• multithreaded request processing;
• high tolerance for configuration and runtime errors;
• built-in HTTPD to serve redirection pages;
• on-the-fly reconfiguration (without Squid restart).

Erjik uses FreeBSD License. You can find full license text on http://www.freebsd.org/copyright/freebsd-license.html or in file LICENSE on the top of the Erjik source tree.

• mime-support;
• erlang;
• erlang-inets.

Build requires:

• make;
• erlang-dev;
• erlang-edoc (optional).

Erlang must be installed to build and run Erjik. You always can obtain latest Erlang version on http://www.erlang.org/download.html or use the one provided by your software repository.

Erjik is developed and tested with Erlang R14A but probably will work with older Erlang versions.

$ make

$ make
$ sudo make install

Use DESTDIR environment variable:

$ DESTDIR=rootfs make install
$ tree rootfs
rootfs/
├── etc
│   └── erjik.conf
├── usr
│   └── lib
│       └── erjik
│           ├── erjik
│           └── erjik-wrapper
└── var
    ├── lib
    │   └── erjik
    │       ├── domains
    │       │   ├── good
    │       │   └── porn
    │       ├── regexps
    │       │   ├── attraction
    │       │   ├── audio-video
    │       │   ├── fileserver
    │       │   ├── good
    │       │   ├── onlinegames
    │       │   ├── porn
    │       │   └── security
    │       └── www
    │           ├── denied-by-ip.html
    │           ├── denied-by-url.html
    │           └── porn.html
    └── log
        └── erjik

To completely remove the Erjik from the system type:

$ sudo make uninstall

Add this lines to your squid.conf:

url_rewrite_program /usr/lib/erjik/erjik-wrapper
url_rewrite_concurrency 1
url_rewrite_children 1

See comments in the /etc/erjik.conf file for description of available configuration parameters.

Normally Erjik is started by Squid so you don't need to launch it manually.

Apply new configuration, reopen log file:

$ /usr/lib/erjik/erjik --hup /etc/erjik.conf

Check if Erjik is alive or not:

$ /usr/lib/erjik/erjik --ping /etc/erjik.conf

Requests must be separated by newlines ('\n' or '\n\r'). Each request must contain at least three tokens, separated by spaces or TABs:

RequestID URL SourceIP

RequestID - is identifier of request to distinguish multiple requests in single pipe. An answer will be prefixed with the same RequestID string. This is the way to use multithreaded URL rewritings with one instance of URL rewriter process started in Squid.

URL - is the URL requested by the user's browser.

SourceIP - is the user's IP address.

The answer will look like:

RequestID FinalURL

FinalURL is an URL of a destination page which will be provided to the user. When the user is allowed to get the source URL, the FinalURL will be the same as the source URL otherwise the FinalURL will point to so called banpage.

In order to test Erjik you can manually start it from the shell and supply test requests on Erjik's stdin. The answers will be send to the stdout:

$ echo '1 http://www.github.com/ 10.0.0.1' | ./erjik ./erjik.conf
1 http://www.github.com/
$ echo '1 http://www.porn.com/ 10.0.0.1' | ./erjik ./erjik.conf
1 http://127.0.0.1:8888/porn.html

Aleksey Morarash [email protected], 2012