Prepare the raw data as follows:

1)  "make" the utilities dbjw and dbabbrev in this directory.

2)  Make symbolic links to USDA data files FOOD_DESC.txt, NUT_DATA.txt,
    NUTR_DEF.txt, FD_GROUP.txt, and WEIGHT.txt in the main nut directory.

3)  Issue command "preprocess/dbfoodname 1>/dev/null" in the main nut
    directory, and correct any long lines with additional editing of
    preprocess/foodedits and/or changes to abbreviations in "abbrev.h",
    and if the latter, "touch dbabbrev.c" and "make" again.

4)  Issue command "preprocess/dbjoin" in the main nut directory.

Here is a brief explanation of the weight processing.  The second field
in the weight record is changed to indicate the appropriateness of the
serving for the food group, in which '0' is best and '9' is worst.

Then all weights are sorted to pick the best weight.  If the best weight
is a '9' then the record is rewritten with a better weight.  If there is
no weight record for a food, then a weight record is written with a '9'.  Thus
the '9' records indicate a new weight not from the USDA and these are
selected to be added to the weight library with the USDA weights.
