CountHTTP 1.0

Short:        (1.0) Reports on AWS logfiles
Author:       mhunter+@andrew.cmu.edu (Matthew Hunter)
Uploader:     mhunter+ andrew cmu edu (Matthew Hunter)
Type:         comm/www
Version:      1.0
Replaces:     CountHTTP.lha
Requires:     OS 2.04
Architecture: m68k-amigaos

CountHTTP 1.0: First public release

This program is designed to read in the logfiles generated by the Amiga Web
Server, and output some interesting and (hopefully) meaningful statistics on
them.  It's designed to be most flexible in how it outputs the generated
information, but not to overanalyze the logfile -- in short, to enable a simple,
easily produced and useful page of statistics, without needing to write another
program to format the output.  If you need extensive load-analysis look
elsewhere.

It accepts the following parameters:

	MATCH/K     == an AmigaDOS pattern to filter all output through
	LOGFILE     == The name of the logfile.  Defaults to LOGS:http_access
	IMGPATTERN  == An AmigaDOS pattern designed to identify images by their
						filenames, so they are not counted as "hits".  Feel free
						to similarly filter out .lha or .ps or whatever else you want.
	HTML        == format output in HTML
	TABLES      == format output with tables.  Implies HTML.
	VERBOSE     == Report the name of every matching site or file logged
	TERSE       == designed for hsc's <$include FILE=counter> style tags, to
						generate a "pseudo-counter". Outputs only the number of
						matching or total hits.
	NOFILES     == Excludes files from being displayed completely.
	NOSITES     == As for NOFILES, but excludes sites.

	If your logfile is large, CountHTTP will require quite a bit of memory.  It
can also take some time to process a large logfile -- a 650k log takes about
10-15 seconds on an 030, in fast ram.

	It will eventually be usable as a CGI script; the changes should be minor.

	If your locale date produces a date string longer than 33 characters (that
AWS uses in its logfile) then this won't work.  Email me the length of your date
string and I will make the (again minor) changes to implement it as a
commandline option.

	All output goes to either stdio or stderr.  stderr is only errors, and stdio
should -always- be a clean output file, suitable for redirection.

	Example output can be seen at
	http://shadow.res.cmu.edu/mhunter/tablestats.html
	http://shadow.res.cmu.edu/mhunter/stats.html