mercurial-crew-with-dirclash: mercurial/dirstate.py comparison

comparison mercurial/dirstate.py @ 1183:d9e85a75dbda

Optimize dirstate walking This generally cuts the time for hg status/diff in half, from 2s down to 1s. The main parts I'm trying to optimize are: 1) os.walk stats every file. dirstate.changes then stats every file again. 2) os.walk yields every file and subdir to dirstate.traverse who yields every file and everything in the dirstate map. dirstate.walk then filters this mass and yields every file to the caller. There should be fewer steps in here, and fewer duplicate strings yielded. 3) dirstate.walk runs util.unique on the results from dirstate.traverse, even though it is also passing things through dirstate.seen to look for duplicates. I've turned os.walk into something hg specific that takes all the dirstate ignore and matching rules into account. The new function also takes an function arg (statmatch()) the caller supplies to help filter out files it doesn't care about. dirstate.changes uses this to update state for each file, avoiding the second stat call. dirstate.walk is changed to turn the match function it is passed into a statmatch function. The only real difference is that a statmatch function takes the stat data as a second parameter. It now calls dirstate.walkhelper, who requires a statmatch function to be passed. This fails test-walk, but right now I think this is from a sorting error fixed by this patch. Index: crew/mercurial/dirstate.py ===================================================================

author	mason@suse.com
date	Thu, 01 Sep 2005 07:34:53 -0700
parents	30ab5b8ee8ec
children	cc61d366bc3b

comparison

equal deleted inserted replaced

-:24d553b598e8
+:d9e85a75dbda
 self.ui = ui
 self.map = None
 self.pl = None
 self.copies = {}
 self.ignorefunc = None
+self.blockignore = False
 def wjoin(self, f):
 return os.path.join(self.root, f)
 def getcwd(self):
 cwd = os.getcwd()
 if cwd == self.root: return ''
 return cwd[len(self.root) + 1:]
 def ignore(self, f):
+if self.blockignore:
+return False
 if not self.ignorefunc:
 bigpat = []
 try:
 l = file(self.wjoin(".hgignore"))
 for pat in l:
 if not dc:
 dc = self.map.copy()
 elif not dc:
 dc = self.filterfiles(files)
+def statmatch(file, stat):
+if file not in dc and self.ignore(file):
+return False
+return match(file)
+return self.walkhelper(files=files, statmatch=statmatch, dc=dc)
+# walk recursively through the directory tree, finding all files
+# matched by the statmatch function
+#
+# results are yielded in a tuple (src, filename), where src is one of:
+# 'f' the file was found in the directory tree
+# 'm' the file was only in the dirstate and not in the tree
+#
+# dc is an optional arg for the current dirstate.  dc is not modified
+# directly by this function, but might be modified by your statmatch call.
+#
+def walkhelper(self, files, statmatch, dc):
+# recursion free walker, faster than os.walk.
+def findfiles(s):
+retfiles = []
+work = [s]
+while work:
+top = work.pop()
+names = os.listdir(top)
+names.sort()
+# nd is the top of the repository dir tree
+nd = util.normpath(top[len(self.root) + 1:])
+if nd == '.': nd = ''
+for f in names:
+np = os.path.join(nd, f)
+if seen(np):
+continue
+p = os.path.join(top, f)
+st = os.stat(p)
+if stat.S_ISDIR(st.st_mode):
+ds = os.path.join(nd, f +'/')
+if statmatch(ds, st):
+work.append(p)
+else:
+if statmatch(np, st):
+yield np
 known = {'.hg': 1}
 def seen(fn):
 if fn in known: return True
 known[fn] = 1
-def traverse():
-for ff in util.unique(files):
+# step one, find all files that match our criteria
-f = os.path.join(self.root, ff)
+files.sort()
-try:
+for ff in util.unique(files):
-st = os.stat(f)
+f = os.path.join(self.root, ff)
-except OSError, inst:
+try:
-if ff not in dc: self.ui.warn('%s: %s\n' % (
+st = os.stat(f)
-util.pathto(self.getcwd(), ff),
+except OSError, inst:
-inst.strerror))
+if ff not in dc: self.ui.warn('%s: %s\n' % (
+util.pathto(self.getcwd(), ff),
+inst.strerror))
+continue
+if stat.S_ISDIR(st.st_mode):
+sorted = [ x for x in findfiles(f) ]
+sorted.sort()
+for fl in sorted:
+yield 'f', fl
+elif stat.S_ISREG(st.st_mode):
+ff = util.normpath(ff)
+if seen(ff):
 continue
-if stat.S_ISDIR(st.st_mode):
+found = False
-for dir, subdirs, fl in os.walk(f):
+self.blockignore = True
-d = dir[len(self.root) + 1:]
+if statmatch(ff, st):
-nd = util.normpath(d)
+found = True
-if nd == '.': nd = ''
+self.blockignore = False
-if seen(nd):
+if found:
-subdirs[:] = []
-continue
-for sd in subdirs:
-ds = os.path.join(nd, sd +'/')
-if self.ignore(ds) or not match(ds):
-subdirs.remove(sd)
-subdirs.sort()
-fl.sort()
-for fn in fl:
-fn = util.pconvert(os.path.join(d, fn))
-yield 'f', fn
-elif stat.S_ISREG(st.st_mode):
 yield 'f', ff
 else:
 kind = 'unknown'
 if stat.S_ISCHR(st.st_mode): kind = 'character device'
 elif stat.S_ISBLK(st.st_mode): kind = 'block device'
 elif stat.S_ISFIFO(st.st_mode): kind = 'fifo'
 elif stat.S_ISLNK(st.st_mode): kind = 'symbolic link'
 elif stat.S_ISSOCK(st.st_mode): kind = 'socket'
 self.ui.warn('%s: unsupported file type (type is %s)\n' % (
 util.pathto(self.getcwd(), ff),
 kind))
-ks = dc.keys()
+# step two run through anything left in the dc hash and yield
-ks.sort()
+# if we haven't already seen it
-for k in ks:
+ks = dc.keys()
+ks.sort()
+for k in ks:
+if not seen(k) and (statmatch(k, None)):
 yield 'm', k
-# yield only files that match: all in dirstate, others only if
-# not in .hgignore
-for src, fn in util.unique(traverse()):
-fn = util.normpath(fn)
-if seen(fn): continue
-if fn not in dc and self.ignore(fn):
-continue
-if match(fn):
-yield src, fn
 def changes(self, files=None, match=util.always):
 self.read()
 if not files:
+files = [self.root]
 dc = self.map.copy()
 else:
 dc = self.filterfiles(files)
 lookup, modified, added, unknown = [], [], [], []
 removed, deleted = [], []
-for src, fn in self.walk(files, match, dc=dc):
+# statmatch function to eliminate entries from the dirstate copy
-try:
+# and put files into the appropriate array.  This gets passed
-s = os.stat(os.path.join(self.root, fn))
+# to the walking code
-except OSError:
+def statmatch(fn, s):
-continue
+def checkappend(l, fn):
+if match is util.always or match(fn):
+l.append(fn)
+if not s or stat.S_ISDIR(s.st_mode):
+return self.ignore(fn) and False or match(fn)
 if not stat.S_ISREG(s.st_mode):
-continue
+return False
-c = dc.get(fn)
+c = dc.pop(fn, None)
 if c:
-del dc[fn]
+type, mode, size, time = c
-if c[0] == 'm':
+# check the common case first
-modified.append(fn)
+if type == 'n':
-elif c[0] == 'a':
+if size != s.st_size or (mode ^ s.st_mode) & 0100:
-added.append(fn)
+checkappend(modified, fn)
-elif c[0] == 'r':
+elif time != s.st_mtime:
+checkappend(lookup, fn)
+elif type == 'm':
+checkappend(modified, fn)
+elif type == 'a':
+checkappend(added, fn)
+elif type == 'r':
+checkappend(unknown, fn)
+else:
+if not self.ignore(fn) and match(fn):
 unknown.append(fn)
-elif c[2] != s.st_size or (c[1] ^ s.st_mode) & 0100:
+# return false because we've already handled all cases above.
-modified.append(fn)
+# there's no need for the walking code to process the file
-elif c[3] != s.st_mtime:
+# any further.
-lookup.append(fn)
+return False
-else:
-unknown.append(fn)
+# because our statmatch always returns false, self.walk will only
+# return files in the dirstate map that are not present in the FS.
+# But, we still need to iterate through the results to force the
+# walk to complete
+for src, fn in self.walkhelper(files, statmatch, dc):
+pass
+# anything left in dc didn't exist in the filesystem
 for fn, c in [(fn, c) for fn, c in dc.items() if match(fn)]:
 if c[0] == 'r':
 removed.append(fn)
 else:
 deleted.append(fn)

Mercurial > hg > mercurial-crew-with-dirclash

comparison mercurial/dirstate.py @ 1183:d9e85a75dbda