##// END OF EJS Templates
xdiff: add a preprocessing step that trims files...
xdiff: add a preprocessing step that trims files xdiff has a `xdl_trim_ends` step that removes common lines, unmatchable lines. That is in theory good, but happens too late - after splitting, hashing, and adjusting the hash values so they are unique. Those splitting, hashing and adjusting hash values steps could have noticeable overhead. Diffing two large files with minor (one-line-ish) changes are not uncommon. In that case, the raw performance of those preparation steps seriously matter. Even allocating an O(N) array and storing line offsets to it is expensive. Therefore my previous attempts [1] [2] cannot be good enough since they do not remove the O(N) array assignment. This patch adds a preprocessing step - `xdl_trim_files` that runs before other preprocessing steps. It counts common prefix and suffix and lines in them (needed for displaying line number), without doing anything else. Testing with a crafted large (169MB) file, with minor change: ``` open('a','w').write(''.join('%s\n' % (i % 100000) for i in xrange(30000000) if i != 6000000)) open('b','w').write(''.join('%s\n' % (i % 100000) for i in xrange(30000000) if i != 6003000)) ``` Running xdiff by a simple binary [3], this patch improves the xdiff perf by more than 10x for the above case: ``` # xdiff before this patch 2.41s user 1.13s system 98% cpu 3.592 total # xdiff after this patch 0.14s user 0.16s system 98% cpu 0.309 total # gnu diffutils 0.12s user 0.15s system 98% cpu 0.272 total # (best of 20 runs) ``` It's still slightly slower than GNU diffutils. But it's pretty close now. Testing with real repo data: For the whole repo, this patch makes xdiff 25% faster: ``` # hg perfbdiff --count 100 --alldata -c d334afc585e2 --blocks [--xdiff] # xdiff, after ! wall 0.058861 comb 0.050000 user 0.050000 sys 0.000000 (best of 100) # xdiff, before ! wall 0.077816 comb 0.080000 user 0.080000 sys 0.000000 (best of 91) # bdiff ! wall 0.117473 comb 0.120000 user 0.120000 sys 0.000000 (best of 67) ``` For files that are long (ex. commands.py), the speedup is more than 3x, very significant: ``` # hg perfbdiff --count 3000 --blocks commands.py.i 1 [--xdiff] # xdiff, after ! wall 0.690583 comb 0.690000 user 0.690000 sys 0.000000 (best of 12) # xdiff, before ! wall 2.240361 comb 2.210000 user 2.210000 sys 0.000000 (best of 4) # bdiff ! wall 2.469852 comb 2.440000 user 2.440000 sys 0.000000 (best of 4) ``` [1]: https://phab.mercurial-scm.org/D2631 [2]: https://phab.mercurial-scm.org/D2634 [3]: ``` // Code to run xdiff from command line. No proper error handling. #include <stdlib.h> #include <unistd.h> #include <sys/types.h> #include <sys/stat.h> #include <fcntl.h> #include "mercurial/thirdparty/xdiff/xdiff.h" #define ensure(x) if (!(x)) exit(255); mmfile_t readfile(const char *path) { struct stat st; int fd = open(path, O_RDONLY); fstat(fd, &st); mmfile_t file = { malloc(st.st_size), st.st_size }; ensure(read(fd, file.ptr, st.st_size) == st.st_size); close(fd); return file; } int main(int argc, char const *argv[]) { mmfile_t a = readfile(argv[1]), b = readfile(argv[2]); xpparam_t xpp = {0}; xdemitconf_t xecfg = {0}; xdemitcb_t ecb = {0}; xdl_diff(&a, &b, &xpp, &xecfg, &ecb); return 0; } ``` Differential Revision: https://phab.mercurial-scm.org/D2686

File last commit:

r36611:900e5ee4 default
r36838:f33a87cf default
Show More
namespaces.py
201 lines | 7.5 KiB | text/x-python | PythonLexer
from __future__ import absolute_import
from .i18n import _
from . import (
registrar,
templatekw,
util,
)
def tolist(val):
"""
a convenience method to return an empty list instead of None
"""
if val is None:
return []
else:
return [val]
class namespaces(object):
"""provides an interface to register and operate on multiple namespaces. See
the namespace class below for details on the namespace object.
"""
_names_version = 0
def __init__(self):
self._names = util.sortdict()
columns = templatekw.getlogcolumns()
# we need current mercurial named objects (bookmarks, tags, and
# branches) to be initialized somewhere, so that place is here
bmknames = lambda repo: repo._bookmarks.keys()
bmknamemap = lambda repo, name: tolist(repo._bookmarks.get(name))
bmknodemap = lambda repo, node: repo.nodebookmarks(node)
n = namespace("bookmarks", templatename="bookmark",
logfmt=columns['bookmark'],
listnames=bmknames,
namemap=bmknamemap, nodemap=bmknodemap,
builtin=True)
self.addnamespace(n)
tagnames = lambda repo: [t for t, n in repo.tagslist()]
tagnamemap = lambda repo, name: tolist(repo._tagscache.tags.get(name))
tagnodemap = lambda repo, node: repo.nodetags(node)
n = namespace("tags", templatename="tag",
logfmt=columns['tag'],
listnames=tagnames,
namemap=tagnamemap, nodemap=tagnodemap,
deprecated={'tip'},
builtin=True)
self.addnamespace(n)
bnames = lambda repo: repo.branchmap().keys()
bnamemap = lambda repo, name: tolist(repo.branchtip(name, True))
bnodemap = lambda repo, node: [repo[node].branch()]
n = namespace("branches", templatename="branch",
logfmt=columns['branch'],
listnames=bnames,
namemap=bnamemap, nodemap=bnodemap,
builtin=True)
self.addnamespace(n)
def __getitem__(self, namespace):
"""returns the namespace object"""
return self._names[namespace]
def __iter__(self):
return self._names.__iter__()
def items(self):
return self._names.iteritems()
iteritems = items
def addnamespace(self, namespace, order=None):
"""register a namespace
namespace: the name to be registered (in plural form)
order: optional argument to specify the order of namespaces
(e.g. 'branches' should be listed before 'bookmarks')
"""
if order is not None:
self._names.insert(order, namespace.name, namespace)
else:
self._names[namespace.name] = namespace
# we only generate a template keyword if one does not already exist
if namespace.name not in templatekw.keywords:
templatekeyword = registrar.templatekeyword(templatekw.keywords)
@templatekeyword(namespace.name, requires={'repo', 'ctx', 'templ'})
def generatekw(context, mapping):
return templatekw.shownames(context, mapping, namespace.name)
def singlenode(self, repo, name):
"""
Return the 'best' node for the given name. Best means the first node
in the first nonempty list returned by a name-to-nodes mapping function
in the defined precedence order.
Raises a KeyError if there is no such node.
"""
for ns, v in self._names.iteritems():
n = v.namemap(repo, name)
if n:
# return max revision number
if len(n) > 1:
cl = repo.changelog
maxrev = max(cl.rev(node) for node in n)
return cl.node(maxrev)
return n[0]
raise KeyError(_('no such name: %s') % name)
class namespace(object):
"""provides an interface to a namespace
Namespaces are basically generic many-to-many mapping between some
(namespaced) names and nodes. The goal here is to control the pollution of
jamming things into tags or bookmarks (in extension-land) and to simplify
internal bits of mercurial: log output, tab completion, etc.
More precisely, we define a mapping of names to nodes, and a mapping from
nodes to names. Each mapping returns a list.
Furthermore, each name mapping will be passed a name to lookup which might
not be in its domain. In this case, each method should return an empty list
and not raise an error.
This namespace object will define the properties we need:
'name': the namespace (plural form)
'templatename': name to use for templating (usually the singular form
of the plural namespace name)
'listnames': list of all names in the namespace (usually the keys of a
dictionary)
'namemap': function that takes a name and returns a list of nodes
'nodemap': function that takes a node and returns a list of names
'deprecated': set of names to be masked for ordinary use
'builtin': bool indicating if this namespace is supported by core
Mercurial.
"""
def __init__(self, name, templatename=None, logname=None, colorname=None,
logfmt=None, listnames=None, namemap=None, nodemap=None,
deprecated=None, builtin=False):
"""create a namespace
name: the namespace to be registered (in plural form)
templatename: the name to use for templating
logname: the name to use for log output; if not specified templatename
is used
colorname: the name to use for colored log output; if not specified
logname is used
logfmt: the format to use for (i18n-ed) log output; if not specified
it is composed from logname
listnames: function to list all names
namemap: function that inputs a name, output node(s)
nodemap: function that inputs a node, output name(s)
deprecated: set of names to be masked for ordinary use
builtin: whether namespace is implemented by core Mercurial
"""
self.name = name
self.templatename = templatename
self.logname = logname
self.colorname = colorname
self.logfmt = logfmt
self.listnames = listnames
self.namemap = namemap
self.nodemap = nodemap
# if logname is not specified, use the template name as backup
if self.logname is None:
self.logname = self.templatename
# if colorname is not specified, just use the logname as a backup
if self.colorname is None:
self.colorname = self.logname
# if logfmt is not specified, compose it from logname as backup
if self.logfmt is None:
# i18n: column positioning for "hg log"
self.logfmt = ("%s:" % self.logname).ljust(13) + "%s\n"
if deprecated is None:
self.deprecated = set()
else:
self.deprecated = deprecated
self.builtin = builtin
def names(self, repo, node):
"""method that returns a (sorted) list of names in a namespace that
match a given node"""
return sorted(self.nodemap(repo, node))
def nodes(self, repo, name):
"""method that returns a list of nodes in a namespace that
match a given name.
"""
return sorted(self.namemap(repo, name))