Homepage(german)     Computer stuff (german) [deutsch] [français]


for tracking time changes in HTTP-server logs


rgxstat-0.43.tgz (77K) [The executable code is stripped about 90K. The package contains the sources and the documentation in english and partially in german.]

rgxstat-0.20.tgz (55K) [The executable code is stripped about 80K. The package contains the sources and html-documentation in english and german.]

(requires the standard header files and GTK. Extremely useful moreover would be the GNU plotutils (see below!).


RGXStat is a program thought for tracking down time changes in the accesses to several areas of a homepage, so that their development can be compared. What this means is best explained by two examples:

statistics 2

This figure shows the number of accesses to html-files in several sub-areas of my homepage since the middle of 2000 until the end of 2001, broken down to weeks. The decline to nearly 0 in the beginning of 2001 is an artefact of the program, probably caused by the weeks at the end and at the beginning of years, consisting of less than 7 days.

statistics 1

This figure shows accesses to several file types, arranged according to the time of day. As can be seen, normally every html-file loads on average about 1.5 jpg-files. But in the night the html-files dominate. This probably is the "background noise" caused by indexing robots, which only request html-files.

Two or three more statistics can be found on the page with additional statistics.

The definition of the areas to compare is done with the help of regular expressions. In the second example those were simply \.html, \.jpg and so on. Additionally regular expressions for referrer, browser, domain a.s.o may be given. So it is e.g. possible to track all accesses made by netscape users coming from google (or things like this). The settings for the time steps, in which access have to be summed up, isn't as flexible, but therefore easier. The log format used by the server also has to be given a modified regular expression. So people wanting to use this program should be firm in regular expressions. This program is not easy to use.

Graphical output as above is produced by the program with help from the libplot, a library contained in the GNU plotutils. Alternatively html-code or ascii-text can be produced. But html doesn't work well, because I haven't used it now for a longer time and didn't care much about the implementation. I would advise against using it. The ascii-output is thought to be input for other programs.

The package contains four more executables, which don't have a GUI. One of it draws queries, leading from search-engines to certain files, out of the log files. Another tries to guess, which areas of a site are related to which other areas (in the sense that the two are often visited together) and draws a treelike structure as ascii-art. The third lists referrers. The last simply is a filter for those two.

In the moment the program can't be installed. It just can be compiled in the directory and then be executed.


All parts of the program by my own are under the BSD-licence. But the GNU plotutil being under GPL, derived works, which use those too, have to be under the GPL too. It is possible to compile RGXStat without libplot. But in this case there is of cause no graphical output any more.

System requirements

The program runs on my Linux boxes (SuSE-6.4 and 7.0) and compiles with gcc. The GUI uses GTK, but the program can also be compiled without GUI. For graphical output the libplot of the GNU plotutils is used. Those are contained in most distributions, but aren't usually installed by default.

Homepage(german)     Computer stuff(german) by Michael Becker, 12/2001. Last modification 11/2002