b>')
The Lua Server Page above should be able to perform at 700,000 responses
per second on a Core i9, without any sort of caching. If you want a Lua
handler that can do 1,000,000 responses per second, then try adding the
following global handler to your /.init.lua file:
function OnHttpRequest()
Write('<b>Hello World</b>')
end
Here's an example of a more typical workflow for Lua Server Pages using
the redbean API:
SetStatus(200)
SetHeader('Content-Type', 'text/plain; charset=utf-8')
Write('<p>Hello ')
Write(EscapeHtml(GetParam('name')))
We didn't need the first two lines in the previous example, because
they're implied by redbean automatically if you don't set them. Responses
are also buffered until the script finishes executing. That enables
redbean to make HTTP as easy as possible. In the future, API capabilities
will be expanded to make possible things like websockets.
redbean embeds the Lua standard library. You can use packages such as io
to persist and share state across requests and connections, as well as the
StoreAsset function, and the lsqlite3 module.
Your Lua interpreter begins its life in the main process at startup in the
.init.lua, which is likely where you'll want to perform all your expensive
one-time operations like importing modules. Then, as requests roll in,
isolated processes are cloned from the blueprint you created.
SPECIAL PATHS
/
redbean will generate a zip central directory listing for this
page, and this page only, but only if there isn't an /index.lua or
/index.html file defined.
/.init.lua
This script is run once in the main process at startup. This lets
you modify the state of the Lua interpreter before connection
processes are forked off. For example, it's a good idea to do
expensive one-time computations here. You can also use this file
to call the ProgramFOO() functions below. The init module load
happens after redbean's arguments and zip assets have been parsed,
but before calling functions like socket() and fork(). Note that
this path is a hidden file so that it can't be unintentionally run
by the network client.
/.reload.lua
This script is run from the main process when SIGHUP is received.
This only applies to redbean when running in daemon mode. Any
changes that are made to the Lua interpreter state will be
inherited by future forked connection processes. Note that this
path is a hidden file so that it can't be unintentionally run by
the network client.
/.lua/...
Your Lua modules go in this directory. The way it works is redbean
sets Lua's package.path to /zip/.lua/?.lua;/zip/.lua/?/init.lua by
default. Cosmopolitan Libc lets system calls like open read from
the ZIP structure, if the filename is prefixed with zip:. So this
works like magic.
/redbean.png
If it exists, it'll be used as the / listing page icon, embedded
as a base64 URI.
/usr/share/ssl/root
This directory contains your root certificate authorities. It is
needed so the Fetch() HTTPS client API can verify that a remote
certificate was signed by a third party. You can add your own
certificate files to this directory within the ZIP executable.
If you enable HTTPS client verification then redbean will check
that HTTPS clients (a) have a certificate and (b) it was signed.
GLOBALS
argv: array[str]
Array of command line arguments, excluding those parsed by
getopt() in the C code, which stops parsing at the first
non-hyphenated arg. In some cases you can use the magic --
argument to delimit C from Lua arguments.
HOOKS
OnHttpRequest()
If this function is defined in the global scope by your /.init.lua
then redbean will call it at the ealiest possible moment to
hand over control for all messages (with the exception of OPTIONS
*). See functions like Route which asks redbean to do its default
thing from the handler.
OnClientConnection(ip:int,port:int,serverip:int,serverport:int) → bool
If this function is defined it'll be called from the main process
each time redbean accepts a new client connection. If it returns
true then redbean will close the connection without calling fork.
OnProcessCreate(pid:int,ip:int,port:int,serverip:int,serverport:int)
If this function is defined it'll be called from the main process
each time redbean forks a connection handler worker process. The
ip/port of the remote client is provided, along with the ip/port
of the listening interface that accepted the connection. This may
be used to create a server activity dashboard, in which case the
data provider handler should set SetHeader('Connection','Close').
This won't be called in uniprocess mode.
OnProcessDestroy(pid:int)
If this function is defined it'll be called from the main process
each time redbean reaps a child connection process using wait4().
This won't be called in uniprocess mode.
OnServerStart()
If this function is defined it'll be called from the main process
right before the main event loop starts.
OnServerStop()
If this function is defined it'll be called from the main process
after all the connection processes have been reaped and exit() is
ready to be called.
OnWorkerStart()
If this function is defined it'll be called from the child worker
process after it's been forked and before messages are handled.
This won't be called in uniprocess mode.
OnWorkerStop()
If this function is defined it'll be called from the child worker
process once _exit() is ready to be called. This won't be called
in uniprocess mode.
FUNCTIONS
Write(data:str)
Appends data to HTTP response payload buffer. This is buffered
independently of headers.
SetStatus(code:int[,reason:str])
Starts an HTTP response, specifying the parameters on its first
line. reason is optional since redbean can fill in the appropriate
text for well-known magic numbers, e.g. 200, 404, etc. This method
will reset the response and is therefore mutually exclusive with
ServeAsset and ServeError. If a status setting function isn't
called, then the default behavior is to send 200 OK.
SetHeader(name:str,value:str)
Appends HTTP header to response header buffer. name is
case-insensitive and restricted to non-space ASCII. value is a
UTF-8 string that must be encodable as ISO-8859-1. Leading and
trailing whitespace is trimmed automatically. Overlong characters
are canonicalized. C0 and C1 control codes are forbidden, with the
exception of tab. This function automatically calls SetStatus(200,
"OK") if a status has not yet been set. The header buffer is
independent of the payload buffer. Neither are written to the wire
until the Lua Server Page has finished executing. This function
disallows the setting of certain headers such as Content-Range and
Date, which are abstracted by the transport layer. In such cases,
consider calling ServeAsset.
SetCookie(name:str,value:str[,options:table])
Appends Set-Cookie HTTP header to the response header buffer.
Several Set-Cookie headers can be added to the same response.
__Host- and __Secure- prefixes are supported and may set or
overwrite some of the options (for example, specifying __Host-
prefix sets the Secure option to true, sets the path to "/", and
removes the Domain option). The following options can be used (their
lowercase equivalents are supported as well):
- Expires: sets the maximum lifetime of the cookie as an HTTP-date
timestamp. Can be specified as a Date in the RFC1123 (string)
format or as a UNIX timestamp (number of seconds).
- MaxAge: sets number of seconds until the cookie expires. A zero
or negative number will expire the cookie immediately. If both
Expires and MaxAge are set, MaxAge has precedence.
- Domain: sets the host to which the cookie will be sent.
- Path: sets the path that must be present in the request URL, or
the client will not send the Cookie header.
- Secure: (bool) requests the cookie to be only send to the
server when a request is made with the https: scheme.
- HttpOnly: (bool) forbids JavaScript from accessing the cookie.
- SameSite: (Strict, Lax, or None) controls whether a cookie is
sent with cross-origin requests, providing some protection
against cross-site request forgery attacks.
GetParam(name:str) → value:str
Returns first value associated with name. name is handled in a
case-sensitive manner. This function checks Request-URL parameters
first. Then it checks application/x-www-form-urlencoded from the
message body, if it exists, which is common for HTML forms sending
POST requests. If a parameter is supplied matching name that has
no value, e.g. foo in ?foo&bar=value, then the returned value will
be nil, whereas for ?foo=&bar=value it would be "". To
differentiate between no-equal and absent, use the HasParam
function. The returned value is decoded from ISO-8859-1 (only in
the case of Request-URL) and we assume that percent-encoded
characters were supplied by the client as UTF-8 sequences, which
are returned exactly as the client supplied them, and may
therefore may contain overlong sequences, control codes, NUL
characters, and even numbers which have been banned by the IETF.
It is the responsibility of the caller to impose further
restrictions on validity, if they're desired.
EscapeHtml(str) → str
Escapes HTML entities: The set of entities is &><"' which become
&amp;&gt;&lt;&quot;&#39;. This function is charset agnostic and
will not canonicalize overlong encodings. It is assumed that a
UTF-8 string will be supplied. See escapehtml.c.
LaunchBrowser([path:str])
Launches web browser on local machine with URL to this redbean
server. This function may be called from your /.init.lua.
CategorizeIp(ip:uint32) → str
Returns a string describing an IP address. This is currently Class
A granular. It can tell you if traffic originated from private
networks, ARIN, APNIC, DOD, etc.
DecodeBase64(ascii:str) → binary:str
Turns ASCII into binary, in a permissive way that ignores
characters outside the base64 alphabet, such as whitespace. See
decodebase64.c.
DecodeLatin1(iso-8859-1:str) → utf-8:str
Turns ISO-8859-1 string into UTF-8.
EncodeBase64(binary:str) → ascii:str
Turns binary into ASCII. This can be used to create HTML data:
URIs that do things like embed a PNG file in a web page. See
encodebase64.c.
EncodeLatin1(utf-8:str[,flags:int]) → iso-8859-1:str
Turns UTF-8 into ISO-8859-1 string.
EscapeFragment(str) → str
Escapes URL #fragment. The allowed characters are
-/?.~_@:!$&'()*+,;=0-9A-Za-z and everything else gets %XX encoded.
Please note that '& can still break HTML and that '() can still
break CSS URLs. This function is charset agnostic and will not
canonicalize overlong encodings. It is assumed that a UTF-8 string
will be supplied. See kescapefragment.c.
EscapeHost(str) → str
Escapes URL host. See kescapeauthority.c
EscapeLiteral(str) → str
Escapes JavaScript or JSON string literal content. The caller is
responsible for adding the surrounding quotation marks. This
implementation \uxxxx sequences for all non-ASCII sequences. HTML
entities are also encoded, so the output doesn't need EscapeHtml.
This function assumes UTF-8 input. Overlong encodings are
canonicalized. Invalid input sequences are assumed to be
ISO-8859-1. The output is UTF-16 since that's what JavaScript
uses. For example, some individual codepoints such as emoji
characters will encode as multiple \uxxxx sequences. Ints that are
impossible to encode as UTF-16 are substituted with the \xFFFD
replacement character. See escapejsstringliteral.c.
EscapeParam(str) → str
Escapes URL parameter name or value. The allowed characters are
-.*_0-9A-Za-z and everything else gets %XX encoded. This function
is charset agnostic and will not canonicalize overlong encodings.
It is assumed that a UTF-8 string will be supplied. See
kescapeparam.c.
EscapePass(str) → str
Escapes URL password. See kescapeauthority.c.
EscapePath(str) → str
Escapes URL path. This is the same as EscapeSegment except slash
is allowed. The allowed characters are -.~_@:!$&'()*+,;=0-9A-Za-z/
and everything else gets %XX encoded. Please note that '& can
still break HTML, so the output may need EscapeHtml too. Also note
that '() can still break CSS URLs. This function is charset
agnostic and will not canonicalize overlong encodings. It is
assumed that a UTF-8 string will be supplied. See kescapepath.c.
EscapeSegment(str) → str
Escapes URL path segment. This is the same as EscapePath except
slash isn't allowed. The allowed characters are
-.~_@:!$&'()*+,;=0-9A-Za-z and everything else gets %XX encoded.
Please note that '& can still break HTML, so the output may need
EscapeHtml too. Also note that '() can still break CSS URLs. This
function is charset agnostic and will not canonicalize overlong
encodings. It is assumed that a UTF-8 string will be supplied. See
kescapesegment.c.
EscapeUser(str) → str
Escapes URL username. See kescapeauthority.c.
Fetch(url:str[,body:str|{method=value:str,body=value:str,...}])
→ status:int,{header:str=value:str,...},body:str
Sends an HTTP/HTTPS request to the specified URL. If only the URL is
provided, then a GET request is sent. If both URL and body parameters
are specified, then a POST request is sent. If any other method needs
to be specified (for example, PUT or DELETE), then passing a table as
the second value allows setting method and body values as well other
options:
- followredirect (default = true): forces temporary and permanent
redirects to be followed. This behavior can be disabled by
passing `false`.
- maxredirects (default = 5): sets the number of allowed redirects
to minimize looping due to misconfigured servers. When the number
is exceeded, the result of the last redirect is returned.
When the redirect is being followed, the same method and body values
are being sent in all cases except when 303 status is returned. In
that case the method is set to GET and the body is removed before the
redirect is followed. Note that if these (method/body) values are
provided as table fields, they will be modified in place.
FormatHttpDateTime(seconds:int) → rfc1123:str
Converts UNIX timestamp to an RFC1123 string that looks like this:
Mon, 29 Mar 2021 15:37:13 GMT. See formathttpdatetime.c.
FormatIp(uint32) → str
Turns integer like 0x01020304 into a string like 1.2.3.4. See also
ParseIp for the inverse operation.
GetAssetMode(path:str) → int
Returns UNIX-style octal mode for ZIP asset (or local file if the
-D flag is used)
GetAssetSize(path:str) → int
Returns byte size of uncompressed contents of ZIP asset (or local
file if the -D flag is used)
GetComment(path:str) → str
Returns comment text associated with asset in the ZIP central
directory.
GetCookie(name:str) → str
Returns cookie value.
GetRemoteAddr() → ip:uint32,port:uint16
Returns client ip4 address and port, e.g. 0x01020304,31337 would
represent 1.2.3.4:31337. This is the same as GetClientAddr except
it will use the ip:port from the X-Forwarded-For header, only if
IsPrivateIp or IsLoopbackIp return true.
GetClientAddr() → ip:uint32,port:uint16
Returns client socket ip4 address and port, e.g. 0x01020304,31337
would represent 1.2.3.4:31337. Please consider using GetRemoteAddr
instead, since the latter takes into consideration reverse proxy
scenarios.
GetServerAddr() → ip:uint32,port:uint16
Returns address to which listening server socket is bound, e.g.
0x01020304,8080 would represent 1.2.3.4:8080. If -p 0 was supplied
as the listening port, then the port in this string will be
whatever number the operating system assigned.
GetDate() → seconds:int
Returns date associated with request that's used to generate the
Date header, which is now, give or take a second. The returned
value is a UNIX timestamp.
GetHeader(name:str) → value:str
Returns HTTP header. name is case-insensitive. The header value is
returned as a canonical UTF-8 string, with leading and trailing
whitespace trimmed, which was decoded from ISO-8859-1, which is
guaranteed to not have C0/C1 control sequences, with the exception
of the tab character. Leading and trailing whitespace is
automatically removed. In the event that the client suplies raw
UTF-8 in the HTTP message headers, the original UTF-8 sequence can
be losslessly restored by counter-intuitively recoding the
returned string back to Latin1. If the requested header is defined
by the RFCs as storing comma-separated values (e.g. Allow,
Accept-Encoding) and the field name occurs multiple times in the
message, then this function will fold those multiple entries into
a single string.
GetHeaders() → {name:str=value:str,...}
Returns HTTP headers as dictionary mapping header key strings to
their UTF-8 decoded values. The ordering of headers from the
request message is not preserved. Whether or not the same key can
repeat depends on whether or not it's a standard header, and if
so, if it's one of the ones that the RFCs define as repeatable.
See khttprepeatable.c. Those headers will not be folded. Standard
headers which aren't on that list, will be overwritten with the
last-occurring one during parsing. Extended headers are always
passed through exactly as they're received. Please consider using
GetHeader API if possible since it does a better job abstracting
these issues.
GetLogLevel() → int
Returns logger verbosity level. Likely return values are kLogDebug
> kLogVerbose > kLogInfo > kLogWarn > kLogError > kLogFatal.
GetHost() → str
Returns host associated with request. This will be the Host
header, if it's supplied. Otherwise it's the bind address.
GetHostOs() → str
Returns string that describes the host OS.
GetMonospaceWidth(str|char) → int
Returns monospace display width of string. This is useful for
fixed-width formatting. For example, CJK characters typically take
up two cells. This function takes into consideration combining
characters, which are discounted, as well as control codes and
ANSI escape sequences.
GetMethod() → str
Returns HTTP method. Normally this will be GET, HEAD, or POST in
which case redbean normalizes this value to its uppercase form.
Anything else that the RFC classifies as a "token" string is
accepted too, which might contain characters like &".
GetParams() → {{name:str[,value:str]},...}
Returns name=value parameters from Request-URL and
application/x-www-form-urlencoded message body in the order they
were received. This may contain duplicates. The inner array will
have either one or two items, depending on whether or not the
equals sign was used.
GetPath() → str
Returns the Request-URL path. This is guaranteed to begin with
"/". It is further guaranteed that no "//" or "/." exists in the
path. The returned value is returned as a UTF-8 string which was
decoded from ISO-8859-1. We assume that percent-encoded characters
were supplied by the client as UTF-8 sequences, which are returned
exactly as the client supplied them, and may therefore may contain
overlong sequences, control codes, NUL characters, and even
numbers which have been banned by the IETF. redbean takes those
things into consideration when performing path safety checks. It
is the responsibility of the caller to impose further restrictions
on validity, if they're desired.
GetEffectivePath() → str
Returns path as it was resolved by the routing algorithms, which
might contain the virtual host prepended if used.
GetScheme() → str
Returns scheme from Request-URL, if any.
GetTime() → seconds:number
Returns current time as a UNIX timestamp with 0.0001s precision.
GetPayload() → str
Returns the request message payload, or empty string if there
isn't one.
GetUrl() → str
Returns the effective Request-URL as an ASCII string, where
illegal characters or UTF-8 is guaranteed to be percent encoded,
and has been normalized to include either the Host or
X-Forwarded-Host headers, if they exist, and possibly a scheme too
if redbean is being used as an HTTP proxy server. In the future
this API might change to return an object instead.
GetHttpVersion() → int
Returns the request HTTP protocol version, which can be 9 for
HTTP/0.9, 10 for HTTP/1.0, or 11 for HTTP/1.1.
GetRandomBytes([length:int]) → str
Returns string with the specified number of random bytes (1..256).
If no length is specified, then a string of length 16 is returned.
GetRedbeanVersion() → int
Returns the Redbean version in the format 0xMMmmpp, with major (MM),
minor (mm), and patch (pp) versions encoded. The version value 1.4
would be represented as 0x010400.
GetZipPaths() → {path:str,...}
Returns paths of all assets in the zip central directory, prefixed
by a slash.
HasParam(name:str) → bool
Returns true if parameter with name was supplied in either the
Request-URL or an application/x-www-form-urlencoded message body.
HidePath(prefix:str)
Programs redbean / listing page to not display any paths beginning
with prefix. This function should only be called from /.init.lua.
IsPublicIp(uint32) → bool
Returns true if IP address is not a private network (10.0.0.0/8,
172.16.0.0/12, 192.168.0.0/16) and is not localhost (127.0.0.0/8).
Note: we intentionally regard TEST-NET IPs as public.
IsPrivateIp(uint32) → bool
Returns true if IP address is part of a private network
(10.0.0.0/8, 172.16.0.0/12, 192.168.0.0/16).
IsLoopbackIp(uint32) → bool
Returns true if IP address is part of the localhost network
(127.0.0.0/8).
IsCompressed(path:str) → bool
Returns true if ZIP artifact at path is stored on disk using
DEFLATE compression.
IndentLines(str[,int]) → str
Adds spaces to beginnings of multiline string. If the int
parameter is not supplied then 1 space will be added.
LoadAsset(path:str) → str
Returns contents of file as string. The asset may be sourced from
either the zip (decompressed) or the local filesystem if the -D
flag was used. If slurping large file into memory is a concern,
then consider using ServeAsset which can serve directly off disk.
StoreAsset(path:str,data:str,mode:int)
Stores asset to executable's ZIP central directory. This currently
happens in an append-only fashion and is still largely in the
proof-of-concept stages. Currently only supported on Linux, XNU,
and FreeBSD.
Log(level:int,message:str)
Emits message string to log, if level is less than or equal to
GetLogLevel. If redbean is running in interactive mode, then this
will log to the console. If redbean is running as a daemon or the
-L LOGFILE flag is passed, then this will log to the file.
Reasonable values for level are kLogDebug > kLogVerbose > kLogInfo
> kLogWarn > kLogError > kLogFatal. The logger emits timestamps in
the local timezone with microsecond precision. If log entries are
emitted more frequently than once per second, then the log entry
will display a delta timestamp, showing how much time has elapsed
since the previous log entry. This behavior is useful for quickly
measuring how long various portions of your code take to execute.
ParseHttpDateTime(rfc1123:str) → seconds:int
Converts RFC1123 string that looks like this: Mon, 29 Mar 2021
15:37:13 GMT to a UNIX timestamp. See parsehttpdatetime.c.
ParseUrl(str) → URL
Parses URL, returning object having the following fields: scheme,
user, pass, host, port, path, params, fragment. This parser is
charset agnostic. Percent encoded bytes are decoded for all
fields. Returned values might contain things like NUL characters,
spaces, control codes, and non-canonical encodings. Absent can be
discerned from empty by checking if the pointer is set. There's no
failure condition for this routine. This is a permissive parser.
This doesn't normalize path segments like `.` or `..` so use
IsAcceptablePath() to check for those. No restrictions are imposed
beyond that which is strictly necessary for parsing. All the data
that is provided will be consumed to the one of the fields. Strict
conformance is enforced on some fields more than others, like
scheme, since it's the most non-deterministically defined field of
them all. Please note this is a URL parser, not a URI parser.
Which means we support everything everything the URI spec says we
should do except for the things we won't do, like tokenizing path
segments into an array and then nesting another array beneath each
of those for storing semicolon parameters. So this parser won't
make SIP easy. What it can do is parse HTTP URLs and most URIs
like data:opaque, better in fact than most things which claim to
be URI parsers.
EncodeUrl(URL) → str
This function is the inverse of ParseUrl. The output will always
be correctly formatted. The exception is if illegal characters are
supplied in the scheme field, since there's no way of escaping
those. Opaque parts are escaped as though they were paths, since
many URI parsers won't understand things like an unescaped
question mark in path.
ParseIp(str) → int
Converts IPv4 address string to integer, e.g. "1.2.3.4" →
0x01020304, or returns -1 for invalid inputs. See also FormatIp
for the inverse operation.
ProgramBrand(str)
Changes HTTP Server header, as well as the <h1> title on the /
listing page. The brand string needs to be a UTF-8 value that's
encodable as ISO-8859-1. If the brand is changed to something
other than redbean, then the promotional links will be removed
from the listing page too. This function should only be called
from /.init.lua.
ProgramCache(seconds:int)
Configures Cache-Control and Expires header generation for static
asset serving. A negative value will disable the headers. Zero
means don't cache. Greater than zero asks public proxies and
browsers to cache for a given number of seconds. This should only
be called from /.init.lua.
ProgramCertificate(pem:str)
Same as the -C flag if called from .init.lua, e.g.
ProgramCertificate(LoadAsset("/.sign.crt")) for zip loading or
ProgramCertificate(Slurp("/etc/letsencrypt.lol/fullchain.pem")) for
local file system only.
ProgramHeader(name:str,value:str)
Appends HTTP header to the header buffer for all responses (whereas
SetHeader only appends a header to the current response buffer).
name is case-insensitive and restricted to non-space ASCII. value
is a UTF-8 string that must be encodable as ISO-8859-1. Leading and
trailing whitespace is trimmed automatically. Overlong characters
are canonicalized. C0 and C1 control codes are forbidden, with the
exception of tab. The header buffer is independent of the payload
buffer. This function disallows the setting of certain headers such
as Content-Range and Date, which are abstracted by the transport
layer.
ProgramPort(uint16)
Hard-codes the port number on which to listen, which can be any
number in the range 1..65535, or alternatively 0 to ask the
operating system to choose a port, which may be revealed later on
by GetServerAddr or the -z flag to stdout.
ProgramPrivateKey(pem:str)
Same as the -K flag if called from .init.lua, e.g.
ProgramPrivateKey(LoadAsset("/.sign.key")) for zip loading or
ProgramPrivateKey(Slurp("/etc/letsencrypt/fullchain.pem")) for
local file system only.
ProgramRedirect(code:int,src:str,location:str)
Configures fallback routing for paths which would otherwise return
404 Not Found. If code is 0 then the path is rewritten internally
as an accelerated redirect. If code is 301, 302, 307, or 308 then
a redirect response will be sent to the client. This should only
be called from /.init.lua.
ProgramSslTicketLifetime(seconds:int)
Defaults to 86400 (24 hours). This may be set to ≤0 to disable
SSL tickets. It's a good idea to use these since it increases
handshake performance 10x and eliminates a network round trip.
EvadeDragnetSurveillance(bool)
If this option is programmed then redbean will not transmit a
Server Name Indicator (SNI) when performing Fetch() requests.
ProgramSslPresharedKey(key:str,identity:str)
This function can be used to enable the PSK ciphersuites
which simplify SSL and enhance its performance in controlled
environments. `key` may contain 1..32 bytes of random binary
data and identity is usually a short plaintext string. The
first time this function is called, the preshared key will
be added to both the client and the server SSL configs. If
it's called multiple times, then the remaining keys will be
added to the server, which is useful if you want to assign
separate keys to each client, each of which needs a separate
identity too. If this function is called multiple times with
the same identity string, then the latter call will overwrite
the prior. If a preshared key is supplied and no certificates
or key-signing-keys are programmed, then redbean won't bother
auto-generating any serving certificates and will instead use
only PSK ciphersuites.
ProgramSslCiphersuite(name:str)
This function may be called multiple times to specify which
ciphersuites should be used in the server and client. The
default list, ordered by preference, is as follows:
ECDHE-ECDSA-AES256-GCM-SHA384
ECDHE-ECDSA-AES128-GCM-SHA256
ECDHE-ECDSA-CHACHA20-POLY1305-SHA256
ECDHE-PSK-AES256-GCM-SHA384
ECDHE-PSK-AES128-GCM-SHA256
ECDHE-PSK-CHACHA20-POLY1305-SHA256
ECDHE-RSA-AES256-GCM-SHA384
ECDHE-RSA-AES128-GCM-SHA256
ECDHE-RSA-CHACHA20-POLY1305-SHA256
DHE-RSA-AES256-GCM-SHA384
DHE-RSA-AES128-GCM-SHA256
DHE-RSA-CHACHA20-POLY1305-SHA256
ECDHE-ECDSA-AES128-CBC-SHA256
ECDHE-RSA-AES256-CBC-SHA384
ECDHE-RSA-AES128-CBC-SHA256
DHE-RSA-AES256-CBC-SHA256
DHE-RSA-AES128-CBC-SHA256
ECDHE-PSK-AES256-CBC-SHA384
ECDHE-PSK-AES128-CBC-SHA256
ECDHE-ECDSA-AES256-CBC-SHA
ECDHE-ECDSA-AES128-CBC-SHA
ECDHE-RSA-AES256-CBC-SHA
ECDHE-RSA-AES128-CBC-SHA
DHE-RSA-AES256-CBC-SHA
DHE-RSA-AES128-CBC-SHA
ECDHE-PSK-AES256-CBC-SHA
ECDHE-PSK-AES128-CBC-SHA
RSA-AES256-GCM-SHA384
RSA-AES128-GCM-SHA256
RSA-AES256-CBC-SHA256
RSA-AES128-CBC-SHA256
RSA-AES256-CBC-SHA
RSA-AES128-CBC-SHA
PSK-AES256-GCM-SHA384
PSK-AES128-GCM-SHA256
PSK-CHACHA20-POLY1305-SHA256
PSK-AES256-CBC-SHA384
PSK-AES128-CBC-SHA256
PSK-AES256-CBC-SHA
PSK-AES128-CBC-SHA
ECDHE-RSA-3DES-EDE-CBC-SHA
DHE-RSA-3DES-EDE-CBC-SHA
ECDHE-PSK-3DES-EDE-CBC-SHA
RSA-3DES-EDE-CBC-SHA
PSK-3DES-EDE-CBC-SHA
The names above are canonical to redbean and were simplified
programmatically from the official IANA names. This function
will accept the IANA names too. In most cases it will accept
the OpenSSL and GnuTLS naming convention as well.
IsDaemon() → bool
Returns true if -d flag was passed to redbean.
ProgramUid(int)
Same as the -U flag if called from .init.lua for setuid()
ProgramGid(int)
Same as the -U flag if called from .init.lua for setgid()
ProgramDirectory(str)
Same as the -D flag if called from .init.lua for overlaying local
file system directories. This may be called multiple times. The
first directory programmed is preferred. These currently do not
show up in the index page listing.
ProgramLogMessages(bool)
Same as the -m flag if called from .init.lua for logging message
headers only.
ProgramLogBodies(bool)
Same as the -b flag if called from .init.lua for logging message
bodies as part of POST / PUT / etc. requests.
ProgramLogPath(str)
Same as the -L flag if called from .init.lua for setting the log
file path on the local file system. It's created if it doesn't
exist. This is called before de-escalating the uesr / group id.
The file is opened in append only mode. If the disk runs out of
space then redbean will truncate the log file if has access to
change the log file after daemonizing.
ProgramPidPath(str)
Same as the -P flag if called from .init.lua for setting the pid
file path on the local file system. It's useful for reloading
daemonized redbean using `kill -HUP $(cat /var/run/redbean.pid)`
or terminating redbean with `kill $(cat /var/run/redbean.pid)`
which will gracefully terminate all clients. Sending the TERM
signal twice will cause a forceful shutdown, which might make
someone with a slow internet connection who's downloading big
files unhappy.
Slurp(filename:str) → str
Reads file data from local file system.
Sleep(seconds:number)
Sleeps the specified number of seconds (can be fractional). The
smallest interval is a microsecond.
Route([host:str,[path:str]])
Instructs redbean to follow the normal HTTP serving path. This
function is useful when writing an OnHttpRequest handler, since
that overrides the serving path entirely. So if the handler
decides it doesn't want to do anything, it can simply call this
function, to hand over control back to the redbean core. By
default, the host and path arguments are supplied from the
resolved GetUrl value. This handler always resolves, since it will
generate a 404 Not Found response if redbean couldn't find an
appropriate endpoint.
RouteHost([host:str,[path:str]]) → bool
This is the same as Route, except it only implements the subset of
request routing needed for serving virtual-hosted assets, where
redbean tries to prefix the path with the hostname when looking up
a file. This function returns true if the request was resolved. If
it was resolved, then your OnHttpRequest request handler can still
set additional headers.
RoutePath([path:str]) → bool
This is the same as Route, except it only implements the subset of
request routing needed for serving assets. This function returns
true if the request was resolved. If it was resolved, then your
OnHttpRequest request handler can still set additional headers.
ServeAsset(path:str)
Instructs redbean to serve static asset at path. This function
causes what would normally happen outside a dynamic handler to
happen. The asset can be sourced from either the zip or local
filesystem if -D is used. This function is mutually exclusive with
SetStatus and other Serve* functions.
ServeError(code:int[,reason:str])
Instructs redbean to serve a boilerplate error page. This takes
care of logging the error, setting the reason phrase, and adding a
payload. This function is mutually exclusive with SetStatus and
other Serve* functions.
ServeRedirect(code:int,location:str)
Instructs redbean to return the specified redirect code along with
the Location header set. This function is mutually exclusive with
SetStatus and other Serve* functions.
SetLogLevel(level:int)
Sets logger verbosity. Reasonable values for level are kLogDebug >
kLogVerbose > kLogInfo > kLogWarn > kLogError > kLogFatal. This is
reset at the end of the http request, so it can be used to disable
access log and message logging.
VisualizeControlCodes(str) → str
Replaces C0 control codes with their UNICODE pictures
representation. This function also canonicalizes overlong
encodings. C1 control codes are replaced with a JavaScript-like
escape sequence.
Underlong(str) → str
Canonicalizes overlong encodings.
Crc32(initial:int,data:str) → int
Computes 32-bit CRC-32 used by zip/zlib/gzip/etc.
Crc32c(initial:int,data:str) → int
Computes 32-bit Castagnoli Cyclic Redundancy Check.
Md5(str) → str
Computes MD5 checksum, returning 16 bytes of binary.
Sha1(str) → str
Computes SHA1 checksum, returning 20 bytes of binary.
Sha224(str) → str
Computes SHA224 checksum, returning 28 bytes of binary.
Sha256(str) → str
Computes SHA256 checksum, returning 32 bytes of binary.
Sha384(str) → str
Computes SHA384 checksum, returning 48 bytes of binary.
Sha512(str) → str
Computes SHA512 checksum, returning 64 bytes of binary.
Bsf(x:int) → int
Returns position of first bit set. Passing 0 will raise an error.
Same as the Intel x86 instruction BSF.
Bsr(x:int) → int
Returns binary logarithm of 𝑥. Passing 0 will raise an error. Same
as the Intel x86 instruction BSR.
Popcnt(x:int) → int
Returns number of bits set in integer.
LSQLITE3 MODULE
Please refer to the LuaSQLite3 Documentation.
For example, you could put the following in your /.init.lua file:
sqlite3 = require "lsqlite3"
db = sqlite3.open_memory()
db:exec[[
CREATE TABLE test (
id INTEGER PRIMARY KEY,
content TEXT
);
INSERT INTO test (content) VALUES ('Hello World');
INSERT INTO test (content) VALUES ('Hello Lua');
INSERT INTO test (content) VALUES ('Hello Sqlite3');
]]
Then, your Lua server pages or OnHttpRequest handler may perform SQL
queries by accessing the db global. The performance is good too, at about
400k qps.
for row in db:nrows("SELECT * FROM test") do
Write(row.id.." "..row.content.."<br>")
end
redbean supports a subset of what's defined in the upstream LuaSQLite3
project. Most of the unsupported APIs relate to pointers and database
notification hooks.
redbean also currently disables SQLite features which don't make sense for
production serving, such as ALTER, VACUUM, ANALYZE, etc. For that reason
we provide an APE build of the SQLite shell which you can use to
administrate your redbean database. See the sqlite3.com download above.
RE MODULE
This module exposes an API for POSIX regular expressions which enable you
to validate input, search for substrings, extract pieces of strings, etc.
Here's a usage example:
# Example IPv4 Address Regular Expression (see also ParseIP)
p = re.compile([[^([0-9]{1,3})\.([0-9]{1,3})\.([0-9]{1,3})\.([0-9]{1,3})$]])
m,a,b,c,d = p:search(𝑠)
if m then
print("ok", tonumber(a), tonumber(b), tonumber(c), tonumber(d))
else
print("not ok")
end
re.search(regex:str,text:str[,flags:int]) → [match[,group_1,...]]
Shortcut for re.compile plus regex_t*:search.
re.compile(regex:str[,flags:int]) → regex_t*
Compiles regular expression, using the POSIX extended syntax. This
has an O(2^𝑛) cost, so it's a good idea to do this from your
/.init.lua file. Flags may contain re.BASIC, re.ICASE, re.NOSUB,
and/or re.NEWLINE. See also regcomp() from libc.
regex_t*:search(text:str[,flags:int]) → [match[,group_1,...]]
Executes regular expression. This has an O(𝑛) cost. This returns
nothing (nil) if the pattern doesn't match anything. Otherwise it
pushes the matched substring and any parenthesis-captured values
too. Flags may contain re.NOTBOL or re.NOTEOL to indicate whether
or not text should be considered at the start and/or end of a
line.
re.BASIC
Use this flag if you prefer the default POSIX regex syntax. We use
extended regex notation by default. For example, an extended
regular expression for matching an IP address might look like
([0-9]*)\.([0-9]*)\.([0-9]*)\.([0-9]*) whereas with basic syntax
it would look like \([0-9]*\)\.\([0-9]*\)\.\([0-9]*\)\.\([0-9]*\).
This flag may only be used with re.compile and re.search.
re.ICASE
Use this flag to make your pattern case ASCII case-insensitive.
This flag may only be used with re.compile and re.search.
re.NEWLINE
Use this flag to make your pattern case ASCII case-insensitive.
This means [a-z] will mean the same thing as [A-Za-z]. This flag
may only be used with re.compile and re.search.
re.NOSUB
Causes re.search to only report success and failure. This is
reported via the API by returning empty string for success. This
flag may only be used with re.compile and re.search.
re.NOTBOL
The first character of the string pointed to by string is not the
beginning of the line. This flag may only be used with re.search
and regex_t*:search
re.NOTEOL
The last character of the string pointed to by string is not the
end of the line. This flag may only be used with re.search and
regex_t*:search
CONSTANTS
kLogDebug
Integer for debug logging level. See Log.
kLogVerbose
Integer for verbose logging level, which is less than kLogDebug.
kLogInfo
Integer for info logging level, which is less than kLogVerbose.
kLogWarn
Integer for warn logging level, which is less than kLogVerbose.
kLogError
Integer for error logging level, which is less than kLogWarn.
kLogFatal
Integer for fatal logging level, which is less than kLogError.
Logging anything at this level will result in a backtrace and
process exit.
SEE ALSO
https://justine.lol/redbean/index.html
https://news.ycombinator.com/item?id=26271117
Reference in a new issue View git blame Copy permalink