How-To: Deduplicating Compound Fields

Compound fields contain multiple pieces of information to report and/or search on, contained within a single field. Alternatively, they may be arrays parsed into an array field within events that then must be summarized.

For example, User Agent data in logs contains browser identifiers separated with spaces that define browser and toolkits used to support them: 

Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/116.0.0.0 Safari/537.36

The following solutions use a variety of methods to extract and aggregate the information.

Click + next to an example below to get the full details.

Deduplicate Compound Field Data With array:union() and split()

Query
logscale
splitString(field=userAgent,by=" ",as=agents)|array:filter(array="agents[]", function={bname=/\//}, var="bname")|array:union(array=agents,as=browsers)| split(browsers)
Introduction

Deduplicating fields of information where there are multiple occurences of a value in a single field, maybe separated by a single character can be achieved in a variety of ways. This solution uses array:union() and split create a unique array and then split the content out to a unique list.

For example, when examining the humio and looking for the browsers or user agents that have used your instance, the UserAgent data will contain the browser and toolkits used to support them, for example:

Raw Events
Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/116.0.0.0 Safari/537.36

The actual names are the Name/Version pairs showing compatibility with different browser standards. Resolving this into a simplified list requires splitting up the list, simplifying (to remove duplicates), filtering, and then summarizing the final list.

Step-by-Step

  1. Starting with the source repository events.

  2. logscale
    splitString(field=userAgent,by=" ",as=agents)

    First we split up the userAgent field using a call to splitString() and place the output into the array field agents

    This will create individual array entries into the agents array for each event:

    agents[0]agents[1]agents[2]agents[3]agents[4]agents[5]agents[6]agents[7]agents[8]agents[9]agents[10]agents[11]agents[12]
    Mozilla/5.0(Macintosh;IntelMacOSX10_15_7)AppleWebKit/537.36(KHTML,likeGecko)Chrome/116.0.0.0Safari/537.36
  3. logscale
    |array:filter(array="agents[]", function={bname=/\//}, var="bname")

  4. logscale
    |array:union(array=agents,as=browsers)

    Using array:union() we aggregate the list of user agents across all the events to create a list of unique entries. This will eliminate duplicates where the value of the user agent is the same value.

    The event data now looks like this:

    browsers[0]browsers[1]browsers[2]
    Gecko/20100101Safari/537.36AppleWebKit/605.1.15

    An array of the individual values.

  5. logscale
    | split(browsers)

    Using the split() will split the array into individual events, turning:

    browsers[0]browsers[1]browsers[2]
    Gecko/20100101Safari/537.36AppleWebKit/605.1.15

    into:

    _indexrow[1]
    0Gecko/20100101
    1Safari/537.36
    2AppleWebKit/605.1.15

  6. Event Result set.

Summary and Results

The resulting output from the query is a list of events with each event containing a matching _index and browser. This can be useful if you want to perform further processing on a list of events rather than an array of values.