Mapping line numbers across two diff files using emacs/python/winmerge

Question

Consider the following two files that are slightly different:

foo (old version):

<Line 1> a
<Line 2> b
<Line 3> c
<Line 4> d

foo (new version):

<Line 1> a
<Line 2> e
<Line 3> b
<Line 4> c
<Line 5> f
<Line 6> d

As you can see, characters e and f are introduced in the new file.

I have a set of line numbers corresponding to the older file…say, 1, 3, and 4 (corresponding to letters a, c, and d).

Is there a way to do a mapping across these two files, so that I can get the line numbers of the corresponding characters in the newer file?

E.G., the result would be:

Old file line numbers (1,3,4) ===> New File line numbers (1,4,6)

Unfortunately I have only emacs (with a working ediff), Python, and winmerge at my disposal.

Answer 1

You can do it all in Emacs:

(defun get-joint-index (file-a index file-b)
  (let ((table (make-hash-table :test #'equal)))
    (flet ((line () (buffer-substring-no-properties
                     (point-at-bol) (point-at-eol))))
      (with-temp-buffer (insert-file file-b)
        (loop for i from 1 do (puthash (line) i table)
              while (zerop (forward-line))))
      (with-temp-buffer (insert-file file-a)
        (loop for i in index do (goto-line i)
              collect (gethash (line) table))))))

To run,

M-:(get-joint-index "/tmp/old" '(1 3 4) "/tmp/new")

-> (1 4 6)

Answer 2

What you need is a string searching algorithm where you have multiple patterns (the lines from the old version of foo) that you want to search for within a text (the new version of foo). The Rabin-Karp algorithm is one such algorithm for this sort of task. I've adapted it to your problem:

def linematcher(haystack, needles, lineNumbers):
    f = open(needles)
    needles = [line.strip() for n, line in enumerate(f, 1) if n in lineNumbers]
    f.close()

    hsubs = set(hash(s) for s in needles)
    for n, lineWithNewline in enumerate(open(haystack), 1):
        line = lineWithNewline.strip()
        hs = hash(line)
        if hs in hsubs and line in needles:
            print "{0} ===> {1}".format(lineNumbers[needles.index(line)], n)

Assuming your two files are called old_foo.txt and new_foo.txt then you would call this function like this:

linematcher('new_foo.txt', 'old_foo.txt', [1, 3, 4])

When I tried in on your data it printed:

1 ===> 1
3 ===> 4
4 ===> 6