Greetings,
The following applies to wget-1.10.2. The problem was noticed with wget
built/run on Solaris-9/SPARC and Solaris-10/x86, but I believe the problem
is independent of the platform.
Certain FTP server directory listings yield file sizes of zero bytes
when interpreted by wget-1.10.2. The problem entries (as seen in the
wget-generated ".listing" file) have only a single blank between the
"group ownership" token and the file-size token, as can occur with
long group names and/or large file sizes. Here is an example:
-r--r--r-- 1 ftp anonymous 48608258 Jan 3 12:01 swissprot.gz
And here is its resulting entry from the wget-generated index.html file:
2006 Jan 03 12:01 File <a href="ftp://ftp.ncbi.nih.gov:21/blast/db/FA
STA/swissprot.gz">swissprot.gz</a> (0 bytes)
Attached below is a patch which fixes the problem, and also adds debug
output in order to help diagnose such problems should they occur with
other directory listing formats.
Regards,
--
Marion Hakanson <[EMAIL PROTECTED]>
--- ./src/ftp-ls.c.orig Thu Aug 4 14:52:33 2005
+++ ./src/ftp-ls.c Wed Jan 4 15:33:08 2006
@@ -228,6 +228,10 @@
error = 1;
break;
}
+ if ( t < ptok && *t == '\0' )
+ /* backed up into former whitespace between tokens */
+ t++;
+
errno = 0;
size = str_to_wgint (t, NULL, 10);
if (size == WGINT_MAX && errno == ERANGE)
@@ -236,6 +240,7 @@
cur.size = 0;
else
cur.size = size;
+ DEBUGP (("size: %s(%s); ", number_to_static_string (cur.size), t));
month = i;
next = 5;
