"""

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 2005-2007 Matt Mackall <mpm@selenic.com>

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 _

import binascii, changegroup, errno, ancestor, mdiff, os

import sha, struct, util, zlib

# revlog version strings

REVLOGV0 = 0

REVLOGNG = 1

# revlog flags

REVLOGNGINLINEDATA = (1 << 16)

REVLOG_DEFAULT_FLAGS = REVLOGNGINLINEDATA

REVLOG_DEFAULT_FORMAT = REVLOGNG

REVLOG_DEFAULT_VERSION = REVLOG_DEFAULT_FORMAT | REVLOG_DEFAULT_FLAGS

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"

v0shaoffset = 56

# 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"

ngshaoffset = 32

versionformat = ">I"

class lazyparser(object):

    """

    this class avoids the need to parse the entirety of large indices

    """

    # lazyparser is not safe to use on windows if win32 extensions not

    # available. it keeps file handle open, which make it not possible

    # to break hardlinks on local cloned repos.

    safe_to_use = os.name != 'nt' or (not util.is_win_9x() and

                                      hasattr(util, 'win32api'))

    def __init__(self, dataf, size, indexformat, shaoffset):

        self.dataf = dataf

        self.format = indexformat

        self.s = struct.calcsize(indexformat)

        self.indexformat = indexformat

        self.datasize = size

        self.l = size/self.s

        self.index = [None] * self.l

        self.map = {nullid: nullrev}

        self.allmap = 0

        self.all = 0

        self.mapfind_count = 0

        self.shaoffset = shaoffset

    def loadmap(self):

        """

        during a commit, we need to make sure the rev being added is

        not a duplicate.  This requires loading the entire index,

        which is fairly slow.  loadmap can load up just the node map,

        which takes much less time.

        """

        if self.allmap: return

        end = self.datasize

        self.allmap = 1

        cur = 0

        count = 0

        blocksize = self.s * 256

        self.dataf.seek(0)

        while cur < end:

            data = self.dataf.read(blocksize)

            off = 0

            for x in xrange(256):

                n = data[off + self.shaoffset:off + self.shaoffset + 20]

                self.map[n] = count

                count += 1

                if count >= self.l:

                    break

                off += self.s

            cur += blocksize

    def loadblock(self, blockstart, blocksize, data=None):

        if self.all: return

        if data is None:

            self.dataf.seek(blockstart)

            if blockstart + blocksize > self.datasize:

                # the revlog may have grown since we've started running,

                # but we don't have space in self.index for more entries.

                # limit blocksize so that we don't get too much data.

                blocksize = max(self.datasize - blockstart, 0)

            data = self.dataf.read(blocksize)

        lend = len(data) / self.s

        i = blockstart / self.s

        off = 0

        # lazyindex supports __delitem__

        if lend > len(self.index) - i:

            lend = len(self.index) - i

        for x in xrange(lend):

            if self.index[i + x] == None:

                b = data[off : off + self.s]

                self.index[i + x] = b

                n = b[self.shaoffset:self.shaoffset + 20]

                self.map[n] = i + x

            off += self.s

    def findnode(self, node):

        """search backwards through the index file for a specific node"""

        if self.allmap: return None

        # hg log will cause many many searches for the manifest

        # nodes.  After we get called a few times, just load the whole

        # thing.

        if self.mapfind_count > 8:

            self.loadmap()

            if node in self.map:

                return node

            return None

        self.mapfind_count += 1

        last = self.l - 1

        while self.index[last] != None:

            if last == 0:

                self.all = 1

                self.allmap = 1

                return None

            last -= 1

        end = (last + 1) * self.s

        blocksize = self.s * 256

        while end >= 0:

            start = max(end - blocksize, 0)

            self.dataf.seek(start)

            data = self.dataf.read(end - start)

            findend = end - start

            while True:

                # we're searching backwards, so weh have to make sure

                # we don't find a changeset where this node is a parent

                off = data.rfind(node, 0, findend)

                findend = off

                if off >= 0:

                    i = off / self.s

                    off = i * self.s

                    n = data[off + self.shaoffset:off + self.shaoffset + 20]

                    if n == node:

                        self.map[n] = i + start / self.s

                        return node

                else:

                    break

            end -= blocksize

        return None

    def loadindex(self, i=None, end=None):

        if self.all: return

        all = False

        if i == None:

            blockstart = 0

            blocksize = (512 / self.s) * self.s

            end = self.datasize

            all = True

        else:

            if end:

                blockstart = i * self.s

                end = end * self.s

                blocksize = end - blockstart

            else:

                blockstart = (i & ~63) * self.s

                blocksize = self.s * 64

                end = blockstart + blocksize

        while blockstart < end:

            self.loadblock(blockstart, blocksize)

            blockstart += blocksize

        if all: self.all = True

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.loadindex(pos)

        return self.p.index[pos]

    def __getitem__(self, pos):

        ret = self.p.index[pos] or self.load(pos)

        if isinstance(ret, str):

            ret = struct.unpack(self.p.indexformat, ret)

        return ret

    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):

        n = self.p.findnode(key)

        if n == None:

            raise KeyError(key)

    def __contains__(self, key):

        if key in self.p.map:

            return True

        self.p.loadmap()

        return key in self.p.map

    def __iter__(self):

        yield nullid

        for i in xrange(self.p.l):

            ret = self.p.index[i]

            if not ret:

                self.p.loadindex(i)

                ret = self.p.index[i]

            if isinstance(ret, str):

                ret = struct.unpack(self.p.indexformat, ret)

            yield ret[-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 LookupError(RevlogError): pass

def getoffset(q):

    if q & 0xFFFF:

        raise RevlogError(_('incompatible revision flag %x') % q)

    return int(q >> 16)

def gettype(q):

    return int(q & 0xFFFF)

def offset_type(offset, type):

    return long(long(offset) << 16 | type)

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):

        """

        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 = indexfile[:-2] + ".d"

        self.opener = opener

        self.indexstat = None

        self.cache = None

        self.defversion = REVLOG_DEFAULT_VERSION

        if hasattr(opener, "defversion"):

            self.defversion = opener.defversion

            if self.defversion & REVLOGNG:

                self.defversion |= REVLOGNGINLINEDATA

        self._load()

    def _load(self):

        v = self.defversion

        try:

            f = self.opener(self.indexfile)

            i = f.read(4)

            f.seek(0)

            if len(i) > 0:

                v = struct.unpack(versionformat, i)[0]

        except IOError, inst:

            if inst.errno != errno.ENOENT:

                raise

            i = ""

        else:

            try:

                st = util.fstat(f)

            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

                    and st.st_size == oldst.st_size):

                    return

                self.indexstat = st

        flags = v & ~0xFFFF

        fmt = v & 0xFFFF

        if fmt == REVLOGV0:

            if flags:

                raise RevlogError(_("index %s unknown flags %#04x for format v0")

                                  % (self.indexfile, flags >> 16))

        elif fmt == REVLOGNG:

            if flags & ~REVLOGNGINLINEDATA:

                raise RevlogError(_("index %s unknown flags %#04x for revlogng")

                                  % (self.indexfile, flags >> 16))

        else:

            raise RevlogError(_("index %s unknown format %d")

                              % (self.indexfile, fmt))

        self.version = v

        self.nodemap = {nullid: nullrev}

        self.index = []

        self.indexformat = indexformatng