##// END OF EJS Templates
revlog: change generaldelta delta parent heuristic...
revlog: change generaldelta delta parent heuristic The old generaldelta heuristic was "if p1 (or p2) was closer than the last full text, use it, otherwise use prev". This was problematic when a repo contained multiple branches that were very different. If commits to branch A were pushed, and the last full text was branch B, it would generate a fulltext. Then if branch B was pushed, it would generate another fulltext. The problem is that the last fulltext (and delta'ing against `prev` in general) has no correlation with the contents of the incoming revision, and therefore will always have degenerate cases. According to the blame, that algorithm was chosen to minimize the chain length. Since there is already code that protects against that (the delta-vs-fulltext code), and since it has been improved since the original generaldelta algorithm went in (2011), I believe the chain length criteria will still be preserved. The new algorithm always diffs against p1 (or p2 if it's closer), unless the resulting delta will fail the delta-vs-fulltext check, in which case we delta against prev. Some before and after stats on manifest.d size. internal large repo old heuristic - 2.0 GB new heuristic - 1.2 GB mozilla-central old heuristic - 242 MB new heuristic - 261 MB The regression in mozilla central is due to the new heuristic choosing p2r as the delta when it's closer to the tip. Switching the algorithm to always prefer p1r brings the size back down (242 MB). This is result of the way in which mozilla does merges and pushes, and the result could easily swing the other direction in other repos (depending on if they merge X into Y or Y into X), but will never be as degenerate as before. I future patch will address the regression by introducing an optional, even more aggressive delta heuristic which will knock the mozilla manifest size down dramatically.

File last commit:

r25661:20de1ace default
r26117:4dc5b51f default
Show More
check-code.py
573 lines | 20.3 KiB | text/x-python | PythonLexer
#!/usr/bin/env python
#
# check-code - a style and portability checker for Mercurial
#
# Copyright 2010 Matt Mackall <mpm@selenic.com>
#
# This software may be used and distributed according to the terms of the
# GNU General Public License version 2 or any later version.
"""style and portability checker for Mercurial
when a rule triggers wrong, do one of the following (prefer one from top):
* do the work-around the rule suggests
* doublecheck that it is a false match
* improve the rule pattern
* add an ignore pattern to the rule (3rd arg) which matches your good line
(you can append a short comment and match this, like: #re-raises, # no-py24)
* change the pattern to a warning and list the exception in test-check-code-hg
* ONLY use no--check-code for skipping entire files from external sources
"""
import re, glob, os, sys
import keyword
import optparse
try:
import re2
except ImportError:
re2 = None
def compilere(pat, multiline=False):
if multiline:
pat = '(?m)' + pat
if re2:
try:
return re2.compile(pat)
except re2.error:
pass
return re.compile(pat)
def repquote(m):
fromc = '.:'
tochr = 'pq'
def encodechr(i):
if i > 255:
return 'u'
c = chr(i)
if c in ' \n':
return c
if c.isalpha():
return 'x'
if c.isdigit():
return 'n'
try:
return tochr[fromc.find(c)]
except (ValueError, IndexError):
return 'o'
t = m.group('text')
tt = ''.join(encodechr(i) for i in xrange(256))
t = t.translate(tt)
return m.group('quote') + t + m.group('quote')
def reppython(m):
comment = m.group('comment')
if comment:
l = len(comment.rstrip())
return "#" * l + comment[l:]
return repquote(m)
def repcomment(m):
return m.group(1) + "#" * len(m.group(2))
def repccomment(m):
t = re.sub(r"((?<=\n) )|\S", "x", m.group(2))
return m.group(1) + t + "*/"
def repcallspaces(m):
t = re.sub(r"\n\s+", "\n", m.group(2))
return m.group(1) + t
def repinclude(m):
return m.group(1) + "<foo>"
def rephere(m):
t = re.sub(r"\S", "x", m.group(2))
return m.group(1) + t
testpats = [
[
(r'pushd|popd', "don't use 'pushd' or 'popd', use 'cd'"),
(r'\W\$?\(\([^\)\n]*\)\)', "don't use (()) or $(()), use 'expr'"),
(r'grep.*-q', "don't use 'grep -q', redirect to /dev/null"),
(r'(?<!hg )grep.*-a', "don't use 'grep -a', use in-line python"),
(r'sed.*-i', "don't use 'sed -i', use a temporary file"),
(r'\becho\b.*\\n', "don't use 'echo \\n', use printf"),
(r'echo -n', "don't use 'echo -n', use printf"),
(r'(^|\|\s*)\bwc\b[^|]*$\n(?!.*\(re\))', "filter wc output"),
(r'head -c', "don't use 'head -c', use 'dd'"),
(r'tail -n', "don't use the '-n' option to tail, just use '-<num>'"),
(r'sha1sum', "don't use sha1sum, use $TESTDIR/md5sum.py"),
(r'ls.*-\w*R', "don't use 'ls -R', use 'find'"),
(r'printf.*[^\\]\\([1-9]|0\d)', "don't use 'printf \NNN', use Python"),
(r'printf.*[^\\]\\x', "don't use printf \\x, use Python"),
(r'\$\(.*\)', "don't use $(expr), use `expr`"),
(r'rm -rf \*', "don't use naked rm -rf, target a directory"),
(r'(^|\|\s*)grep (-\w\s+)*[^|]*[(|]\w',
"use egrep for extended grep syntax"),
(r'/bin/', "don't use explicit paths for tools"),
(r'[^\n]\Z', "no trailing newline"),
(r'export.*=', "don't export and assign at once"),
(r'^source\b', "don't use 'source', use '.'"),
(r'touch -d', "don't use 'touch -d', use 'touch -t' instead"),
(r'ls +[^|\n-]+ +-', "options to 'ls' must come before filenames"),
(r'[^>\n]>\s*\$HGRCPATH', "don't overwrite $HGRCPATH, append to it"),
(r'^stop\(\)', "don't use 'stop' as a shell function name"),
(r'(\[|\btest\b).*-e ', "don't use 'test -e', use 'test -f'"),
(r'\[\[\s+[^\]]*\]\]', "don't use '[[ ]]', use '[ ]'"),
(r'^alias\b.*=', "don't use alias, use a function"),
(r'if\s*!', "don't use '!' to negate exit status"),
(r'/dev/u?random', "don't use entropy, use /dev/zero"),
(r'do\s*true;\s*done', "don't use true as loop body, use sleep 0"),
(r'^( *)\t', "don't use tabs to indent"),
(r'sed (-e )?\'(\d+|/[^/]*/)i(?!\\\n)',
"put a backslash-escaped newline after sed 'i' command"),
(r'^diff *-\w*u.*$\n(^ \$ |^$)', "prefix diff -u with cmp"),
(r'seq ', "don't use 'seq', use $TESTDIR/seq.py")
],
# warnings
[
(r'^function', "don't use 'function', use old style"),
(r'^diff.*-\w*N', "don't use 'diff -N'"),
(r'\$PWD|\${PWD}', "don't use $PWD, use `pwd`"),
(r'^([^"\'\n]|("[^"\n]*")|(\'[^\'\n]*\'))*\^', "^ must be quoted"),
(r'kill (`|\$\()', "don't use kill, use killdaemons.py")
]
]
testfilters = [
(r"( *)(#([^\n]*\S)?)", repcomment),
(r"<<(\S+)((.|\n)*?\n\1)", rephere),
]
winglobmsg = "use (glob) to match Windows paths too"
uprefix = r"^ \$ "
utestpats = [
[
(r'^(\S.*|| [$>] .*)[ \t]\n', "trailing whitespace on non-output"),
(uprefix + r'.*\|\s*sed[^|>\n]*\n',
"use regex test output patterns instead of sed"),
(uprefix + r'(true|exit 0)', "explicit zero exit unnecessary"),
(uprefix + r'.*(?<!\[)\$\?', "explicit exit code checks unnecessary"),
(uprefix + r'.*\|\| echo.*(fail|error)',
"explicit exit code checks unnecessary"),
(uprefix + r'set -e', "don't use set -e"),
(uprefix + r'(\s|fi\b|done\b)', "use > for continued lines"),
(uprefix + r'.*:\.\S*/', "x:.y in a path does not work on msys, rewrite "
"as x://.y, or see `hg log -k msys` for alternatives", r'-\S+:\.|' #-Rxxx
'# no-msys'), # in test-pull.t which is skipped on windows
(r'^ saved backup bundle to \$TESTTMP.*\.hg$', winglobmsg),
(r'^ changeset .* references (corrupted|missing) \$TESTTMP/.*[^)]$',
winglobmsg),
(r'^ pulling from \$TESTTMP/.*[^)]$', winglobmsg,
'\$TESTTMP/unix-repo$'), # in test-issue1802.t which skipped on windows
(r'^ reverting (?!subrepo ).*/.*[^)]$', winglobmsg),
(r'^ cloning subrepo \S+/.*[^)]$', winglobmsg),
(r'^ pushing to \$TESTTMP/.*[^)]$', winglobmsg),
(r'^ pushing subrepo \S+/\S+ to.*[^)]$', winglobmsg),
(r'^ moving \S+/.*[^)]$', winglobmsg),
(r'^ no changes made to subrepo since.*/.*[^)]$', winglobmsg),
(r'^ .*: largefile \S+ not available from file:.*/.*[^)]$', winglobmsg),
(r'^ .*file://\$TESTTMP',
'write "file:/*/$TESTTMP" + (glob) to match on windows too'),
(r'^ (cat|find): .*: No such file or directory',
'use test -f to test for file existence'),
],
# warnings
[
(r'^ [^*?/\n]* \(glob\)$',
"glob match with no glob character (?*/)"),
]
]
for i in [0, 1]:
for tp in testpats[i]:
p = tp[0]
m = tp[1]
if p.startswith(r'^'):
p = r"^ [$>] (%s)" % p[1:]
else:
p = r"^ [$>] .*(%s)" % p
utestpats[i].append((p, m) + tp[2:])
utestfilters = [
(r"<<(\S+)((.|\n)*?\n > \1)", rephere),
(r"( *)(#([^\n]*\S)?)", repcomment),
]
pypats = [
[
(r'\([^)]*\*\w[^()]+\w+=', "can't pass varargs with keyword in Py2.5"),
(r'^\s*def\s*\w+\s*\(.*,\s*\(',
"tuple parameter unpacking not available in Python 3+"),
(r'lambda\s*\(.*,.*\)',
"tuple parameter unpacking not available in Python 3+"),
(r'import (.+,[^.]+\.[^.]+|[^.]+\.[^.]+,)',
'2to3 can\'t always rewrite "import qux, foo.bar", '
'use "import foo.bar" on its own line instead.'),
(r'(?<!def)\s+(cmp)\(', "cmp is not available in Python 3+"),
(r'\breduce\s*\(.*', "reduce is not available in Python 3+"),
(r'dict\(.*=', 'dict() is different in Py2 and 3 and is slower than {}',
'dict-from-generator'),
(r'\.has_key\b', "dict.has_key is not available in Python 3+"),
(r'\s<>\s', '<> operator is not available in Python 3+, use !='),
(r'^\s*\t', "don't use tabs"),
(r'\S;\s*\n', "semicolon"),
(r'[^_]_\([ \t\n]*(?:"[^"]+"[ \t\n+]*)+%', "don't use % inside _()"),
(r"[^_]_\([ \t\n]*(?:'[^']+'[ \t\n+]*)+%", "don't use % inside _()"),
(r'(\w|\)),\w', "missing whitespace after ,"),
(r'(\w|\))[+/*\-<>]\w', "missing whitespace in expression"),
(r'^\s+(\w|\.)+=\w[^,()\n]*$', "missing whitespace in assignment"),
(r'.{81}', "line too long"),
(r' x+[xo][\'"]\n\s+[\'"]x', 'string join across lines with no space'),
(r'[^\n]\Z', "no trailing newline"),
(r'(\S[ \t]+|^[ \t]+)\n', "trailing whitespace"),
# (r'^\s+[^_ \n][^_. \n]+_[^_\n]+\s*=',
# "don't use underbars in identifiers"),
(r'^\s+(self\.)?[A-za-z][a-z0-9]+[A-Z]\w* = ',
"don't use camelcase in identifiers"),
(r'^\s*(if|while|def|class|except|try)\s[^[\n]*:\s*[^\\n]#\s]+',
"linebreak after :"),
(r'class\s[^( \n]+:', "old-style class, use class foo(object)"),
(r'class\s[^( \n]+\(\):',
"class foo() creates old style object, use class foo(object)"),
(r'\b(%s)\(' % '|'.join(k for k in keyword.kwlist
if k not in ('print', 'exec')),
"Python keyword is not a function"),
(r',]', "unneeded trailing ',' in list"),
# (r'class\s[A-Z][^\(]*\((?!Exception)',
# "don't capitalize non-exception classes"),
# (r'in range\(', "use xrange"),
# (r'^\s*print\s+', "avoid using print in core and extensions"),
(r'[\x80-\xff]', "non-ASCII character literal"),
(r'("\')\.format\(', "str.format() has no bytes counterpart, use %"),
(r'^\s*(%s)\s\s' % '|'.join(keyword.kwlist),
"gratuitous whitespace after Python keyword"),
(r'([\(\[][ \t]\S)|(\S[ \t][\)\]])', "gratuitous whitespace in () or []"),
# (r'\s\s=', "gratuitous whitespace before ="),
(r'[^>< ](\+=|-=|!=|<>|<=|>=|<<=|>>=|%=)\S',
"missing whitespace around operator"),
(r'[^>< ](\+=|-=|!=|<>|<=|>=|<<=|>>=|%=)\s',
"missing whitespace around operator"),
(r'\s(\+=|-=|!=|<>|<=|>=|<<=|>>=|%=)\S',
"missing whitespace around operator"),
(r'[^^+=*/!<>&| %-](\s=|=\s)[^= ]',
"wrong whitespace around ="),
(r'\([^()]*( =[^=]|[^<>!=]= )',
"no whitespace around = for named parameters"),
(r'raise Exception', "don't raise generic exceptions"),
(r'raise [^,(]+, (\([^\)]+\)|[^,\(\)]+)$',
"don't use old-style two-argument raise, use Exception(message)"),
(r' is\s+(not\s+)?["\'0-9-]', "object comparison with literal"),
(r' [=!]=\s+(True|False|None)',
"comparison with singleton, use 'is' or 'is not' instead"),
(r'^\s*(while|if) [01]:',
"use True/False for constant Boolean expression"),
(r'(?:(?<!def)\s+|\()hasattr',
'hasattr(foo, bar) is broken, use util.safehasattr(foo, bar) instead'),
(r'opener\([^)]*\).read\(',
"use opener.read() instead"),
(r'opener\([^)]*\).write\(',
"use opener.write() instead"),
(r'[\s\(](open|file)\([^)]*\)\.read\(',
"use util.readfile() instead"),
(r'[\s\(](open|file)\([^)]*\)\.write\(',
"use util.writefile() instead"),
(r'^[\s\(]*(open(er)?|file)\([^)]*\)',
"always assign an opened file to a variable, and close it afterwards"),
(r'[\s\(](open|file)\([^)]*\)\.',
"always assign an opened file to a variable, and close it afterwards"),
(r'(?i)descend[e]nt', "the proper spelling is descendAnt"),
(r'\.debug\(\_', "don't mark debug messages for translation"),
(r'\.strip\(\)\.split\(\)', "no need to strip before splitting"),
(r'^\s*except\s*:', "naked except clause", r'#.*re-raises'),
(r'^\s*except\s([^\(,]+|\([^\)]+\))\s*,',
'legacy exception syntax; use "as" instead of ","'),
(r':\n( )*( ){1,3}[^ ]', "must indent 4 spaces"),
(r'ui\.(status|progress|write|note|warn)\([\'\"]x',
"missing _() in ui message (use () to hide false-positives)"),
(r'release\(.*wlock, .*lock\)', "wrong lock release order"),
(r'\b__bool__\b', "__bool__ should be __nonzero__ in Python 2"),
(r'os\.path\.join\(.*, *(""|\'\')\)',
"use pathutil.normasprefix(path) instead of os.path.join(path, '')"),
(r'\s0[0-7]+\b', 'legacy octal syntax; use "0o" prefix instead of "0"'),
],
# warnings
[
(r'(^| )pp +xxxxqq[ \n][^\n]', "add two newlines after '.. note::'"),
]
]
pyfilters = [
(r"""(?msx)(?P<comment>\#.*?$)|
((?P<quote>('''|\"\"\"|(?<!')'(?!')|(?<!")"(?!")))
(?P<text>(([^\\]|\\.)*?))
(?P=quote))""", reppython),
]
txtfilters = []
txtpats = [
[
('\s$', 'trailing whitespace'),
('.. note::[ \n][^\n]', 'add two newlines after note::')
],
[]
]
cpats = [
[
(r'//', "don't use //-style comments"),
(r'^ ', "don't use spaces to indent"),
(r'\S\t', "don't use tabs except for indent"),
(r'(\S[ \t]+|^[ \t]+)\n', "trailing whitespace"),
(r'.{81}', "line too long"),
(r'(while|if|do|for)\(', "use space after while/if/do/for"),
(r'return\(', "return is not a function"),
(r' ;', "no space before ;"),
(r'[^;] \)', "no space before )"),
(r'[)][{]', "space between ) and {"),
(r'\w+\* \w+', "use int *foo, not int* foo"),
(r'\W\([^\)]+\) \w+', "use (int)foo, not (int) foo"),
(r'\w+ (\+\+|--)', "use foo++, not foo ++"),
(r'\w,\w', "missing whitespace after ,"),
(r'^[^#]\w[+/*]\w', "missing whitespace in expression"),
(r'^#\s+\w', "use #foo, not # foo"),
(r'[^\n]\Z', "no trailing newline"),
(r'^\s*#import\b', "use only #include in standard C code"),
],
# warnings
[]
]
cfilters = [
(r'(/\*)(((\*(?!/))|[^*])*)\*/', repccomment),
(r'''(?P<quote>(?<!")")(?P<text>([^"]|\\")+)"(?!")''', repquote),
(r'''(#\s*include\s+<)([^>]+)>''', repinclude),
(r'(\()([^)]+\))', repcallspaces),
]
inutilpats = [
[
(r'\bui\.', "don't use ui in util"),
],
# warnings
[]
]
inrevlogpats = [
[
(r'\brepo\.', "don't use repo in revlog"),
],
# warnings
[]
]
webtemplatefilters = []
webtemplatepats = [
[],
[
(r'{desc(\|(?!websub|firstline)[^\|]*)+}',
'follow desc keyword with either firstline or websub'),
]
]
checks = [
('python', r'.*\.(py|cgi)$', r'^#!.*python', pyfilters, pypats),
('test script', r'(.*/)?test-[^.~]*$', '', testfilters, testpats),
('c', r'.*\.[ch]$', '', cfilters, cpats),
('unified test', r'.*\.t$', '', utestfilters, utestpats),
('layering violation repo in revlog', r'mercurial/revlog\.py', '',
pyfilters, inrevlogpats),
('layering violation ui in util', r'mercurial/util\.py', '', pyfilters,
inutilpats),
('txt', r'.*\.txt$', '', txtfilters, txtpats),
('web template', r'mercurial/templates/.*\.tmpl', '',
webtemplatefilters, webtemplatepats),
]
def _preparepats():
for c in checks:
failandwarn = c[-1]
for pats in failandwarn:
for i, pseq in enumerate(pats):
# fix-up regexes for multi-line searches
p = pseq[0]
# \s doesn't match \n
p = re.sub(r'(?<!\\)\\s', r'[ \\t]', p)
# [^...] doesn't match newline
p = re.sub(r'(?<!\\)\[\^', r'[^\\n', p)
pats[i] = (re.compile(p, re.MULTILINE),) + pseq[1:]
filters = c[3]
for i, flt in enumerate(filters):
filters[i] = re.compile(flt[0]), flt[1]
_preparepats()
class norepeatlogger(object):
def __init__(self):
self._lastseen = None
def log(self, fname, lineno, line, msg, blame):
"""print error related a to given line of a given file.
The faulty line will also be printed but only once in the case
of multiple errors.
:fname: filename
:lineno: line number
:line: actual content of the line
:msg: error message
"""
msgid = fname, lineno, line
if msgid != self._lastseen:
if blame:
print "%s:%d (%s):" % (fname, lineno, blame)
else:
print "%s:%d:" % (fname, lineno)
print " > %s" % line
self._lastseen = msgid
print " " + msg
_defaultlogger = norepeatlogger()
def getblame(f):
lines = []
for l in os.popen('hg annotate -un %s' % f):
start, line = l.split(':', 1)
user, rev = start.split()
lines.append((line[1:-1], user, rev))
return lines
def checkfile(f, logfunc=_defaultlogger.log, maxerr=None, warnings=False,
blame=False, debug=False, lineno=True):
"""checks style and portability of a given file
:f: filepath
:logfunc: function used to report error
logfunc(filename, linenumber, linecontent, errormessage)
:maxerr: number of error to display before aborting.
Set to false (default) to report all errors
return True if no error is found, False otherwise.
"""
blamecache = None
result = True
try:
fp = open(f)
except IOError as e:
print "Skipping %s, %s" % (f, str(e).split(':', 1)[0])
return result
pre = post = fp.read()
fp.close()
for name, match, magic, filters, pats in checks:
if debug:
print name, f
fc = 0
if not (re.match(match, f) or (magic and re.search(magic, f))):
if debug:
print "Skipping %s for %s it doesn't match %s" % (
name, match, f)
continue
if "no-" "check-code" in pre:
print "Skipping %s it has no-" "check-code" % f
return "Skip" # skip checking this file
for p, r in filters:
post = re.sub(p, r, post)
nerrs = len(pats[0]) # nerr elements are errors
if warnings:
pats = pats[0] + pats[1]
else:
pats = pats[0]
# print post # uncomment to show filtered version
if debug:
print "Checking %s for %s" % (name, f)
prelines = None
errors = []
for i, pat in enumerate(pats):
if len(pat) == 3:
p, msg, ignore = pat
else:
p, msg = pat
ignore = None
if i >= nerrs:
msg = "warning: " + msg
pos = 0
n = 0
for m in p.finditer(post):
if prelines is None:
prelines = pre.splitlines()
postlines = post.splitlines(True)
start = m.start()
while n < len(postlines):
step = len(postlines[n])
if pos + step > start:
break
pos += step
n += 1
l = prelines[n]
if ignore and re.search(ignore, l, re.MULTILINE):
if debug:
print "Skipping %s for %s:%s (ignore pattern)" % (
name, f, n)
continue
bd = ""
if blame:
bd = 'working directory'
if not blamecache:
blamecache = getblame(f)
if n < len(blamecache):
bl, bu, br = blamecache[n]
if bl == l:
bd = '%s@%s' % (bu, br)
errors.append((f, lineno and n + 1, l, msg, bd))
result = False
errors.sort()
for e in errors:
logfunc(*e)
fc += 1
if maxerr and fc >= maxerr:
print " (too many errors, giving up)"
break
return result
if __name__ == "__main__":
parser = optparse.OptionParser("%prog [options] [files]")
parser.add_option("-w", "--warnings", action="store_true",
help="include warning-level checks")
parser.add_option("-p", "--per-file", type="int",
help="max warnings per file")
parser.add_option("-b", "--blame", action="store_true",
help="use annotate to generate blame info")
parser.add_option("", "--debug", action="store_true",
help="show debug information")
parser.add_option("", "--nolineno", action="store_false",
dest='lineno', help="don't show line numbers")
parser.set_defaults(per_file=15, warnings=False, blame=False, debug=False,
lineno=True)
(options, args) = parser.parse_args()
if len(args) == 0:
check = glob.glob("*")
else:
check = args
ret = 0
for f in check:
if not checkfile(f, maxerr=options.per_file, warnings=options.warnings,
blame=options.blame, debug=options.debug,
lineno=options.lineno):
ret = 1
sys.exit(ret)