Year in date field is truncated

[dstillman]

On a few different library catalogs, I've seen years get truncated to between 1 and 3 characters. I don't see anything obviously wrong with the MARC records, but I don't really know what to look for, either.

Instead of "1994", I get "1":

​http://ctwweb.wesleyan.edu:7003/cgi-bin/Pwebrecon.cgi?v1=18&ti=1,18&Search%5FArg=postmodern&Search%5FCode=TALL&CNT=25&PID=17006&SEQ=20061102164523&SID=8

Instead of "1996", I get "199":

​http://ctwweb.wesleyan.edu:7003/cgi-bin/Pwebrecon.cgi?v1=2&ti=1,2&Search%5FArg=sex&SL=None&Search%5FCode=TALL&CNT=25&PID=17006&SEQ=20061102164956&SID=9

[dstillman]

I get the same results for the same items at different libraries.

​http://clio.cul.columbia.edu/cgi-bin/Pwebrecon.cgi?DB=local&Search_Arg=ISBN+%22081956303X%22&Search_Code=CMD&CNT=10
​http://clio.cul.columbia.edu/cgi-bin/Pwebrecon.cgi?DB=local&Search_Arg=ISBN+%220816611734%22&Search_Code=CMD&CNT=10

[stakats]

Fixed in [900]. We were adding null characters to the MARC string to get the directory to line up with content but were not stripping those nulls back out. Updated translator is attached for Dan S. to use to refresh the current repository.

[dstillman]

Way too big to send to 1.0.0b2.r2, unfortunately.

[simon]

I reversed this patch in r1272 because the nulls are stripped later (in getField()) and can't be stripped before then because the MARC directory is based on field offsets. (The behavior this patch introduced results in issues with non-ASCII characters; see #567.) The year truncation issues do not seem to have reappeared, but if they do, this is the first place to look.