Reverse HTTP proxy to filter requests by different rules. Can be used between production webserver and the application server to prevent abuse of the application backend.
The original purpose of this program was to defend searx, but it can be used to guard any web application.
$ go install github.com/asciimoo/filtron
$ "$GOPATH/bin/filtron" --help
A rule has two required attributes: name
and actions
A rule can contain all of the following attributes:
limit
integer - Defines how many matching requests allowed to access the application withininterval
seconds. (Can be omitted if0
)interval
integer - Time range in seconds to reset rule numbers (Can be omitted iflimit
is0
)filters
list of selectorsaggregations
list of selectors (iffilters
specified it activates only in case of the filter matches)subrules
list of rules (iffilters
specified it activates only in case of the filter matches)disabled
bool - Disable a rule (default isfalse
)stop
bool - Finish request validation immediately and skip remaining rules (default isfalse
)
JSON representation of a rule:
{
"name": "example rule",
"interval": 60,
"limit": 10,
"filters": ["GET:q", "Header:User-Agent=^curl"],
"actions": [
{"name": "log",
"params": {"destination": "stderr"}},
{"name": "block",
"params": {"message": "Not allowed"}}
]
}
Explanation: Allow only 10 requests a minute where q
represented as GET parameter and the user agent header starts with curl
. Request is logged to STDERR and blocked with a custom error message if limit is exceeded. See more examples here.
Rule's actions are sequentially activated if a request exceeds rule's limit
Note: Only the rule's first action will be executed that serves custom response
Log the request
Serve HTTP 429 response instead of passing the request to the application
Execute a shell command. cmd
(string) and args
(list of selectors) are required params (Example: {"name": "shell", "params": {"cmd": "echo %v is the IP", "args": ["IP"]}}
)
If all the selectors found, it increments a counter. Rule blocks the request if counter reaches limit
Counts the values returned by selectors. Rule blocks the request if any value's number reaches limit
Each rule can contain any number of subrules. Activates on parent rule's filter match.
Request's different parts can be extracted using selector expressions.
Selectors are strings that can match any attribute of a HTTP request with the following syntax:
[!]RequestAttribute[:SubAttribute][=Expression]
!
can negate the selectorRequestAttribute
(required) selects specific part of a request - possible values:- Single value
IP
Host
Path
Method
- Multiple values
GET
POST
Param
- it is an alias for bothGET
andPOST
Cookie
Header
- Single value
SubAttribute
ifRequestAttribute
is not a single value, this can specify the inner attributeExpression
possible value:- a regular expression to filter the selected attribute values.
nslookup(Hostname)
to filter the selected attribute values with the IP addresses ofHostname
. Filtron resolvesHostname
to its IP addresses when the rule is loaded (IPv4 and IPv6).
IP
returns the client's IP address
GET:x
returns the x
GET parameter if exists
!Header:Accept-Language
returns true if there is no Accept-Language
HTTP header
Path=^/(x|y)$
matches if the path is /x
or /y
IP=nslookup(example.com)
matches if the client's IP address is one of the IP addresses of example.com.
Filtron can be configured through its REST API which listens on 127.0.0.1:4005
by default.
Loaded rules in JSON format
Reload the rule file specified at startup
UI built on the API
Bugs or suggestions? Visit the issue tracker.