ExternalReferrals

MeatballWiki | RecentChanges | Random Page | Indices | Categories

An IndexingScheme that tracks the number of external (or foreign) links into the site. It tracks the referral URL in the HTTP request. This is useful for knowing what information is generally useful to the rest of the world.

: http://sunir.org/apps/refers.pl

The source is posted at http://sunir.org/src/meatball/IndexingSchemes.

Some useful examples:

Detailed "most referred" pages: http://sunir.org/apps/refers.pl?full=1&targets=1&orphans=1&search=1&invert=1

Detailed "most referred" pages again, without search engines (useful for tracking blog activity): http://sunir.org/apps/refers.pl?full=1&targets=1&orphans=1&invert=1

Chronological log of all referrers in the last 12 hours: http://sunir.org/apps/refers.pl?full=1&hours=12&targets=1&chrono=1

Script to do

bug: https referring URLs don't collapse to just domain with full=0, instead showing entire URL

[CategoryToDo]

Script updates

[January 22, 2003] Spelling corrections. Now search hits are elided, although the search parameter will include them again. Reverse chronological order is now supported through the chrono display option. The invert parameter will key the data structure off the targets. This last feature is quite useful. Try http://sunir.org/apps/refers.pl?invert=1, or the more detailed, http://sunir.org/apps/refers.pl?full=1&targets=1&orphans=1&invert=1.

[February 26, 2002] Use the optimized log fetch. A couple weeks ago I also made it sort alphabetically as well as based on the number of hits. This groups together referrals from the same domain.

[February 24, 2001] Upgraded to use Standard.pm. Now the targets sort by hit counts instead of alphabetically.

[August 28, 2000] Fixed the caching code.

[August 19, 2000] Scan the cache directory for available logs and use those if usemod.com has deleted them (actually, this was done much earlier). Added counts after targets.

[August 10, 2000] Completed:
- Cache the log files.
- Add "total."

[August 9, 2000] Fixed the timezone logic (now that the clock on the server is correct). Now it correctly calculates the seconds since 1 January 1970 00:00 UTC from a localized timestamp.

: Fixed the regex that matched against the logs so it didn't exclude valid entries (from HTTP/1.0 clients).

[August 2, 2000] Created.

CategoryIndexingScheme

ExternalReferrals

Discussion