[Rd] Problems Reading from Socket

[Mike Schaffer]

My followup question went unanswered in another thread, so I thought  
I'd start a new one and rephrase it.


I need to use CGI POST to retrieve data from a server, so thanks to  
Duncan's suggestion, I used the httpRequest package as example code  
for doing so.  BUT, I have another problem.  The data retrieved from  
the socket has occasional problems.  I include example code below to  
demonstrate the problem.  The problems occur in R v2.3.1 (2006-06-01)  
on both Linux and OS X.


For comparison, here is code that produces the correct output (I used  
paste on the URL to prevent word-wrapping from clobbering it):

full.url<-paste("http://genomics11.bu.edu/cgi-bin/",
"Tractor_dev/external/get_msa.cgi?user_id=0&",
"table=seqs_ucsc_hg18&len=350&gene_set_ids=",
"NM_000029,NM_000064,NM_000066&orgs=Hs,",
"mm8,canFam2",sep="")

readLines(full.url)




To reproduce the incorrect results, the socket-based code is:

host<-"genomics11.bu.edu"
path<-"/cgi-bin/Tractor_dev/external/get_msa.cgi"
dat<-paste("user_id=0&table=seqs_ucsc_hg18&len=350&",
"gene_set_ids=NM_000029,NM_000064,",
"NM_000066&orgs=Hs,mm8,canFam2",sep="")

len <- length( strsplit(dat,"")[[1]])

request<-paste("POST ",path," HTTP/1.1\nHost: ",host,
"\nReferer:\nContent-type: application/x-www-form-urlencoded\nContent- 
length: ",
len,"\nConnection: Keep-Alive\n\n",dat,sep="")

fp <- socketConnection(host=host,port=80,server=FALSE,blocking=TRUE)
write(request,fp)
socketSelect(list(fp)) # Wait until results are ready
readLines(fp)
close(fp)



So my question is why do I get lines that are split up by hexadecimal  
characters (e.g. element 13 -- 'a1'-- splits what should be one line:  
lines 12 + 14)?  Am I misunderstanding what is needed to correctly  
read from a socket or is this a bug?  Can anyone offer assistance?

Thanks.

--
Mike