# What the log line actually contains

Checked rather than assumed, on 9 file(s) from practice-a.

| property | value |
|---|---|
| lines parsed | 343962 |
| lines not matching the combined format | 0 |
| lines carrying a user agent | 343962 |
| lines carrying a referrer | 223026 |
| requests attributed to a known crawler | 26646 |
| of those, served with a 200 or 206 | 21580 |
| of those, a page rather than an asset | 10342 |
| requests not attributed to a crawler | 317316 |
| arrivals from an answer engine | 117 |
| distinct page paths the site served to anybody | 981 |

Requests are counted three ways because they mean three different things. Every request is what the bandwidth bill sees. A served request is one the server actually answered, so a redirect chain or a run of 404s does not count as content taken. A page is a served request whose path is not a stylesheet, script, font or image, which is what the argument about content is actually about.

A referrer count of zero would mean the ratio this paper reports has no
divisor and the paper reports an asymmetry instead. That is a finding
about the log format as much as about the traffic, so it is recorded here
before any analysis is written.