Read file from bottom

[Brian Birkinbine]

I'm fairly new to python as well.  I was brainstorming a non tail specific way of attacking this
without actually reading through the entire file (even with caching to conserve memory)

Assumptions:
	linesize throughout file is uniform/close enough (line 1 is similarly long as last line)
		and doesn't have to start on a newline boundary (but I guess you could pick an
		arbitrary point (10% from the end), read into buffer, then search forward to the
		newline, then start normal line processing.

Python example code:
	import os

	# add error checking, etc to make production ready

	filename = "YOURFILENAMEGOESHERE"

	inputfd = file(filename,'rb')

	# read in one line, calculate line size
	# can be hardcoded if data is uniform or a percentage of the total filesize, etc...
	linesize = len(inputfd.readline())

	#grab filesize from os.stat
	st = os.stat(filename)
	filesize = st.st_size

	# num of lines backward
	numlines = 20
	# calculate start point from end of file
	startpoint = filesize - (numlines * linesize)

	# seek to calculated startpoint
	inputfd.seek(startpoint)

	# read file from startpoint to end into buffer
	# could also just do readline.
	buffer = inputfd.read()
	inputfd.close()

	#YOUR PROCESSING GOES HERE ...

On Mon, May 13, 2002 at 05:20:34PM -0400, Hardy Merrill wrote:
> You probably already know this, but on the chance you don't,
> you can add a parameter to tail to tell it how many lines from
> the bottom you want.  For example, if I want 35 lines from the
> bottom I can do
> 
>    tail -35 <filename> | python script.py
> 
> Julia Bell [julia.bell at jpl.nasa.gov] wrote:
> > Good idea; I never thought of something like that.
> > 
> > My initial reaction is that I don't know I exactly how many lines frm the
> > bottom I want.  (I currently read the lines, search for data on the lines
> > that meets some trigger criterion, and then start processing the data from
> > there.)  But, there still might be something worth thinking about regarding
> > whether it is more efficient to pre-process the data (similar to using
> > 'tail') than to process it as part of my script.
> > 
> > Thanks.
> > 
> > Julia Bell
> > 
> > 
> > William Park wrote:
> > 
> > > Julia Bell <julia.bell at jpl.nasa.gov> wrote:
> > >
> > > >
> > > > Is there an efficient (and relatively easy - I'm new to Python) way of
> > > > reading the file from the last line backward instead of the first line
> > > > forward?
> > >
> > >
> > > tail | python script.py
> > >