"""

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.

"""

import zlib, struct, sha, binascii, heapq

from mercurial import mdiff

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 %s" % t)

indexformat = ">4l20s20s20s"

class lazyparser:

    """

    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):

        self.data = data

        self.s = struct.calcsize(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:

            d = self.data[i * self.s: (i + 1) * self.s]

            e = struct.unpack(indexformat, d)

            self.index[i] = e

            self.map[e[6]] = i

            i += 1

class lazyindex:

    """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):

        self.p.load(pos)

        return self.p.index[pos]

    def __getitem__(self, pos):

        return self.p.index[pos] or self.load(pos)

    def append(self, e):

        self.p.index.append(e)

class lazymap:

    """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("node " + hex(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][6]

            except:

                self.p.load(i)

                yield self.p.index[i][6]

    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

class RevlogError(Exception): pass

class revlog:

    """

    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):

        """

        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.cache = None

        try:

            i = self.opener(self.indexfile).read()

        except IOError:

            i = ""

        if len(i) > 10000:

            # big index, let's parse it on demand

            parser = lazyparser(i, self)

            self.index = lazyindex(parser)

            self.nodemap = lazymap(parser)

        else:

            s = struct.calcsize(indexformat)

            l = len(i) / s

            self.index = [None] * l

            m = [None] * l

            n = 0

            for f in xrange(0, len(i), s):

                # offset, size, base, linkrev, p1, p2, nodeid

                e = struct.unpack(indexformat, i[f:f + s])

                m[n] = (e[6], n)

                self.index[n] = e

                n += 1

            self.nodemap = dict(m)

            self.nodemap[nullid] = -1

    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][6]

    def rev(self, node): return self.nodemap[node]

    def linkrev(self, node): return self.index[self.nodemap[node]][3]

    def parents(self, node):

        if node == nullid: return (nullid, nullid)

        return self.index[self.nodemap[node]][4:6]

    def start(self, rev): return self.index[rev][0]

    def length(self, rev): return self.index[rev][1]

    def end(self, rev): return self.start(rev) + self.length(rev)

    def base(self, rev): return self.index[rev][2]

    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 heads(self, stop=None):

        """return the list of all nodes that have no children"""

        p = {}

        h = []

        stoprev = 0

        if stop and stop in self.nodemap:

            stoprev = self.rev(stop)

        for r in range(self.count() - 1, -1, -1):

            n = self.node(r)

            if n not in p:

                h.append(n)

            if n == stop:

                break

            if r < stoprev:

                break

            for pn in self.parents(n):

                p[pn] = 1

        return h

    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 KeyError("Ambiguous identifier")

            if len(c) < 1: raise KeyError("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 delta(self, node):

        """return or calculate a delta between a node and its predecessor"""

        r = self.rev(node)

        b = self.base(r)

        if r == b:

            return self.diff(self.revision(self.node(r - 1)),

                             self.revision(node))

        else:

            f = self.opener(self.datafile)

            f.seek(self.start(r))

            data = f.read(self.length(r))

        return decompress(data)

    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)

        start, length, base, link, p1, p2, node = self.index[rev]

        end = start + length

        if base != rev: start = self.start(base)

        # do we have useful data cached?

        if self.cache and self.cache[1] >= base and self.cache[1] < rev:

            base = self.cache[1]

            start = self.start(base + 1)

            text = self.cache[2]

            last = 0

        f = self.opener(self.datafile)

        f.seek(start)

        data = f.read(end - start)

        if text is None:

            last = self.length(base)

            text = decompress(data[:last])

        bins = []

        for r in xrange(base + 1, rev + 1):

            s = self.length(r)

            bins.append(decompress(data[last:last + s]))

            last = last + s

        text = mdiff.patches(text, bins)

        if node != hash(text, p1, p2):

            raise IOError("integrity check failed on %s:%d"

                          % (self.datafile, rev))

        self.cache = (node, rev, text)

        return text

    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)