AnsweredAssumed Answered

URL as a feed/indicator

Question asked by Jay Watson on May 10, 2018
Latest reply on May 17, 2018 by Dion Stempfley

Like • Show 0 Likes0
Comment • 5

I would like to add a custom feed that focuses on URL as an indicator. Using domain and mapping them to alias.host does not seem to be overly beneficial as it creates a lot noise and triggers on potentially a non-indicator. As I do not see URL as being indexed I can only think that this would be a problem and would not work as expected. Does anyone have a clever way of pulling in a feed focusing on a list of specific URLs rather then purely IPs and domains?

'http://www.example1234.com/content-access/article.html'

'http://www.example1234.com/content-access/eng/access-in/index.html'

vs.

'www.example1234.com'

No one else has this question

Outcomes

5 Replies

Christopher Ahearn May 10, 2018 11:03 AM
Mark Correct
Correct Answer
Jay,

You can create feeds based off of non-indexed meta keys. Essentially, you can create a feed from almost ANY meta key in the solution. You would just use that meta key as the meta callback.

If you currently do not have URL meta, obviously you would need that first. I know some proxy logs will supply the entire URL. A while back I wrote a parser to concatenate meta values together to build the URL. It was posted to this community site.

Please note that any TEXT formatted meta key is limited to 256 bytes of data. Therefore, if the url becomes too big, it may not match. Feeds, as you may know, are exact matches.

There may be some less efficient ways of doing the match with a parser, but I haven't explored it.

Chris
Like • Show 0 Likes0
Actions

William Motley

May 10, 2018 11:32 AM

There is an option to HTTP_lua to register url. There are some caveats:

Don't index urls.
If registerUrl is enabled, then registerComponents defaults to false. If you want both you have to enable both explicitly.
Especially don't index urls if also registering components
An actual url - e.g., the href that a user clicks on - is not in the http request. The registered url is an approximation at best.

Actions

Dion Stempfley @ William Motley on May 17, 2018 2:53 PM

In your second bullet you reference registerComponents. I don't see that function in the http_lua_options.lua. Does this mean I need to write a function registerComponents() in http_lua_options.lua that returns true?

/Dion

Actions

William Motley

@ Dion Stempfley on May 17, 2018 3:31 PM

It's there, or at least should be, as the first listed option. Perhaps you have an old version of the options file?

Here's the latest from Live:

module("HTTP_lua_options")
-- 2018.05.02.1
function registerComponents()
    --[=[
        "Register Path Components" : default TRUE (or FALSE if "Register URL" enabled)
            IMPORTANT:  It is strongly advised to not enable both this option and
            "Register URL".
            Register directory, filename, extension, query (from the request path) and
            host (from the HOST: header) as discrete meta.
            For example,
                alias.host:  www.example.com
                directory:   /someDir/
                filename:    somefile.html
                extension:   html
                query:       ?foo=bar
            If the "Register URL" option below is enabled, then this option defaults
            to FALSE.  To register both discrete meta from the path components and
            a reconstructed url, both options must be explicitly enabled.  It is
            strongly advised not to enable both options.
    --]=]
    --return true
end
function registerUrl()
    --[=[
        "Register URL" : default FALSE
            IMPORTANT:  It is strongly advised to not enable both this option and
            "Register Path Components".
            Default behavior is to register directory, filename, extension, query
            (from the request path) and host (from the HOST: header) as discrete meta.
            This option will instead register them as a single meta value (individual keys
            will not be registered due to redundancy),
                url:  www.example.com/someDir/someFile.html?foo=bar
            Note that the registered URL is a reconstructed approximation - it may not be the
            exact URL that was 'clicked on'.
    --]=]
    return false
end
function splitQuery()
    --[=[
        "Split Query String" : default false
            Default behavior is for the entire querystring from a request to be registered as an
            single meta value:
                query: alpha=one&beta=two&gamma=three
            If this option is enabled, then each element of a querystring will be registered as
            individual meta values:
                query: alpha=one
                query: beta=two
                query: gamma=three
    --]=]
    return false
end
function useOrigIP()
    --[=[
        "Use orig_ip" : default TRUE
            Default behavior is to register values from x-forwarded-for headers and the like
            with index key "orig_ip".
            If this option is disabled, then values will be registered as following:
                hostnames      "alias.host"
                IPv4           "alias.ip"
                IPv6           "alias.ipv6"
                email address  "email"
                other          "alias.host"
    --]=]
    return true
end
function refererPath()
    --[=[
        "Referer Path" : default FALSE
            Default behavior is to register the value of a "Referer:" header as "referer" meta.
            If this option is enabled, then the host, directory, filename, extension, and querystring
            values will be broken out from Referer and registered individually.  In order to
            avoid duplication, the entire Referer value will not be registered.
            For example, given the header:
                Referer: http://www.example.com/hello/world.html?foo=bar&one=two
            If this option is disabled (default), then the following meta will be registered:
                referer:  http://www.example.com/hello/world.html?foo=bar&one=two
            If enabled, then the following meta will be registered:
                alias.host:  www.example.com
                directory:   /hello/
                filename:    world.html
                extension:   html
                query:       foo=bar&one=two
            Note that if the "Split Query String" option is also enabled then the querystring
            will instead be registered individually (see above).
    --]=]
    return false
end
function userAgent()
    --[=[
        "User-Agent Key" : default "client"
            Default behavior is to register the value of User-Agent headers with the 'client'
            index key.
            Modifying this value will cause User-Agent values to additionally be registered
            with the specified key.  If the key does not already exist it will be created - normal
            key name restrictions apply.
            Note that this will result in duplication of meta.  User-agent will be registered
            to both "client" and the specified key.
    --]=]
    return "client"
end
function respReason()
    --[=[
        "Response Code Reason" : default TRUE
            For reponse codes other than 2xx, default behavior is to register both the status
            code and reason phrase together as error meta.  For example,
                error:  404 Not Found
            Disasbling this option (setting to false) will cause only the response code to be
            registered.  For example,
                error:  404
    --]=]
    return true
end
function decompress()
    --[=[
        "Decompress" : default 0
            Decompress content-encoded HTTP responses.  Encodings gzip,
            deflate, and chunked are supported.  Enabling this provides
            visibility into such responses to other parsers.
            Decompression incurs a performance penalty which will vary
            depending upon the prevalence of compressed or encoded HTTP
            responses seen in the environment.  This can be ameliorated
            to some extent by choosing to only decompress specific content
            types.
            This is a bit-packed value representing the content types to
            decompress, where:
                1    application/*
                2    audio/*
                4    font/*
                8    image/*
                16   message/*
                32   model/*
                64   text/*
                128  video/*
            The default value of 0 means that decompression will not be
            performed for any content type, which maximizes performance.
            A value of 65 specifies that content-types "application" and
            "text" will be decompressed.  This should provide a good
            balance of visibility and performance.
            To maximize visibility, a value of 255 will enable decompression
            of all content types.
            Enabling a content-type enables all constituent sub-types.  For
            example, "application" includes "application/octet-stream",
            "application/javascript", etc.
            NOTES:
                Only valid for versions 11.0+.  This option has no effect on
                versions 10.x or older as they do not have the capability to
                decompress encoded HTTP responses.
                Has no effect on instances of compression which are not HTTP
                responses, such as compressed archive files (zip, rar, et al),
                LZMA streams, etc.
    --]=]
    return 0
end
function advanced()
    --[=[
        "Advanced Analysis" : default FALSE
            Perform advanced analysis of HTTP characteristics.  Analysis includes only the first
            request and first response.  Meta is registered to the key "analysis.service".
    --]=]
    return false
end
function customHeaders()
    --[=[
        "Custom Headers" : default NONE
    
        Beware of excessive duplication, which will impact performance and retention.  Meta
        registered will be in addition to, not replacement of, standard meta registration.
        In other words, if you specify "user-agent" headers be registered to key "foo", it
        will still also be registered to alias.host (or alias.ip / alias.ipv6 if appropriate).
        
        Syntax is,
        
            ["header"] = "key",
            
        Where,
        
            "header" is the desired HTTP header in lowercase.  Do not included spaces, colons, etc.
            
            "key" is the desired meta key with which to register the value of that header
            
        Key names must be 16 characters or less, and consist only of alphanumeric, dots, and
        hyphens.  Keys specified that do not meet these requirements will be modified in order
        to conform.
        
        Keys listed here are registered as format="Text".  Don't use keys indexed in other formats.
        
    --]=]
    return {
        --["origin"] = "referer",
    }
end

Actions

Dion Stempfley @ William Motley on May 17, 2018 4:07 PM
Mark Correct
Correct Answer
Yep. I had an older version of http_lua_options.lua. Thanks for the reply.

/Dion
Like • Show 0 Likes0
Actions

URL as a feed/indicator

Outcomes

Related Content

Recommended Content