# bundle2.py - generic container format to transmit arbitrary data.

#

# Copyright 2013 Facebook, Inc.

#

# This software may be used and distributed according to the terms of the

# GNU General Public License version 2 or any later version.

"""Handling of the new bundle2 format

The goal of bundle2 is to act as an atomically packet to transmit a set of

payloads in an application agnostic way. It consist in a sequence of "parts"

that will be handed to and processed by the application layer.

General format architecture

===========================

The format is architectured as follow

 - magic string

 - stream level parameters

 - payload parts (any number)

 - end of stream marker.

the Binary format

============================

All numbers are unsigned and big-endian.

stream level parameters

------------------------

Binary format is as follow

:params size: int32

  The total number of Bytes used by the parameters

:params value: arbitrary number of Bytes

  A blob of `params size` containing the serialized version of all stream level

  parameters.

  The blob contains a space separated list of parameters. Parameters with value

  are stored in the form `<name>=<value>`. Both name and value are urlquoted.

  Empty name are obviously forbidden.

  Name MUST start with a letter. If this first letter is lower case, the

  parameter is advisory and can be safely ignored. However when the first

  letter is capital, the parameter is mandatory and the bundling process MUST

  stop if he is not able to proceed it.

  Stream parameters use a simple textual format for two main reasons:

  - Stream level parameters should remain simple and we want to discourage any

    crazy usage.

  - Textual data allow easy human inspection of a bundle2 header in case of

    troubles.

  Any Applicative level options MUST go into a bundle2 part instead.

Payload part

------------------------

Binary format is as follow

:header size: int32

  The total number of Bytes used by the part headers. When the header is empty

  (size = 0) this is interpreted as the end of stream marker.

:header:

    The header defines how to interpret the part. It contains two piece of

    data: the part type, and the part parameters.

    The part type is used to route an application level handler, that can

    interpret payload.

    Part parameters are passed to the application level handler.  They are

    meant to convey information that will help the application level object to

    interpret the part payload.

    The binary format of the header is has follow

    :typesize: (one byte)

    :parttype: alphanumerical part name (restricted to [a-zA-Z0-9_:-]*)

    :partid: A 32bits integer (unique in the bundle) that can be used to refer

             to this part.

    :parameters:

        Part's parameter may have arbitrary content, the binary structure is::

            <mandatory-count><advisory-count><param-sizes><param-data>

        :mandatory-count: 1 byte, number of mandatory parameters

        :advisory-count:  1 byte, number of advisory parameters

        :param-sizes:

            N couple of bytes, where N is the total number of parameters. Each

            couple contains (<size-of-key>, <size-of-value) for one parameter.

        :param-data:

            A blob of bytes from which each parameter key and value can be

            retrieved using the list of size couples stored in the previous

            field.

            Mandatory parameters comes first, then the advisory ones.

            Each parameter's key MUST be unique within the part.

:payload:

    payload is a series of `<chunksize><chunkdata>`.

    `chunksize` is an int32, `chunkdata` are plain bytes (as much as

    `chunksize` says)` The payload part is concluded by a zero size chunk.

    The current implementation always produces either zero or one chunk.

    This is an implementation limitation that will ultimately be lifted.

    `chunksize` can be negative to trigger special case processing. No such

    processing is in place yet.

Bundle processing

============================

Each part is processed in order using a "part handler". Handler are registered

for a certain part type.

The matching of a part to its handler is case insensitive. The case of the

part type is used to know if a part is mandatory or advisory. If the Part type

contains any uppercase char it is considered mandatory. When no handler is

known for a Mandatory part, the process is aborted and an exception is raised.

If the part is advisory and no handler is known, the part is ignored. When the

process is aborted, the full bundle is still read from the stream to keep the

channel usable. But none of the part read from an abort are processed. In the

future, dropping the stream may become an option for channel we do not care to

preserve.

"""

import errno

import sys

import util

import struct

import urllib

import string

import obsolete

import pushkey

import url

import re

import changegroup, error

from i18n import _

_pack = struct.pack

_unpack = struct.unpack

_fstreamparamsize = '>i'

_fpartheadersize = '>i'

_fparttypesize = '>B'

_fpartid = '>I'

_fpayloadsize = '>i'

_fpartparamcount = '>BB'

preferedchunksize = 4096

_parttypeforbidden = re.compile('[^a-zA-Z0-9_:-]')

def outdebug(ui, message):

    """debug regarding output stream (bundling)"""

    ui.debug('bundle2-output: %s\n' % message)

def indebug(ui, message):

    """debug on input stream (unbundling)"""

    ui.debug('bundle2-input: %s\n' % message)

def validateparttype(parttype):

    """raise ValueError if a parttype contains invalid character"""

    if _parttypeforbidden.search(parttype):

        raise ValueError(parttype)

def _makefpartparamsizes(nbparams):

    """return a struct format to read part parameter sizes

    The number parameters is variable so we need to build that format

    dynamically.

    """

    return '>'+('BB'*nbparams)

parthandlermapping = {}

def parthandler(parttype, params=()):

    """decorator that register a function as a bundle2 part handler

    eg::

        @parthandler('myparttype', ('mandatory', 'param', 'handled'))

        def myparttypehandler(...):

            '''process a part of type "my part".'''

            ...

    """

    validateparttype(parttype)

    def _decorator(func):

        lparttype = parttype.lower() # enforce lower case matching.

        assert lparttype not in parthandlermapping

        parthandlermapping[lparttype] = func

        func.params = frozenset(params)

        return func

    return _decorator

class unbundlerecords(object):

    """keep record of what happens during and unbundle

    New records are added using `records.add('cat', obj)`. Where 'cat' is a

    category of record and obj is an arbitrary object.

    `records['cat']` will return all entries of this category 'cat'.

    Iterating on the object itself will yield `('category', obj)` tuples

    for all entries.

    All iterations happens in chronological order.

    """

    def __init__(self):

        self._categories = {}

        self._sequences = []

        self._replies = {}

    def add(self, category, entry, inreplyto=None):

        """add a new record of a given category.

        The entry can then be retrieved in the list returned by

        self['category']."""

        self._categories.setdefault(category, []).append(entry)

        self._sequences.append((category, entry))

        if inreplyto is not None:

            self.getreplies(inreplyto).add(category, entry)

    def getreplies(self, partid):

        """get the records that are replies to a specific part"""

        return self._replies.setdefault(partid, unbundlerecords())

    def __getitem__(self, cat):

        return tuple(self._categories.get(cat, ()))

    def __iter__(self):

        return iter(self._sequences)

    def __len__(self):

        return len(self._sequences)

    def __nonzero__(self):

        return bool(self._sequences)

class bundleoperation(object):

    """an object that represents a single bundling process

    Its purpose is to carry unbundle-related objects and states.

    A new object should be created at the beginning of each bundle processing.

    The object is to be returned by the processing function.

    The object has very little content now it will ultimately contain:

    * an access to the repo the bundle is applied to,

    * a ui object,

    * a way to retrieve a transaction to add changes to the repo,

    * a way to record the result of processing each part,

    * a way to construct a bundle response when applicable.

    """

    def __init__(self, repo, transactiongetter, captureoutput=True):

        self.repo = repo

        self.ui = repo.ui

        self.records = unbundlerecords()

        self.gettransaction = transactiongetter

        self.reply = None

        self.captureoutput = captureoutput

class TransactionUnavailable(RuntimeError):

    pass

def _notransaction():

    """default method to get a transaction while processing a bundle

    Raise an exception to highlight the fact that no transaction was expected

    to be created"""

    raise TransactionUnavailable()

def processbundle(repo, unbundler, transactiongetter=None, op=None):

    """This function process a bundle, apply effect to/from a repo

    It iterates over each part then searches for and uses the proper handling

    code to process the part. Parts are processed in order.

    This is very early version of this function that will be strongly reworked

    before final usage.

    Unknown Mandatory part will abort the process.

    It is temporarily possible to provide a prebuilt bundleoperation to the

    function. This is used to ensure output is properly propagated in case of

    an error during the unbundling. This output capturing part will likely be

    reworked and this ability will probably go away in the process.

    """

    if op is None:

        if transactiongetter is None:

            transactiongetter = _notransaction

        op = bundleoperation(repo, transactiongetter)

    # todo:

    # - replace this is a init function soon.

    # - exception catching

    unbundler.params

    iterparts = unbundler.iterparts()

    part = None

    try:

        for part in iterparts:

            _processpart(op, part)

    except BaseException, exc:

        for part in iterparts:

            # consume the bundle content

            part.seek(0, 2)

        # Small hack to let caller code distinguish exceptions from bundle2

        # processing from processing the old format. This is mostly

        # needed to handle different return codes to unbundle according to the

        # type of bundle. We should probably clean up or drop this return code

        # craziness in a future version.

        exc.duringunbundle2 = True

        salvaged = []

        if op.reply is not None:

            salvaged = op.reply.salvageoutput()

        exc._bundle2salvagedoutput = salvaged

        raise

    return op

def _processpart(op, part):

    """process a single part from a bundle

    The part is guaranteed to have been fully consumed when the function exits

    (even if an exception is raised)."""

    try:

        try:

            handler = parthandlermapping.get(part.type)

            if handler is None:

                raise error.UnsupportedPartError(parttype=part.type)

            indebug(op.ui, 'found a handler for part %r' % part.type)

            unknownparams = part.mandatorykeys - handler.params

            if unknownparams:

                unknownparams = list(unknownparams)

                unknownparams.sort()

                raise error.UnsupportedPartError(parttype=part.type,

                                               params=unknownparams)

        except error.UnsupportedPartError, exc:

            if part.mandatory: # mandatory parts

                raise

            indebug(op.ui, 'ignoring unsupported advisory part %s' % exc)

            return # skip to part processing

        # handler is called outside the above try block so that we don't

        # risk catching KeyErrors from anything other than the

        # parthandlermapping lookup (any KeyError raised by handler()

        # itself represents a defect of a different variety).

        output = None

        if op.captureoutput and op.reply is not None:

            op.ui.pushbuffer(error=True, subproc=True)

            output = ''

        try:

            handler(op, part)

        finally:

            if output is not None:

                output = op.ui.popbuffer()

            if output:

                outpart = op.reply.newpart('output', data=output,

                                           mandatory=False)

                outpart.addparam('in-reply-to', str(part.id), mandatory=False)

    finally:

        # consume the part content to not corrupt the stream.

        part.seek(0, 2)

def decodecaps(blob):

    """decode a bundle2 caps bytes blob into a dictionary

    The blob is a list of capabilities (one per line)

    Capabilities may have values using a line of the form::

        capability=value1,value2,value3

    The values are always a list."""

    caps = {}

    for line in blob.splitlines():

        if not line:

            continue

        if '=' not in line:

            key, vals = line, ()

        else:

            key, vals = line.split('=', 1)

            vals = vals.split(',')

        key = urllib.unquote(key)

        vals = [urllib.unquote(v) for v in vals]

        caps[key] = vals

    return caps

def encodecaps(caps):

    """encode a bundle2 caps dictionary into a bytes blob"""

    chunks = []

    for ca in sorted(caps):

        vals = caps[ca]

        ca = urllib.quote(ca)

        vals = [urllib.quote(v) for v in vals]

        if vals:

            ca = "%s=%s" % (ca, ','.join(vals))

        chunks.append(ca)

    return '\n'.join(chunks)