view hgext/convert/p4.py @ 33649:377e8ddaebef stable

pathauditor: disable cache of audited paths by default (issue5628)

The initial attempt was to discard cache when appropriate, but it appears
to be error prone. We had to carefully inspect all places where audit() is
called e.g. without actually updating filesystem, before removing files and
directories, etc.

So, this patch disables the cache of audited paths by default, and enables
it only for the following cases:

 - short-lived auditor objects
 - repo.vfs, repo.svfs, and repo.cachevfs, which are managed directories
   and considered sort of append-only (a file/directory would never be
   replaced with a symlink)

There would be more cacheable vfs objects (e.g. mq.queue.opener), but I
decided not to inspect all of them in this patch. We can make them cached
later.

Benchmark result:

- using old clone of http://selenic.com/repo/linux-2.6/ (38319 files)
- on tmpfs
- run HGRCPATH=/dev/null hg up -q --time tip && hg up -q null
- try 4 times and take the last three results

original:
real 7.480 secs (user 1.140+22.760 sys 0.150+1.690)
real 8.010 secs (user 1.070+22.280 sys 0.170+2.120)
real 7.470 secs (user 1.120+22.390 sys 0.120+1.910)

clearcache (the other series):
real 7.680 secs (user 1.120+23.420 sys 0.140+1.970)
real 7.670 secs (user 1.110+23.620 sys 0.130+1.810)
real 7.740 secs (user 1.090+23.510 sys 0.160+1.940)

enable cache only for vfs and svfs (this series):
real 8.730 secs (user 1.500+25.190 sys 0.260+2.260)
real 8.750 secs (user 1.490+25.170 sys 0.250+2.340)
real 9.010 secs (user 1.680+25.340 sys 0.280+2.540)

remove cache function at all (for reference):
real 9.620 secs (user 1.440+27.120 sys 0.250+2.980)
real 9.420 secs (user 1.400+26.940 sys 0.320+3.130)
real 9.760 secs (user 1.530+27.270 sys 0.250+2.970)

# Perforce source for convert extension.
#
# Copyright 2009, Frank Kingswood <[e-mail]>
#
# This software may be used and distributed according to the terms of the
# GNU General Public License version 2 or any later version.
from __future__ import absolute_import

import marshal
import re

from mercurial.i18n import _
from mercurial import (
    error,
    util,
)

from . import common

def loaditer(f):
    "Yield the dictionary objects generated by p4"
    try:
        while True:
            d = marshal.load(f)
            if not d:
                break
            yield d
    except EOFError:
        pass

def decodefilename(filename):
    """Perforce escapes special characters @, #, *, or %
    with %40, %23, %2A, or %25 respectively

    >>> decodefilename('portable-net45%252Bnetcore45%252Bwp8%252BMonoAndroid')
    'portable-net45%2Bnetcore45%2Bwp8%2BMonoAndroid'
    >>> decodefilename('//Depot/Directory/%2525/%2523/%23%40.%2A')
    '//Depot/Directory/%25/%23/#@.*'
    """
    replacements = [('%2A', '*'), ('%23', '#'), ('%40', '@'), ('%25', '%')]
    for k, v in replacements:
        filename = filename.replace(k, v)
    return filename

class p4_source(common.converter_source):
    def __init__(self, ui, path, revs=None):
        # avoid import cycle
        from . import convcmd

        super(p4_source, self).__init__(ui, path, revs=revs)

        if "/" in path and not path.startswith('//'):
            raise common.NoRepo(_('%s does not look like a P4 repository') %
                                path)

        common.checktool('p4', abort=False)

        self.revmap = {}
        self.encoding = self.ui.config('convert', 'p4.encoding',
                                       default=convcmd.orig_encoding)
        self.re_type = re.compile(
            "([a-z]+)?(text|binary|symlink|apple|resource|unicode|utf\d+)"
            "(\+\w+)?$")
        self.re_keywords = re.compile(
            r"\$(Id|Header|Date|DateTime|Change|File|Revision|Author)"
            r":[^$\n]*\$")
        self.re_keywords_old = re.compile("\$(Id|Header):[^$\n]*\$")

        if revs and len(revs) > 1:
            raise error.Abort(_("p4 source does not support specifying "
                               "multiple revisions"))

    def setrevmap(self, revmap):
        """Sets the parsed revmap dictionary.

        Revmap stores mappings from a source revision to a target revision.
        It is set in convertcmd.convert and provided by the user as a file
        on the commandline.

        Revisions in the map are considered beeing present in the
        repository and ignored during _parse(). This allows for incremental
        imports if a revmap is provided.
        """
        self.revmap = revmap

    def _parse_view(self, path):
        "Read changes affecting the path"
        cmd = 'p4 -G changes -s submitted %s' % util.shellquote(path)
        stdout = util.popen(cmd, mode='rb')
        p4changes = {}
        for d in loaditer(stdout):
            c = d.get("change", None)
            if c:
                p4changes[c] = True
        return p4changes

    def _parse(self, ui, path):
        "Prepare list of P4 filenames and revisions to import"
        p4changes = {}
        changeset = {}
        files_map = {}
        copies_map = {}
        localname = {}
        depotname = {}
        heads = []

        ui.status(_('reading p4 views\n'))

        # read client spec or view
        if "/" in path:
            p4changes.update(self._parse_view(path))
            if path.startswith("//") and path.endswith("/..."):
                views = {path[:-3]:""}
            else:
                views = {"//": ""}
        else:
            cmd = 'p4 -G client -o %s' % util.shellquote(path)
            clientspec = marshal.load(util.popen(cmd, mode='rb'))

            views = {}
            for client in clientspec:
                if client.startswith("View"):
                    sview, cview = clientspec[client].split()
                    p4changes.update(self._parse_view(sview))
                    if sview.endswith("...") and cview.endswith("..."):
                        sview = sview[:-3]
                        cview = cview[:-3]
                    cview = cview[2:]
                    cview = cview[cview.find("/") + 1:]
                    views[sview] = cview

        # list of changes that affect our source files
        p4changes = p4changes.keys()
        p4changes.sort(key=int)

        # list with depot pathnames, longest first
        vieworder = views.keys()
        vieworder.sort(key=len, reverse=True)

        # handle revision limiting
        startrev = self.ui.config('convert', 'p4.startrev', default=0)

        # now read the full changelists to get the list of file revisions
        ui.status(_('collecting p4 changelists\n'))
        lastid = None
        for change in p4changes:
            if startrev and int(change) < int(startrev):
                continue
            if self.revs and int(change) > int(self.revs[0]):
                continue
            if change in self.revmap:
                # Ignore already present revisions, but set the parent pointer.
                lastid = change
                continue

            if lastid:
                parents = [lastid]
            else:
                parents = []

            d = self._fetch_revision(change)
            c = self._construct_commit(d, parents)

            descarr = c.desc.splitlines(True)
            if len(descarr) > 0:
                shortdesc = descarr[0].rstrip('\r\n')
            else:
                shortdesc = '**empty changelist description**'

            t = '%s %s' % (c.rev, repr(shortdesc)[1:-1])
            ui.status(util.ellipsis(t, 80) + '\n')

            files = []
            copies = {}
            copiedfiles = []
            i = 0
            while ("depotFile%d" % i) in d and ("rev%d" % i) in d:
                oldname = d["depotFile%d" % i]
                filename = None
                for v in vieworder:
                    if oldname.lower().startswith(v.lower()):
                        filename = decodefilename(views[v] + oldname[len(v):])
                        break
                if filename:
                    files.append((filename, d["rev%d" % i]))
                    depotname[filename] = oldname
                    if (d.get("action%d" % i) == "move/add"):
                        copiedfiles.append(filename)
                    localname[oldname] = filename
                i += 1

            # Collect information about copied files
            for filename in copiedfiles:
                oldname = depotname[filename]

                flcmd = 'p4 -G filelog %s' \
                      % util.shellquote(oldname)
                flstdout = util.popen(flcmd, mode='rb')

                copiedfilename = None
                for d in loaditer(flstdout):
                    copiedoldname = None

                    i = 0
                    while ("change%d" % i) in d:
                        if (d["change%d" % i] == change and
                            d["action%d" % i] == "move/add"):
                            j = 0
                            while ("file%d,%d" % (i, j)) in d:
                                if d["how%d,%d" % (i, j)] == "moved from":
                                    copiedoldname = d["file%d,%d" % (i, j)]
                                    break
                                j += 1
                        i += 1

                    if copiedoldname and copiedoldname in localname:
                        copiedfilename = localname[copiedoldname]
                        break

                if copiedfilename:
                    copies[filename] = copiedfilename
                else:
                    ui.warn(_("cannot find source for copied file: %s@%s\n")
                            % (filename, change))

            changeset[change] = c
            files_map[change] = files
            copies_map[change] = copies
            lastid = change

        if lastid and len(changeset) > 0:
            heads = [lastid]

        return {
            'changeset': changeset,
            'files': files_map,
            'copies': copies_map,
            'heads': heads,
            'depotname': depotname,
        }

    @util.propertycache
    def _parse_once(self):
        return self._parse(self.ui, self.path)

    @util.propertycache
    def copies(self):
        return self._parse_once['copies']

    @util.propertycache
    def files(self):
        return self._parse_once['files']

    @util.propertycache
    def changeset(self):
        return self._parse_once['changeset']

    @util.propertycache
    def heads(self):
        return self._parse_once['heads']

    @util.propertycache
    def depotname(self):
        return self._parse_once['depotname']

    def getheads(self):
        return self.heads

    def getfile(self, name, rev):
        cmd = 'p4 -G print %s' \
            % util.shellquote("%s#%s" % (self.depotname[name], rev))

        lasterror = None
        while True:
            stdout = util.popen(cmd, mode='rb')

            mode = None
            contents = []
            keywords = None

            for d in loaditer(stdout):
                code = d["code"]
                data = d.get("data")

                if code == "error":
                    # if this is the first time error happened
                    # re-attempt getting the file
                    if not lasterror:
                        lasterror = IOError(d["generic"], data)
                        # this will exit inner-most for-loop
                        break
                    else:
                        raise lasterror

                elif code == "stat":
                    action = d.get("action")
                    if action in ["purge", "delete", "move/delete"]:
                        return None, None
                    p4type = self.re_type.match(d["type"])
                    if p4type:
                        mode = ""
                        flags = ((p4type.group(1) or "")
                               + (p4type.group(3) or ""))
                        if "x" in flags:
                            mode = "x"
                        if p4type.group(2) == "symlink":
                            mode = "l"
                        if "ko" in flags:
                            keywords = self.re_keywords_old
                        elif "k" in flags:
                            keywords = self.re_keywords

                elif code == "text" or code == "binary":
                    contents.append(data)

                lasterror = None

            if not lasterror:
                break

        if mode is None:
            return None, None

        contents = ''.join(contents)

        if keywords:
            contents = keywords.sub("$\\1$", contents)
        if mode == "l" and contents.endswith("\n"):
            contents = contents[:-1]

        return contents, mode

    def getchanges(self, rev, full):
        if full:
            raise error.Abort(_("convert from p4 does not support --full"))
        return self.files[rev], self.copies[rev], set()

    def _construct_commit(self, obj, parents=None):
        """
        Constructs a common.commit object from an unmarshalled
        `p4 describe` output
        """
        desc = self.recode(obj.get("desc", ""))
        date = (int(obj["time"]), 0)     # timezone not set
        if parents is None:
            parents = []

        return common.commit(author=self.recode(obj["user"]),
            date=util.datestr(date, '%Y-%m-%d %H:%M:%S %1%2'),
            parents=parents, desc=desc, branch=None, rev=obj['change'],
            extra={"p4": obj['change'], "convert_revision": obj['change']})

    def _fetch_revision(self, rev):
        """Return an output of `p4 describe` including author, commit date as
        a dictionary."""
        cmd = "p4 -G describe -s %s" % rev
        stdout = util.popen(cmd, mode='rb')
        return marshal.load(stdout)

    def getcommit(self, rev):
        if rev in self.changeset:
            return self.changeset[rev]
        elif rev in self.revmap:
            d = self._fetch_revision(rev)
            return self._construct_commit(d, parents=None)
        raise error.Abort(
            _("cannot find %s in the revmap or parsed changesets") % rev)

    def gettags(self):
        return {}

    def getchangedfiles(self, rev, i):
        return sorted([x[0] for x in self.files[rev]])