"odbc_lob_read(sth, colno, buf, length, \%attrs) where \%attrs can contain TYPE => SQL_type." http://www.nntp.perl.org/group/perl.dbi.dev/2010/07/msg6140.html
where Type should be SQLBINARY http://www.nntp.perl.org/group/perl.dbi.dev/2010/09/msg6258.html "It turns out I was mistaken as DBD::Oracle DOES implement blob read although I've not tested it works. Someone, in the distant past attempted an implementation in DBD::ODBC but it does NOT work and is seriously flawed." http://www.nntp.perl.org/group/perl.dbi.dev/2010/07/msg6144.html Martin ______________________________________________ "Confusion and Obfuscation directives from Corporate can be disregarded for OpenSource implementations"..Latest Memo > Date: Fri, 6 Sep 2013 09:18:57 +0100 > From: martin.ev...@easysoft.com > To: dbi-users@perl.org > Subject: Re: odbc_lob_read chunk is short > > On 05/09/13 20:28, Mike Grau wrote: > > Hello, > > > > I am completely new at this, but I am trying to read a BLOB in chunks > > via ODBC in Oracle 11g using DBD::ODBC-1.43, unixODBC-2.3.1 and > > oracle-instantclient11.2-odbc-11.2. I can read an entire BLOB in one > > read and write it to the filesystem, but I really need to be able to do > > it in chunks, I think. > > > > Reading the BLOB in chunks, but I am always getting one byte less than > > requested, losing the last byte, which results in a corrupt document (a > > PDF) when written to the filesystem. I'm at a loss as to why this is > > happening and how to correct it. > > > > Using this snippet based on the example provided by DBD::ODBC... > > > > while($len = $s->odbc_lob_read(1, \my $x, 8, {TYPE => 999})) { > > print "len=$len, x=$x\n"; > > } > > > > ... produces this output: > > > > len=7, x=435886 > > len=7, x=0000 n > > len=7, x=0001282 > > len=7, x=60 0000 > > > > And a tracefile shows the same. > > > > SQLGetData(col=1,type=-2)=1 (retlen=2046252) > > <- odbc_lob_read= ( 7 ) [1 items] at /root/lob_read.pl line 45 > > >> odbc_lob_read DISPATCH (DBI::st=HASH(0x203c940) rc1/1 @5 g2 ima0 > > pid#11494) at /root/lob_read.pl line 45 > > -> odbc_lob_read for DBD::ODBC::st (DBI::st=HASH(0x203c940)~0x203c9d0 1 > > SCALAR(0x1e0a440) 8 HASH(0x203c970)) thr#1cea010 > > SQLGetData(col=1,type=-2)=1 (retlen=2046244) > > <- odbc_lob_read= ( 7 ) [1 items] at /root/lob_read.pl line 45 > > >> odbc_lob_read DISPATCH (DBI::st=HASH(0x203c940) rc1/1 @5 g2 ima0 > > pid#11494) at /root/lob_read.pl line 45 > > -> odbc_lob_read for DBD::ODBC::st (DBI::st=HASH(0x203c940)~0x203c9d0 1 > > SCALAR(0x1e0a440) 8 HASH(0x203c9a0)) thr#1cea010 > > > > Can someone please give me some guidance how to get the full chunk > > requested? > > > > I'm using > > unixODBC-2.3.1 - built from the tarball > > DBD-ODBC-1.43 - installed from the tarball (perl Makefile.PL; make; > > make install) > > oracle-instantclient11.2-odbc-11.2.0.3 installed from RPM > > DBI is the distro's perl-DBI-1.617-4.1.1.x86_64 (openSUSE 12.3) > > The database and client are both x86_64 > > > > The instant client is link against libodbcinst.so.1 and unixODBC now > > provides libodbcinst.so.2, so I symlinked them: > > libodbcinst.so.1 -> /usr/local/lib64/libodbcinst.so.2 > > Ugly, but seems to work. > > > > Ultimately I want to index documents stored in Oracle with Sphinx and so > > must use ODBC. I'd like to know that this much is working before moving > > on to Sphinx. > > > > TIA -- Mike G. > > > > Bear with me and I'll take another look at that code. If I cannot see > anything I might need to send you a new version with slightly different > logging. By all means nag me if I don't get back to you by tomorrow. > > Martin > -- > Martin J. Evans > Easysoft Limited > http://www.easysoft.com
