Skip to content

HTTPS clone URL

Subversion checkout URL

You can clone with
or
.
Download ZIP
Fetching contributors…

Cannot retrieve contributors at this time

3218 lines (2229 sloc) 133.359 kb
= Name =
ngx_lua - Embed the power of Lua into Nginx
''This module is not distributed with the Nginx source.'' See [[#Installation|the installation instructions]].
= Status =
This module is under active development and is production ready.
= Version =
This document describes ngx_lua [https://github.com/chaoslawful/lua-nginx-module/tags v0.3.1rc37] released on 29 November 2011.
= Synopsis =
<geshi lang="nginx">
# set search paths for pure Lua external libraries (';;' is the default path):
lua_package_path '/foo/bar/?.lua;/blah/?.lua;;';
# set search paths for Lua external libraries written in C (can also use ';;'):
lua_package_cpath '/bar/baz/?.so;/blah/blah/?.so;;';
server {
location /inline_concat {
# MIME type determined by default_type:
default_type 'text/plain';
set $a "hello";
set $b "world";
# inline lua script
set_by_lua $res "return ngx.arg[1]..ngx.arg[2]" $a $b;
echo $res;
}
location /rel_file_concat {
set $a "foo";
set $b "bar";
# script path relative to nginx prefix
# $ngx_prefix/conf/concat.lua contents:
#
# return ngx.arg[1]..ngx.arg[2]
#
set_by_lua_file $res conf/concat.lua $a $b;
echo $res;
}
location /abs_file_concat {
set $a "fee";
set $b "baz";
# absolute script path not modified
set_by_lua_file $res /usr/nginx/conf/concat.lua $a $b;
echo $res;
}
location /lua_content {
# MIME type determined by default_type:
default_type 'text/plain';
content_by_lua "ngx.say('Hello,world!')"
}
location /nginx_var {
# MIME type determined by default_type:
default_type 'text/plain';
# try access /nginx_var?a=hello,world
content_by_lua "ngx.print(ngx.var['arg_a'], '\\n')";
}
location /request_body {
# force reading request body (default off)
lua_need_request_body on;
client_max_body_size 50k;
client_body_buffer_size 50k;
content_by_lua 'ngx.print(ngx.var.request_body)';
}
# transparent non-blocking I/O in Lua via subrequests
location /lua {
# MIME type determined by default_type:
default_type 'text/plain';
content_by_lua '
local res = ngx.location.capture("/some_other_location")
if res.status == 200 then
ngx.print(res.body)
end';
}
# GET /recur?num=5
location /recur {
# MIME type determined by default_type:
default_type 'text/plain';
content_by_lua '
local num = tonumber(ngx.var.arg_num) or 0
if num > 50 then
ngx.say("num too big")
return
end
ngx.say("num is: ", num)
if num > 0 then
res = ngx.location.capture("/recur?num=" .. tostring(num - 1))
ngx.print("status=", res.status, " ")
ngx.print("body=", res.body)
else
ngx.say("end")
end
';
}
location /foo {
rewrite_by_lua '
res = ngx.location.capture("/memc",
{ args = { cmd = 'incr', key = ngx.var.uri } }
)
';
proxy_pass http://blah.blah.com;
}
location /blah {
access_by_lua '
local res = ngx.location.capture("/auth")
if res.status == ngx.HTTP_OK then
return
end
if res.status == ngx.HTTP_FORBIDDEN then
ngx.exit(res.status)
end
ngx.exit(ngx.HTTP_INTERNAL_SERVER_ERROR)
';
# proxy_pass/fastcgi_pass/postgres_pass/...
}
location /mixed {
rewrite_by_lua_file /path/to/rewrite.lua;
access_by_lua_file /path/to/access.lua;
content_by_lua_file /path/to/content.lua;
}
# use nginx var in code path
# WARN: contents in nginx var must be carefully filtered,
# otherwise there'll be great security risk!
location ~ ^/app/(.+) {
content_by_lua_file /path/to/lua/app/root/$1.lua;
}
location / {
lua_need_request_body on;
client_max_body_size 100k;
client_body_buffer_size 100k;
access_by_lua '
-- check the client IP addr is in our black list
if ngx.var.remote_addr == "132.5.72.3" then
ngx.exit(ngx.HTTP_FORBIDDEN)
end
-- check if the request body contains bad words
if ngx.var.request_body and
string.match(ngx.var.request_body, "fsck")
then
return ngx.redirect("/terms_of_use.html")
end
-- tests passed
';
# proxy_pass/fastcgi_pass/etc settings
}
}
</geshi>
= Description =
This module embeds the Lua interpreter or LuaJIT into the nginx core and integrates the powerful Lua threads (aka Lua coroutines) into the nginx event model
by means of nginx subrequests.
Unlike [http://httpd.apache.org/docs/2.3/mod/mod_lua.html Apache's mod_lua] and [http://redmine.lighttpd.net/wiki/1/Docs:ModMagnet Lighttpd's mod_magnet], Lua code written atop this module can be ''100% non-blocking'' on network traffic
as long as you use the [[#ngx.location.capture|ngx.location.capture]] or
[[#ngx.location.capture_multi|ngx.location.capture_multi]] interfaces
to let the nginx core do all your
requests to mysql, postgresql, memcached, redis,
upstream http web services, and etc etc etc (see
[[HttpDrizzleModule]], [http://github.com/FRiCKLE/ngx_postgres/ ngx_postgres], [[HttpMemcModule]], [[HttpRedis2Module]] and [[HttpProxyModule]] modules for details).
The Lua interpreter instance is shared across all
the requests in a single nginx worker process.
Request contexts are isolated from each other
by means of Lua (lightweight) threads (aka Lua coroutines).
And Lua modules loaded are persistent on
the nginx worker process level. So the memory
footprint is quite small even when your
nginx worker process is handling 10K requests at the same time.
= Directives =
== lua_code_cache ==
'''syntax:''' ''lua_code_cache on | off''
'''default:''' ''lua_code_cache on''
'''context:''' ''main, server, location, location if''
Enable or disable the Lua code cache for [[#set_by_lua_file|set_by_lua_file]],
[[#content_by_lua_file|content_by_lua_file]], [[#rewrite_by_lua_file|rewrite_by_lua_file]], and
[[#access_by_lua_file|access_by_lua_file]], and also force Lua module reloading on a per-request basis.
The Lua files referenced in [[#set_by_lua_file|set_by_lua_file]],
[[#content_by_lua_file|content_by_lua_file]], [[#access_by_lua_file|access_by_lua_file]],
and [[#rewrite_by_lua_file|rewrite_by_lua_file]] will not be cached at all,
and Lua's <code>package.loaded</code> table will be cleared
at the entry point of every request (such that Lua modules
will not be cached either). With this in place, developers can follow
the PHP way, i.e., edit-and-refresh.
Please note however, that Lua code inlined into nginx.conf
such as those specified by [[#set_by_lua|set_by_lua]], [[#content_by_lua|content_by_lua]],
[[#access_by_lua|access_by_lua]], and [[#rewrite_by_lua|rewrite_by_lua]] will ''always'' be
cached because only the Nginx config file parser can correctly parse the <code>nginx.conf</code>
file and the only ways to to reload the config file
are to send a <code>HUP</code> signal or to restart Nginx.
The ngx_lua module currently does not support the "stat" mode like
Apache's <code>mod_lua</code> does but this is planned for implementation in the future.
Disabling the Lua code cache is strongly
discouraged for production use and should only be used during
development as it has a significant impact on overall performance.
In addition, race conditions when reloading Lua modules are common for concurrent requests
when the code cache is disabled.
== lua_regex_cache_max_entries ==
'''syntax:''' ''lua_regex_cache_max_entries <num>''
'''default:''' ''lua_regex_cache_max_entries 1024''
'''context:''' ''http''
Specifies the maximum number of entries allowed in the worker process level compiled regex cache.
The regular expressions used in [[#ngx.re.match|ngx.re.match]], [[#ngx.re.gmatch|ngx.re.gmatch]], [[#ngx.re.sub|ngx.re.sub]], and [[#ngx.re.gsub|ngx.re.gsub]] will be cached within this cache if the regex option <code>o</code> (i.e., compile-once flag) is specified.
The default number of entries allowed is 1024 and when this limit is reached, new regexes will not be cached (as if the <code>o</code> option was not specified) and there will be one, and only one, warning in the <code>error.log</code> file:
<geshi lang="text">
2011/08/27 23:18:26 [warn] 31997#0: *1 lua exceeding regex cache max entries (1024), ...
</geshi>
You should not activate the <code>o</code> option for regexes (and/or <code>replace</code> string arguments for [[#ngx.re.sub|ngx.re.sub]] and [[#ngx.re.gsub|ngx.re.gsub]]) that are generated ''on the fly'' and give rise to infinite variations to avoid hitting the specified limit.
== lua_package_path ==
'''syntax:''' ''lua_package_path <lua-style-path-str>''
'''default:''' ''The content of LUA_PATH environ variable or Lua's compiled-in defaults.''
'''context:''' ''main''
Sets the Lua module search path used by scripts specified by [[#set_by_lua|set_by_lua]],
[[#content_by_lua|content_by_lua]] and others. The path string is in standard Lua path form, and <code>;;</code>
can be used to stand for the original path.
== lua_package_cpath ==
'''syntax:''' ''lua_package_cpath <lua-style-cpath-str>''
'''default:''' ''The content of LUA_CPATH environ variable or Lua's compiled-in defaults.''
'''context:''' ''main''
Sets the Lua C-module search path used by scripts specified by [[#set_by_lua|set_by_lua]],
[[#content_by_lua|content_by_lua]] and others. The cpath string is in standard Lua cpath form, and <code>;;</code>
can be used to stand for the original cpath.
== set_by_lua ==
'''syntax:''' ''set_by_lua $res <lua-script-str> [$arg1 $arg2 ...]''
'''context:''' ''main, server, location, server if, location if''
'''phase:''' ''rewrite''
Execute user code specified by <code><lua-script-str></code> with input arguments <code>$arg1 $arg2 ...</code>, and set the script's return value to <code>$res</code> in string form. In
<code><lua-script-str></code> code the input arguments can be retrieved from <code>ngx.arg</code>
table (index starts from <code>1</code> and increased sequentially).
[[#set_by_lua|set_by_lua]] directive is designed to execute short, fast running code blocks. The Nginx
event loop is blocked during code execution and time consuming code sequences should be avoided.
Note that [[#set_by_lua|set_by_lua]] can only output a value to a single Nginx variable at
a time but a workaround is possible by using the [[#ngx.var.VARIABLE|ngx.var.VARIABLE]] interface.
For example,
<geshi lang="nginx">
location /foo {
set $diff ''; # we have to predefine the $diff variable here
set_by_lua $sum '
local a = 32
local b = 56
ngx.var.diff = a - b; -- write to $diff directly
return a + b; -- return the $sum value normally
';
echo "sum = $sum, diff = $diff";
}
</geshi>
This directive can be freely mixed with all the directives of [[HttpRewriteModule]], [[HttpSetMiscModule]], and [[HttpArrayVarModule]]. All of these directives will run in exactly the same order that they are written in the config file. For example,
<geshi lang="nginx">
set $foo 32;
set_by_lua $bar 'tonumber(ngx.var.foo) + 1';
set $baz "bar: $bar"; # $baz == "bar: 33"
</geshi>
This directive requires the [https://github.com/simpl/ngx_devel_kit ngx_devel_kit] module.
== set_by_lua_file ==
'''syntax:''' ''set_by_lua_file $res <path-to-lua-script> [$arg1 $arg2 ...]''
'''context:''' ''main, server, location, server if, location if''
'''phase:''' ''rewrite''
Basically the same as [[#set_by_lua|set_by_lua]], except the code to be executed is in the
file specified by <code><path-lua-script></code>.
When the Lua code cache is on (default state), the user code is loaded once at the first request and cached
and the Nginx config must be reloaded each time you modify the lua code file.
You can temporarily disable the Lua code cache during development by
switching [[#lua_code_cache|lua_code_cache]] <code>off</code> in your <code>nginx.conf</code> to avoid reloading Nginx.
This directive requires the [https://github.com/simpl/ngx_devel_kit ngx_devel_kit] module.
== content_by_lua ==
'''syntax:''' ''content_by_lua <lua-script-str>''
'''context:''' ''location, location if''
'''phase:''' ''content''
Acts as a "content handler" and executes lua code string specified in <code><lua-script-str></code> for every request.
The lua code may call predefined APIs to generate response content and is executed in a new spawned coroutine in an independent global environment (i.e. a sandbox).
Do not use this directive and other content handler directives in a same location. For example, this directive and the [[HttpProxyModule#proxy_pass|proxy_pass]] directive should not be used in the same location.
== content_by_lua_file ==
'''syntax:''' ''content_by_lua_file <path-to-lua-script>''
'''context:''' ''location, location if''
'''phase:''' ''content''
Same as [[#content_by_lua|content_by_lua]], except the code to be executed is in the file
specified by <code><path-lua-script></code>.
Nginx variables can be used in <code><path-to-lua-script></code> string, in order to provide
greater flexibility in practice. This however carries some risks and is not ordinarily recommended.
When the Lua code cache is on (default state), the user code is loaded once at the first request and cached
and the Nginx config must be reloaded each time you modify the lua code file.
You can temporarily disable the Lua code cache during development by
switching [[#lua_code_cache|lua_code_cache]] <code>off</code> in your <code>nginx.conf</code> to avoid reloading Nginx.
== rewrite_by_lua ==
'''syntax:''' ''rewrite_by_lua <lua-script-str>''
'''context:''' ''http, server, location, location if''
'''phase:''' ''post-rewrite''
Acts as a rewrite phase handler and executes lua code string specified in <code><lua-script-str></code> for every request. The lua code may call predefined APIs to generate response content and is executed in a new spawned coroutine in an independent global environment (i.e. a sandbox).
This directive uses exactly the same mechamism as [[#content_by_lua|content_by_lua]] so all the nginx APIs defined there are also available here.
Note that this handler always runs ''after'' the standard [[HttpRewriteModule]]. So the following will work as expected:
<geshi lang="nginx">
location /foo {
set $a 12; # create and initialize $a
set $b ""; # create and initialize $b
rewrite_by_lua 'ngx.var.b = tonumber(ngx.var.a) + 1';
echo "res = $b";
}
</geshi>
because <code>set $a 12</code> and <code>set $b ""</code> run ''before'' [[#rewrite_by_lua|rewrite_by_lua]].
On the other hand, the following will not work as expected:
<geshi lang="nginx">
? location /foo {
? set $a 12; # create and initialize $a
? set $b ''; # create and initialize $b
? rewrite_by_lua 'ngx.var.b = tonumber(ngx.var.a) + 1';
? if ($b = '13') {
? rewrite ^ /bar redirect;
? break;
? }
?
? echo "res = $b";
? }
</geshi>
because <code>if</code> runs ''before'' [[#rewrite_by_lua|rewrite_by_lua]] even if it is put after [[#rewrite_by_lua|rewrite_by_lua]] in the config.
The right way of doing this is as follows:
<geshi lang="nginx">
location /foo {
set $a 12; # create and initialize $a
set $b ''; # create and initialize $b
rewrite_by_lua '
ngx.var.b = tonumber(ngx.var.a) + 1
if tonumber(ngx.var.b) == 13 then
return ngx.redirect("/bar");
end
';
echo "res = $b";
}
</geshi>
It is worth mentioning that, the <code>ngx_eval</code> module can be approximately implemented by [[#rewrite_by_lua|rewrite_by_lua]]. For example,
<geshi lang="nginx">
location / {
eval $res {
proxy_pass http://foo.com/check-spam;
}
if ($res = 'spam') {
rewrite ^ /terms-of-use.html redirect;
}
fastcgi_pass ...;
}
</geshi>
can be implemented in terms of <code>ngx_lua</code> like this
<geshi lang="nginx">
location = /check-spam {
internal;
proxy_pass http://foo.com/check-spam;
}
location / {
rewrite_by_lua '
local res = ngx.location.capture("/check-spam")
if res.body == "spam" then
ngx.redirect("/terms-of-use.html")
end
';
fastcgi_pass ...;
}
</geshi>
Just as any other rewrite phase handlers, [[#rewrite_by_lua|rewrite_by_lua]] also runs in subrequests.
Note that when calling <code>ngx.exit(ngx.OK)</code> within a [[#rewrite_by_lua|rewrite_by_lua]] handler, the nginx request processing control flow will still continue to the content handler. To terminate the current request from within a [[#rewrite_by_lua|rewrite_by_lua]] handler, calling [[#ngx.exit|ngx.exit]] with status >= 200 (<code>ngx.HTTP_OK</code>) and status < 300 (<code>ngx.HTTP_SPECIAL_RESPONSE</code>) for successful quits and <code>ngx.exit(ngx.HTTP_INTERNAL_SERVER_ERROR)</code> (or its friends) for failures.
If the [[HttpRewriteModule]]'s [[HttpRewriteModule#rewrite|rewrite]] directive is used to change the URI and initiate location re-lookups (internal redirections), then any [[#rewrite_by_lua|rewrite_by_lua]] or [[#rewrite_by_lua_file|rewrite_by_lua_file]] code sequences within the current location will not be executed. For example,
<geshi lang="nginx">
location /foo {
rewrite ^ /bar;
rewrite_by_lua 'ngx.exit(503)';
}
location /bar {
...
}
</geshi>
Here the Lua code <code>ngx.exit(503)</code> will never run. This will be the case if <code>rewrite ^ /bar last</code> is used as this will similarly initiate an internal redirection. If the <code>break</code> modifier is used instead, there will be no internal rediction and the rewrite_by_lua code will be executed.
== rewrite_by_lua_file ==
'''syntax:''' ''rewrite_by_lua_file <path-to-lua-script>''
'''context:''' ''http, server, location, location if''
'''phase:''' ''post-rewrite''
Same as [[#rewrite_by_lua|rewrite_by_lua]], except the code to be executed is in the file specified by <code><path-lua-script></code>.
Nginx variables can be used in <code><path-to-lua-script></code> string, in order to provide greater flexibility in practice. This however carries some risks and is not ordinarily recommended.
When the Lua code cache is on (default state), the user code is loaded once at the first request and cached and the Nginx config must be reloaded each time you modify the Lua source file. You can temporarily disable the Lua code cache during development by switching [[#lua_code_cache|lua_code_cache]] <code>off</code> in your <code>nginx.conf</code> to avoid reloading Nginx.
== access_by_lua ==
'''syntax:''' ''access_by_lua <lua-script-str>''
'''context:''' ''http, server, location, location if''
'''phase:''' ''post-access''
Acts as an access phase handler and executes lua code string specified in <code><lua-script-str></code> for every request. The user code may call predefined APIs to generate response content.
This directive uses exactly the same mechanism as [[#content_by_lua|content_by_lua]] so all the nginx APIs defined there are also available here.
Note that this handler always runs ''after'' the standard [[HttpAccessModule]]. So the following will work as expected:
<geshi lang="nginx">
location / {
deny 192.168.1.1;
allow 192.168.1.0/24;
allow 10.1.1.0/16;
deny all;
access_by_lua '
local res = ngx.location.capture("/mysql", { ... })
...
';
# proxy_pass/fastcgi_pass/...
}
</geshi>
That is, if a client address appears in the blacklist, then we do not have to bother sending a MySQL query to do more advanced authentication in [[#access_by_lua|access_by_lua]].
It is worth mentioning that, the <code>ngx_auth_request</code> module can be approximately implemented by [[#access_by_lua|access_by_lua]]. For example,
<geshi lang="nginx">
location / {
auth_request /auth;
# proxy_pass/fastcgi_pass/postgres_pass/...
}
</geshi>
can be implemented in terms of <code>ngx_lua</code> like this
<geshi lang="nginx">
location / {
access_by_lua '
local res = ngx.location.capture("/auth")
if res.status == ngx.HTTP_OK then
return
end
if res.status == ngx.HTTP_FORBIDDEN then
ngx.exit(res.status)
end
ngx.exit(ngx.HTTP_INTERNAL_SERVER_ERROR)
';
# proxy_pass/fastcgi_pass/postgres_pass/...
}
</geshi>
Just as any other access phase handlers, [[#access_by_lua|access_by_lua]] will ''not'' run in subrequests.
Note that when calling <code>ngx.exit(ngx.OK)</code> within a [[#access_by_lua|access_by_lua]] handler, the nginx request processing control flow will still continue to the content handler. To terminate the current request from within a [[#access_by_lua|access_by_lua]] handler, calling [[#ngx.exit|ngx.exit]] with status >= 200 (<code>ngx.HTTP_OK</code>) and status < 300 (<code>ngx.HTTP_SPECIAL_RESPONSE</code>) for successful quits and <code>ngx.exit(ngx.HTTP_INTERNAL_SERVER_ERROR)</code> (or its friends) for failures.
== access_by_lua_file ==
'''syntax:''' ''access_by_lua_file <path-to-lua-script>''
'''context:''' ''http, server, location, location if''
'''phase:''' ''post-access''
Same as [[#access_by_lua|access_by_lua]], except the code to be executed is in the file
specified by <code><path-lua-script></code>.
Nginx variables can be used in <code><path-to-lua-script></code> string, in order to provide
greater flexibility in practice. This however carries some risks and is not ordinarily recommended.
When the Lua code cache is on (default state), the user code is loaded once at the first request and cached
and the Nginx config must be reloaded each time you modify the lua code file.
You can temporarily disable the Lua code cache during development by
switching [[#lua_code_cache|lua_code_cache]] <code>off</code> in your <code>nginx.conf</code> to avoid reloading Nginx.
== header_filter_by_lua ==
'''syntax:''' ''header_filter_by_lua <lua-script-str>''
'''context:''' ''http, server, location, location if''
'''phase:''' ''output-header-filter''
Uses Lua code specified in <code><lua-script-str></code> to define an output header filter. Note that the following Nginx Lua APIs are currently disabled within this context:
* Output API (e.g., [[#ngx.say|ngx.say]] and [[#ngx.send_headers|ngx.send_headers]])
* Control APIs (e.g., [[#ngx.exit|ngx.exit]])
* Subrequest APIs (e.g., [[#ngx.location.capture|ngx.location.capture]] and [[#ngx.location.capture_multi|ngx.location.capture_multi]])
Here is an example of overriding a response header (or adding one if absent) in our Lua header filter:
<geshi lang="nginx">
location / {
proxy_pass http://mybackend;
header_filter_by_lua 'ngx.header.Foo = "blah"';
}
</geshi>
This directive was first introduced in the <code>v0.2.1rc20</code> release.
== header_filter_by_lua_file ==
'''syntax:''' ''header_filter_by_lua_file <path-to-lua-script-file>''
'''context:''' ''http, server, location, location if''
'''phase:''' ''output-header-filter''
Use Lua code defined in a separate file specified by <code><path-to-lua-script-file></code> to define an output header filter.
This is very much like [[#header_filter_by_lua|header_filter_by_lua]] except that it loads Lua code from an external Lua source file.
This directive was first introduced in the <code>v0.2.1rc20</code> release.
== lua_need_request_body ==
'''syntax:''' ''lua_need_request_body <on | off>''
'''default:''' ''off''
'''context:''' ''main | server | location''
'''phase:''' ''depends on usage''
Determines whether to force reading of the request body data before running rewrite/access/access_by_lua* or not. The Nginx core does not read the client request body by default and if you need the request body then you must turn this directive <code>on</code> or call the [[#ngx.req.read_body|ngx.req.read_body]] function from within your Lua code.
To read the request body data within the [[HttpCoreModule#$request_body|$request_body]] variable,
[[HttpCoreModule#client_body_buffer_size|client_body_buffer_size]] must have the same value as [[HttpCoreModule#client_max_body_size|client_max_body_size]]. Because when the content length exceeds [[HttpCoreModule#client_body_buffer_size|client_body_buffer_size]] but less than [[HttpCoreModule#client_max_body_size|client_max_body_size]], Nginx will automatically buffer the data into a temporary file on the disk, which will lead to empty value in the [[HttpCoreModule#$request_body|$request_body]] variable.
If the current location includes [[#rewrite_by_lua|rewrite_by_lua]] or [[#rewrite_by_lua_file|rewrite_by_lua_file]] directives,
then the request body will be read just before the [[#rewrite_by_lua|rewrite_by_lua]] or [[#rewrite_by_lua_file|rewrite_by_lua_file]] code is run (and also at the
<code>rewrite</code> phase). Similarly, if only [[#content_by_lua|content_by_lua]] is specified,
the request body will not be read until the content handler's Lua code is
about to run (i.e., the request body will be read during the content phase).
It is recommended however, to use the [[#ngx.req.read_body|ngx.req.read_body]] and [[#ngx.req.discard_body|ngx.req.discard_body]] functions for finer control over the request body reading process instead.
This also applies to [[#access_by_lua|access_by_lua]] and [[#access_by_lua_file|access_by_lua_file]].
== lua_shared_dict ==
'''syntax:''' ''lua_shared_dict <name> <size>''
'''default:''' ''no''
'''context:''' ''main''
'''phase:''' ''depends on usage''
Declares a shared memory zone, <code><name></code>, to serve as storage for the shm-based Lua dictionary <code>ngx.shared.<name></code>.
The <code><size></code> argument accepts size units such as <code>k</code> and <code>m</code>:
<geshi lang="nginx">
http {
lua_shared_dict dogs 10m;
...
}
</geshi>
See [[#ngx.shared.DICT|ngx.shared.DICT]] for details.
This directive was first introduced in the <code>v0.3.1rc22</code> release.
= Nginx API for Lua =
== General Considerations ==
The Nginx API exposed to the Lua land is provided in the form of two standard packages <code>ngx</code> and <code>ndk</code>. These packages are in the default global scope within <code>ngx_lua</code> and can also be introduced to external Lua modules by using the [http://www.lua.org/manual/5.1/manual.html#pdf-package.seeall package.seeall] option:
<geshi lang="lua">
module("my_module", package.seeall)
function say(a) ngx.say(a) end
</geshi>
Alternatively, they can be imported external Lua modules by using file scoped local Lua variables:
<geshi lang="lua">
local ngx = ngx
module("my_module")
function say(a) ngx.say(a) end
</geshi>
It is also possible to directly require the packages:
<geshi lang="lua">
local ngx = require "ngx"
local ndk = require "ndk"
</geshi>
The ability to require these packages was introduced in the <code>v0.2.1rc19</code> release.
Network I/O operations in user code should only be done through Nginx API calls as the Nginx event loop may be blocked and performance may drop off dramatically otherwise. Minor disk file operations may be done via Lua's standard <code>io</code> and <code>file</code> libraries but these should be avoided wherever possible as these also block the Nginx process. Delegating all network and disk I/O operations to Nginx subrequests (via the [[#ngx.location.catpure|ngx.location.capture]] method and its friends) is strongly recommended for maximum performance.
== ngx.arg ==
'''syntax:''' ''val = ngx.arg[index]''
'''context:''' ''set_by_lua*''
Index the input arguments to the [[#set_by_lua|set_by_lua]] and [[#set_by_lua_file|set_by_lua_file]] directives:
<geshi lang="lua">
value = ngx.arg[n]
</geshi>
Here is an example
<geshi lang="nginx">
location /foo {
set $a 32;
set $b 56;
set_by_lua $res
'return tonumber(ngx.arg[1]) + tonumber(ngx.arg[2])'
$a $b;
echo $sum;
}
</geshi>
that outputs <code>88</code>, the sum of <code>32</code> and <code>56</code>.
== ngx.var.VARIABLE ==
'''syntax:''' ''ngx.var.VAR_NAME''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
<geshi lang="nginx">
value = ngx.var.some_nginx_variable_name
ngx.var.some_nginx_variable_name = value
</geshi>
Note that you can only write to nginx variables that are already defined.
For example:
<geshi lang="nginx">
location /foo {
set $my_var ''; # this line is required to create $my_var at config time
content_by_lua '
ngx.var.my_var = 123;
...
';
}
</geshi>
That is, nginx variables cannot be created on-the-fly.
Some special nginx variables like <code>$args</code> and <code>$limit_rate</code> can be assigned a value,
some are not, like <code>$arg_PARAMETER</code>.
Nginx regex group capturing variables <code>$1</code>, <code>$2</code>, <code>$3</code>, and etc, can be read by this
interface as well, by writing <code>ngx.var[1]</code>, <code>ngx.var[2]</code>, <code>ngx.var[3]</code>, and etc.
Setting <code>nil</code> values to <code>ngx.var.Foo</code> will effectively make Nginx variable <code>$Foo</code> undefined. For instance,
<geshi lang="lua">
ngx.var.args = nil
</geshi>
== Core constants ==
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
<geshi lang="lua">
ngx.OK (0)
ngx.ERROR (-1)
ngx.AGAIN (-2)
ngx.DONE (-4)
</geshi>
They take the same values of <code>NGX_OK</code>, <code>NGX_AGAIN</code>, <code>NGX_DONE</code>, <code>NGX_ERROR</code>, and etc. But now
only [[#ngx.exit|ngx.exit]] only take two of these values, i.e., <code>NGX_OK</code> and <code>NGX_ERROR</code>.
== HTTP method constants ==
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
<geshi lang="lua">
ngx.HTTP_GET
ngx.HTTP_HEAD
ngx.HTTP_PUT
ngx.HTTP_POST
ngx.HTTP_DELETE
</geshi>
These constants are usually used in [[#ngx.location.capture|ngx.location.catpure]] and [[#ngx.location.capture_multi|ngx.location.capture_multi]] method calls.
== HTTP status constants ==
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
<geshi lang="nginx">
value = ngx.HTTP_OK (200)
value = ngx.HTTP_CREATED (201)
value = ngx.HTTP_SPECIAL_RESPONSE (300)
value = ngx.HTTP_MOVED_PERMANENTLY (301)
value = ngx.HTTP_MOVED_TEMPORARILY (302)
value = ngx.HTTP_SEE_OTHER (303)
value = ngx.HTTP_NOT_MODIFIED (304)
value = ngx.HTTP_BAD_REQUEST (400)
value = ngx.HTTP_UNAUTHORIZED (401)
value = ngx.HTTP_FORBIDDEN (403)
value = ngx.HTTP_NOT_FOUND (404)
value = ngx.HTTP_NOT_ALLOWED (405)
value = ngx.HTTP_GONE (410)
value = ngx.HTTP_INTERNAL_SERVER_ERROR (500)
value = ngx.HTTP_METHOD_NOT_IMPLEMENTED (501)
value = ngx.HTTP_SERVICE_UNAVAILABLE (503)
</geshi>
== Nginx log level constants ==
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
<geshi lang="lua">
ngx.STDERR
ngx.EMERG
ngx.ALERT
ngx.CRIT
ngx.ERR
ngx.WARN
ngx.NOTICE
ngx.INFO
ngx.DEBUG
</geshi>
These constants are usually used by the [[#ngx.log|ngx.log]] method.
== print ==
'''syntax:''' ''print(...)''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Emit args concatenated to nginx's <code>error.log</code> file, with log level <code>ngx.NOTICE</code> and prefix <code>lua print: </code>.
It is equivalent to
<geshi lang="lua">
ngx.log(ngx.NOTICE, 'lua print: ', a, b, ...)
</geshi>
Lua <code>nil</code> arguments are accepted and result in literal <code>"nil"</code>, and Lua booleans result in <code>"true"</code> or <code>"false"</code>.
== ngx.ctx ==
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
This table can be used to store per-request context data for Lua programmers.
This table has a liftime identical to the current request (just like Nginx variables). Consider the following example,
<geshi lang="nginx">
location /test {
rewrite_by_lua '
ngx.say("foo = ", ngx.ctx.foo)
ngx.ctx.foo = 76
';
access_by_lua '
ngx.ctx.foo = ngx.ctx.foo + 3
';
content_by_lua '
ngx.say(ngx.ctx.foo)
';
}
</geshi>
Then <code>GET /test</code> will yield the output
<geshi lang="bash">
foo = nil
79
</geshi>
That is, the <code>ngx.ctx.foo</code> entry persists across the rewrite, access, and content phases of a request.
Also, every request has its own copy, include subrequests, for example:
<geshi lang="nginx">
location /sub {
content_by_lua '
ngx.say("sub pre: ", ngx.ctx.blah)
ngx.ctx.blah = 32
ngx.say("sub post: ", ngx.ctx.blah)
';
}
location /main {
content_by_lua '
ngx.ctx.blah = 73
ngx.say("main pre: ", ngx.ctx.blah)
local res = ngx.location.capture("/sub")
ngx.print(res.body)
ngx.say("main post: ", ngx.ctx.blah)
';
}
</geshi>
Then <code>GET /main</code> will give the output
<geshi lang="bash">
main pre: 73
sub pre: nil
sub post: 32
main post: 73
</geshi>
We can see that modification of the <code>ngx.ctx.blah</code> entry in the subrequest does not affect the one in its parent request. They do have two separate versions of <code>ngx.ctx.blah</code> per se.
Internal redirection will destroy the original request's <code>ngx.ctx</code> data (if any) and the new request will have an emptied <code>ngx.ctx</code> table. For instance,
<geshi lang="nginx">
location /new {
content_by_lua '
ngx.say(ngx.ctx.foo)
';
}
location /orig {
content_by_lua '
ngx.ctx.foo = "hello"
ngx.exec("/new")
';
}
</geshi>
Then <code>GET /orig</code> will give you
<geshi lang="bash">
nil
</geshi>
rather than the original <code>"hello"</code> value.
Arbitrary data values can be inserted into this "matic" table, including Lua closures and nested tables. You can also register your own meta methods with it.
Overriding <code>ngx.ctx</code> with a new Lua table is also supported, for example,
<geshi lang="lua">
ngx.ctx = { foo = 32, bar = 54 }
</geshi>
== ngx.location.capture ==
'''syntax:''' ''res = ngx.location.capture(uri, options?)''
'''context:''' ''rewrite_by_lua*, access_by_lua*, content_by_lua*''
Issue a synchronous but still non-blocking ''Nginx Subrequest'' using <code>uri</code>.
Nginx subrequests provide a powerful way to make non-blocking internal requests to other locations configured with disk file directory or ''any'' other nginx C modules like <code>ngx_proxy</code>, <code>ngx_fastcgi</code>, <code>ngx_memc</code>,
<code>ngx_postgres</code>, <code>ngx_drizzle</code>, and even <code>ngx_lua</code> itself and etc etc etc.
Also note that subrequests just mimic the HTTP interface but there is ''no'' extra HTTP/TCP traffic ''nor'' IPC involved. Everything works internally, efficiently, on the C level.
Subrequests are completely different from HTTP 301/302 redirection (via [[#ngx.redirect|ngx.redirect]]) and internal redirection (via [[#ngx.exec|ngx.exec]]).
Here is a basic example:
<geshi lang="lua">
res = ngx.location.capture(uri)
</geshi>
Returns a Lua table with three slots (<code>res.status</code>, <code>res.header</code>, and <code>res.body</code>).
<code>res.header</code> holds all the response headers of the
subrequest and it is a normal Lua table. For multi-value response headers,
the value is a Lua (array) table that holds all the values in the order that
they appear. For instance, if the subrequest response headers contains the following
lines:
<geshi lang="bash">
Set-Cookie: a=3
Set-Cookie: foo=bar
Set-Cookie: baz=blah
</geshi>
Then <code>res.header["Set-Cookie"]</code> will be evaluted to the table value
<code>{"a=3", "foo=bar", "baz=blah"}</code>.
URI query strings can be concatenated to URI itself, for instance,
<geshi lang="lua">
res = ngx.location.capture('/foo/bar?a=3&b=4')
</geshi>
Named locations like <code>@foo</code> are not allowed due to a limitation in
the nginx core. Use normal locations combined with the <code>internal</code> directive to
prepare internal-only locations.
An optional option table can be fed as the second
argument, which support the options:
* <code>method</code>
: specify the subrequest's request method, which only accepts constants like <code>ngx.HTTP_POST</code>.
* <code>body</code>
: specify the subrequest's request body (string value only).
* <code>args</code>
: specify the subrequest's URI query arguments (both string value and Lua tables are accepted)
* <code>ctx</code>
: specify a Lua table to be the [[#ngx.ctx|ngx.ctx]] table for the subrequest. It can be the current request's [[#ngx.ctx|ngx.ctx]] table, which effectively make the parent and its subrequest to share exactly the same context table. This option was first introduced in the <code>v0.3.1rc25</code> release.
* <code>vars</code>
: take a Lua table which holds the values to set the specified Nginx variables in the subrequest as this option's value. This option was first introduced in the <code>v0.3.1rc31</code> release.
* <code>copy_all_vars</code>
: specify whether to copy over all the Nginx variable values of the current request to the subrequest in question. modifications of the nginx variables in the subrequest will not affect the current (parent) request. This option was first introduced in the <code>v0.3.1rc31</code> release.
* <code>share_all_vars</code>
: specify whether to share all the Nginx variables of the subrequest with the current (parent) request. modifications of the Nginx variables in the subrequest will affect the current (parent) request.
Issuing a POST subrequest, for example, can be done as follows
<geshi lang="lua">
res = ngx.location.capture(
'/foo/bar',
{ method = ngx.HTTP_POST, body = 'hello, world' }
)
</geshi>
See HTTP method constants methods other than POST.
The <code>method</code> option is <code>ngx.HTTP_GET</code> by default.
The <code>args</code> option can specify extra URI arguments, for instance,
<geshi lang="lua">
ngx.location.capture('/foo?a=1',
{ args = { b = 3, c = ':' } }
)
</geshi>
is equivalent to
<geshi lang="lua">
ngx.location.capture('/foo?a=1&b=3&c=%3a')
</geshi>
that is, this method will automatically escape argument keys and values according to URI rules and
concatenating them together into a complete query string. The format for the Lua table passed as the <code>args</code> argument is identical to the format used in the [[#ngx.encode_args|ngx.encode_args]] method.
The <code>args</code> option can also take plain query strings:
<geshi lang="lua">
ngx.location.capture('/foo?a=1',
{ args = 'b=3&c=%3a' } }
)
</geshi>
This is functionally identical to the previous examples.
The <code>share_all_vars</code> option controls whether to share nginx variables among the current request and new subrequests.
If this option is set to <code>true</code>, then the current request and associated subrequests will share the same Nginx variable scope. Hence, changes to Nginx variables made by a subrequest will affect the current request.
Care should be taken in using this option as variable scope sharing can have unexpected side effects. The <code>args</code>, <code>vars</code>, or <code>copy_all_vars</code> options are generally preferable instead.
This option is set to <code>false</code> by default
<geshi lang="nginx">
location /other {
set $dog "$dog world";
echo "$uri dog: $dog";
}
location /lua {
set $dog 'hello';
content_by_lua '
res = ngx.location.capture("/other",
{ share_all_vars = true });
ngx.print(res.body)
ngx.say(ngx.var.uri, ": ", ngx.var.dog)
';
}
</geshi>
Accessing location <code>/lua</code> gives
<geshi lang="text">
/other dog: hello world
/lua: hello world
</geshi>
The <code>copy_all_vars</code> option provides a copy of the parent request's Nginx variables to subrequests when such subrequests are issued. Changes made to these variables by such subrequests will not affect the parent request or any other subrequests sharing the parent request's variables.
<geshi lang="nginx">
location /other {
set $dog "$dog world";
echo "$uri dog: $dog";
}
location /lua {
set $dog 'hello';
content_by_lua '
res = ngx.location.capture("/other",
{ copy_all_vars = true });
ngx.print(res.body)
ngx.say(ngx.var.uri, ": ", ngx.var.dog)
';
}
</geshi>
Request <code>GET /lua</code> will give the output
<geshi lang="text">
/other dog: hello world
/lua: hello
</geshi>
Note that if both <code>share_all_vars</code> and <code>copy_all_vars</code> are set to true,
then the variable scope will be shared.
In addition to the two settings above, it is possible to specify
values for variables in the subrequest using the <code>vars</code> option. These
variables are set after the sharing or copying of variables has been
evaluated, and provides a more efficient method of passing specific
values to a subrequest over encoding them as URL arguments and
unescaping them in the Nginx config file.
<geshi lang="nginx">
location /other {
content_by_lua '
ngx.say("dog = ", ngx.var.dog)
ngx.say("cat = ", ngx.var.cat)
';
}
location /lua {
set $dog '';
set $cat '';
content_by_lua '
res = ngx.location.capture("/other",
{ vars = { dog = "hello", cat = 32 }});
ngx.print(res.body)
';
}
</geshi>
Accessing <code>/lua</code> will yield the output
<geshi lang="text">
dog = hello
cat = 32
</geshi>
The <code>ctx</code> option can be used to specify a custom Lua table to serve as the [[#ngx.ctx|ngx.ctx]] table for the subrequest.
<geshi lang="nginx">
location /sub {
content_by_lua '
ngx.ctx.foo = "bar";
';
}
location /lua {
content_by_lua '
local ctx = {}
res = ngx.location.capture("/sub", { ctx = ctx })
ngx.say(ctx.foo);
ngx.say(ngx.ctx.foo);
';
}
</geshi>
Then request <code>GET /lua</code> gives
<geshi lang="text">
bar
nil
</geshi>
It is also possible to use this <code>ctx</code> option to share the same [[#ngx.ctx|ngx.ctx]] table between the current (parent) request and the subrequest:
<geshi lang="nginx">
location /sub {
content_by_lua '
ngx.ctx.foo = "bar";
';
}
location /lua {
content_by_lua '
res = ngx.location.capture("/sub", { ctx = ngx.ctx })
ngx.say(ngx.ctx.foo);
';
}
</geshi>
Request <code>GET /lua</code> yields the output
<geshi lang="text">
bar
</geshi>
Note that, by default, subrequests issued by [[#ngx.location.capture|ngx.location.capture]] inherit all the
request headers of the current request. This may have unexpected side effects on the
subrequest responses. For example, when you are using the standard <code>ngx_proxy</code> module to serve
your subrequests, then an "Accept-Encoding: gzip" header in your main request may result
in gzipped responses that your Lua code will not able to handle properly. So always set
[[HttpProxyModule#proxy_pass_request_headers|proxy_pass_request_headers]] <code>off</code> in your subrequest location to ignore the original request headers.
Please also refer to restrictions on [[#Locations_With_HttpEchoModule_Directives|capturing locations that include Echo Module directives]].
== ngx.location.capture_multi ==
'''syntax:''' ''res1, res2, ... = ngx.location.capture_multi({ {uri, options?}, {uri, options?}, ... })''
'''context:''' ''rewrite_by_lua*, access_by_lua*, content_by_lua*''
Just like [[#ngx.location.capture|ngx.location.capture]], but supports multiple subrequests running in parallel.
This function issue several parallel subrequests specified by the input table, and returns their results in the same order. For example,
<geshi lang="lua">
res1, res2, res3 = ngx.location.capture_multi{
{ "/foo", { args = "a=3&b=4" } },
{ "/bar" },
{ "/baz", { method = ngx.HTTP_POST, body = "hello" } },
}
if res1.status == ngx.HTTP_OK then
...
end
if res2.body == "BLAH" then
...
end
</geshi>
This function will not return until all the subrequests terminate.
The total latency is the longest latency of the subrequests, instead of their sum.
When you do not know how many subrequests you want to issue in advance,
you can use Lua tables for both requests and responses. For instance,
<geshi lang="lua">
-- construct the requests table
local reqs = {}
table.insert(reqs, { "/mysql" })
table.insert(reqs, { "/postgres" })
table.insert(reqs, { "/redis" })
table.insert(reqs, { "/memcached" })
-- issue all the requests at once and wait until they all return
local resps = { ngx.location.capture_multi(reqs) }
-- loop over the responses table
for i, resp in ipairs(resps) do
-- process the response table "resp"
end
</geshi>
The [[#ngx.location.capture|ngx.location.capture]] function is just a special form
of this function. Logically speaking, the [[#ngx.location.capture|ngx.location.capture]] can be implemented like this
<geshi lang="lua">
ngx.location.capture =
function (uri, args)
return ngx.location.capture_multi({ {uri, args} })
end
</geshi>
Please also refer to restrictions on [[#Locations_With_HttpEchoModule_Directives|capturing locations that include Echo Module directives]].
== ngx.status ==
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Read and write the current request's response status. This should be called
before sending out the response headers.
<geshi lang="lua">
ngx.status = ngx.HTTP_CREATED
status = ngx.status
</geshi>
== ngx.header.HEADER ==
'''syntax:''' ''ngx.header.HEADER = VALUE''
'''syntax:''' ''value = ngx.header.HEADER''
'''context:''' ''rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
When assigning to <code>ngx.header.HEADER</code> will set, add, or clear the current request's response header named <code>HEADER</code>. Underscores (<code>_</code>) in the header names will be replaced by dashes (<code>-</code>) and the header names will be matched case-insensitively.
<geshi lang="lua">
-- equivalent to ngx.header["Content-Type"] = 'text/plain'
ngx.header.content_type = 'text/plain';
ngx.header["X-My-Header"] = 'blah blah';
</geshi>
Multi-value headers can be set this way:
<geshi lang="lua">
ngx.header['Set-Cookie'] = {'a=32; path=/', 'b=4; path=/'}
</geshi>
will yield
<geshi lang="bash">
Set-Cookie: a=32; path=/
Set-Cookie: b=4; path=/
</geshi>
in the response headers. Only array-like tables are accepted.
Note that, for those standard headers that only accepts a single value, like <code>Content-Type</code>, only the last element
in the (array) table will take effect. So
<geshi lang="lua">
ngx.header.content_type = {'a', 'b'}
</geshi>
is equivalent to
<geshi lang="lua">
ngx.header.content_type = 'b'
</geshi>
Setting a slot to <code>nil</code> effectively removes it from the response headers:
<geshi lang="lua">
ngx.header["X-My-Header"] = nil;
</geshi>
same does assigning an empty table:
<geshi lang="lua">
ngx.header["X-My-Header"] = {};
</geshi>
Setting <code>ngx.header.HEADER</code> after sending out response headers (either explicitly with [[#ngx.send_headers|ngx.send_headers]] or implicitly with [[#ngx.print|ngx.print]] and its friends) will throw out a Lua exception.
Reading <code>ngx.header.HEADER</code> will return the value of the response header named <code>HEADER</code>. Underscores (<code>_</code>) in the header names will also be replaced by dashes (<code>-</code>) and the header names will be matched case-insensitively. If the response header is not present at all, <code>nil</code> will be returned.
This is particularly useful in the context of [[#filter_header_by_lua|filter_header_by_lua]] and [[#filter_header_by_lua_file|filter_header_by_lua_file]], for example,
<geshi lang="nginx">
location /test {
set $footer '';
proxy_pass http://some-backend;
header_filter_by_lua '
if ngx.header["X-My-Header"] == "blah" then
ngx.var.footer = "some value"
end
';
echo_after_body $footer;
}
</geshi>
For multi-value headers, all of the values of header will be collected in order and returned as a Lua table. For example, response headers
<geshi lang="text">
Foo: bar
Foo: baz
</geshi>
will result in
<geshi lang="lua">
{"bar", "baz"}
</geshi>
to be returned when reading <code>ngx.header.Foo</code>.
Note that <code>ngx.header</code> is not a normal Lua table so you cannot iterate through it using Lua's <code>ipairs</code> function.
For reading ''request'' headers, use the [[#ngx.req.get_headers|ngx.req.get_headers]] function instead.
== ngx.req.set_uri ==
'''syntax:''' ''ngx.req.set_uri(uri, jump?)''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Rewrite the current request's (parsed) URI by the <code>uri</code> argument. The <code>uri</code> argument must be a Lua string and cannot be of zero length, or a Lua exception will be thrown.
The optional boolean <code>jump</code> argument can trigger location rematch (or location jump) as [[HttpRewriteModule]]'s [[HttpRewriteModule#rewrite|rewrite]] directive, that is, when <code>jump</code> is <code>true</code> (default to <code>false</code>), this function will never return and it will tell Nginx to try re-searching locations with the new URI value at the later <code>post-rewrite</code> phase and jumping to the new location. Location jump will not be triggered otherwise, and only the current request's URI will be modified, which is also the default behavior. This function will return but with no returned values when the <code>jump</code> argument is <code>false</code> or absent altogether.
For example, the following nginx config snippet
<geshi lang="nginx">
rewrite ^ /foo last;
</geshi>
can be coded in Lua like this:
<geshi lang="lua">
ngx.req.set_uri("/foo", true)
</geshi>
Similarly, Nginx config
<geshi lang="nginx">
rewrite ^ /foo break;
</geshi>
can be coded in Lua as
<geshi lang="lua">
ngx.req.set_uri("/foo", false)
</geshi>
or equivalently,
<geshi lang="lua">
ngx.req.set_uri("/foo")
</geshi>
The <code>jump</code> can only be set to <code>true</code> in [[#rewrite_by_lua|rewrite_by_lua]] and [[#rewrite_by_lua_file|rewrite_by_lua_file]]. Use of jump in other contexts is prohibited and will throw out a Lua exception.
A more sophisticated example involving regex substitutions is as follows
<geshi lang="nginx">
location /test {
rewrite_by_lua '
local uri = ngx.re.sub(ngx.var.uri, "^/test/(.*)", "$1", "o")
ngx.req.set_uri(uri)
';
proxy_pass http://my_backend;
}
</geshi>
which is functionally equivalent to
<geshi lang="nginx">
location /test {
rewrite ^/test/(.*) /$1 break;
proxy_pass http://my_backend;
}
</geshi>
Note that you cannot use this interface to rewrite URI arguments, and you need to use [[#ngx.req.set_uri_args|ngx.req.set_uri_args]] for that. For instance, Nginx config
<geshi lang="nginx">
rewrite ^ /foo?a=3? last;
</geshi>
can be coded as
<geshi lang="nginx">
ngx.req.set_uri_args("a=3")
ngx.req.set_uri("/foo", true)
</geshi>
or
<geshi lang="nginx">
ngx.req.set_uri_args({a = 3})
ngx.req.set_uri("/foo", true)
</geshi>
This interface was first introduced in the <code>v0.3.1rc14</code> release.
== ngx.req.set_uri_args ==
'''syntax:''' ''ngx.req.set_uri_args(args)''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Rewrite the current request's URI query arguments by the <code>args</code> argument. The <code>args</code> argument can be either a Lua string, as in
<geshi lang="lua">
ngx.req.set_uri_args("a=3&b=hello%20world")
</geshi>
or a Lua table holding the query arguments' key-value pairs, as in
<geshi lang="lua">
ngx.req.set_uri_args({ a = 3, b = "hello world" })
</geshi>
where in the latter case, this method will automatically escape argument keys and values according to the URI escaping rule.
This interface was first introduced in the <code>v0.3.1rc13</code> release.
See also [[#ngx.req.set_uri|ngx.req.set_uri]].
== ngx.req.get_uri_args ==
'''syntax:''' ''args = ngx.req.get_uri_args()''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Returns a Lua table holds all of the current request's request URL query arguments.
Here is an example,
<geshi lang="nginx">
location = /test {
content_by_lua '
local args = ngx.req.get_uri_args()
for key, val in pairs(args) do
if type(val) == "table" then
ngx.say(key, ": ", table.concat(val, ", "))
else
ngx.say(key, ": ", val)
end
end
';
}
</geshi>
Then <code>GET /test?foo=bar&bar=baz&bar=blah</code> will yield the response body
<geshi lang="bash">
foo: bar
bar: baz, blah
</geshi>
Multiple occurrences of an argument key will result in a table value holding all of the values for that key in order.
Keys and values will be automatically unescaped according to URI escaping rules. For example, in the above settings, <code>GET /test?a%20b=1%61+2</code> will yield the output
<geshi lang="bash">
a b: 1a 2
</geshi>
Arguments without the <code>=<value></code> parts are treated as boolean arguments. For example, <code>GET /test?foo&bar</code> will yield the outputs
<geshi lang="bash">
foo: true
bar: true
</geshi>
That is, they will take Lua boolean values <code>true</code>. However, they're different from arguments taking empty string values. For example, <code>GET /test?foo=&bar=</code> will give something like
<geshi lang="bash">
foo:
bar:
</geshi>
Empty key arguments are discarded, for instance, <code>GET /test?=hello&=world</code> will yield empty outputs.
Updating query arguments via the nginx variable <code>$args</code> (or <code>ngx.var.args</code> in Lua) at runtime are also supported:
<geshi lang="lua">
ngx.var.args = "a=3&b=42"
local args = ngx.req.get_uri_args()
</geshi>
Here the <code>args</code> table will always look like
<geshi lang="lua">
{a = 3, b = 42}
</geshi>
regardless of the actual request query string.
== ngx.req.get_post_args ==
'''syntax:''' ''ngx.req.get_post_args()''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Returns a Lua table holds all of the current request's POST query arguments (of the MIME type <code>application/x-www-form-urlencoded</code>). It is required to read the request body first by calling [[#ngx.req.read_body|ngx.req.read_body]] or to turn on the [[#lua_need_request_body|lua_need_request_body]] directive, or a Lua exception will be thrown.
Here is an example,
<geshi lang="nginx">
location = /test {
content_by_lua '
ngx.req.read_body()
local args = ngx.req.get_post_args()
for key, val in pairs(args) do
if type(val) == "table" then
ngx.say(key, ": ", table.concat(val, ", "))
else
ngx.say(key, ": ", val)
end
end
';
}
</geshi>
Then
<geshi lang="bash">
# Post request with the body 'foo=bar&bar=baz&bar=blah'
$ curl --data 'foo=bar&bar=baz&bar=blah' localhost/test
</geshi>
will yield the response body like
<geshi lang="bash">
foo: bar
bar: baz, blah
</geshi>
Multiple occurrences of an argument key will result in a table value holding all of the values for that key in order.
Keys and values will be automatically unescaped according to URI escaping rules. For example, in the above settings,
<geshi lang="bash">
# POST request with body 'a%20b=1%61+2'
$ curl -d 'a%20b=1%61+2' localhost/test
</geshi>
will yield the output
<geshi lang="bash">
a b: 1a 2
</geshi>
Arguments without the <code>=<value></code> parts are treated as boolean arguments. For example, <code>GET /test?foo&bar</code> will yield the outputs
<geshi lang="bash">
foo: true
bar: true
</geshi>
That is, they will take Lua boolean values <code>true</code>. However, they're different from arguments taking empty string values. For example, <code>POST /test</code> with request body <code>foo=&bar=</code> will give something like
<geshi lang="bash">
foo:
bar:
</geshi>
Empty key arguments are discarded, for instance, <code>POST /test</code> with body <code>=hello&=world</code> will yield empty outputs.
== ngx.req.get_headers ==
'''syntax:''' ''headers = ngx.req.get_headers()''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Returns a Lua table holds all of the current request's request headers.
Here is an example,
<geshi lang="lua">
local h = ngx.req.get_headers()
for k, v in pairs(h) do
...
end
</geshi>
To read an individual header:
<geshi lang="lua">
ngx.say("Host: ", ngx.req.get_headers()["Host"])
</geshi>
For multiple instances of request headers like
<geshi lang="bash">
Foo: foo
Foo: bar
Foo: baz
</geshi>
the value of <code>ngx.req.get_headers()["Foo"]</code> will be a Lua (array) table like this:
<geshi lang="lua">
{"foo", "bar", "baz"}
</geshi>
Another way to read individual request headers is to use <code>ngx.var.http_HEADER</code>, that is, nginx's standard [[HttpCoreModule#$http_HEADER|$http_HEADER]] variables.
== ngx.req.set_header ==
'''syntax:''' ''ngx.req.set_header(header_name, header_value)''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Set the current request's request header named <code>header_name</code> to value <code>header_value</code>, overriding any existing ones.
None of the current request's subrequests will be affected.
Here is an example of setting the <code>Content-Length</code> header:
<geshi lang="lua">
ngx.req.set_header("Content-Type", "text/css")
</geshi>
The <code>header_value</code> can take an array list of values,
for example,
<geshi lang="lua">
ngx.req.set_header("Foo", {"a", "abc"})
</geshi>
will produce two new request headers:
<geshi lang="bash">
Foo: a
Foo: abc
</geshi>
and old <code>Foo</code> headers will be overridden if there is any.
When the <code>header_value</code> argument is <code>nil</code>, the request header will be removed. So
<geshi lang="lua">
ngx.req.set_header("X-Foo", nil)
</geshi>
is equivalent to
<geshi lang="lua">
ngx.req.clear_header("X-Foo")
</geshi>
== ngx.req.read_body ==
'''syntax:''' ''ngx.req.read_body()''
'''context:''' ''rewrite_by_lua*, access_by_lua*, content_by_lua*''
Read the client request body synchronously but still non-blockingly.
If the request body is already read previously by turning on [[#lua_need_request_body|lua_need_request_body]] or by using other modules, then this function is a no-op and returns immediately.
If the request body has already been explicitly discarded, either by this module's [[#ngx.req.discard_body|ngx.req.discard_body]] or other modules, this function is a no-op and returns immediately.
In case of errors, like connection errors while reading the data, this method will throw out a Lua exception ''or'' terminate the current request with the 500 status code immediately.
You can later either retrieve the request body data via [[#ngx.req.get_body_data|ngx.req.get_body_data]] or retrieve the temporary file name for the body data cached to disk via [[#ngx.req.get_body_file|ngx.req.get_body_file]], depending on
# whether the current request body is already exceeding your [[HttpCoreModule#client_body_buffer_size|client_body_buffer_size]],
# and whether you have turned on [[HttpCoreModule#client_body_in_file_only|client_body_in_file_only]].
In case that you do not want to read the request body and the current request may have a request body, then it is crucial to use the [[#ngx.req.discard_body|ngx.req.discard_body]] function to explicitly discard the request body, or you'll break HTTP 1.1 keepalive and HTTP 1.1 pipelining.
Here is a small example:
<geshi lang="lua">
ngx.req.read_body()
local args = ngx.req.get_post_args()
</geshi>
This function was first introduced in the <code>v0.3.1rc17</code> release.
== ngx.req.discard_body ==
'''syntax:''' ''ngx.req.discard_body()''
'''context:''' ''rewrite_by_lua*, access_by_lua*, content_by_lua*''
Explicitly discard the request body, i.e., read the data on the connection and throw it away immediately. Please note that, simply ignoring request body is not the right way to discard it, you need to call this function, or you'll break things under HTTP 1.1 keepalive or HTTP 1.1 pipelining.
This function is an asynchronous call and returns immediately.
If the request body has already been read, this function does nothing and returns immediately.
This function was first introduced in the <code>v0.3.1rc17</code> release.
See also [[#ngx.req.read_body|ngx.req.read_body]].
== ngx.req.get_body_data ==
'''syntax:''' ''data = ngx.req.get_body_data()''
'''context:''' ''rewrite_by_lua*, access_by_lua*, content_by_lua*''
Retrieves the in-memory request body data. It returns a Lua string rather than a Lua table holding all the parsed query arguments. If you want the latter, use [[#ngx.req.get_post_args|ngx.req.get_post_args]] instead.
This function returns <code>nil</code> if
# the request body has not been read,
# the request body has been read into disk temporary files,
# or the request body has zero size.
If the request body has not been read yet, call [[#ngx.req.read_body|ngx.req.read_body]] first (or turned on [[#lua_need_request_body|lua_need_request_body]] to force this module to read the request body automatically, but this is not recommended).
If the request body has been read into disk files, try calling the [[#ngx.req.get_body_file|ngx.req.get_body_file]] function instead.
In case that you want to enforce in-memory request bodies, try setting [[HttpCoreModule#client_body_buffer_size|client_body_buffer_size]] to the same size value in [[HttpCoreModule#client_max_body_size|client_max_body_size]].
Note that calling this function instead of using <code>ngx.var.request_body</code> or <code>ngx.var.echo_request-body</code> is more efficient because it can save one dynamic memory allocation and one data copy.
This function was first introduced in the <code>v0.3.1rc17</code> release.
See also [[#ngx.req.get_body_file|ngx.req.get_body_file]].
== ngx.req.get_body_file ==
'''syntax:''' ''file_name = ngx.req.get_body_file()''
'''context:''' ''rewrite_by_lua*, access_by_lua*, content_by_lua*''
Retrieves the file name for the in-file request body data. Returns <code>nil</code> if the request body has not been read or has been read into memory.
The returned file is read only and is usually cleaned up automatically by Nginx's memory pool. It should not be modified, renamed, or removed by your own Lua code.
If the request body has not been read yet, call [[#ngx.req.read_body|ngx.req.read_body]] first (or turned on [[#lua_need_request_body|lua_need_request_body]] to force this module to read the request body automatically, but this is not recommended).
If the request body has been read into memory, try calling the [[#ngx.req.get_body_data|ngx.req.get_body_data]] function instead.
In case that you want to enforce in-file request bodies, try turning on [[HttpCoreModule#client_body_in_file_only|client_body_in_file_only]].
This function was first introduced in the <code>v0.3.1rc17</code> release.
See also [[#ngx.req.get_body_data|ngx.req.get_body_data]].
== ngx.req.set_body_data ==
'''syntax:''' ''ngx.req.set_body_data(data)''
'''context:''' ''rewrite_by_lua*, access_by_lua*, content_by_lua*''
Set the current request's request body using the in-memory data specified by the <code>data</code> argument.
If the current request's request body has not been read, then it will be properly discarded. When the current request's request body has been read into memory or buffered into a disk file, then the old request body's memory will be freed or the disk file will be cleaned up immediately, respectively.
This function requires patching the Nginx core to function properly because the Nginx core does not allow modifying request bodies by the current design. Here is a patch for Nginx 1.0.9: [https://github.com/agentzh/ngx_openresty/blob/master/patches/nginx-1.0.9-allow_request_body_updating.patch nginx-1.0.9-allow_request_body_updating.patch], and this patch should be applied cleanly to other releases of Nginx as well.
If you're using [http://openresty.org/ ngx_openresty] 1.0.8.17+, then you've already had this patch applied.
This function was first introduced in the <code>v0.3.1rc18</code> release.
See also [[#ngx.req.set_body_file|ngx.req.set_body_file]].
== ngx.req.set_body_file ==
'''syntax:''' ''ngx.req.set_body_file(file_name, auto_clean?)''
'''context:''' ''rewrite_by_lua*, access_by_lua*, content_by_lua*''
Set the current request's request body using the in-file data specified by the <code>file_name</code> argument.
If the optional <code>auto_clean</code> argument is given a <code>true</code> value, then this file will be automatically removed at request completion or the next time this function or [[#ngx.req.set_body_data|ngx.req.set_body_data]] are called in the same request. The <code>auto_clean</code> is default to <code>false</code>.
You must ensure that the file specified by the <code>file_name</code> argument exists and is readable by an Nginx worker process by setting its permission properly. Otherwise a Lua exception will be thrown.
If the current request's request body has not been read, then it will be properly discarded. When the current request's request body has been read into memory or buffered into a disk file, then the old request body's memory will be freed or the disk file will be cleaned up immediately, respectively.
This function requires patching the Nginx core to function properly because the Nginx core does not allow modifying request bodies by the current design. Here is a patch for Nginx 1.0.9: [https://github.com/agentzh/ngx_openresty/blob/master/patches/nginx-1.0.9-allow_request_body_updating.patch nginx-1.0.9-allow_request_body_updating.patch], and this patch should be applied cleanly to other releases of Nginx as well.
If you're using [http://openresty.org/ ngx_openresty] 1.0.8.17+, then you've already had this patch applied.
This function was first introduced in the <code>v0.3.1rc18</code> release.
See also [[#ngx.req.set_body_data|ngx.req.set_body_data]].
== ngx.req.clear_header ==
'''syntax:''' ''ngx.req.clear_header(header_name)''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Clear the current request's request header named <code>header_name</code>. None of the current request's subrequests will be affected.
== ngx.exec ==
'''syntax:''' ''ngx.exec(uri, args?)''
'''context:''' ''rewrite_by_lua*, access_by_lua*, content_by_lua*''
Does an internal redirect to <code>uri</code> with <code>args</code>.
<geshi lang="lua">
ngx.exec('/some-location');
ngx.exec('/some-location', 'a=3&b=5&c=6');
ngx.exec('/some-location?a=3&b=5', 'c=6');
</geshi>
Named locations are also supported, but query strings are ignored. For example,
<geshi lang="nginx">
location /foo {
content_by_lua '
ngx.exec("@bar");
';
}
location @bar {
...
}
</geshi>
The optional second <code>args</code> can be used to specify extra URI query arguments, for example:
<geshi lang="lua">
ngx.exec("/foo", "a=3&b=hello%20world")
</geshi>
Alternatively, you can pass a Lua table for the <code>args</code> argument and let ngx_lua do URI escaping and string concatenation automatically for you, for instance,
<geshi lang="lua">
ngx.exec("/foo", { a = 3, b = "hello world" })
</geshi>
The result is exactly the same as the previous example. The format for the Lua table passed as the <code>args</code> argument is identical to the format used in the [[#ngx.encode_args|ngx.encode_args]] method.
Note that this is very different from [[#ngx.redirect|ngx.redirect]] in that
it is just an internal redirect and no new HTTP traffic is involved.
This method never returns.
This method ''must'' be called before [[#ngx.send_headers|ngx.send_headers]] or explicit response body
outputs by either [[#ngx.print|ngx.print]] or [[#ngx.say|ngx.say]].
This method is very much like the [[HttpEchoModule#echo_exec|echo_exec]] directive in [[HttpEchoModule]].
== ngx.redirect ==
'''syntax:''' ''ngx.redirect(uri, status?)''
'''context:''' ''rewrite_by_lua*, access_by_lua*, content_by_lua*''
Issue an <code>HTTP 301<code> or <code>302</code> redirection to <code>uri</code>.
The optional <code>status</code> parameter specifies whether
<code>301</code> or <code>302</code> to be used. It is <code>302</code> (<code>ngx.HTTP_MOVED_TEMPORARILY</code>) by default.
Here is an example assuming the current server name is <code>localhost</code> and that it is listening on Port 1984:
<geshi lang="lua">
return ngx.redirect("/foo")
</geshi>
which is equivalent to
<geshi lang="lua">
return ngx.redirect("http://localhost:1984/foo", ngx.HTTP_MOVED_TEMPORARILY)
</geshi>
We can also use the numberical code directly as the second <code>status</code> argument:
<geshi lang="lua">
return ngx.redirect("/foo", 301)
</geshi>
This method ''must'' be called before [[#ngx.send_headers|ngx.send_headers]] or explicit response body outputs by either [[#ngx.print|ngx.print]] or [[#ngx.say|ngx.say]].
This method never returns.
This method is very much like the [[HttpRewriteModule#rewrite|rewrite]] directive with the <code>redirect</code> modifier in the standard
[[HttpRewriteModule]], for example, this <code>nginx.conf</code> snippet
<geshi lang="nginx">
rewrite ^ /foo? redirect; # nginx config
</geshi>
is equivalent to the following Lua code
<geshi lang="lua">
return ngx.redirect('/foo'); -- lua code
</geshi>
while
<geshi lang="nginx">
rewrite ^ /foo? permanent; # nginx config
</geshi>
is equivalent to
<geshi lang="lua">
return ngx.redirect('/foo', ngx.HTTP_MOVED_PERMANENTLY) -- Lua code
</geshi>
URI arguments can be specified as well, for example:
<geshi lang="lua">
return ngx.redirect('/foo?a=3&b=4')
</geshi>
== ngx.send_headers ==
'''syntax:''' ''ngx.send_headers()''
'''context:''' ''rewrite_by_lua*, access_by_lua*, content_by_lua*''
Explicitly send out the response headers.
Usually you do not have to send headers yourself. <code>ngx_lua</code> will automatically send out headers right before you
output contents via [[#ngx.say|ngx.say]] or [[#ngx.print|ngx.print]].
Headers will also be sent automatically when [[#content_by_lua|content_by_lua]] exits normally.
== ngx.headers_sent ==
'''syntax:''' ''value = ngx.headers_sent''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Returns <code>true</code> if the response headers have been sent (by ngx_lua), and <code>false</code> otherwise.
This API was first introduced in ngx_lua v0.3.1rc6.
== ngx.print ==
'''syntax:''' ''ngx.print(...)''
'''context:''' ''rewrite_by_lua*, access_by_lua*, content_by_lua*''
Emit arguments concatenated to the HTTP client (as response body). If response headers have not been sent yet, this function will first send the headers out, and then output the body data.
Lua <code>nil</code> value will result in outputing <code>"nil"</code>, and Lua boolean values will emit literal <code>"true"</code> or <code>"false"</code>, accordingly.
Also, nested arrays of strings are also allowed. The elements in the arrays will be sent one by one. For example
<geshi lang="lua">
local table = {
"hello, ",
{"world: ", true, " or ", false,
{": ", nil}}
}
ngx.print(table)
</geshi>
will yield the output
<geshi lang="bash">
hello, world: true or false: nil
</geshi>
Non-array table arguments will cause a Lua exception to be thrown.
This is an asynchronous call, that is, this function will return immediately without waiting for all the data has actually been written into the system send buffer. If you want to wait for the data to be flushed before proceeding, you should call <code>ngx.flush(true)</code> right after this call. See [[#ngx.flush|ngx.flush]] for more details.
== ngx.say ==
'''syntax:''' ''ngx.say(...)''
'''context:''' ''rewrite_by_lua*, access_by_lua*, content_by_lua*''
Just as [[#ngx.print|ngx.print]] but also emit a trailing newline.
== ngx.log ==
'''syntax:''' ''ngx.log(log_level, ...)''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Log arguments concatenated to error.log with the given logging level.
Lua <code>nil</code> arguments are accepted and result in literal <code>"nil"</code>, and Lua booleans result in literal <code>"true"</code> or <code>"false"</code> outputs.
The <code>log_level</code> argument can take constants like <code>ngx.ERR</code> and <code>ngx.WARN</code>. Check out [[#Nginx log level constants|Nginx log level constants]] for details.
== ngx.flush ==
'''syntax:''' ''ngx.flush(wait?)''
'''context:''' ''rewrite_by_lua*, access_by_lua*, content_by_lua*''
Flushing the response outputs. This operation has no effect in HTTP 1.0 buffering output mode. See [[#HTTP 1.0 support|HTTP 1.0 support]].
By default, this function is an asynchronous call, that is, it returns immediately without waiting for the ouptuts to be actually flushed into the system send buffer.
This function takes an optional boolean <code>wait</code> argument. When it is given <code>true</code> (which is default to <code>false</code>), this function becomes a synchronous call, that is, it will not return until all the outputs have actually been flushed into the system send buffer or the [[HttpCoreModule#send_timeout|send_timeout]] setting has expired. Note that, even in the synchronous mode, this function still works nonblockingly (thanks to the Lua coroutine mechanism!).
Synchronous flushing is very useful for streaming output in Lua.
The <code>wait</code> argument was first introduced in the <code>v0.3.1rc34</code> release.
== ngx.exit ==
'''syntax:''' ''ngx.exit(status)''
'''context:''' ''rewrite_by_lua*, access_by_lua*, content_by_lua*''
When <code>status >= 200</code> (i.e., <code>ngx.HTTP_OK</code> and above), it will interrupt the execution of the current request and return status code to nginx.
When <code>status == 0</code> (i.e., <code>ngx.OK</code>), it will only quit the current phase handler (or the content handler if the [[#content_by_lua|content_by_lua]] directive is used) and continue to run laster phases (if any) for the current request.
The <code>status</code> argument can be <code>ngx.OK</code>, <code>ngx.ERROR</code>, <code>ngx.HTTP_NOT_FOUND</code>,
<code>ngx.HTTP_MOVED_TEMPORARILY</code>, or other [[#HTTP status constants|HTTP status constants]].
To return an error page with custom contents, use code snippets like this:
<geshi lang="lua">
ngx.status = ngx.HTTP_GONE
ngx.say("This is our own content")
-- to cause quit the whole request rather than the current phase handler
ngx.exit(ngx.HTTP_OK)
</geshi>
The effect in action:
<geshi lang="bash">
$ curl -i http://localhost/test
HTTP/1.1 410 Gone
Server: nginx/1.0.6
Date: Thu, 15 Sep 2011 00:51:48 GMT
Content-Type: text/plain
Transfer-Encoding: chunked
Connection: keep-alive
This is our own content
</geshi>
Number literals can be used directly as the argument, for instance,
<geshi lang="lua">
ngx.exit(501)
</geshi>
== ngx.eof ==
'''syntax:''' ''ngx.eof()''
'''context:''' ''rewrite_by_lua*, access_by_lua*, content_by_lua*''
Explicitly specify the end of the response output stream.
== ngx.escape_uri ==
'''syntax:''' ''newstr = ngx.escape_uri(str)''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Escape <code>str</code> as a URI component.
== ngx.unescape_uri ==
'''syntax:''' ''newstr = ngx.unescape_uri(str)''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Unescape <code>str</code> as an escaped URI component.
For example,
<geshi lang="lua">
ngx.say(ngx.unescape_uri("b%20r56+7"))
</geshi>
gives the output
<geshi lang="text">
b r56 7
</geshi>
== ngx.encode_args ==
'''syntax:''' ''str = ngx.encode_args(table)''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Encode the Lua table to a query args string according to the URI encoded rules.
For example,
<geshi lang="lua">
ngx.encode_args({foo = 3, ["b r"] = "hello world"})
</geshi>
yields
<geshi lang="text">
foo=3&b%20r=hello%20world
</geshi>
The table keys must be Lua strings.
Multi-value query args are also supported. Just use a Lua table for the arg's value, for example:
<geshi lang="lua">
ngx.encode_args({baz = {32, "hello"}})
</geshi>
gives
<geshi lang="text">
baz=32&baz=hello
</geshi>
If the value table is empty and the effect is equivalent to the <code>nil</code> value.
Boolean argument values are also supported, for instance,
<geshi lang="lua">
ngx.encode_args({a = true, b = 1})
</geshi>
yields
<geshi lang="text">
a&b=1
</geshi>
If the argument value is <code>false</code>, then the effect is equivalent to the <code>nil</code> value.
This method was first introduced in the <code>v0.3.1rc27</code> release.
== ngx.encode_base64 ==
'''syntax:''' ''newstr = ngx.encode_base64(str)''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Encode <code>str</code> to a base64 digest.
== ngx.decode_base64 ==
'''syntax:''' ''newstr = ngx.decode_base64(str)''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Decodes the <code>str</code> argument as a base64 digest to the raw form. Returns <code>nil</code> if <code>str</code> is not well formed.
== ngx.crc32_short ==
'''syntax:''' ''intval = ngx.crc32_short(str)''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Calculates the CRC-32 (Cyclic Redundancy Code) digest for the <code>str</code> argument.
This method performs better on relatively short <code>str</code> inputs (i.e., less than 30 ~ 60 bytes), as compared to [[#ngx.crc32_long|ngx.crc32_long]]. The result is exactly the same as [[#ngx.crc32_long|ngx.crc32_long]].
Behind the scene, it is just a thin wrapper around the <code>ngx_crc32_short</code> function defined in the Nginx core.
This API was first introduced in the <code>v0.3.1rc8</code> release.
== ngx.crc32_long ==
'''syntax:''' ''intval = ngx.crc32_long(str)''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Calculates the CRC-32 (Cyclic Redundancy Code) digest for the <code>str</code> argument.
This method performs better on relatively long <code>str</code> inputs (i.e., longer than 30 ~ 60 bytes), as compared to [[#ngx.crc32_short|ngx.crc32_short]]. The result is exactly the same as [[#ngx.crc32_short|ngx.crc32_short]].
Behind the scene, it is just a thin wrapper around the <code>ngx_crc32_long</code> function defined in the Nginx core.
This API was first introduced in the <code>v0.3.1rc8</code> release.
== ngx.hmac_sha1 ==
'''syntax:''' ''digest = ngx.hmac_sha1(secret_key, str)''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Computes the [http://en.wikipedia.org/wiki/HMAC HMAC-SHA1] digest of the argument <code>str</code> and turns the result using the secret key <code><secret_key></code>.
The raw binary form of the <code>HMAC-SHA1</code> digest will be generated, use [[#ngx.encode_base64|ngx.encode_base64]], for example, to encode the result to a textual representation if desired.
For example,
<geshi lang="lua">
local key = "thisisverysecretstuff"
local src = "some string we want to sign"
local digest = ngx.hmac_sha1(key, src)
ngx.say(ngx.encode_base64(digest))
</geshi>
yields the output
<geshi lang="text">
R/pvxzHC4NLtj7S+kXFg/NePTmk=
</geshi>
This API requires the OpenSSL library enabled in your Nignx build (usually by passing the <code>--with-http_ssl_module</code> option to the <code>./configure</code> script).
This function was first introduced in the <code>v0.3.1rc29</code> release.
== ngx.md5 ==
'''syntax:''' ''digest = ngx.md5(str)''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Returns the hexadecimal representation of the MD5 digest of the <code>str</code> argument.
For example,
<geshi lang="nginx">
location = /md5 {
content_by_lua 'ngx.say(ngx.md5("hello"))';
}
</geshi>
yields the output
<geshi lang="text">
5d41402abc4b2a76b9719d911017c592
</geshi>
See also [[#ngx.md5_bin|ngx.md5_bin]] if you want the raw binary MD5 digest.
== ngx.md5_bin ==
'''syntax:''' ''digest = ngx.md5_bin(str)''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Returns the binary form of the MD5 digest of the <code>str</code> argument.
See also [[#ngx.md5|ngx.md5]] if you want the hexadecimal form of the MD5 digest.
== ngx.today ==
'''syntax:''' ''str = ngx.today()''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Returns current date (in the format <code>yyyy-mm-dd</code>) from the nginx cached time (no syscall involved unlike Lua's date library).
This is the local time.
== ngx.time ==
'''syntax:''' ''secs = ngx.time()''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Returns the elapsed seconds from the epoch for the current timestamp from the nginx cached time (no syscall involved unlike Lua's date library).
You can enforce updating the Nginx time cache by calling [[#ngx.update_time|ngx.update_time]] first.
== ngx.now ==
'''syntax:''' ''secs = ngx.now()''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Returns a floating-point number for the elapsed time in seconds (including microseconds as the decimal part) from the epoch for the current timestamp from the nginx cached time (no syscall involved unlike Lua's date library).
You can use the standard Nginx directive [[CoreModule#timer_resolution|timer_resolution]] to adjust the accuracy or forcibly updating the Nginx time cache by calling [[#ngx.update_time|ngx.update_time]] first.
This API was first introduced in <code>v0.3.1rc32</code>.
== ngx.update_time ==
'''syntax:''' ''ngx.update_time()''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Forcibly updating the Nginx current time cache. This call involves a syscall and thus has some overhead, so do not abuse it.
This API was first introduced in <code>v0.3.1rc32</code>.
== ngx.localtime ==
'''syntax:''' ''str = ngx.localtime()''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Returns the current timestamp (in the format <code>yyyy-mm-dd hh:mm:ss</code>) of the nginx cached time (no syscall involved unlike Lua's [http://www.lua.org/manual/5.1/manual.html#pdf-os.date os.date] function).
This is the local time.
== ngx.utctime ==
'''syntax:''' ''str = ngx.utctime()''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Returns the current timestamp (in the format <code>yyyy-mm-dd hh:mm:ss</code>) of the nginx cached time (no syscall involved unlike Lua's [http://www.lua.org/manual/5.1/manual.html#pdf-os.date os.date] function).
This is the UTC time.
== ngx.cookie_time ==
'''syntax:''' ''str = ngx.cookie_time(sec)''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Returns a formated string can be used as the cookie expiration time. The parameter <code>sec</code> is the timestamp in seconds (like those returned from [[#ngx.time|ngx.time]]).
<geshi lang="nginx">
ngx.say(ngx.cookie_time(1290079655))
-- yields "Thu, 18-Nov-10 11:27:35 GMT"
</geshi>
== ngx.http_time ==
'''syntax:''' ''str = ngx.http_time(sec)''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Returns a formated string can be used as the http header time (for example, being used in <code>Last-Modified</code> header). The parameter <code>sec</code> is the timestamp in seconds (like those returned from [[#ngx.time|ngx.time]]).
<geshi lang="nginx">
ngx.say(ngx.http_time(1290079655))
-- yields "Thu, 18 Nov 10 11:27:35 GMT"
</geshi>
== ngx.parse_http_time ==
'''syntax:''' ''sec = ngx.parse_http_time(str)''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Parse the http time string (as returned by [[#ngx.http_time|ngx.http_time]]) into seconds. Returns the seconds or <code>nil</code> if the input string is in bad forms.
<geshi lang="nginx">
local time = ngx.parse_http_time("Thu, 18 Nov 10 11:27:35 GMT")
if time == nil then
...
end
</geshi>
== ngx.is_subrequest ==
'''syntax:''' ''value = ngx.is_subrequest''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Returns <code>true</code> if the current request is an nginx subrequest, or <code>false</code> otherwise.
== ngx.re.match ==
'''syntax:''' ''captures = ngx.re.match(subject, regex, options?, ctx?)''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Matches the <code>subject</code> string using the Perl-compatible regular expression <code>regex</code> with the optional <code>options</code>.
Only the first occurrence of the match is returned, or <code>nil</code> if no match is found. In case of fatal errors, like seeing bad <code>UTF-8</code> sequences in <code>UTF-8</code> mode, a Lua exception will be raised.
When a match is found, a Lua table <code>captures</code> is returned, where <code>captures[0]</code> holds the whole substring being matched, and <code>captures[1]</code> holds the first parenthesized subpattern's capturing, <code>captures[2]</code> the second, and so on.
<geshi lang="lua">
local m = ngx.re.match("hello, 1234", "[0-9]+")
-- m[0] == "1234"
</geshi>
<geshi lang="lua">
local m = ngx.re.match("hello, 1234", "([0-9])[0-9]+")
-- m[0] == "1234"
-- m[1] == "1"
</geshi>
Unmatched subpatterns will have <code>nil</code> values in their <code>captures</code> table fields.
<geshi lang="lua">
local m = ngx.re.match("hello, world", "(world)|(hello)")
-- m[0] == "hello"
-- m[1] == nil
-- m[2] == "hello"
</geshi>
You can also specify <code>options</code> to control how the match will be performed. The following option characters are supported:
<geshi lang="text">
a anchored mode (only match from the beginning)
d enable the DFA mode (or the longest token match semantics).
this requires PCRE 6.0+ or else a Lua exception will be thrown.
first introduced in ngx_lua v0.3.1rc30.
i caseless mode (similar to Perl's /i modifier)
j enable PCRE JIT compilation, this requires PCRE 8.20+ and
PCRE must be built with the --enable-jit option, or it is
a no-op. this should always be used along with the <code>o</code>
option to gain the full performance benefit.
first introduced in ngx_lua v0.3.1rc30.
m multi-line mode (similar to Perl's /m modifier)
o compile-once mode (similar to Perl's /o modifer),
to enable the worker-process-level compiled-regex cache
s single-line mode (similar to Perl's /s modifier)
u UTF-8 mode. this requires PCRE to be built with
the --enable-utf8 option or else a Lua exception will be thrown.
x extended mode (similar to Perl's /x modifier)
</geshi>
These options can be combined:
<geshi lang="nginx">
local m = ngx.re.match("hello, world", "HEL LO", "ix")
-- m[0] == "hello"
</geshi>
<geshi lang="nginx">
local m = ngx.re.match("hello, 美好生活", "HELLO, (.{2})", "iu")
-- m[0] == "hello, 美好"
-- m[1] == "美好"
</geshi>
The <code>o</code> option is useful for performance tuning, because the regex pattern in question will only be compiled once, cached in the worker-process level, and shared among all requests in the current Nginx worker process. You can tune the upper limit of the regex cache via the [[#lua_regex_cache_max_entries|lua_regex_cache_max_entries]] directive.
The optional fourth argument, <code>ctx</code>, can be a Lua table holding an optional <code>pos</code> field. When the <code>pos</code> field in the <code>ctx</code> table argument is specified, <code>ngx.re.match</code> will start matching from that offset. Regardless of the presence of the <code>pos</code> field in the <code>ctx</code> table, <code>ngx.re.match</code> will always set this <code>pos</code> field to the position ''after'' the substring matched by the whole pattern in case of a successful match. When match fails, the <code>ctx</code> table will be left intact.
<geshi lang="lua">
local ctx = {}
local m = ngx.re.match("1234, hello", "[0-9]+", "", ctx)
-- m[0] = "1234"
-- ctx.pos == 4
</geshi>
<geshi lang="lua">
local ctx = { pos = 2 }
local m = ngx.re.match("1234, hello", "[0-9]+", "", ctx)
-- m[0] = "34"
-- ctx.pos == 4
</geshi>
The <code>ctx</code> table argument combined with the <code>a</code> regex modifier can be used to construct a lexer atop <code>ngx.re.match</code>.
Note that, the <code>options</code> argument is not optional when the <code>ctx</code> argument is specified; use the empty Lua string (<code>""</code>) as the placeholder for <code>options</code> if you do not want to specify any regex options.
This method requires the PCRE library enabled in your Nginx build. ([[#Special PCRE Sequences|Known Issue With Special PCRE Sequences]]).
This feature was introduced in the <code>v0.2.1rc11</code> release.
== ngx.re.gmatch ==
'''syntax:''' ''iterator = ngx.re.gmatch(subject, regex, options?)''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Similar to [[#ngx.re.match|ngx.re.match]], but returns a Lua iterator instead, so as to let the user programmer iterate all the matches over the <code><subject></code> string argument with the PCRE <code>regex</code>.
Here is a small exmple to demonstrate its basic usage:
<geshi lang="lua">
local iterator = ngx.re.gmatch("hello, world!", "([a-z]+)", "i")
local m
m = iterator() -- m[0] == m[1] == "hello"
m = iterator() -- m[0] == m[1] == "world"
m = iterator() -- m == nil
</geshi>
More often we just put it into a Lua <code>for</code> loop:
<geshi lang="lua">
for m in ngx.re.gmatch("hello, world!", "([a-z]+)", "i")
ngx.say(m[0])
ngx.say(m[1])
end
</geshi>
The optional <code>options</code> argument takes exactly the same semantics as the [[#ngx.re.match|ngx.re.match]] method.
The current implementation requires that the iterator returned should only be used in a single request. That is, one should ''not'' assign it to a variable belonging to persistent namespace like a Lua package.
This method requires the PCRE library enabled in your Nginx build. ([[#Special PCRE Sequences|Known Issue With Special PCRE Sequences]]).
This feature was first introduced in the <code>v0.2.1rc12</code> release.
== ngx.re.sub ==
'''syntax:''' ''newstr, n = ngx.re.sub(subject, regex, replace, options?)''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Substitutes the first match of the Perl-compatible regular expression <code>regex</code> on the <code>subject</code> argument string with the string or function argument <code>replace</code>. The optional <code>options</code> argument has exactly the same meaning as in [[#ngx.re.match|ngx.re.match]].
This method returns the resulting new string as well as the number of successful substitutions, or throw out a Lua exception when an error occurred (syntax errors in the <code><replace></code> string argument, for example).
When the <code>replace</code> is a string, then it is treated as a special template for string replacement. For example,
<geshi lang="lua">
local newstr, n = ngx.re.sub("hello, 1234", "([0-9])[0-9]", "[$0][$1]")
-- newstr == "hello, [12][1]34"
-- n == 1
</geshi>
where <code>$0</code> referring to the whole substring matched by the pattern and <code>$1</code> referring to the first parenthesized capturing substring.
You can also use curly braces to disambiguate variable names from the background string literals:
<geshi lang="lua">
local newstr, n = ngx.re.sub("hello, 1234", "[0-9]", "${0}00")
-- newstr == "hello, 10034"
-- n == 1
</geshi>
Literal dollar sign characters (<code>$</code>) in the <code>replace</code> string argument can be escaped by another dollar sign, for instance,
<geshi lang="lua">
local newstr, n = ngx.re.sub("hello, 1234", "[0-9]", "$$")
-- newstr == "hello, $234"
-- n == 1
</geshi>
Do not use backlashes to escape dollar signs; it will not work as expected.
When the <code>replace</code> argument is of type "function", then it will be invoked with the "match table" as the argument to generate the replace string literal for substitution. The "match table" fed into the <code>replace</code> function is exactly the same as the return value of [[#ngx.re.match|ngx.re.match]]. Here is an example:
<geshi lang="lua">
local func = function (m)
return "[" .. m[0] .. "][" .. m[1] .. "]"
end
local newstr, n = ngx.re.sub("hello, 1234", "( [0-9] ) [0-9]", func, "x")
-- newstr == "hello, [12][1]34"
-- n == 1
</geshi>
The dollar sign characters in the return value of the <code>replace</code> function argument are not special at all.
This method requires the PCRE library enabled in your Nginx build. ([[#Special PCRE Sequences|Known Issue With Special PCRE Sequences]]).
This feature was first introduced in the <code>v0.2.1rc13</code> release.
== ngx.re.gsub ==
'''syntax:''' ''newstr, n = ngx.re.gsub(subject, regex, replace, options?)''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Just like [[#ngx.re.sub|ngx.re.sub]], but does global substitution.
Here is some examples:
<geshi lang="lua">
local newstr, n = ngx.re.gsub("hello, world", "([a-z])[a-z]+", "[$0,$1]", "i")
-- newstr == "[hello,h], [world,w]"
-- n == 2
</geshi>
<geshi lang="lua">
local func = function (m)
return "[" .. m[0] .. "," .. m[1] .. "]"
end
local newstr, n = ngx.re.gsub("hello, world", "([a-z])[a-z]+", func, "i")
-- newstr == "[hello,h], [world,w]"
-- n == 2
</geshi>
This method requires the PCRE library enabled in your Nginx build. ([[#Special PCRE Sequences|Known Issue With Special PCRE Sequences]]).
This feature was first introduced in the <code>v0.2.1rc15</code> release.
== ngx.shared.DICT ==
'''syntax:''' ''dict = ngx.shared.DICT''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Fetching the shm-based Lua dictionary object for the shared memory zone named <code>DICT</code> defined by the [[#lua_shared_dict|lua_shared_dict]] directive.
The resulting object <code>dict</code> has the following methods:
* [[#ngx.shared.DICT.get|get]]
* [[#ngx.shared.DICT.set|set]]
* [[#ngx.shared.DICT.add|add]]
* [[#ngx.shared.DICT.replace|replace]]
* [[#ngx.shared.DICT.incr|incr]]
* [[#ngx.shared.DICT.delete|delete]]
Here is an example:
<geshi lang="nginx">
http {
lua_shared_dict dogs 10m;
server {
location /set {
content_by_lua '
local dogs = ngx.shared.dogs
dogs:set("Jim", 8)
ngx.say("STORED")
';
}
location /get {
content_by_lua '
local dogs = ngx.shared.dogs
ngx.say(dogs:get("Jim"))
';
}
}
}
</geshi>
Let us test it:
<geshi lang="bash">
$ curl localhost/set
STORED
$ curl localhost/get
8
$ curl localhost/get
8
</geshi>
You will consistently get the output <code>8</code> when accessing <code>/get</code> regardless how many Nginx workers there are because the <code>dogs</code> dictionary resides in the shared memory and visible to ''all'' of the worker processes.
The shared dictionary will retain its contents through a server config reload (either by sending the <code>HUP</code> signal to the Nginx process or by using the <code>-s reload</code> command-line option).
The contents in the dictionary storage will be lost, however, when the Nginx server quits.
This feature was first introduced in the <code>v0.3.1rc22</code> release.
== ngx.shared.DICT.get ==
'''syntax:''' ''value = ngx.shared.DICT:get(key)''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Retrieving the value in the dictionary [[#ngx.shared.DICT|ngx.shared.DICT]] for the key <code>key</code>. If the key does not exist or has been expired, then <code>nil</code> will be returned.
The value returned will have the original data type when they were inserted into the dictionary, for example, Lua booleans, numbers, or strings.
The first argument to this method must be the dictionary object itself, for example,
<geshi lang="lua">
local cats = ngx.shared.cats
local value = cats.get(cats, "Marry")
</geshi>
or use Lua's syntactic sugar for method calls:
<geshi lang="lua">
local cats = ngx.shared.cats
local value = cats:get("Marry")
</geshi>
These two forms are fundamentally equivalent.
This feature was first introduced in the <code>v0.3.1rc22</code> release.
See also [[#ngx.shared.DICT|ngx.shared.DICT]].
== ngx.shared.DICT.set ==
'''syntax:''' ''success, err, forcible = ngx.shared.DICT:set(key, value, exptime?)''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Unconditionally sets a key-value pair into the shm-based dictionary [[#ngx.shared.DICT|ngx.shared.DICT]]. Returns three values:
* <code>success</code>: boolean value to indicate whether the key-value pair is stored or not.
* <code>err</code>: textual error message, can be <code>"no memory"</code>.
* <code>forcible</code>: a boolean value to indicate whether other valid items have been removed forcibly when out of storage in the shared memory zone.
The <code>value</code> argument inserted can be Lua booleans, numbers, strings, or <code>nil</code>. Their value type will also be stored into the dictionary, thus you can get exactly the same data type when later retrieving the value out of the dictionary via the [[#ngx.shared.DICT.get|get]] method.
The optional <code>exptime</code> argument specifies expiration time (in seconds) for the inserted key-value pair. The time resolution is <code>0.001</code> seconds. If the <code>exptime</code> takes the value <code>0</code> (which is the default), then the item will never be expired.
When it fails to allocate memory for the current key-value item, then <code>set</code> will try removing existing items in the storage according to the Least-Recently Used (LRU) algorithm. Note that, LRU takes priority over expiration time here. If up to tens of existing items have been removed and the storage left is still insufficient (either due to the total capacity limit specified by [[#lua_shared_dict|lua_shared_dict]] or memory segmentation), then the <code>err</code> return value will be <code>no memory</code> and <code>success</code> will be <code>false</code>.
If this method succeeds in storing the current item by forcibly removing other not-yet-expired items in the dictionary via LRU, the <code>forcible</code> return value will be <code>true</code>. If it stores the item without forcibly removing other valid items, then the return value <code>forcible</code> will be <code>false</code>.
The first argument to this method must be the dictionary object itself, for example,
<geshi lang="lua">
local cats = ngx.shared.cats
local succ, err, forcible = cats.set(cats, "Marry", "it is a nice cat!")
</geshi>
or use Lua's syntactic sugar for method calls:
<geshi lang="lua">
local cats = ngx.shared.cats
local succ, err, forcible = cats:set("Marry", "it is a nice cat!")
</geshi>
These two forms are fundamentally equivalent.
This feature was first introduced in the <code>v0.3.1rc22</code> release.
See also [[#ngx.shared.DICT|ngx.shared.DICT]].
== ngx.shared.DICT.add ==
'''syntax:''' ''success, err, forcible = ngx.shared.DICT:add(key, value, exptime?)''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Just like the [[#ngx.shared.DICT.set|set]] method, but only stores the key-value pair into the dictionary [[#ngx.shared.DICT|ngx.shared.DICT]] if the key does ''not'' exist.
If the <code>key</code> argument already exists in the dictionary (and not expired for sure), the <code>success</code> return value will be <code>false</code> and the <code>err</code> return value will be <code>"exists"</code>.
This feature was first introduced in the <code>v0.3.1rc22</code> release.
See also [[#ngx.shared.DICT|ngx.shared.DICT]].
== ngx.shared.DICT.replace ==
'''syntax:''' ''success, err, forcible = ngx.shared.DICT:replace(key, value, exptime?)''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Just like the [[#ngx.shared.DICT.set|set]] method, but only stores the key-value pair into the dictionary [[#ngx.shared.DICT|ngx.shared.DICT]] if the key ''does'' exist.
If the <code>key</code> argument does ''not'' exist in the dictionary (or expired already), the <code>success</code> return value will be <code>false</code> and the <code>err</code> return value will be <code>"not found"</code>.
This feature was first introduced in the <code>v0.3.1rc22</code> release.
See also [[#ngx.shared.DICT|ngx.shared.DICT]].
== ngx.shared.DICT.delete ==
'''syntax:''' ''ngx.shared.DICT:delete(key)''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Unconditionally removes the key-value pair from the shm-based dictionary [[#ngx.shared.DICT|ngx.shared.DICT]].
It is equivalent to <code>ngx.shared.DICT:set(key, nil)</code>.
This feature was first introduced in the <code>v0.3.1rc22</code> release.
See also [[#ngx.shared.DICT|ngx.shared.DICT]].
== ngx.shared.DICT.incr ==
'''syntax:''' ''newval, err = ngx.shared.DICT:incr(key, value)''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
Increments the (numerical) value for <code>key</code> in the shm-based dictionary [[#ngx.shared.DICT|ngx.shared.DICT]] by the step value <code>value</code>. Returns the new resulting number if the operation is successfully completed or <code>nil</code> and an error message otherwise.
The key must already exist in the dictionary, otherwise it will return <code>nil</code> and <code>"not found"</code>.
If the original value is not a valid Lua number in the dictionary, it will return <code>nil</code> and <code>"not a number"</code>.
The <code>value</code> argument can be any valid Lua numbers, like negative numbers or floating-point numbers.
This feature was first introduced in the <code>v0.3.1rc22</code> release.
See also [[#ngx.shared.DICT|ngx.shared.DICT]].
== ndk.set_var.DIRECTIVE ==
'''syntax:''' ''res = ndk.set_var.DIRECTIVE_NAME''
'''context:''' ''set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*''
This mechanism allows calling other nginx C modules' directives that are implemented by [https://github.com/simpl/ngx_devel_kit Nginx Devel Kit] (NDK)'s set_var submodule's <code>ndk_set_var_value</code>.
For example, the following [[HttpSetMiscModule]] directives can be invoked this way:
* [[HttpSetMiscModule#set_quote_sql_str|set_quote_sql_str]]
* [[HttpSetMiscModule#set_quote_pgsql_str|set_quote_pgsql_str]]
* [[HttpSetMiscModule#set_quote_json_str|set_quote_json_str]]
* [[HttpSetMiscModule#set_unescape_uri|set_unescape_uri]]
* [[HttpSetMiscModule#set_escape_uri|set_escape_uri]]
* [[HttpSetMiscModule#set_encode_base32|set_encode_base32]]
* [[HttpSetMiscModule#set_decode_base32|set_decode_base32]]
* [[HttpSetMiscModule#set_encode_base64|set_encode_base64]]
* [[HttpSetMiscModule#set_decode_base64|set_decode_base64]]
* [[HttpSetMiscModule#set_encode_base64|set_encode_hex]]
* [[HttpSetMiscModule#set_decode_base64|set_decode_hex]]
* [[HttpSetMiscModule#set_encode_base64|set_sha1]]
* [[HttpSetMiscModule#set_decode_base64|set_md5]]
For instance,
<geshi lang="lua">
local res = ndk.set_var.set_escape_uri('a/b');
-- now res == 'a%2fb'
</geshi>
Similarly, the following directives provided by [[HttpEncryptedSessionModule]] can be invoked from within Lua too:
* [[HttpEncryptedSessionModule#set_encrypt_session|set_encrypt_session]]
* [[HttpEncryptedSessionModule#set_decrypt_session|set_decrypt_session]]
This feature requires the [https://github.com/simpl/ngx_devel_kit ngx_devel_kit] module.
= HTTP 1.0 support =
The HTTP 1.0 protocol does not support chunked outputs and always requires an
explicit <code>Content-Length</code> header when the response body is non-empty. So when
an HTTP 1.0 request is present, This module will automatically buffer all the
outputs of user calls of [[#ngx.say|ngx.say]] and [[#ngx.print|ngx.print]] and
postpone sending response headers until it sees all the outputs in the response
body, and at that time ngx_lua can calculate the total length of the body and
construct a proper <code>Content-Length</code> header for the HTTP 1.0 client.
Note that, common HTTP benchmark tools like <code>ab</code> and <code>http_load</code> always issue
HTTP 1.0 requests by default. To force <code>curl</code> to send HTTP 1.0 requests, use
the <code>-0</code> option.
= Data Sharing within an Nginx Worker =
'''NOTE: This mechanism behaves differently when code cache is turned off, and should be considered as a DIRTY TRICK. Backward compatibility is NOT guaranteed. Use at your own risk! We're going to design a whole new data-sharing mechanism.'''
If you want to globally share user data among all the requests handled by the same nginx worker process, you can encapsulate your shared data into a Lua module, require the module in your code, and manipulate shared data through it. It works because required Lua modules are loaded only once, and all coroutines will share the same copy of the module. Note however that Lua global variables WILL NOT persist between requests because of the one-coroutine-per-request isolation design.
Here is a complete small example:
<geshi lang="lua">
-- mydata.lua
module("mydata", package.seeall)
local data = {
dog = 3,
cat = 4,
pig = 5,
}
function get_age(name)
return data[name]
end
</geshi>
and then accessing it from your nginx.conf:
<geshi lang="nginx">
location /lua {
content_lua_by_lua '
local mydata = require("mydata")
ngx.say(mydata.get_age("dog"))
';
}
</geshi>
Your <code>mydata</code> module in this example will only be loaded
and run on the first request to the location <code>/lua</code>,
and all those subsequent requests to the same nginx
worker process will use the reloaded instance of the
module as well as the same copy of the data in it,
until you send a <code>HUP</code> signal to the nginx master
process to enforce a reload.
This data sharing technique is essential for high-performance Lua apps built atop this module. It is common to cache reusable data globally.
It is worth noting that this is ''per-worker'' sharing, not ''per-server'' sharing. That is, when you have multiple nginx worker processes under an nginx master, this data sharing cannot pass process boundary. If you indeed need server-wide data sharing, you can
# Use only a single nginx worker and a single server. This is not recommended when you have a multi-core CPU or multiple CPUs in a single machine.
# Use some true backend storage like <code>memcached</code>, <code>redis</code>, or an RDBMS like <code>mysql</code>.
= Known Issues =
== Lua Coroutine Yielding/Resuming ==
* As the module's predefined Nginx I/O API uses the coroutine yielding/resuming mechanism, user code should not call any Lua modules that use the Lua coroutine mechanism in order to prevent conflicts with the module's predefined Nginx API methods such as [[#ngx.location.capture|ngx.location.capture]] (Actually, coroutine modules have been masked off in [[#content_by_lua|content_by_lua]] directives and others). This limitation is significant and work is ongoing on an alternative coroutine implementation that can fit into the Nginx event model to address this. When this is done, it will be possible to use the Lua coroutine mechanism freely as it is in standard Lua implementations.
* Lua's <code>dofile</code> builtin is implemented as a C function in both Lua 5.1 and LuaJIT 2.0 and when you call [[#ngx.location.capture|ngx.location.capture]], [[#ngx.exec|ngx.exec]], [[#ngx.exit|ngx.exit]] or [[#ngx.req.read_body|ngx.req.read_body]] or similar in the file to be loaded by <code>dofile</code>, a coroutine yield across the C function boundary will be initiated. This however is not allowed within ngx_lua and will usually result in error messages like <code>lua handler aborted: runtime error: attempt to yield across C-call boundary</code>. To avoid this, define a real Lua module in your <code>.lua</code> file and use Lua's <code>require</code> builtin instead.
* Because the standard Lua 5.1 interpreter's VM is not fully resumable, the methods [[#ngx.location.capture|ngx.location.capture]], [[#ngx.location.capture_multi|ngx.location.capture_multi]], [[#ngx.redirect|ngx.redirect]], [[#ngx.exec|ngx.exec]], and [[#ngx.exit|ngx.exit]] cannot be used within the context of a Lua [http://www.lua.org/manual/5.1/manual.html#pdf-pcall pcall()] or [http://www.lua.org/manual/5.1/manual.html#pdf-xpcall xpcall()] when the standard Lua 5.1 interpreter is used; you'll get the error <code>attempt to yield across metamethod/C-call boundary</code>. To fix this, please use LuaJIT 2.0 instead, because LuaJIT 2.0 supports a fully resume-able VM.
== Lua Variable Scope ==
Care should be taken when importing modules and this form should be used:
<geshi lang="nginx">
local xxx = require('xxx')
</geshi>
: instead of the old deprecated form:
<geshi lang="nginx">
require('xxx')
</geshi>
: If you have to use the old form, force reload the module for every request by using the <code>package.loaded.<module></code> command:
<geshi lang="nginx">
package.loaded.xxx = nil
require('xxx')
</geshi>
It is recommended to always place the following piece of code at the end of Lua modules that use the [[#ngx.location.capture|ngx.location.capture]] or [[#ngx.location.capture_multi|ngx.location.capture_multi]] directives to prevent casual use of module-level global variables that are shared among ''all'' requests:
<geshi lang="nginx">
getmetatable(foo.bar).__newindex = function (table, key, val)
error('Attempt to write to undeclared variable "' .. key .. '": '
.. debug.traceback())
end
</geshi>
Assuming your current Lua module is named <code>foo.bar</code>, this will guarantee that local variables in module <code>foo.bar</code> functions have been declared as "local". It prevents undesirable race conditions while accessing such variables. See [[#Data_Sharing_within_an_Nginx_Worker|Data Sharing within an Nginx Worker]] for the reasons behind this.
== Locations With [[HttpEchoModule]] Directives ==
The [[#ngx.location.capture|ngx.location.capture]] and [[#ngx.location.capture_multi|ngx.location.capture_multi]] directives cannot capture locations that include the [[HttpEchoModule#echo_location|echo_location]], [[HttpEchoModule#echo_location_async|echo_location_async]], [[HttpEchoModule#echo_subrequest|echo_subrequest]], or [[HttpEchoModule#echo_subrequest_async|echo_subrequest_async]] directives.
<geshi lang="nginx">
location /foo {
content_by_lua '
res = ngx.location.capature("/bar")
';
}
location /bar {
echo_location /blah;
}
location /blah {
echo "Success!";
}
</geshi>
<geshi lang="nginx">
$ curl -i http://example.com/foo
</geshi>
will not work as expected.
== Special PCRE Sequences ==
PCRE sequences such as <code>\d</code>, <code>\s</code>, or <code>\w</code>, require special attention because in string literals, the backslash character, <code>\</code>, is stripped out by both the Lua language parser and by the Nginx config file parser before processing. So the following snippet will not work as expected:
<geshi lang="nginx">
# nginx.conf
? location /test {
? content_by_lua '
? local regex = "\d+" -- THIS IS WRONG!!
? local m = ngx.re.match("hello, 1234", regex)
? if m then ngx.say(m[0]) else ngx.say("not matched!") end
? ';
? }
# evaluates to "not matched!"
</geshi>
To avoid this, you need to ''double'' escape the backslash:
<geshi lang="nginx">
# nginx.conf
location /test {
content_by_lua '
local regex = "\\\\d+"
local m = ngx.re.match("hello, 1234", regex)
if m then ngx.say(m[0]) else ngx.say("not matched!") end
';
}
# evaluates to "1234"
</geshi>
Here, <code>\\\\d+</code> is stripped down to <code>\\d+</code> by the Nginx config file parser and this is further stripped down to <code>\d+</code> by the Lua language parser before running.
Alternatively, you can present the regex pattern as a long-bracketed lua string literal by encasing it in "long brackets", <code>&#91;[...]]</code>, in which case backslashes have to only be escaped once for the Nginx config file parser.
<geshi lang="nginx">
# nginx.conf
location /test {
content_by_lua '
local regex = [[\\d+]]
local m = ngx.re.match("hello, 1234", regex)
if m then ngx.say(m[0]) else ngx.say("not matched!") end
';
}
# evaluates to "1234"
</geshi>
Here, <code>&#91;[\\d+]]</code> is stripped down to <code>&#91;[\d+]]</code> by the Nginx config file parser and this is processed correctly.
Note that you may need to use a longer from of the long bracket, <code>[=[...]=]</code>, if the regex pattern contains <code>&#91;...]</code> sequences.
You can also, if you wish, use the <code>[=[...]=]</code> form as your default form and it may help with readability if you put a space between your long brackets and your regex patterns.
<geshi lang="nginx">
# nginx.conf
location /test {
content_by_lua '
local regex = [=[ [0-9]+ ]=]
local m = ngx.re.match("hello, 1234", regex)
if m then ngx.say(m[0]) else ngx.say("not matched!") end
';
}
# evaluates to "1234"
</geshi>
An alternative approach to escaping PCRE sequences is to ensure that Lua code is placed in external script files and executed using the various <code>*_by_lua_file</code> directives.
With this approach, the backslashes are only stripped by the Lua language parser and therefore only need to be escaped once each.
<geshi lang="lua">
-- test.lua
local regex = "\\d+"
local m = ngx.re.match("hello, 1234", regex)
if m then ngx.say(m[0]) else ngx.say("not matched!") end
-- evaluates to "1234"
</geshi>
Within external script files, PCRE sequences presented as long-bracketed lua string literals do not require modification.
<geshi lang="lua">
-- test.lua
local regex = [[\d+]]
local m = ngx.re.match("hello, 1234", regex)
if m then ngx.say(m[0]) else ngx.say("not matched!") end
-- evaluates to "1234"
</geshi>
= Performance =
The Lua state (aka the Lua vm instance) is shared across all the requests
handled by a single nginx worker process to miminize memory use.
On a ThinkPad T400 2.80 GHz laptop, the HelloWorld example is easy to achieve 28k req/sec using <code>http_load -p 10</code>. In contrast, Nginx + php-fpm 5.2.8 + Unix Domain Socket yields 6k req/sec and [http://nodejs.org/ Node.js] v0.6.1 yields 10.2k req/sec for their HelloWorld equivalents.
You can get better performance when building this module with [http://luajit.org/luajit.html LuaJIT 2.0].
= Typical Use Cases =
Just to name a few:
* Mashup'ing and processing outputs of various nginx upstream outputs (proxy, drizzle, postgres, redis, memcached, and etc) in Lua,
* doing arbitrarily complex access control and security checks in Lua before requests actually reach the upstream backends,
* manipulating response headers in an arbitrary way (by Lua)
* fetching backend information from external storage backends (like redis, memcached, mysql, postgresql) and use that information to choose which upstream backend to access on-the-fly,
* coding up arbitrarily complex web applications in a content handler using synchronous but still non-blocking access to the database backends and other storage,
* doing very complex URL dispatch in Lua at rewrite phase,
* using Lua to implement advanced caching mechanism for nginx subrequests and arbitrary locations.
The possibilities are unlimited as the module allows bringing together various elements within Nginx as well as exposing the power of the Lua language to the user. The module provides the full flexibility of scripting while offering performance levels comparable with native C language programs both in terms of CPU time as well as memory footprint. This is particularly the case when LuaJIT 2.0 is enabled.
Other scripting language implementations typically struggle to match this performance level.
= Installation =
You're recommended to install this module as well as the Lua 5.1 interpreter or LuaJIT 2.0 (with many other good stuffs) via the ngx_openresty bundle:
http://openresty.org
The installation steps are usually as simple as <code>./configure && make && make install</code>.
Alternatively, you can compile this module with nginx core's source by hand:
# Install Lua 5.1 or LuaJIT 2.0 into your system. Lua can be obtained freely from its project [http://www.lua.org/ homepage]. For Ubuntu/Debian users, just install the liblua5.1-0-dev package (or something like that).
# Download the latest version of the release tarball of the ngx_devel_kit (NDK) module from lua-nginx-module [http://github.com/simpl/ngx_devel_kit/tags file list].
# Download the latest version of the release tarball of this module from lua-nginx-module [http://github.com/chaoslawful/lua-nginx-module/tags file list].
# Grab the nginx source code from [http://nginx.org/ nginx.org], for example, the version 1.0.10 (see nginx compatibility), and then build the source with this module:
<geshi lang="bash">
wget 'http://nginx.org/download/nginx-1.0.10.tar.gz'
tar -xzvf nginx-1.0.10.tar.gz
cd nginx-1.0.10/
# tell nginx's build system where to find lua:
export LUA_LIB=/path/to/lua/lib
export LUA_INC=/path/to/lua/include
# or tell where to find LuaJIT when you want to use JIT instead
# export LUAJIT_LIB=/path/to/luajit/lib
# export LUAJIT_INC=/path/to/luajit/include/luajit-2.0
# Here we assume you would install you nginx under /opt/nginx/.
./configure --prefix=/opt/nginx \
--add-module=/path/to/ngx_devel_kit \
--add-module=/path/to/lua-nginx-module
make -j2
make install
</geshi>
= Compatibility =
The module is compatible with the following versions of Nginx:
* 1.1.x (last tested: 1.1.5)
* 1.0.x (last tested: 1.0.10)
* 0.9.x (last tested: 0.9.4)
* 0.8.x >= 0.8.54 (last tested: 0.8.54)
The module '''may''' work with Nginx Versions 0.7.x and 0.8.x < 0.8.54 but these have not been tested.
Nginx Versions < 0.7.0 are '''not''' compatible.
Please consider filing a bug report if you find that any particular version of Nginx above Version 0.8.54 does not
work with this module.
= Report Bugs =
Although a lot of effort has been put into testing and code tuning, there must be some serious bugs lurking somewhere in this module. So whenever you are bitten by any quirks, please do not hesitate to
# create a ticket on the [http://github.com/chaoslawful/lua-nginx-module/issues issue tracking interface] provided by GitHub,
# or send a bug report or even patches to the [http://mailman.nginx.org/mailman/listinfo/nginx nginx mailing list].
= Source Repository =
Available on github at [http://github.com/chaoslawful/lua-nginx-module chaoslawful/lua-nginx-module].
= Test Suite =
To run the test suite, you also need the following dependencies:
* Nginx version >= 0.8.54
* Perl modules:
** test-nginx: http://github.com/agentzh/test-nginx
* Nginx modules:
** echo-nginx-module: http://github.com/agentzh/echo-nginx-module
** drizzle-nginx-module: http://github.com/chaoslawful/drizzle-nginx-module
** rds-json-nginx-module: http://github.com/agentzh/rds-json-nginx-module
** set-misc-nginx-module: http://github.com/agentzh/set-misc-nginx-module
** headers-more-nginx-module: http://github.com/agentzh/headers-more-nginx-module
** memc-nginx-module: http://github.com/agentzh/memc-nginx-module
** srcache-nginx-module: http://github.com/agentzh/srcache-nginx-module
** ngx_auth_request: http://mdounin.ru/hg/ngx_http_auth_request_module/
* C libraries:
** yajl: https://github.com/lloyd/yajl
* Lua modules:
** lua-yajl: https://github.com/brimworks/lua-yajl
*** Note: the compiled module has to be placed in '/usr/local/lib/lua/5.1/'
* Applications:
** mysql: create database 'ngx_test', grant all privileges to user 'ngx_test', password is 'ngx_test'
** memcached
The adding order of these modules is important as the position of any filter module in the
filtering chain determines the final output. The correct adding order during configuration is:
# ngx_devel_kit
# set-misc-nginx-module
# ngx_http_auth_request_module
# echo-nginx-module
# memc-nginx-module
# lua-nginx-module (i.e. this module)
# headers-more-nginx-module
# srcache-nginx-module
# drizzle-nginx-module
# rds-json-nginx-module
= TODO =
* add <code>ignore_resp_headers</code>, <code>ignore_resp_body</code>, and <code>ignore_resp</code> options to [[#ngx.location.capture|ngx.location.capture]] and ngx.location.capture_multi` methods, to allow micro performance tuning on the user side.
* add directives to run lua codes when nginx stops/reloads.
* deal with TCP 3-second delay problem under great connection harness.
* add options to [[#ngx.location.capture|ngx.location.capture]] and [[#ngx.location.capture_multi|ngx.location.capture_multi]] in order to share and copy a particular set of nginx variables with subrequests, specified by the user.
* add an option to [[#ngx.location.capture|ngx.location.capture]] and [[#ngx.location.capture_multi|ngx.location.capture_multi]] so as to specify the [[#ngx.ctx|ngx.ctx]] table for subrequests.
* expose nginx's shared memory facility to the Lua land.
* add support for multi-value arguments to [[#ngx.req.set_uri_args]] if its <code>args</code> argument is a Lua table.
= Future Plans =
* add the <code>lua_require</code> directive to load module into main thread's globals.
* add the "cosocket" mechamism that will emulate a common set of Lua socket API that will give you totally transparently non-blocking capability out of the box by means of a completely new upstream layer atop the nginx event model and no nginx subrequest overheads.
* add Lua code automatic time slicing support by yielding and resuming the Lua VM actively via Lua's debug hooks.
* make set_by_lua using the same mechanism as content_by_lua.
* add coroutine API back to the Lua land.
= Changes =
== v0.3.0 ==
'''New features'''
* added the [[#header_filter_by_lua|header_filter_by_lua]] and [[#header_filter_by_lua_file|header_filter_by_lua_file]] directives. thanks Liseen Wan (万珣新).
* implemented the PCRE regex API for Lua: [[#ngx.re.match|ngx.re.match]], [[#ngx.re.gmatch|ngx.re.gmatch]], [[#ngx.re.sub|ngx.re.sub]], and [[#ngx.re.gsub|ngx.re.gsub]].
* now we add the <code>ngx</code> and <code>ndk</code> table into <code>package.loaded</code> such that the user can write <code>local ngx = require 'ngx'</code> and <code>local ndk = require 'ndk'</code>. thanks @Lance.
* added new directive [[#lua_regex_cache_max_entries|lua_regex_cache_max_entries]] to control the upper limit of the worker-process-level compiled-regex cache enabled by the <code>o</code> regex option.
* implemented the special [[#ngx.ctx|ngx.ctx]] Lua table for user programmers to store per-request Lua context data for their applications. thanks 欧远宁 for suggesting this feature.
* now [[#ngx.print|ngx.print]] and [[#ngx.say|ngx.say]] allow (nested) array-like table arguments. the array elements in them will be sent piece by piece. this will avoid string concatenation for templating engines like [http://www.savarese.com/software/ltp/ ltp].
* implemented the [[#ngx.req.get_post_args|ngx.req.get_post_args]] method for fetching url-encoded POST query arguments from within Lua.
* implemented the [[#ngx.req.get_uri_args|ngx.req.get_uri_args]] method to fetch parsed URL query arguments from within Lua. thanks Bertrand Mansion (golgote).
* added new function [[#ngx.parse_http_time|ngx.parse_http_time]], thanks James Hurst.
* now we allow Lua boolean and <code>nil</code> values in arguments to [[#ngx.say|ngx.say]], [[#ngx.print|ngx.print]], [[#ngx.log|ngx.log]] and [[#print|print]].
* added support for user C macros <code>LUA_DEFAULT_PATH</code> and <code>LUA_DEFAULT_CPATH</code>. for now we can only define them in <code>ngx_lua</code>'s <code>config</code> file because nginx <code>configure</code>'s <code>--with-cc-opt</code> option hates values with double-quotes in them. sigh. [http://openresty.org/ ngx_openresty] is already using this feature to bundle 3rd-party Lua libraries.
'''Bug fixes'''
* worked-around the "stack overflow" issue while using <code>luarocks.loader</code> and disabling [[#lua_code_cache|lua_code_cache]], as described as github issue #27. thanks Patrick Crosby.
* fixed the <code>zero size buf in output</code> alert while combining [[#lua_need_request_body|lua_need_request_body]] on + [[#access_by_lua|access_by_lua]]/[[#rewrite_by_lua|rewrite_by_lua]] + [[HttpProxyModule#proxy_pass|proxy_pass]]/[[HttpFcgiModule#fastcgi_pass|fastcgi_pass]]. thanks Liseen Wan (万珣新).
* fixed issues with HTTP 1.0 HEAD requests.
* made setting <code>ngx.header.HEADER</code> after sending out response headers throw out a Lua exception to help debugging issues like github issue #49. thanks Bill Donahue (ikhoyo).
* fixed an issue regarding defining global variables in C header files: we should have defined the global <code>ngx_http_lua_exception</code> in a single compilation unit. thanks @姜大炮.
= Authors =
* chaoslawful (王晓哲) <chaoslawful at gmail dot com>
* Zhang "agentzh" Yichun (章亦春) <agentzh at gmail dot com>
= Copyright & License =
This module is licenced under the BSD license.
Copyright (C) 2009-2011, by Xiaozhe Wang (chaoslawful) <chaoslawful@gmail.com>.
Copyright (C) 2009-2011, by Zhang "agentzh" Yichun (章亦春) <agentzh@gmail.com>.
All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
= See Also =
* [http://openresty.org/#RoutingMySQLQueriesBasedOnURIArgs Routing requests to different MySQL queries based on URI arguments]
* [http://openresty.org/#DynamicRoutingBasedOnRedis Dynamic Routing Based on Redis and Lua]
* [http://openresty.org/#UsingLuaRocks Using LuaRocks with ngx_lua]
* [https://github.com/chaoslawful/lua-nginx-module/wiki/Introduction Introduction to ngx_lua]
* [http://github.com/simpl/ngx_devel_kit ngx_devel_kit]
* [[HttpEchoModule]]
* [[HttpDrizzleModule]]
* [http://github.com/FRiCKLE/ngx_postgres postgres-nginx-module]
* [[HttpMemcModule]]
* [http://openresty.org The ngx_openresty bundle]
Jump to Line
Something went wrong with that request. Please try again.