Hi,

I have a problem while joining two sorted files with "join".

File 1.txt
Alnus|123
ALO140102|234
ALO 1401 02|345
ALO-1401-02|456
Alobar Holoprosencephalies|567

File 2.txt
1|Alnus|
1|ALO 1401 02|
1|ALO-1401-02|
1|Alobar Holoprosencephalies|

If I join the files as follows:
join -i -t '|' -1 1 -2 2 file1.txt file2.txt

this doesn't work because the join command ignores punctuation i.e. it checks ALO140102 against file 2 and when it doesn't find a match it moves on to Alobar Holoprosencephalies. If ALO140102 IS present in file 2 then the match works fine. Therefore I need to get the join command to recognise non-alphanumerics.

Any ideas?!!

I've done it a long winded way by replacing punctuation with alphanumeric tags (e.g. REMOVE1) resorted the files and and then do the join. This works fine as the tags are matched exactly whereas the punctuation was not. However, this seems a ridiculous way to do it - there must be a better one!

I think it may be to do with the way UNIX matches which I think you can change with the LC_COLLATE variable but I'm not sure.