带有伪条目的Unix连接命令

Question

I have a poor understanding of the semantics behind the join command. I would like to merge through a shell scrip and add dummy values.

I have two files that I want to merge together. File A has 4 columns with a KEY column. File B has 60K+ columns, with the very first column being the KEY column.

Both keys overlap at ~80%.

Goal: create File C, which is every entry from File A and the matching rows from File B. If A did not match in B, I'd like dummy value "0" to be inserted into every missing field (60K+ fields)

Approach:

As a newbie to shell scripting, I figured a simple join would be effective. I sorted File A and B by the KEY value first using sort -k# appropriately.

join -a1 -1 2 -2 1 -e "0" file.A file.B > file.C

Now, how does join see the fields/columns its looking at? File B has 60k-1 columns that are spaced as:

KEY 1 0     1 1     2 4     0 1 ...  

Now, when I tried my command, file C has the correct number of entries, but I couldn't figure out how to add the missing values. File A has entries that file B does not have, and I'd like to place the null value 0 in every column that was not matched in file A by file B.

Thus, in file C, the result should be (according to my understanding of join ):

KEY A1 A2 A3 A4 1 0     1 1     2 4     0 1 ... 
KEY A1 A2 A3 A4 0 0     0 0     0 0     0 0 ...

The spacing AFTER the joining doesn't matter to me, but the file B is created with the alternating tab-space-tab-space format.

Why isn't join -e "0" adding in my dummy values when I asked it to? I would also appreciate any other shell strategies to do this. I know I can merge in perl by running it through line by line (or R if it didn't take so long to load), but I feel shell is more powerfully equipped for this.

EDIT

The data in teh file is mixed. The first 5 columns are identifying strings in file A, and in file B there is a key string and single alphanumeric characters in each of the additional. File A will always be small (no more than 1 MB), but file B can stretch up to 2+ GB.

Attempted R: df <- read.table("file.B", header=FALSE, fill=TRUE)

Answer 1

I read in the join info page:

`-e STRING'
     Replace those output fields that are missing in the input with
     STRING.  I.E. missing fields specified with the `-12jo' options.

I inferred that -o was required. Try this:

join -a1 -1 2 -2 1 -o auto -e "0" file.A file.B > file.C