[R] How to read last (incomplete) line from gzfile()?

[Prof Brian Ripley]

I realize I didn't give a solution:

scan(con, "", sep="\n")

should do it.


On Sun, 20 Apr 2008, Prof Brian Ripley wrote:

> On Sun, 20 Apr 2008, Stephen Eglen wrote:
>
>> Hi,
>>
>> I have some text files that do not have trailing \n on the last
>> (incomplete) line; how do I read in the last line?
>>
>> e.g. here is a test case: [linux + R version 2.6.1 (2007-11-26) ]
>>
>>
>> z <- gzfile("short.gz", open="w")
>> cat("7\n5\n3", file=z)
>> close(z)
>>
>> z <- gzfile('short.gz')
>> readLines(z)
>>
>> [1] "7" "5"
>>
>> readLines would indicate that for blocking connections (which I assume
>> this is), the last line (containing 3) would be read in, and a warning
>> would be generated.  I get the desired behaviour if I use a file() on
>> a file that is not compressed:
>>
>> y <- file("short", open="w")
>> cat("7\n5\n3", file=y)
>> close(y)
>>
>> y <- file('short')
>> readLines(y)
>>
>> [1] "7" "5" "3"
>> Warning message:
>> In readLines(y) : incomplete final line found on 'short'
>>
>> What am I missing?
>
> gzfile is not a text-mode connection, and the help is wrong for those.
> E.g.
>
>> readLines(pipe("gzip -dc short.gz"))
> [1] "7" "5"
>
> does the same.  It is only text-mode blocking connections which do this.
>
> (I just looked at the code, and you might have found that illuminating.)

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595