23 KiB
Overview
Mongoose is small and easy to use web server. It is self-contained, and does not require any external software to run.
On Windows, mongoose iconifies itself to the system tray icon when started.
Right-click on the icon pops up a menu, where it is possible to stop
mongoose, or configure it, or install it as Windows service. The easiest way
to share a folder on Windows is to copy mongoose.exe
to a folder,
double-click the exe, and launch a browser at
http://localhost:8080. Note that 'localhost' should
be changed to a machine's name if a folder is accessed from other computer.
On UNIX and Mac, mongoose is a command line utility. Running mongoose
in
terminal, optionally followed by configuration parameters
(mongoose [OPTIONS]
) or configuration file name
(mongoose [config_file_name]
) starts the
web server. Mongoose does not detach from terminal. Pressing Ctrl-C
keys
would stop the server.
When started, mongoose first searches for the configuration file.
If configuration file is specified explicitly in the command line, i.e.
mongoose path_to_config_file
, then specified configuration file is used.
Otherwise, mongoose would search for file mongoose.conf
in the same directory
where binary is located, and use it. Configuration file can be absent.
Configuration file is a sequence of lines, each line containing
command line argument name and it's value. Empty lines, and lines beginning
with #
, are ignored. Here is the example of mongoose.conf
file:
document_root c:\www
listening_ports 8080,8043s
ssl_certificate c:\mongoose\ssl_cert.pem
When configuration file is processed, mongoose process command line arguments,
if they are specified. Command line arguments therefore can override
configuration file settings. Command line arguments must start with -
.
For example, if mongoose.conf
has line
document_root /var/www
, and mongoose has been started as
mongoose -document_root /etc
, then /etc
directory will be served as
document root, because command line options take priority over
configuration file. Configuration options section below provide a good
overview of Mongoose features.
Note that configuration options on the command line must start with -
,
but their names are the same as in the config file. All option names are
listed in the next section. Thus, the following two setups are equivalent:
# Using command line arguments
$ mongoose -listening_ports 1234 -document_root /var/www
# Using config file
$ cat mongoose.conf
listening_ports 1234
document_root /var/www
$ mongoose
Mongoose can also be used to modify .htpasswd
passwords file:
mongoose -A <htpasswd_file> <realm> <user> <passwd>
Unlike other web servers, mongoose does not require CGI scripts be located in a special directory. CGI scripts can be anywhere. CGI (and SSI) files are recognized by the file name pattern. Mongoose uses shell-like glob patterns. Pattern match starts at the beginning of the string, so essentially patterns are prefix patterns. Syntax is as follows:
** Matches everything
* Matches everything but slash character, '/'
? Matches any character
$ Matches the end of the string
| Matches if pattern on the left side or the right side matches.
All other characters in the pattern match themselves. Examples:
**.cgi$ Any string that ends with .cgi
/foo Any string that begins with /foo
**a$|**b$ Any string that ends with a or b
Configuration Options
Below is a list of configuration options Mongoose understands. Every option is followed by it's default value. If default value is not present, then it is empty.
cgi_pattern **.cgi$|**.pl$|**.php$
All files that match cgi_pattern
are treated as CGI files. Default pattern
allows CGI files be anywhere. To restrict CGIs to a certain directory,
use /path/to/cgi-bin/**.cgi
as pattern. Note that full file path is
matched against the pattern, not the URI.
cgi_environment
Extra environment variables to be passed to the CGI script in
addition to standard ones. The list must be comma-separated list
of name=value pairs, like this: VARIABLE1=VALUE1,VARIABLE2=VALUE2
.
put_delete_passwords_file
Passwords file for PUT and DELETE requests. Without it, PUT and DELETE requests will fail.
cgi_interpreter
Path to an executable to use as CGI interpreter for all CGI scripts regardless script extension. If this option is not set (which is a default), Mongoose looks at first line of a CGI script, shebang line, for an interpreter.
For example, if both PHP and perl CGIs are used, then
#!/path/to/php-cgi.exe
and #!/path/to/perl.exe
must be first lines of the
respective CGI scripts. Note that paths should be either full file paths,
or file paths relative to the current working directory of mongoose server.
If mongoose is started by mouse double-click on Windows, current working
directory is a directory where mongoose executable is located.
If all CGIs use the same interpreter, for example they are all PHP, then
cgi_interpreter
can be set to the path to php-cgi.exe
executable and
shebang line in the CGI scripts can be omitted.
Note that PHP scripts must use php-cgi.exe
executable, not php.exe
.
protect_uri
Comma separated list of URI=PATH pairs, specifying that given URIs must be protected with respected password files. Paths must be full file paths.
authentication_domain mydomain.com
Authorization realm used in .htpasswd
authorization.
ssi_pattern **.shtml$|**.shtm$
All files that match ssi_pattern
are treated as SSI.
Server Side Includes (SSI) is a simple interpreted server-side scripting language which is most commonly used to include the contents of a file into a web page. It can be useful when it is desirable to include a common piece of code throughout a website, for example, headers and footers.
In order for a webpage to recognize an SSI-enabled HTML file, the filename
should end with a special extension, by default the extension should be
either .shtml
or .shtm
.
Unknown SSI directives are silently ignored by mongoose. Currently, two SSI
directives are supported, <!--#include ...>
and
<!--#exec "command">
. Note that <!--#include ...>
directive supports
three path specifications:
<!--#include virtual="path"> Path is relative to web server root
<!--#include file="path"> Path is relative to web server working dir
<!--#include "path"> Path is relative to current document
The include
directive may be used to include the contents of a file or the
result of running a CGI script. The exec
directive is used to execute a
command on a server, and show command's output. Example:
<!--#exec "ls -l" -->
For more information on Server Side Includes, take a look at the Wikipedia: Server Side Includes
throttle
Limit download speed for clients. throttle
is a comma-separated
list of key=value pairs, where key could be:
* limit speed for all connections
x.x.x.x/mask limit speed for specified subnet
uri_prefix_pattern limit speed for given URIs
The value is a floating-point number of bytes per second, optionally
followed by a k
or m
character, meaning kilobytes and
megabytes respectively. A limit of 0 means unlimited rate. The
last matching rule wins. Examples:
*=1k,10.0.0.0/8=0 limit all accesses to 1 kilobyte per second,
but give connections from 10.0.0.0/8 subnet
unlimited speed
/downloads/=5k limit accesses to all URIs in `/downloads/` to
5 kilobytes per secods. All other accesses are unlimited
access_log_file
Path to a file for access logs. Either full path, or relative to current working directory. If absent (default), then accesses are not logged.
error_log_file
Path to a file for error logs. Either full path, or relative to current working directory. If absent (default), then errors are not logged.
enable_directory_listing yes
Enable directory listing, either yes
or no
.
global_passwords_file
Path to a global passwords file, either full path or relative to the current
working directory. If set, per-directory .htpasswd
files are ignored,
and all requests are authorised against that file.
index_files index.html,index.htm,index.cgi,index.shtml,index.php
Comma-separated list of files to be treated as directory index files.
access_control_list
An Access Control List (ACL) allows restrictions to be put on the list of IP
addresses which have access to the web server. In the case of the Mongoose
web server, the ACL is a comma separated list of IP subnets, where each
subnet is prepended by either a -
or a +
sign. A plus sign means allow,
where a minus sign means deny. If a subnet mask is omitted, such as -1.2.3.4
,
this means to deny only that single IP address.
Subnet masks may vary from 0 to 32, inclusive. The default setting is to allow all accesses. On each request the full list is traversed, and the last match wins. Examples:
-0.0.0.0/0,+192.168/16 deny all acccesses, only allow 192.168/16 subnet
To learn more about subnet masks, see the Wikipedia page on Subnetwork
extra_mime_types
Extra mime types to recognize, in form extension1=type1,exten- sion2=type2,...
. Extension must include dot. Example:
.cpp=plain/text,.java=plain/text
listening_ports 8080
Comma-separated list of ports to listen on. If the port is SSL, a
letter s
must be appeneded, for example, 80,443s
will open
port 80 and port 443, and connections on port 443 will be SSL-ed.
For non-SSL ports, it is allowed to append letter r
, meaning 'redirect'.
Redirect ports will redirect all their traffic to the first configured
SSL port. For example, if listening_ports
is 80r,443s
, then all
HTTP traffic coming at port 80 will be redirected to HTTPS port 443.
It is possible to specify an IP address to bind to. In this case,
an IP address and a colon must be prepended to the port number.
For example, to bind to a loopback interface on port 80 and to
all interfaces on HTTPS port 443, use 127.0.0.1:80,443s
.
document_root .
A directory to serve. By default, currect directory is served. Current
directory is commonly referenced as dot (.
).
ssl_certificate
Path to SSL certificate file. This option is only required when at least one
of the listening_ports
is SSL.
num_threads 50
Number of worker threads. Mongoose handles each incoming connection in a separate thread. Therefore, the value of this option is effectively a number of concurrent HTTP connections Mongoose can handle.
run_as_user
Switch to given user credentials after startup. Usually, this option is required when mongoose needs to bind on privileged port on UNIX. To do that, mongoose needs to be started as root. But running as root is a bad idea, therefore this option can be used to drop privileges. Example:
mongoose -listening_ports 80 -run_as_user nobody
url_rewrite_patterns
Comma-separated list of URL rewrites in the form of
uri_pattern=file_or_directory_path
. When Mongoose receives the request,
it constructs the file name to show by combining document_root
and the URI.
However, if the rewrite option is used and uri_pattern
matches the
requested URI, then document_root
is ignored. Insted,
file_or_directory_path
is used, which should be a full path name or
a path relative to the web server's current working directory. Note that
uri_pattern
, as all mongoose patterns, is a prefix pattern.
This makes it possible to serve many directories outside from document_root
,
redirect all requests to scripts, and do other tricky things. For example,
to redirect all accesses to .doc
files to a special script, do:
mongoose -url_rewrite_patterns **.doc$=/path/to/cgi-bin/handle_doc.cgi
Or, to imitate user home directories support, do:
mongoose -url_rewrite_patterns /~joe/=/home/joe/,/~bill=/home/bill/
hide_files_patterns
A pattern for the files to hide. Files that match the pattern will not
show up in directory listing and return 404 Not Found
if requested. Pattern
must be for a file name only, not including directory name. Example:
mongoose -hide_files_patterns secret.txt|even_more_secret.txt
Lua Server Pages
Pre-built Windows and Mac mongoose binaries have built-in Lua Server Pages support. That means it is possible to write PHP-like scripts with mongoose, using Lua programming language instead of PHP. Lua is known for it's speed and small size. Mongoose uses Lua version 5.2.1, the documentation for it can be found at Lua 5.2 reference manual.
To create a Lua Page, make sure a file has .lp
extension. For example,
let's say it is going to be my_page.lp
. The contents of the file, just like
with PHP, is HTML with embedded Lua code. Lua code must be enclosed in
<? ?>
blocks, and can appear anywhere on the page. For example, to
print current weekday name, one can write:
<p>
<span>Today is:</span>
<? print(os.date("%A")) ?>
</p>
Note that this example uses function print()
, which prints data to the
web page. Using function print()
is the way to generate web content from
inside Lua code. In addition to print()
, all standard library functions
are accessible from the Lua code (please check reference manual for details),
and also information about the request is available in request_info
object,
like request method, all headers, etcetera. Please refer to
struct mg_request_info
definition in
mongoose.h
to see what kind of information is present in request_info
object. Also,
page.lp and
prime_numbers.lp
contains some example code that uses request_info
and other functions(form submitting for example).
One substantial difference of mongoose's Lua Pages and PHP is that Mongoose expects Lua page to output HTTP headers. Therefore, at the very beginning of every Lua Page must be a Lua block that outputs HTTP headers, like this:
<? print('HTTP/1.0 200 OK\r\nContent-Type: text/html\r\n\r\n') ?>
<html><body>
... the rest of the web page ...
It is easy to do things like redirects, for example:
<? print('HTTP/1.0 302 Found\r\nLocation: http://google.com\r\n\r\n') ?>
To serve Lua Page, mongoose creates Lua context. That context is used for all Lua blocks within the page. That means, all Lua blocks on the same page share the same context. If one block defines a variable, for example, that variable is visible in the block that follows.
Common Problems
-
PHP doesn't work - getting empty page, or 'File not found' error. The reason for that is wrong paths to the interpreter. Remember that with PHP, correct interpreter is
php-cgi.exe
(php-cgi
on UNIX). Solution: specify full path to the PHP interpreter, e.g.:mongoose -cgi_interpreter /full/path/to/php-cgi
-
Mongoose fails to start. If Mongoose exits immediately when run, this usually indicates a syntax error in the configuration file (named
mongoose.conf
by default) or the command-line arguments. Syntax checking is omitted from Mongoose to keep its size low. However, the Manual should be of help. Note: the syntax changes from time to time, so updating the config file might be necessary after executable update.
Embedding
Embedding Mongoose is easy. Copy
mongoose.c and
mongoose.h
to your application's source tree and include them in the build. For
example, your application's code lives in C++ file my_app.cpp
, then on UNIX
this command embeds Mongoose:
$ ls
my_app.cpp mongoose.c mongoose.h
$ g++ my_app.cc mongoose.c -o my_app
Somewhere in the application code, call mg_start()
to start the server.
Pass configuration options and event handlers to mg_start()
.
Mongoose then calls handlers when certain events happen.
For example, when new request arrives, Mongoose calls begin_request
handler function to let user handle the request. In the handler, user code
can get all information about the request -- parsed headers, etcetera.
Mongoose API is logically divided in three categories: server setup/shutdown functions, functions to be used by user-written event handlers, and convenience utility functions.
Starting and stopping embedded web server
To start the embedded web server, call mg_start()
. To stop it, call
mg_stop()
.
// This structure needs to be passed to mg_start(), to let mongoose know
// which callbacks to invoke. For detailed description, see
// https://github.com/valenok/mongoose/blob/master/UserManual.md
struct mg_callbacks {
int (*begin_request)(struct mg_connection *);
void (*end_request)(const struct mg_connection *, int reply_status_code);
int (*log_message)(const struct mg_connection *, const char *message);
int (*init_ssl)(void *ssl_context);
int (*websocket_connect)(const struct mg_connection *);
void (*websocket_ready)(struct mg_connection *);
int (*websocket_data)(struct mg_connection *);
const char * (*open_file)(const struct mg_connection *,
const char *path, size_t *data_len);
void (*init_lua)(struct mg_connection *, void *lua_context);
void (*upload)(struct mg_connection *, const char *file_name);
int (*http_error)(struct mg_connection *, int status);
};
hello.c provides a minimalistic example.
Common pattern is to implement begin_request
callback, and serve static files
from memory, and/or construct dynamic replies on the fly. Here is
my embed.c gist
that shows how to easily any data can be embedded
directly into the executable. If such data needs to be encrypted, then
encrypted database or encryption dongles would be a better choice.
Build on Android
This is a small guide to help you run mongoose on Android. Currently it is tested on the HTC Wildfire. If you have managed to run it on other devices as well, please comment or drop an email in the mailing list. Note : You dont need root access to run mongoose on Android.
- Download the source from the Downloads page.
- Download the Android NDK from here
- Make a folder (e.g. mongoose) and inside that make a folder named "jni".
- Add
mongoose.h
,mongoose.c
andmain.c
from the source to the jni folder. - Make a new file in the jni folder named "Android.mk". This is the make file for ndk-build.
Android.mk:
LOCAL_PATH := $(call my-dir)
include $(CLEAR_VARS)
LOCAL_MODULE := mongoose
LOCAL_SRC_FILES := main.c mongoose.c
include $(BUILD_EXECUTABLE)
- Run
./ndk-build -C /path/to/mongoose/
. This should generate mongoose/lib/armeabi/mongoose - Using the adb tool, push the generated mongoose binary to
/data/local
folder on device. - From adb shell, navigate to
/data/local
and execute./mongoose
. - To test if the server is running fine, visit your web-browser and
navigate to
http://127.0.0.1:8080
You should see theIndex of /
page.
Notes:
- jni stands for Java Native Interface. Read up on Android NDK if you want to know how to interact with the native C functions of mongoose in Android Java applications.
- Download android-sdk for the adb tool.
- TODO: A Java application that interacts with the native binary or a shared library.
Mongoose internals
Mongoose is multithreaded web server. mg_start()
function allocates
web server context (struct mg_context
), which holds all information
about web server instance:
- configuration options. Note that mongoose makes internal copies of passed options.
- SSL context, if any
- user-defined callbacks
- opened listening sockets
- a queue for accepted sockets
- mutexes and condition variables for inter-thread synchronization
When mg_start()
returns, all initialization is quaranteed to be complete
(e.g. listening ports are opened, SSL is initialized, etc). mg_start()
starts
two threads: a master thread, that accepts new connections, and several
worker threads, that process accepted connections. The number of worker threads
is configurable via num_threads
configuration option. That number puts a
limit on number of simultaneous requests that can be handled by mongoose.
When master thread accepts new connection, a new accepted socket (described by
struct socket
) it placed into the accepted sockets queue,
which has size of 20 (see code). Any idle worker thread
can grab accepted sockets from that queue. If all worker threads are busy,
master thread can accept and queue up to 20 more TCP connections,
filling up the queue.
In the attempt to queue next accepted connection, master thread blocks
until there is space in a queue. When master thread is blocked on a
full queue, TCP layer in OS can also queue incoming connection.
The number is limited by the listen()
call parameter on listening socket,
which is SOMAXCONN
in case of Mongoose, and depends on a platform.
Worker threads are running in an infinite loop, which in simplified form looks something like this:
static void *worker_thread() {
while (consume_socket()) {
process_new_connection();
}
}
Function consume_socket()
gets new accepted socket from the mongoose socket
queue, atomically removing it from the queue. If the queue is empty,
consume_socket()
blocks and waits until new sockets are placed in a queue
by the master thread. process_new_connection()
actually processes the
connection, i.e. reads the request, parses it, and performs appropriate action
depending on a parsed request.
Master thread uses poll()
and accept()
to accept new connections on
listening sockets. poll()
is used to avoid FD_SETSIZE
limitation of
select()
. Since there are only a few listening sockets, there is no reason
to use hi-performance alternatives like epoll()
or kqueue()
. Worker
threads use blocking IO on accepted sockets for reading and writing data.
All accepted sockets have SO_RCVTIMEO
and SO_SNDTIMEO
socket options set
(controlled by request_timeout_ms
mongoose option, 30 seconds default) which
specify read/write timeout on client connection.
Other Resources
- Presentation made by Arnout Vandecappelle at FOSDEM 2011 on 2011-02-06 in Brussels, Belgium, called "Creating secure web based user interfaces for Embedded Devices" (pdf | odp)
- Linux Journal article by Michel J.Hammel, 2010-04-01, called Mongoose: an Embeddable Web Server in C