Aggregators

Aggregators represent different ways to count and output data as it is processed in rare. Aggregation takes in different formats of the matched and extracted expression, and either counts or analyzes the values.

More Examples

More examples of each can be found in examples. For CLI documentation, run rare help

Filter

rare help filter

Summary

Filter is a command used to match and (optionally) extract that match without any aggregation. It's effectively a grep or a combination of grep, awk, and/or sed.

Example

Extract out two numbers from access.log

$ rare filter -n 4 -m "(\d{3}) (\d+)" -e "{1} {2}" access.log
404 169
404 169
404 571
404 571
Matched: 4 / 4

Histogram

rare help histogram

Summary

The histogram format outputs an aggregation by counting the occurences of an extracted match. That is to say, on every line a regex will be matched (or not), and the matched groups can be used to extract and build a key, that will act as the bucketing name.

Example

Extract HTTP verb, URL and status code. Key off of status code and verb.

Tip

Use -x to display percentages and a simple bargraph.

$ rare histo -m '"(\w{3,4}) ([A-Za-z0-9/.]+).*" (\d{3})' -e '{3} {1}' access.log
200 GET                          160663
404 GET                          857
304 GET                          53
200 HEAD                         18
403 GET                          14

Bar Graph

rare help bargraph

Summary

Similar to histogram or table, bargraph can generate a stacked or grouped bargraph by one or two keys.

Example

Color Coded Keys

When run in terminal, below will be color-coded keys. Alternatively, you can leave off -s (stacking) to see each key displayed vertically.

$ rare bars -sz -m "\[(.+?)\].*\" (\d+)" \
    -e "{$ {buckettime {1} year nginx} {bucket {2} {multi 10 10}}}" \
    testdata/*

        0 200  1 300  2 400
2019  000000000222222222222222222222222222222  458,136
2020  0000000000000000002222222222222222222222222222222  576,030
Matched: 1,034,166 / 1,034,166

Numerical Analysis

rare help analyze

Summary

This command will extract a number from the match and run basic analysis on that number (Such as mean, median, mode, and quantiles).

Example

Note

-x or --extra will capture more information (Median, Mode, and Percentiles), but dramatically slows down the analysis.

$ rare analyze --extra \
    -m '"(\w{3,4}) ([A-Za-z0-9/.@_-]+).*" (\d{3}) (\d+)' \
    -e "{4}" testdata/access.log

Samples:  161,622
Mean:     2,566,283.9616
Min:      0.0000
Max:      1,198,677,592.0000

Median:   1,021.0000
Mode:     1,021.0000
P90:      19,506.0000
P99:      64,757,808.0000
P99.9:    395,186,166.0000
Matched: 161,622 / 161,622

Table

rare help table

Summary

Create a 2D view (table) of data extracted from a file. Expression needs to yield a two dimensions. Can either use \x00 or the {$ a b} helper. First element is the column name, followed by the row name.

Example

$ rare tabulate -m "(\d{3}) (\d+)" \
    -e "{$ {1} {bucket {2} 100000}}" -sk access.log

         200      404      304      403      301      206      
0        153,271  860      53       14       12       2                 
1000000  796      0        0        0        0        0                 
2000000  513      0        0        0        0        0                 
7000000  262      0        0        0        0        0                 
4000000  257      0        0        0        0        0                 
6000000  221      0        0        0        0        0                 
5000000  218      0        0        0        0        0                 
9000000  206      0        0        0        0        0                 
3000000  202      0        0        0        0        0                 
10000000 201      0        0        0        0        0                 
11000000 190      0        0        0        0        0                 
21000000 142      0        0        0        0        0                 
15000000 138      0        0        0        0        0                 
8000000  137      0        0        0        0        0                 
22000000 123      0        0        0        0        0                 
14000000 121      0        0        0        0        0                 
16000000 110      0        0        0        0        0                 
17000000 99       0        0        0        0        0                 
34000000 91       0        0        0        0        0                 
Matched: 161,622 / 161,622
Rows: 223; Cols: 6

Heatmap

rare help heatmap

Summary

Create a dense, color-coded, version of table-data by using cells to display the strength of a value. Can either use \x00 or the {$ a b} helper. First element is the column name, followed by the row name.

Example

$ rare heatmap -m '\[(.+?)\].*" (\d+)' \
    -e "{timeattr {time {1}} yearweek}" -e "{2}" access.log

        - 0    5 22,602    9 45,204
     2019-34..2019-41..2019-50..2020-15..2020-23..2020-31...2020-9
200  1111111111111111111111111111111111111111111111111111111-11111
206  -------------------------------------------------------------
301  -------------------------------------------------------------
304  -------------------------------------------------------------
400  -------------------------------------------------------------
404  33516265914153253212111-1511-13-141-1412-132111--14-1-1-13211
405  -------------------------------------------------------------
408  -------------------------------------------------------------
Matched: 1,035,666 / 1,035,666 (R: 8; C: 61)

Reduce

rare help reduce

Summary

Create a set of values or table based on an expression that accumulates results. More powerful than table or histogram because it can interpret data in multiple ways in a single output. Also can group and sort results.

Usage:

1. Extract data from a regex using --match (-m)
2. Optionally group the data into buckets using --group (-g). Can use key=value format to give column a name
3. Specify one or more "accumulators", which are expressions. {.} represents the current value, so for example, {sumi {.} {1}} adds the match {1} to the current value.
   1. Can specify only expression, key=expression format or key:initial=expression.
   2. Can reference past accumulators by key, eg {divi {key1} {key2}}
4. Optionally --sort the data based on an expression or reference to an accumulator. eg. --sort {key1}. Can reverse with --sort-reverse flag

Example

$ rare reduce -m "(\d{3}) (\d+)" -g "http={1}" -a "total={sumi {.} {2}}" \
    -a "count={sumi {.} 1}" -a "avg={divi {total} {count}}" \
    --sort="-{avg}" access.log 
http total      count  avg
206  260976     6      43496
200  1481979390 318003 4660
404  275969007  686541 401
405  279264     708    394
301  3885       21     185
400  5027469    30165  166
304  0          48     0
408  0          174    0
Matched: 1,035,666 / 1,035,666

---

Sorting

Many of the aggregators support changing the order in which the data is displayed in. You can change this from default either by setting the --sort flag or --sort-rows and --sort-cols flags for tables.

These are the supported sorters:

• text -- Pure alphanumeric sort. Fastest, but can sort numbers oddly (eg. would sort 1, 11, 2, ...)
• numeric -- Attempts to parse the value as numeric. If unable to parse, falls back to alphanumeric (Default)
• contextual -- Tries to use context to be smart about sorting, eg if it sees a month or weekday name, will sort by that. Falls back to numeric
• date -- Parses the value as if it were a date. Falls back to contextual
• value -- Orders the results based on their aggregated value. eg. would put the most frequent item at the top. Defaults to descending order

Modifiers

In addition to the sorting method, you can also modify the sort by adding a colon and the modifier, eg: numeric:desc

These are the supported modifiers:

• :reverse -- Reverse of the "default"
• :asc -- Ascending order
• :desc -- Descending order