##// END OF EJS Templates
hgweb: tweak zlib chunking behavior...
hgweb: tweak zlib chunking behavior When doing streaming compression with zlib, zlib appears to emit chunks with data after ~20-30kb on average is available. In other words, most calls to compress() return an empty string. On the mozilla-unified repo, only 48,433 of 921,167 (5.26%) of calls to compress() returned data. In other words, we were sending hundreds of thousands of empty chunks via a generator where they touched who knows how many frames (my guess is millions). Filtering out the empty chunks from the generator cuts down on overhead. In addition, we were previously feeding 8kb chunks into zlib compression. Since this function tends to emit *compressed* data after 20-30kb is available, it would take several calls before data was produced. We increase the amount of data fed in at a time to 32kb. This reduces the number of calls to compress() from 921,167 to 115,146. It also reduces the number of output chunks from 48,433 to 31,377. This does increase the average output chunk size by a little. But I don't think this will matter in most scenarios. The combination of these 2 changes appears to shave ~6s CPU time or ~3% from a server serving the mozilla-unified repo.

File last commit:

r28389:9ab45fbe stable
r29792:58467204 default
Show More
bdiff.py
98 lines | 2.5 KiB | text/x-python | PythonLexer
Martin Geisler
pure Python implementation of bdiff.c
r7703 # bdiff.py - Python implementation of bdiff.c
#
# Copyright 2009 Matt Mackall <mpm@selenic.com> and others
#
Martin Geisler
updated license to be explicit about GPL version 2
r8225 # This software may be used and distributed according to the terms of the
Matt Mackall
Update license to GPLv2+
r10263 # GNU General Public License version 2 or any later version.
Martin Geisler
pure Python implementation of bdiff.c
r7703
Gregory Szorc
bdiff: use absolute_import
r27335 from __future__ import absolute_import
timeless
bdiff: (pure) support array.array arrays (issue5130)
r28389 import array
Gregory Szorc
bdiff: use absolute_import
r27335 import difflib
import re
import struct
Matt Mackall
pure/bdiff: fix circular import
r7944
def splitnewlines(text):
'''like str.splitlines, but only split on newlines.'''
lines = [l + '\n' for l in text.split('\n')]
if lines:
if lines[-1] == '\n':
lines.pop()
else:
lines[-1] = lines[-1][:-1]
return lines
Martin Geisler
pure Python implementation of bdiff.c
r7703
def _normalizeblocks(a, b, blocks):
prev = None
Dan Villiom Podlaski Christiansen
pure bdiff: don't use a generator...
r14066 r = []
Martin Geisler
pure Python implementation of bdiff.c
r7703 for curr in blocks:
if prev is None:
prev = curr
continue
shift = 0
a1, b1, l1 = prev
a1end = a1 + l1
b1end = b1 + l1
a2, b2, l2 = curr
a2end = a2 + l2
b2end = b2 + l2
if a1end == a2:
Matt Mackall
many, many trivial check-code fixups
r10282 while (a1end + shift < a2end and
a[a1end + shift] == b[b1end + shift]):
Martin Geisler
pure Python implementation of bdiff.c
r7703 shift += 1
elif b1end == b2:
Matt Mackall
many, many trivial check-code fixups
r10282 while (b1end + shift < b2end and
a[a1end + shift] == b[b1end + shift]):
Martin Geisler
pure Python implementation of bdiff.c
r7703 shift += 1
Dan Villiom Podlaski Christiansen
pure bdiff: don't use a generator...
r14066 r.append((a1, b1, l1 + shift))
Matt Mackall
many, many trivial check-code fixups
r10282 prev = a2 + shift, b2 + shift, l2 - shift
Dan Villiom Podlaski Christiansen
pure bdiff: don't use a generator...
r14066 r.append(prev)
return r
Martin Geisler
pure Python implementation of bdiff.c
r7703
timeless
bdiff: (pure) support array.array arrays (issue5130)
r28389 def _tostring(c):
if type(c) is array.array:
# this copy overhead isn't ideal
return c.tostring()
return str(c)
Martin Geisler
pure Python implementation of bdiff.c
r7703 def bdiff(a, b):
timeless
bdiff: (pure) support array.array arrays (issue5130)
r28389 a = _tostring(a).splitlines(True)
b = _tostring(b).splitlines(True)
Martin Geisler
pure Python implementation of bdiff.c
r7703
if not a:
s = "".join(b)
return s and (struct.pack(">lll", 0, 0, len(s)) + s)
bin = []
p = [0]
for i in a: p.append(p[-1] + len(i))
d = difflib.SequenceMatcher(None, a, b).get_matching_blocks()
d = _normalizeblocks(a, b, d)
la = 0
lb = 0
for am, bm, size in d:
s = "".join(b[lb:bm])
if am > la or s:
bin.append(struct.pack(">lll", p[la], p[am], len(s)) + s)
la = am + size
lb = bm + size
return "".join(bin)
def blocks(a, b):
Matt Mackall
pure/bdiff: fix circular import
r7944 an = splitnewlines(a)
bn = splitnewlines(b)
Martin Geisler
pure Python implementation of bdiff.c
r7703 d = difflib.SequenceMatcher(None, an, bn).get_matching_blocks()
d = _normalizeblocks(an, bn, d)
return [(i, i + n, j, j + n) for (i, j, n) in d]
Patrick Mezard
mdiff: replace wscleanup() regexps with C loops...
r15530 def fixws(text, allws):
if allws:
text = re.sub('[ \t\r]+', '', text)
else:
text = re.sub('[ \t\r]+', ' ', text)
text = text.replace(' \n', '\n')
return text