|Did you know ...||Search Documentation:|
|Packs (add-ons) for SWI-Prolog|
|Title:||Analyse SWI-Prolog HTTP server log files|
|Rating:||Not rated. Create the first rating!|
|Author:||Jan Wielemaker <firstname.lastname@example.org>|
No reviews. Create the first review!.
The SWI-Prolog HTTP server can be asked to write a logfile by loading the library(http/http_log). The resulting file is a file holding Prolog terms. This package allows for loading the logfile into the Prolog database and ask queries about it.
Before you can analyse a logfile it must be read into the Prolog
database. Assuming the default name
httpd.log, this is achieved using
the following command:
% swipl ?- use_module(library(http/logstat)). ?- read_log('httpd.log').
In the examples below, we assume a logfile as created by ClioPatria. Adjust the paths and query parameter to match your web server. Once loaded, the database can be queried using logrecord/1. A simple example is:
?- logrecord([path('/sparql/'), search([query=Query])]).
Suppose you want to make a CSV file holding the IP addresses and the search executed from there. This can be achieved using:
?- findall(row(IP,Q), logrecord([path('/browse/search'), search([q=Q]), ip(IP)]), Rows), csv_write_file('queries.csv', Rows).
The Prolog script file script/log2clf.pl can be used to convert SWI-Prolog logfiles into the webserver Common Log Format syntax, so you can run common tools such as webalizer on them. This script is executed as:
% log2clf [-o out] [-a date] [-b date] file ...
Pack contains 4 files holding a total of 16.9K bytes.