Opposite of projectHTML.

The basic case (for prog lang comparison and similar tables):

Take multiple XHTML files and extract one given table (by ID, or the nth one,...?) from all.
Copy the first file, but for each row in the table, splice in the others.

What data structure?

for each non-first file
    scan to the table
    collect an array of the rows
    for each row, cat the desired td/hds
        (what about colspans?)
    end
end

run the first file
   when a row in the table begins
       splice in the right row's data from each other file
            (in what order?)




------------

general XML table util:
    handle/convert html, cals, etc.
    sort rows
    reorder columns
    transpose
    determinant, projection, etc.
    insert row (with uniform, numbered, or spliced content
    delete rows/columns
    convert to/from csv
    delete/expand colspans/rowspans
    [gnrl: factor out styles into CSS and/or colspecs]
    drop/normalize style info
    create class by column
    project (choose columns)
    join (combine from multiple files)
    select (choose rows)
    validate:
        datatype/regex per column (or row?)
        number of cols, incl. colspan
        coherent rowspans
        tfoot/thead
        limitations on cell content (nested table, block items, etc.)