Initial commit of 'wizard remaster'.

Signed-off-by: Edward Z. Yang <ezyang@mit.edu>

diff --git a/wizard/command/remaster.py b/wizard/command/remaster.py
new file mode 100644 (file)
index 0000000..e601500
--- /dev/null
+++ b/wizard/command/remaster.py
@@ -0,0 +1,110 @@
+import logging
+import tempfile
+import os
+import sys
+
+from wizard import command, shell
+
+def main(argv, baton):
+    options, args = parse_args(argv, baton)
+    # Determine upstream commit history
+    # XXX This is a little sloppy (since it also pulls in remotes and
+    # originals), but if the upstream repo has a clean set of remotes,
+    # that shouldn't be a problem.
+    excludes = map(lambda line: line.partition("\t")[0],
+                   shell.eval("git", "ls-remote", "origin").splitlines())
+    # Determine local commits and their parents
+    local_commits = set()
+    all_parents = set()
+    for line in shell.eval("git", "rev-list", "--parents", "--branches", "--not", *excludes).split("\n"):
+        (commit, _, parent_string) = line.partition(' ')
+        local_commits.add(commit)
+        parents = parent_string.split()
+        all_parents.update(parents)
+    # Determine what commits need mapping
+    needs_map = all_parents - local_commits
+    # Determine the new commits for these maps
+    mapping = {}
+    for hash in needs_map:
+        summary    = shell.eval("git", "log", "-1", "--pretty=format:%s", hash)
+        # Find the corresponding commit by grepping for the summary from
+        # "live" tags (which should have been updated to the new history
+        # we are remastering to.)  -F == fixed string (no regexing).
+        candidates = shell.eval("git", "rev-list", "-F", "--grep=" + summary, "--tags").splitlines()
+        if len(candidates) != 1:
+            raise "Failed looking for " + hash
+        mapping[hash] = candidates[0]
+
+    # XXX Make this more robust: given our pre-processing, there is a
+    # very specific set of parent IDs we expect to see (not necessarily
+    # the ones in our mapping so far: those are precisely the IDs that
+    # may change, but some will stay the same.)  Consider nops.  This
+    # might be hard since git-filter-branch manufactures hashes as it
+    # goes along.
+
+    # Prepare the parent filter script
+    t = tempfile.NamedTemporaryFile(delete=False)
+    try:
+        t.write("#!/bin/sed -f\n")
+        for search, replace in mapping.items():
+            t.write("s/%s/%s/g\n" % (search, replace))
+        t.close()
+        shell.call("chmod", "a+x", t.name) # necessary?
+        logging.info("Sed script %s", t.name)
+        # Do the rewrite
+        shell.call("git", "filter-branch", "--parent-filter", t.name, "--",
+                   "--branches", "--not", *excludes,
+                   stdout=sys.stdout, stderr=sys.stderr)
+    finally:
+        # Cleanup
+        os.unlink(t.name)
+
+def parse_args(argv, baton):
+    usage = """usage: %prog remaster [ARGS]
+
+Reconciles divergent commit histories by rewriting all parent links to
+point to the new commits.  This only works if we are able to construct a
+one-to-one correspondence between the old and new commits.  This should
+be automatically invoked by 'wizard upgrade' if a remastering is
+necessary.
+
+Consider this history:
+
+      D-----E   heads/master
+     /     /
+    A--B--C     remotes/origin/master
+
+Suppose on a fetch, we discover that origin/master has been rebased, and
+replaced with 'old-master':
+
+      D-----E   heads/master
+     /     /
+    A--B--C     remotes/origin/old-master
+
+    A'-B'-C'    remotes/origin/master
+
+We would like to construct a new tree as follows:
+
+      D'----E'  heads/master
+     /     /
+    A'-B'-C'    remotes/origin/master
+
+Where D/D' and E/E' have identical trees, just different parent commit
+pointers.  This is what 'wizard remaster' does.
+
+In order to do this, we need to know two things: (1) which commits in
+the old history were not provided by the user (the ones to rewrite are
+'git log master ^origin/master', in the old history before the force
+update of branch locations), and (2) what the correspondence between
+the old commits and the new commits are.  (1) is determined by looking
+at all references in the remote repository.  (2) is determined by
+comparing commit messages; a user can also manually add extra mappings
+if this heuristic fails (not implemented yet).
+"""
+    parser = command.WizardOptionParser(usage)
+    parser.add_option("-f", "--force", dest="force", action="store_true",
+            default=False, help="Force overwriting.")
+    options, args = parser.parse_all(argv)
+    if len(args) > 0:
+        parser.error("too many arguments")
+    return options, args