"""

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 <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 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 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