view contrib/undumprevlog @ 33269:ead6749354e1

phabricator: try to fetch differential revisions in batch

Previously, we read Differential Revisions one by one by calling
`differential.query`.

Fetching them one by one is suboptimal. Unfortunately, there is no Conduit
API that allows us to get a stack of diffids using a single API call.

This patch tries to be smarter using a simple heuristic: when fetching D59
as a stack, previous IDs like D51, D52, D53, ..., D58 are likely belonging
to a same stack so just fetch them as well. Since `differential.query` only
returns cheap metadata without expensive diff content, it shouldn't be a big
problem for the server.

Using a test Phabricator instance, this patch reduces `phabread` reading a
10 patch stack from about 13 to 30 seconds to 8 seconds.

#!/usr/bin/env python
# Undump a dump from dumprevlog
# $ hg init
# $ undumprevlog < repo.dump

from __future__ import absolute_import

import sys
from mercurial import (
    node,
    revlog,
    transaction,
    util,
    vfs as vfsmod,
)

for fp in (sys.stdin, sys.stdout, sys.stderr):
    util.setbinary(fp)

opener = vfsmod.vfs('.', False)
tr = transaction.transaction(sys.stderr.write, opener, {'store': opener},
                             "undump.journal")
while True:
    l = sys.stdin.readline()
    if not l:
        break
    if l.startswith("file:"):
        f = l[6:-1]
        r = revlog.revlog(opener, f)
        print f
    elif l.startswith("node:"):
        n = node.bin(l[6:-1])
    elif l.startswith("linkrev:"):
        lr = int(l[9:-1])
    elif l.startswith("parents:"):
        p = l[9:-1].split()
        p1 = node.bin(p[0])
        p2 = node.bin(p[1])
    elif l.startswith("length:"):
        length = int(l[8:-1])
        sys.stdin.readline() # start marker
        d = sys.stdin.read(length)
        sys.stdin.readline() # end marker
        r.addrevision(d, tr, lr, p1, p2)

tr.close()