Fill in the uncompressed size during revlog.addgroup

This uses code from Matt to calculate the size change that
would result from applying a delta to keep an accurate running
total of the text size during revlog.addgroup

mercurial/mdiff.py

@@ -1,195 +1,196 b''
 # mdiff.py - diff and patch routines for mercurial
 #
 # Copyright 2005 Matt Mackall <[e-mail]>
 #
 # This software may be used and distributed according to the terms
 # of the GNU General Public License, incorporated herein by reference.
 
 from demandload import demandload
 import struct, bdiff, util, mpatch
 demandload(globals(), "re")
 
 
 def unidiff(a, ad, b, bd, fn, r=None, text=False,
             showfunc=False, ignorews=False):
 
     if not a and not b: return ""
     epoch = util.datestr((0, 0))
 
     if not text and (util.binary(a) or util.binary(b)):
         l = ['Binary file %s has changed\n' % fn]
     elif not a:
         b = b.splitlines(1)
         if a is None:
             l1 = "--- %s\t%s\n" % ("/dev/null", epoch)
         else:
             l1 = "--- %s\t%s\n" % ("a/" + fn, ad)
         l2 = "+++ %s\t%s\n" % ("b/" + fn, bd)
         l3 = "@@ -0,0 +1,%d @@\n" % len(b)
         l = [l1, l2, l3] + ["+" + e for e in b]
     elif not b:
         a = a.splitlines(1)
         l1 = "--- %s\t%s\n" % ("a/" + fn, ad)
         if b is None:
             l2 = "+++ %s\t%s\n" % ("/dev/null", epoch)
         else:
             l2 = "+++ %s\t%s\n" % ("b/" + fn, bd)
         l3 = "@@ -1,%d +0,0 @@\n" % len(a)
         l = [l1, l2, l3] + ["-" + e for e in a]
     else:
         al = a.splitlines(1)
         bl = b.splitlines(1)
         l = list(bunidiff(a, b, al, bl, "a/" + fn, "b/" + fn,
                           showfunc=showfunc, ignorews=ignorews))
         if not l: return ""
         # difflib uses a space, rather than a tab
         l[0] = "%s\t%s\n" % (l[0][:-2], ad)
         l[1] = "%s\t%s\n" % (l[1][:-2], bd)
 
     for ln in xrange(len(l)):
         if l[ln][-1] != '\n':
             l[ln] += "\n\ No newline at end of file\n"
 
     if r:
         l.insert(0, "diff %s %s\n" %
                     (' '.join(["-r %s" % rev for rev in r]), fn))
 
     return "".join(l)
 
 # somewhat self contained replacement for difflib.unified_diff
 # t1 and t2 are the text to be diffed
 # l1 and l2 are the text broken up into lines
 # header1 and header2 are the filenames for the diff output
 # context is the number of context lines
 # showfunc enables diff -p output
 # ignorews ignores all whitespace changes in the diff
 def bunidiff(t1, t2, l1, l2, header1, header2, context=3, showfunc=False,
              ignorews=False):
     def contextend(l, len):
         ret = l + context
         if ret > len:
             ret = len
         return ret
 
     def contextstart(l):
         ret = l - context
         if ret < 0:
             return 0
         return ret
 
     def yieldhunk(hunk, header):
         if header:
             for x in header:
                 yield x
         (astart, a2, bstart, b2, delta) = hunk
         aend = contextend(a2, len(l1))
         alen = aend - astart
         blen = b2 - bstart + aend - a2
 
         func = ""
         if showfunc:
             # walk backwards from the start of the context
             # to find a line starting with an alphanumeric char.
             for x in xrange(astart, -1, -1):
                 t = l1[x].rstrip()
                 if funcre.match(t):
                     func = ' ' + t[:40]
                     break
 
         yield "@@ -%d,%d +%d,%d @@%s\n" % (astart + 1, alen,
                                            bstart + 1, blen, func)
         for x in delta:
             yield x
         for x in xrange(a2, aend):
             yield ' ' + l1[x]
 
     header = [ "--- %s\t\n" % header1, "+++ %s\t\n" % header2 ]
 
     if showfunc:
         funcre = re.compile('\w')
     if ignorews:
         wsre = re.compile('[ \t]')
 
     # bdiff.blocks gives us the matching sequences in the files.  The loop
     # below finds the spaces between those matching sequences and translates
     # them into diff output.
     #
     diff = bdiff.blocks(t1, t2)
     hunk = None
     for i in xrange(len(diff)):
         # The first match is special.
         # we've either found a match starting at line 0 or a match later
         # in the file.  If it starts later, old and new below will both be
         # empty and we'll continue to the next match.
         if i > 0:
             s = diff[i-1]
         else:
             s = [0, 0, 0, 0]
         delta = []
         s1 = diff[i]
         a1 = s[1]
         a2 = s1[0]
         b1 = s[3]
         b2 = s1[2]
 
         old = l1[a1:a2]
         new = l2[b1:b2]
 
         # bdiff sometimes gives huge matches past eof, this check eats them,
         # and deals with the special first match case described above
         if not old and not new:
             continue
 
         if ignorews:
             wsold = wsre.sub('', "".join(old))
             wsnew = wsre.sub('', "".join(new))
             if wsold == wsnew:
                 continue
 
         astart = contextstart(a1)
         bstart = contextstart(b1)
         prev = None
         if hunk:
             # join with the previous hunk if it falls inside the context
             if astart < hunk[1] + context + 1:
                 prev = hunk
                 astart = hunk[1]
                 bstart = hunk[3]
             else:
                 for x in yieldhunk(hunk, header):
                     yield x
                 # we only want to yield the header if the files differ, and
                 # we only want to yield it once.
                 header = None
         if prev:
             # we've joined the previous hunk, record the new ending points.
             hunk[1] = a2
             hunk[3] = b2
             delta = hunk[4]
         else:
             # create a new hunk
             hunk = [ astart, a2, bstart, b2, delta ]
 
         delta[len(delta):] = [ ' ' + x for x in l1[astart:a1] ]
         delta[len(delta):] = [ '-' + x for x in old ]
         delta[len(delta):] = [ '+' + x for x in new ]
 
     if hunk:
         for x in yieldhunk(hunk, header):
             yield x
 
 def patchtext(bin):
     pos = 0
     t = []
     while pos < len(bin):
         p1, p2, l = struct.unpack(">lll", bin[pos:pos + 12])
         pos += 12
         t.append(bin[pos:pos + l])
         pos += l
     return "".join(t)
 
 def patch(a, bin):
     return mpatch.patches(a, [bin])
 
 patches = mpatch.patches
+patchedsize = mpatch.patchedsize
 textdiff = bdiff.bdiff

mercurial/mpatch.c

@@ -1,368 +1,404 b''
 /*
  mpatch.c - efficient binary patching for Mercurial
 
  This implements a patch algorithm that's O(m + nlog n) where m is the
  size of the output and n is the number of patches.
 
  Given a list of binary patches, it unpacks each into a hunk list,
  then combines the hunk lists with a treewise recursion to form a
  single hunk list. This hunk list is then applied to the original
  text.
 
  The text (or binary) fragments are copied directly from their source
  Python objects into a preallocated output string to avoid the
  allocation of intermediate Python objects. Working memory is about 2x
  the total number of hunks.
 
  Copyright 2005 Matt Mackall <[e-mail]>
 
  This software may be used and distributed according to the terms
  of the GNU General Public License, incorporated herein by reference.
 */
 
 #include <Python.h>
 #include <stdlib.h>
 #include <string.h>
 #ifdef _WIN32
 #ifdef _MSC_VER
 #define inline __inline
 typedef unsigned long uint32_t;
 #else
 #include <stdint.h>
 #endif
 static uint32_t ntohl(uint32_t x)
 {
 	return ((x & 0x000000ffUL) << 24) |
 		((x & 0x0000ff00UL) <<  8) |
 		((x & 0x00ff0000UL) >>  8) |
 		((x & 0xff000000UL) >> 24);
 }
 #else
 #include <sys/types.h>
 #include <arpa/inet.h>
 #endif
 
 static char mpatch_doc[] = "Efficient binary patching.";
 static PyObject *mpatch_Error;
 
 struct frag {
 	int start, end, len;
 	char *data;
 };
 
 struct flist {
 	struct frag *base, *head, *tail;
 };
 
 static struct flist *lalloc(int size)
 {
 	struct flist *a = NULL;
 
 	a = (struct flist *)malloc(sizeof(struct flist));
 	if (a) {
 		a->base = (struct frag *)malloc(sizeof(struct frag) * size);
 		if (!a->base) {
 			free(a);
 			a = NULL;
 		} else
 			a->head = a->tail = a->base;
 		return a;
 	}
 	if (!PyErr_Occurred())
 		PyErr_NoMemory();
 	return NULL;
 }
 
 static void lfree(struct flist *a)
 {
 	if (a) {
 		free(a->base);
 		free(a);
 	}
 }
 
 static int lsize(struct flist *a)
 {
 	return a->tail - a->head;
 }
 
 /* move hunks in source that are less cut to dest, compensating
    for changes in offset. the last hunk may be split if necessary.
 */
 static int gather(struct flist *dest, struct flist *src, int cut, int offset)
 {
 	struct frag *d = dest->tail, *s = src->head;
 	int postend, c, l;
 
 	while (s != src->tail) {
 		if (s->start + offset >= cut)
 			break; /* we've gone far enough */
 
 		postend = offset + s->start + s->len;
 		if (postend <= cut) {
 			/* save this hunk */
 			offset += s->start + s->len - s->end;
 			*d++ = *s++;
 		}
 		else {
 			/* break up this hunk */
 			c = cut - offset;
 			if (s->end < c)
 				c = s->end;
 			l = cut - offset - s->start;
 			if (s->len < l)
 				l = s->len;
 
 			offset += s->start + l - c;
 
 			d->start = s->start;
 			d->end = c;
 			d->len = l;
 			d->data = s->data;
 			d++;
 			s->start = c;
 			s->len = s->len - l;
 			s->data = s->data + l;
 
 			break;
 		}
 	}
 
 	dest->tail = d;
 	src->head = s;
 	return offset;
 }
 
 /* like gather, but with no output list */
 static int discard(struct flist *src, int cut, int offset)
 {
 	struct frag *s = src->head;
 	int postend, c, l;
 
 	while (s != src->tail) {
 		if (s->start + offset >= cut)
 			break;
 
 		postend = offset + s->start + s->len;
 		if (postend <= cut) {
 			offset += s->start + s->len - s->end;
 			s++;
 		}
 		else {
 			c = cut - offset;
 			if (s->end < c)
 				c = s->end;
 			l = cut - offset - s->start;
 			if (s->len < l)
 				l = s->len;
 
 			offset += s->start + l - c;
 			s->start = c;
 			s->len = s->len - l;
 			s->data = s->data + l;
 
 			break;
 		}
 	}
 
 	src->head = s;
 	return offset;
 }
 
 /* combine hunk lists a and b, while adjusting b for offset changes in a/
    this deletes a and b and returns the resultant list. */
 static struct flist *combine(struct flist *a, struct flist *b)
 {
 	struct flist *c = NULL;
 	struct frag *bh, *ct;
 	int offset = 0, post;
 
 	if (a && b)
 		c = lalloc((lsize(a) + lsize(b)) * 2);
 
 	if (c) {
 
 		for (bh = b->head; bh != b->tail; bh++) {
 			/* save old hunks */
 			offset = gather(c, a, bh->start, offset);
 
 			/* discard replaced hunks */
 			post = discard(a, bh->end, offset);
 
 			/* insert new hunk */
 			ct = c->tail;
 			ct->start = bh->start - offset;
 			ct->end = bh->end - post;
 			ct->len = bh->len;
 			ct->data = bh->data;
 			c->tail++;
 			offset = post;
 		}
 
 		/* hold on to tail from a */
 		memcpy(c->tail, a->head, sizeof(struct frag) * lsize(a));
 		c->tail += lsize(a);
 	}
 
 	lfree(a);
 	lfree(b);
 	return c;
 }
 
 /* decode a binary patch into a hunk list */
 static struct flist *decode(char *bin, int len)
 {
 	struct flist *l;
 	struct frag *lt;
 	char *end = bin + len;
 	char decode[12]; /* for dealing with alignment issues */
 
 	/* assume worst case size, we won't have many of these lists */
 	l = lalloc(len / 12);
 	if (!l)
 		return NULL;
 
 	lt = l->tail;
 
 	while (bin < end) {
 		memcpy(decode, bin, 12);
 		lt->start = ntohl(*(uint32_t *)decode);
 		lt->end = ntohl(*(uint32_t *)(decode + 4));
 		lt->len = ntohl(*(uint32_t *)(decode + 8));
 		lt->data = bin + 12;
 		bin += 12 + lt->len;
 		lt++;
 	}
 
 	if (bin != end) {
 		if (!PyErr_Occurred())
 			PyErr_SetString(mpatch_Error, "patch cannot be decoded");
 		lfree(l);
 		return NULL;
 	}
 
 	l->tail = lt;
 	return l;
 }
 
 /* calculate the size of resultant text */
 static int calcsize(int len, struct flist *l)
 {
 	int outlen = 0, last = 0;
 	struct frag *f = l->head;
 
 	while (f != l->tail) {
 		if (f->start < last || f->end > len) {
 			if (!PyErr_Occurred())
 				PyErr_SetString(mpatch_Error,
 				                "invalid patch");
 			return -1;
 		}
 		outlen += f->start - last;
 		last = f->end;
 		outlen += f->len;
 		f++;
 	}
 
 	outlen += len - last;
 	return outlen;
 }
 
 static int apply(char *buf, char *orig, int len, struct flist *l)
 {
 	struct frag *f = l->head;
 	int last = 0;
 	char *p = buf;
 
 	while (f != l->tail) {
 		if (f->start < last || f->end > len) {
 			if (!PyErr_Occurred())
 				PyErr_SetString(mpatch_Error,
 				                "invalid patch");
 			return 0;
 		}
 		memcpy(p, orig + last, f->start - last);
 		p += f->start - last;
 		memcpy(p, f->data, f->len);
 		last = f->end;
 		p += f->len;
 		f++;
 	}
 	memcpy(p, orig + last, len - last);
 	return 1;
 }
 
 /* recursively generate a patch of all bins between start and end */
 static struct flist *fold(PyObject *bins, int start, int end)
 {
 	int len;
 
 	if (start + 1 == end) {
 		/* trivial case, output a decoded list */
 		PyObject *tmp = PyList_GetItem(bins, start);
 		if (!tmp)
 			return NULL;
 		return decode(PyString_AsString(tmp), PyString_Size(tmp));
 	}
 
 	/* divide and conquer, memory management is elsewhere */
 	len = (end - start) / 2;
 	return combine(fold(bins, start, start + len),
 		       fold(bins, start + len, end));
 }
 
 static PyObject *
 patches(PyObject *self, PyObject *args)
 {
 	PyObject *text, *bins, *result;
 	struct flist *patch;
 	char *in, *out;
 	int len, outlen;
 
 	if (!PyArg_ParseTuple(args, "SO:mpatch", &text, &bins))
 		return NULL;
 
 	len = PyList_Size(bins);
 	if (!len) {
 		/* nothing to do */
 		Py_INCREF(text);
 		return text;
 	}
 
 	patch = fold(bins, 0, len);
 	if (!patch)
 		return NULL;
 
 	outlen = calcsize(PyString_Size(text), patch);
 	if (outlen < 0) {
 		result = NULL;
 		goto cleanup;
 	}
 	result = PyString_FromStringAndSize(NULL, outlen);
 	if (!result) {
 		result = NULL;
 		goto cleanup;
 	}
 	in = PyString_AsString(text);
 	out = PyString_AsString(result);
 	if (!apply(out, in, PyString_Size(text), patch)) {
 		Py_DECREF(result);
 		result = NULL;
 	}
 cleanup:
 	lfree(patch);
 	return result;
 }
 
+/* calculate size of a patched file directly */
+static PyObject *
+patchedsize(PyObject *self, PyObject *args)
+{
+	long orig, start, end, len, outlen = 0, last = 0;
+	int patchlen;
+	char *bin, *binend;
+	char decode[12]; /* for dealing with alignment issues */
+
+	if (!PyArg_ParseTuple(args, "ls#", &orig, &bin, &patchlen))
+		return NULL;
+
+	binend = bin + patchlen;
+
+	while (bin < binend) {
+		memcpy(decode, bin, 12);
+		start = ntohl(*(uint32_t *)decode);
+		end = ntohl(*(uint32_t *)(decode + 4));
+		len = ntohl(*(uint32_t *)(decode + 8));
+		bin += 12 + len;
+		outlen += start - last;
+		last = end;
+		outlen += len;
+	}
+
+	if (bin != binend) {
+		if (!PyErr_Occurred())
+			PyErr_SetString(mpatch_Error, "patch cannot be decoded");
+		return NULL;
+	}
+
+	outlen += orig - last;
+	return Py_BuildValue("l", outlen);
+}
+
 static PyMethodDef methods[] = {
 	{"patches", patches, METH_VARARGS, "apply a series of patches\n"},
+	{"patchedsize", patchedsize, METH_VARARGS, "calculed patched size\n"},
 	{NULL, NULL}
 };
 
 PyMODINIT_FUNC
 initmpatch(void)
 {
 	Py_InitModule3("mpatch", methods, mpatch_doc);
 	mpatch_Error = PyErr_NewException("mpatch.mpatchError", NULL, NULL);
 }
 

mercurial/revlog.py

@@ -1,1068 +1,1101 b''
 """
 revlog.py - storage back-end for mercurial
 
 This provides efficient delta storage with O(1) retrieve and append
 and O(changes) merge between branches
 
 Copyright 2005 Matt Mackall <[e-mail]>
 
 This software may be used and distributed according to the terms
 of the GNU General Public License, incorporated herein by reference.
 """
 
 from node import *
 from i18n import gettext as _
 from demandload import demandload
 demandload(globals(), "binascii changegroup errno heapq mdiff os")
 demandload(globals(), "sha struct zlib")
 
 # revlog version strings
 REVLOGV0 = 0
 REVLOGNG = 1
 
 # revlog flags
 REVLOGNGINLINEDATA = (1 << 16)
 
 def flagstr(flag):
     if flag == "inline":
         return REVLOGNGINLINEDATA
     raise RevlogError(_("unknown revlog flag %s" % flag))
 
 def hash(text, p1, p2):
     """generate a hash from the given text and its parent hashes
 
     This hash combines both the current file contents and its history
     in a manner that makes it easy to distinguish nodes with the same
     content in the revision graph.
     """
     l = [p1, p2]
     l.sort()
     s = sha.new(l[0])
     s.update(l[1])
     s.update(text)
     return s.digest()
 
 def compress(text):
     """ generate a possibly-compressed representation of text """
     if not text: return ("", text)
     if len(text) < 44:
         if text[0] == '\0': return ("", text)
         return ('u', text)
     bin = zlib.compress(text)
     if len(bin) > len(text):
         if text[0] == '\0': return ("", text)
         return ('u', text)
     return ("", bin)
 
 def decompress(bin):
     """ decompress the given input """
     if not bin: return bin
     t = bin[0]
     if t == '\0': return bin
     if t == 'x': return zlib.decompress(bin)
     if t == 'u': return bin[1:]
     raise RevlogError(_("unknown compression type %r") % t)
 
 indexformatv0 = ">4l20s20s20s"
 # index ng:
 # 6 bytes offset
 # 2 bytes flags
 # 4 bytes compressed length
 # 4 bytes uncompressed length
 # 4 bytes: base rev
 # 4 bytes link rev
 # 4 bytes parent 1 rev
 # 4 bytes parent 2 rev
 # 32 bytes: nodeid
 indexformatng = ">Qiiiiii20s12x"
 versionformat = ">i"
 
 class lazyparser(object):
     """
     this class avoids the need to parse the entirety of large indices
 
     By default we parse and load 1000 entries at a time.
 
     If no position is specified, we load the whole index, and replace
     the lazy objects in revlog with the underlying objects for
     efficiency in cases where we look at most of the nodes.
     """
     def __init__(self, data, revlog, indexformat):
         self.data = data
         self.s = struct.calcsize(indexformat)
         self.indexformat = indexformat
         self.l = len(data)/self.s
         self.index = [None] * self.l
         self.map = {nullid: -1}
         self.all = 0
         self.revlog = revlog
 
     def load(self, pos=None):
         if self.all: return
         if pos is not None:
             block = pos / 1000
             i = block * 1000
             end = min(self.l, i + 1000)
         else:
             self.all = 1
             i = 0
             end = self.l
             self.revlog.index = self.index
             self.revlog.nodemap = self.map
 
         while i < end:
             if not self.index[i]:
                 d = self.data[i * self.s: (i + 1) * self.s]
                 e = struct.unpack(self.indexformat, d)
                 self.index[i] = e
                 self.map[e[-1]] = i
             i += 1
 
 class lazyindex(object):
     """a lazy version of the index array"""
     def __init__(self, parser):
         self.p = parser
     def __len__(self):
         return len(self.p.index)
     def load(self, pos):
         if pos < 0:
             pos += len(self.p.index)
         self.p.load(pos)
         return self.p.index[pos]
     def __getitem__(self, pos):
         return self.p.index[pos] or self.load(pos)
     def __setitem__(self, pos, item):
         self.p.index[pos] = item
     def __delitem__(self, pos):
         del self.p.index[pos]
     def append(self, e):
         self.p.index.append(e)
 
 class lazymap(object):
     """a lazy version of the node map"""
     def __init__(self, parser):
         self.p = parser
     def load(self, key):
         if self.p.all: return
         n = self.p.data.find(key)
         if n < 0:
             raise KeyError(key)
         pos = n / self.p.s
         self.p.load(pos)
     def __contains__(self, key):
         self.p.load()
         return key in self.p.map
     def __iter__(self):
         yield nullid
         for i in xrange(self.p.l):
             try:
                 yield self.p.index[i][-1]
             except:
                 self.p.load(i)
                 yield self.p.index[i][-1]
     def __getitem__(self, key):
         try:
             return self.p.map[key]
         except KeyError:
             try:
                 self.load(key)
                 return self.p.map[key]
             except KeyError:
                 raise KeyError("node " + hex(key))
     def __setitem__(self, key, val):
         self.p.map[key] = val
     def __delitem__(self, key):
         del self.p.map[key]
 
 class RevlogError(Exception): pass
 
 class revlog(object):
     """
     the underlying revision storage object
 
     A revlog consists of two parts, an index and the revision data.
 
     The index is a file with a fixed record size containing
     information on each revision, includings its nodeid (hash), the
     nodeids of its parents, the position and offset of its data within
     the data file, and the revision it's based on. Finally, each entry
     contains a linkrev entry that can serve as a pointer to external
     data.
 
     The revision data itself is a linear collection of data chunks.
     Each chunk represents a revision and is usually represented as a
     delta against the previous chunk. To bound lookup time, runs of
     deltas are limited to about 2 times the length of the original
     version data. This makes retrieval of a version proportional to
     its size, or O(1) relative to the number of revisions.
 
     Both pieces of the revlog are written to in an append-only
     fashion, which means we never need to rewrite a file to insert or
     remove data, and can use some simple techniques to avoid the need
     for locking while reading.
     """
     def __init__(self, opener, indexfile, datafile, defversion=0):
         """
         create a revlog object
 
         opener is a function that abstracts the file opening operation
         and can be used to implement COW semantics or the like.
         """
         self.indexfile = indexfile
         self.datafile = datafile
         self.opener = opener
 
         self.indexstat = None
         self.cache = None
         self.chunkcache = None
         self.defversion = defversion
         self.load()
 
     def load(self):
         v = self.defversion
         try:
             f = self.opener(self.indexfile)
             i = f.read()
         except IOError, inst:
             if inst.errno != errno.ENOENT:
                 raise
             i = ""
         else:
             try:
                 st = os.fstat(f.fileno())
             except AttributeError, inst:
                 st = None
             else:
                 oldst = self.indexstat
                 if (oldst and st.st_dev == oldst.st_dev
                     and st.st_ino == oldst.st_ino
                     and st.st_mtime == oldst.st_mtime
                     and st.st_ctime == oldst.st_ctime):
                     return
                 self.indexstat = st
                 if len(i) > 0:
                     v = struct.unpack(versionformat, i[:4])[0]
         flags = v & ~0xFFFF
         fmt = v & 0xFFFF
         if fmt == 0:
             if flags:
                 raise RevlogError(_("index %s invalid flags %x for format v0" %
                                    (self.indexfile, flags)))
         elif fmt == REVLOGNG:
             if flags & ~REVLOGNGINLINEDATA:
                 raise RevlogError(_("index %s invalid flags %x for revlogng" %
                                    (self.indexfile, flags)))
         else:
             raise RevlogError(_("index %s invalid format %d" %
                                (self.indexfile, fmt)))
         self.version = v
         if v == 0:
             self.indexformat = indexformatv0
         else:
             self.indexformat = indexformatng
 
         if i:
             if not self.inlinedata() and st and st.st_size > 10000:
                 # big index, let's parse it on demand
                 parser = lazyparser(i, self, self.indexformat)
                 self.index = lazyindex(parser)
                 self.nodemap = lazymap(parser)
             else:
                 self.parseindex(i)
             if self.inlinedata():
                 # we've already got the entire data file read in, save it
                 # in the chunk data
                 self.chunkcache = (0, i)
             if self.version != 0:
                 e = list(self.index[0])
                 type = self.ngtype(e[0])
                 e[0] = self.offset_type(0, type)
                 self.index[0] = e
         else:
             self.nodemap = { nullid: -1}
             self.index = []
 
 
     def parseindex(self, data):
         s = struct.calcsize(self.indexformat)
         l = len(data)
         self.index = []
         self.nodemap =  {nullid: -1}
         inline = self.inlinedata()
         off = 0
         n = 0
         while off < l:
             e = struct.unpack(self.indexformat, data[off:off + s])
             self.index.append(e)
             self.nodemap[e[-1]] = n
             n += 1
             off += s
             if inline:
                 off += e[1]
 
     def ngoffset(self, q):
         if q & 0xFFFF:
             raise RevlogError(_('%s: incompatible revision flag %x') %
                                (self.indexfile, type))
         return long(q >> 16)
 
     def ngtype(self, q):
         return int(q & 0xFFFF)
 
     def offset_type(self, offset, type):
         return long(long(offset) << 16 | type)
 
     def loadindexmap(self):
         """loads both the map and the index from the lazy parser"""
         if isinstance(self.index, lazyindex):
             p = self.index.p
             p.load()
 
     def inlinedata(self): return self.version & REVLOGNGINLINEDATA
     def tip(self): return self.node(len(self.index) - 1)
     def count(self): return len(self.index)
     def node(self, rev):
         return (rev < 0) and nullid or self.index[rev][-1]
     def rev(self, node):
         try:
             return self.nodemap[node]
         except KeyError:
             raise RevlogError(_('%s: no node %s') % (self.indexfile, hex(node)))
     def linkrev(self, node): return self.index[self.rev(node)][-4]
     def parents(self, node):
         if node == nullid: return (nullid, nullid)
         r = self.rev(node)
         d = self.index[r][-3:-1]
         if self.version == 0:
             return d
         return [ self.node(x) for x in d ]
     def start(self, rev):
         if rev < 0:
             return -1
         if self.version != 0:
             return self.ngoffset(self.index[rev][0])
         return self.index[rev][0]
+
     def end(self, rev): return self.start(rev) + self.length(rev)
 
+    def size(self, rev):
+        """return the length of the uncompressed text for a given revision"""
+        l = -1
+        if self.version != 0:
+            l = self.index[rev][2]
+        if l >= 0:
+            return l
+
+        t = self.revision(self.node(rev))
+        return len(t)
+
+        # alternate implementation, The advantage to this code is it
+        # will be faster for a single revision.  But, the results are not
+        # cached, so finding the size of every revision will be slower.
+        """
+        if self.cache and self.cache[1] == rev:
+            return len(self.cache[2])
+
+        base = self.base(rev)
+        if self.cache and self.cache[1] >= base and self.cache[1] < rev:
+            base = self.cache[1]
+            text = self.cache[2]
+        else:
+            text = self.revision(self.node(base))
+
+        l = len(text)
+        for x in xrange(base + 1, rev + 1):
+            l = mdiff.patchedsize(l, self.chunk(x))
+        return l
+        """
+
     def length(self, rev):
         if rev < 0:
             return 0
         else:
             return self.index[rev][1]
     def base(self, rev): return (rev < 0) and rev or self.index[rev][-5]
 
     def reachable(self, rev, stop=None):
         reachable = {}
         visit = [rev]
         reachable[rev] = 1
         if stop:
             stopn = self.rev(stop)
         else:
             stopn = 0
         while visit:
             n = visit.pop(0)
             if n == stop:
                 continue
             if n == nullid:
                 continue
             for p in self.parents(n):
                 if self.rev(p) < stopn:
                     continue
                 if p not in reachable:
                     reachable[p] = 1
                     visit.append(p)
         return reachable
 
     def nodesbetween(self, roots=None, heads=None):
         """Return a tuple containing three elements. Elements 1 and 2 contain
         a final list bases and heads after all the unreachable ones have been
         pruned.  Element 0 contains a topologically sorted list of all
 
         nodes that satisfy these constraints:
         1. All nodes must be descended from a node in roots (the nodes on
            roots are considered descended from themselves).
         2. All nodes must also be ancestors of a node in heads (the nodes in
            heads are considered to be their own ancestors).
 
         If roots is unspecified, nullid is assumed as the only root.
         If heads is unspecified, it is taken to be the output of the
         heads method (i.e. a list of all nodes in the repository that
         have no children)."""
         nonodes = ([], [], [])
         if roots is not None:
             roots = list(roots)
             if not roots:
                 return nonodes
             lowestrev = min([self.rev(n) for n in roots])
         else:
             roots = [nullid] # Everybody's a descendent of nullid
             lowestrev = -1
         if (lowestrev == -1) and (heads is None):
             # We want _all_ the nodes!
             return ([self.node(r) for r in xrange(0, self.count())],
                     [nullid], list(self.heads()))
         if heads is None:
             # All nodes are ancestors, so the latest ancestor is the last
             # node.
             highestrev = self.count() - 1
             # Set ancestors to None to signal that every node is an ancestor.
             ancestors = None
             # Set heads to an empty dictionary for later discovery of heads
             heads = {}
         else:
             heads = list(heads)
             if not heads:
                 return nonodes
             ancestors = {}
             # Start at the top and keep marking parents until we're done.
             nodestotag = heads[:]
             # Turn heads into a dictionary so we can remove 'fake' heads.
             # Also, later we will be using it to filter out the heads we can't
             # find from roots.
             heads = dict.fromkeys(heads, 0)
             # Remember where the top was so we can use it as a limit later.
             highestrev = max([self.rev(n) for n in nodestotag])
             while nodestotag:
                 # grab a node to tag
                 n = nodestotag.pop()
                 # Never tag nullid
                 if n == nullid:
                     continue
                 # A node's revision number represents its place in a
                 # topologically sorted list of nodes.
                 r = self.rev(n)
                 if r >= lowestrev:
                     if n not in ancestors:
                         # If we are possibly a descendent of one of the roots
                         # and we haven't already been marked as an ancestor
                         ancestors[n] = 1 # Mark as ancestor
                         # Add non-nullid parents to list of nodes to tag.
                         nodestotag.extend([p for p in self.parents(n) if
                                            p != nullid])
                     elif n in heads: # We've seen it before, is it a fake head?
                         # So it is, real heads should not be the ancestors of
                         # any other heads.
                         heads.pop(n)
             if not ancestors:
                 return nonodes
             # Now that we have our set of ancestors, we want to remove any
             # roots that are not ancestors.
 
             # If one of the roots was nullid, everything is included anyway.
             if lowestrev > -1:
                 # But, since we weren't, let's recompute the lowest rev to not
                 # include roots that aren't ancestors.
 
                 # Filter out roots that aren't ancestors of heads
                 roots = [n for n in roots if n in ancestors]
                 # Recompute the lowest revision
                 if roots:
                     lowestrev = min([self.rev(n) for n in roots])
                 else:
                     # No more roots?  Return empty list
                     return nonodes
             else:
                 # We are descending from nullid, and don't need to care about
                 # any other roots.
                 lowestrev = -1
                 roots = [nullid]
         # Transform our roots list into a 'set' (i.e. a dictionary where the
         # values don't matter.
         descendents = dict.fromkeys(roots, 1)
         # Also, keep the original roots so we can filter out roots that aren't
         # 'real' roots (i.e. are descended from other roots).
         roots = descendents.copy()
         # Our topologically sorted list of output nodes.
         orderedout = []
         # Don't start at nullid since we don't want nullid in our output list,
         # and if nullid shows up in descedents, empty parents will look like
         # they're descendents.
         for r in xrange(max(lowestrev, 0), highestrev + 1):
             n = self.node(r)
             isdescendent = False
             if lowestrev == -1:  # Everybody is a descendent of nullid
                 isdescendent = True
             elif n in descendents:
                 # n is already a descendent
                 isdescendent = True
                 # This check only needs to be done here because all the roots
                 # will start being marked is descendents before the loop.
                 if n in roots:
                     # If n was a root, check if it's a 'real' root.
                     p = tuple(self.parents(n))
                     # If any of its parents are descendents, it's not a root.
                     if (p[0] in descendents) or (p[1] in descendents):
                         roots.pop(n)
             else:
                 p = tuple(self.parents(n))
                 # A node is a descendent if either of its parents are
                 # descendents.  (We seeded the dependents list with the roots
                 # up there, remember?)
                 if (p[0] in descendents) or (p[1] in descendents):
                     descendents[n] = 1
                     isdescendent = True
             if isdescendent and ((ancestors is None) or (n in ancestors)):
                 # Only include nodes that are both descendents and ancestors.
                 orderedout.append(n)
                 if (ancestors is not None) and (n in heads):
                     # We're trying to figure out which heads are reachable
                     # from roots.
                     # Mark this head as having been reached
                     heads[n] = 1
                 elif ancestors is None:
                     # Otherwise, we're trying to discover the heads.
                     # Assume this is a head because if it isn't, the next step
                     # will eventually remove it.
                     heads[n] = 1
                     # But, obviously its parents aren't.
                     for p in self.parents(n):
                         heads.pop(p, None)
         heads = [n for n in heads.iterkeys() if heads[n] != 0]
         roots = roots.keys()
         assert orderedout
         assert roots
         assert heads
         return (orderedout, roots, heads)
 
     def heads(self, start=None):
         """return the list of all nodes that have no children
 
         if start is specified, only heads that are descendants of
         start will be returned
 
         """
         if start is None:
             start = nullid
         reachable = {start: 1}
         heads = {start: 1}
         startrev = self.rev(start)
 
         for r in xrange(startrev + 1, self.count()):
             n = self.node(r)
             for pn in self.parents(n):
                 if pn in reachable:
                     reachable[n] = 1
                     heads[n] = 1
                 if pn in heads:
                     del heads[pn]
         return heads.keys()
 
     def children(self, node):
         """find the children of a given node"""
         c = []
         p = self.rev(node)
         for r in range(p + 1, self.count()):
             n = self.node(r)
             for pn in self.parents(n):
                 if pn == node:
                     c.append(n)
                     continue
                 elif pn == nullid:
                     continue
         return c
 
     def lookup(self, id):
         """locate a node based on revision number or subset of hex nodeid"""
         try:
             rev = int(id)
             if str(rev) != id: raise ValueError
             if rev < 0: rev = self.count() + rev
             if rev < 0 or rev >= self.count(): raise ValueError
             return self.node(rev)
         except (ValueError, OverflowError):
             c = []
             for n in self.nodemap:
                 if hex(n).startswith(id):
                     c.append(n)
             if len(c) > 1: raise RevlogError(_("Ambiguous identifier"))
             if len(c) < 1: raise RevlogError(_("No match found"))
             return c[0]
 
         return None
 
     def diff(self, a, b):
         """return a delta between two revisions"""
         return mdiff.textdiff(a, b)
 
     def patches(self, t, pl):
         """apply a list of patches to a string"""
         return mdiff.patches(t, pl)
 
     def chunk(self, rev, df=None, cachelen=4096):
         start, length = self.start(rev), self.length(rev)
         inline = self.inlinedata()
         if inline:
             start += (rev + 1) * struct.calcsize(self.indexformat)
         end = start + length
         def loadcache(df):
             cache_length = max(cachelen, length) # 4k
             if not df:
                 if inline:
                     df = self.opener(self.indexfile)
                 else:
                     df = self.opener(self.datafile)
             df.seek(start)
             self.chunkcache = (start, df.read(cache_length))
 
         if not self.chunkcache:
             loadcache(df)
 
         cache_start = self.chunkcache[0]
         cache_end = cache_start + len(self.chunkcache[1])
         if start >= cache_start and end <= cache_end:
             # it is cached
             offset = start - cache_start
         else:
             loadcache(df)
             offset = 0
 
         #def checkchunk():
         #    df = self.opener(self.datafile)
         #    df.seek(start)
         #    return df.read(length)
         #assert s == checkchunk()
         return decompress(self.chunkcache[1][offset:offset + length])
 
     def delta(self, node):
         """return or calculate a delta between a node and its predecessor"""
         r = self.rev(node)
         return self.revdiff(r - 1, r)
 
     def revdiff(self, rev1, rev2):
         """return or calculate a delta between two revisions"""
         b1 = self.base(rev1)
         b2 = self.base(rev2)
         if b1 == b2 and rev1 + 1 == rev2:
             return self.chunk(rev2)
         else:
             return self.diff(self.revision(self.node(rev1)),
                              self.revision(self.node(rev2)))
 
     def revision(self, node):
         """return an uncompressed revision of a given"""
         if node == nullid: return ""
         if self.cache and self.cache[0] == node: return self.cache[2]
 
         # look up what we need to read
         text = None
         rev = self.rev(node)
         base = self.base(rev)
 
         if self.inlinedata():
             # we probably have the whole chunk cached
             df = None
         else:
             df = self.opener(self.datafile)
 
         # do we have useful data cached?
         if self.cache and self.cache[1] >= base and self.cache[1] < rev:
             base = self.cache[1]
             text = self.cache[2]
         else:
             text = self.chunk(base, df=df)
 
         bins = []
         for r in xrange(base + 1, rev + 1):
             bins.append(self.chunk(r, df=df))
 
         text = self.patches(text, bins)
 
         p1, p2 = self.parents(node)
         if node != hash(text, p1, p2):
             raise RevlogError(_("integrity check failed on %s:%d")
                           % (self.datafile, rev))
 
         self.cache = (node, rev, text)
         return text
 
     def checkinlinesize(self, tr, fp=None):
         if not self.inlinedata():
             return
         if not fp:
             fp = self.opener(self.indexfile, 'r')
         size = fp.tell()
         if size < 131072:
             return
         tr.add(self.datafile, 0)
         df = self.opener(self.datafile, 'w')
         calc = struct.calcsize(self.indexformat)
         for r in xrange(self.count()):
             start = self.start(r) + (r + 1) * calc
             length = self.length(r)
             fp.seek(start)
             d = fp.read(length)
             df.write(d)
         fp.close()
         df.close()
         fp = self.opener(self.indexfile, 'w', atomictemp=True)
         self.version &= ~(REVLOGNGINLINEDATA)
         if self.count():
             x = self.index[0]
             e = struct.pack(self.indexformat, *x)[4:]
             l = struct.pack(versionformat, self.version)
             fp.write(l)
             fp.write(e)
 
         for i in xrange(1, self.count()):
             x = self.index[i]
             e = struct.pack(self.indexformat, *x)
             fp.write(e)
 
         # if we don't call rename, the temp file will never replace the
         # real index
         fp.rename()
         self.chunkcache = None
 
     def addrevision(self, text, transaction, link, p1=None, p2=None, d=None):
         """add a revision to the log
 
         text - the revision data to add
         transaction - the transaction object used for rollback
         link - the linkrev data to add
         p1, p2 - the parent nodeids of the revision
         d - an optional precomputed delta
         """
         if text is None: text = ""
         if p1 is None: p1 = self.tip()
         if p2 is None: p2 = nullid
 
         node = hash(text, p1, p2)
 
         if node in self.nodemap:
             return node
 
         n = self.count()
         t = n - 1
 
         if n:
             base = self.base(t)
             start = self.start(base)
             end = self.end(t)
             if not d:
                 prev = self.revision(self.tip())
                 d = self.diff(prev, str(text))
             data = compress(d)
             l = len(data[1]) + len(data[0])
             dist = end - start + l
 
         # full versions are inserted when the needed deltas
         # become comparable to the uncompressed text
         if not n or dist > len(text) * 2:
             data = compress(text)
             l = len(data[1]) + len(data[0])
             base = n
         else:
             base = self.base(t)
 
         offset = 0
         if t >= 0:
             offset = self.end(t)
 
         if self.version == 0:
             e = (offset, l, base, link, p1, p2, node)
         else:
             e = (self.offset_type(offset, 0), l, len(text),
                  base, link, self.rev(p1), self.rev(p2), node)
 
         self.index.append(e)
         self.nodemap[node] = n
         entry = struct.pack(self.indexformat, *e)
 
         if not self.inlinedata():
             transaction.add(self.datafile, offset)
             transaction.add(self.indexfile, n * len(entry))
             f = self.opener(self.datafile, "a")
             if data[0]:
                 f.write(data[0])
             f.write(data[1])
             f = self.opener(self.indexfile, "a")
         else:
             f = self.opener(self.indexfile, "a+")
             f.seek(0, 2)
             transaction.add(self.indexfile, f.tell())
 
         if len(self.index) == 1 and self.version != 0:
             l = struct.pack(versionformat, self.version)
             f.write(l)
             entry = entry[4:]
 
         f.write(entry)
 
         if self.inlinedata():
             f.write(data[0])
             f.write(data[1])
             self.checkinlinesize(transaction, f)
 
         self.cache = (node, n, text)
         return node
 
     def ancestor(self, a, b):
         """calculate the least common ancestor of nodes a and b"""
         # calculate the distance of every node from root
         dist = {nullid: 0}
         for i in xrange(self.count()):
             n = self.node(i)
             p1, p2 = self.parents(n)
             dist[n] = max(dist[p1], dist[p2]) + 1
 
         # traverse ancestors in order of decreasing distance from root
         def ancestors(node):
             # we store negative distances because heap returns smallest member
             h = [(-dist[node], node)]
             seen = {}
             while h:
                 d, n = heapq.heappop(h)
                 if n not in seen:
                     seen[n] = 1
                     yield (-d, n)
                     for p in self.parents(n):
                         heapq.heappush(h, (-dist[p], p))
 
         def generations(node):
             sg, s = None, {}
             for g,n in ancestors(node):
                 if g != sg:
                     if sg:
                         yield sg, s
                     sg, s = g, {n:1}
                 else:
                     s[n] = 1
             yield sg, s
 
         x = generations(a)
         y = generations(b)
         gx = x.next()
         gy = y.next()
 
         # increment each ancestor list until it is closer to root than
         # the other, or they match
         while 1:
             #print "ancestor gen %s %s" % (gx[0], gy[0])
             if gx[0] == gy[0]:
                 # find the intersection
                 i = [ n for n in gx[1] if n in gy[1] ]
                 if i:
                     return i[0]
                 else:
                     #print "next"
                     gy = y.next()
                     gx = x.next()
             elif gx[0] < gy[0]:
                 #print "next y"
                 gy = y.next()
             else:
                 #print "next x"
                 gx = x.next()
 
     def group(self, nodelist, lookup, infocollect=None):
         """calculate a delta group
 
         Given a list of changeset revs, return a set of deltas and
         metadata corresponding to nodes. the first delta is
         parent(nodes[0]) -> nodes[0] the receiver is guaranteed to
         have this parent as it has all history before these
         changesets. parent is parent[0]
         """
         revs = [self.rev(n) for n in nodelist]
 
         # if we don't have any revisions touched by these changesets, bail
         if not revs:
             yield changegroup.closechunk()
             return
 
         # add the parent of the first rev
         p = self.parents(self.node(revs[0]))[0]
         revs.insert(0, self.rev(p))
 
         # build deltas
         for d in xrange(0, len(revs) - 1):
             a, b = revs[d], revs[d + 1]
             nb = self.node(b)
 
             if infocollect is not None:
                 infocollect(nb)
 
             d = self.revdiff(a, b)
             p = self.parents(nb)
             meta = nb + p[0] + p[1] + lookup(nb)
             yield changegroup.genchunk("%s%s" % (meta, d))
 
         yield changegroup.closechunk()
 
     def addgroup(self, revs, linkmapper, transaction, unique=0):
         """
         add a delta group
 
         given a set of deltas, add them to the revision log. the
         first delta is against its parent, which should be in our
         log, the rest are against the previous delta.
         """
 
         #track the base of the current delta log
         r = self.count()
         t = r - 1
         node = None
 
         base = prev = -1
-        start = end = measure = 0
+        start = end = textlen = 0
         if r:
             end = self.end(t)
 
         ifh = self.opener(self.indexfile, "a+")
         ifh.seek(0, 2)
         transaction.add(self.indexfile, ifh.tell())
         if self.inlinedata():
             dfh = None
         else:
             transaction.add(self.datafile, end)
             dfh = self.opener(self.datafile, "a")
 
         # loop through our set of deltas
         chain = None
         for chunk in revs:
             node, p1, p2, cs = struct.unpack("20s20s20s20s", chunk[:80])
             link = linkmapper(cs)
             if node in self.nodemap:
                 # this can happen if two branches make the same change
                 # if unique:
                 #    raise RevlogError(_("already have %s") % hex(node[:4]))
                 chain = node
                 continue
             delta = chunk[80:]
 
             for p in (p1, p2):
                 if not p in self.nodemap:
                     raise RevlogError(_("unknown parent %s") % short(p1))
 
             if not chain:
                 # retrieve the parent revision of the delta chain
                 chain = p1
                 if not chain in self.nodemap:
                     raise RevlogError(_("unknown base %s") % short(chain[:4]))
 
             # full versions are inserted when the needed deltas become
             # comparable to the uncompressed text or when the previous
             # version is not the one we have a delta against. We use
             # the size of the previous full rev as a proxy for the
             # current size.
 
             if chain == prev:
                 tempd = compress(delta)
                 cdelta = tempd[0] + tempd[1]
+                textlen = mdiff.patchedsize(textlen, delta)
 
-            if chain != prev or (end - start + len(cdelta)) > measure * 2:
+            if chain != prev or (end - start + len(cdelta)) > textlen * 2:
                 # flush our writes here so we can read it in revision
                 if dfh:
                     dfh.flush()
                 ifh.flush()
                 text = self.revision(chain)
                 text = self.patches(text, [delta])
                 chk = self.addrevision(text, transaction, link, p1, p2)
                 if chk != node:
                     raise RevlogError(_("consistency error adding group"))
-                measure = len(text)
+                textlen = len(text)
             else:
                 if self.version == 0:
                     e = (end, len(cdelta), base, link, p1, p2, node)
                 else:
-                    e = (self.offset_type(end, 0), len(cdelta), -1, base,
+                    e = (self.offset_type(end, 0), len(cdelta), textlen, base,
                          link, self.rev(p1), self.rev(p2), node)
                 self.index.append(e)
                 self.nodemap[node] = r
                 if self.inlinedata():
                     ifh.write(struct.pack(self.indexformat, *e))
                     ifh.write(cdelta)
                     self.checkinlinesize(transaction, ifh)
                     if not self.inlinedata():
                         dfh = self.opener(self.datafile, "a")
                         ifh = self.opener(self.indexfile, "a")
                 else:
                     if not dfh:
                         # addrevision switched from inline to conventional
                         # reopen the index
                         dfh = self.opener(self.datafile, "a")
                         ifh = self.opener(self.indexfile, "a")
                     dfh.write(cdelta)
                     ifh.write(struct.pack(self.indexformat, *e))
 
             t, r, chain, prev = r, r + 1, node, node
             base = self.base(t)
             start = self.start(base)
             end = self.end(t)
 
         if node is None:
             raise RevlogError(_("group to be added is empty"))
         return node
 
     def strip(self, rev, minlink):
         if self.count() == 0 or rev >= self.count():
             return
 
         if isinstance(self.index, lazyindex):
             self.loadindexmap()
 
         # When stripping away a revision, we need to make sure it
         # does not actually belong to an older changeset.
         # The minlink parameter defines the oldest revision
         # we're allowed to strip away.
         while minlink > self.index[rev][-4]:
             rev += 1
             if rev >= self.count():
                 return
 
         # first truncate the files on disk
         end = self.start(rev)
         if not self.inlinedata():
             df = self.opener(self.datafile, "a")
             df.truncate(end)
             end = rev * struct.calcsize(self.indexformat)
         else:
             end += rev * struct.calcsize(self.indexformat)
 
         indexf = self.opener(self.indexfile, "a")
         indexf.truncate(end)
 
         # then reset internal state in memory to forget those revisions
         self.cache = None
         self.chunkcache = None
         for x in xrange(rev, self.count()):
             del self.nodemap[self.node(x)]
 
         del self.index[rev:]
 
     def checksize(self):
         expected = 0
         if self.count():
             expected = self.end(self.count() - 1)
 
         try:
             f = self.opener(self.datafile)
             f.seek(0, 2)
             actual = f.tell()
             dd = actual - expected
         except IOError, inst:
             if inst.errno != errno.ENOENT:
                 raise
             dd = 0
 
         try:
             f = self.opener(self.indexfile)
             f.seek(0, 2)
             actual = f.tell()
             s = struct.calcsize(self.indexformat)
             i = actual / s
             di = actual - (i * s)
             if self.inlinedata():
                 databytes = 0
                 for r in xrange(self.count()):
                     databytes += self.length(r)
                 dd = 0
                 di = actual - self.count() * s - databytes
         except IOError, inst:
             if inst.errno != errno.ENOENT:
                 raise
             di = 0
 
         return (dd, di)