view doc/gendoc.py @ 29830:92ac2baaea86

revlog: use an LRU cache for delta chain bases

Profiling using statprof revealed a hotspot during changegroup
application calculating delta chain bases on generaldelta repos.
Essentially, revlog._addrevision() was performing a lot of redundant
work tracing the delta chain as part of determining when the chain
distance was acceptable. This was most pronounced when adding
revisions to manifests, which can have delta chains thousands of
revisions long.

There was a delta chain base cache on revlogs before, but it only
captured a single revision. This was acceptable before generaldelta,
when _addrevision would build deltas from the previous revision and
thus we'd pretty much guarantee a cache hit when resolving the delta
chain base on a subsequent _addrevision call. However, it isn't
suitable for generaldelta because parent revisions aren't necessarily
the last processed revision.

This patch converts the delta chain base cache to an LRU dict cache.
The cache can hold multiple entries, so generaldelta repos have a
higher chance of getting a cache hit.

The impact of this change when processing changegroup additions is
significant. On a generaldelta conversion of the "mozilla-unified"
repo (which contains heads of the main Firefox repositories in
chronological order - this means there are lots of transitions between
heads in revlog order), this change has the following impact when
performing an `hg unbundle` of an uncompressed bundle of the repo:

before: 5:42 CPU time
after:  4:34 CPU time

Most of this time is saved when applying the changelog and manifest
revlogs:

before: 2:30 CPU time
after:  1:17 CPU time

That nearly a 50% reduction in CPU time applying changesets and
manifests!

Applying a gzipped bundle of the same repo (effectively simulating a
`hg clone` over HTTP) showed a similar speedup:

before: 5:53 CPU time
after:  4:46 CPU time

Wall time improvements were basically the same as CPU time.

I didn't measure explicitly, but it feels like most of the time
is saved when processing manifests. This makes sense, as large
manifests tend to have very long delta chains and thus benefit the
most from this cache.

So, this change effectively makes changegroup application (which is
used by `hg unbundle`, `hg clone`, `hg pull`, `hg unshelve`, and
various other commands) significantly faster when delta chains are
long (which can happen on repos with large numbers of files and thus
large manifests).

In theory, this change can result in more memory utilization. However,
we're caching a dict of ints. At most we have 200 ints + Python object
overhead per revlog. And, the cache is really only populated when
performing read-heavy operations, such as adding changegroups or
scanning an individual revlog. For memory bloat to be an issue, we'd
need to scan/read several revisions from several revlogs all while
having active references to several revlogs. I don't think there are
many operations that do this, so I don't think memory bloat from the
cache will be an issue.

#!/usr/bin/env python
"""usage: %s DOC ...

where DOC is the name of a document
"""

from __future__ import absolute_import

import os
import sys
import textwrap

# This script is executed during installs and may not have C extensions
# available. Relax C module requirements.
os.environ['HGMODULEPOLICY'] = 'allow'
# import from the live mercurial repo
sys.path.insert(0, "..")
from mercurial import demandimport; demandimport.enable()
from mercurial import (
    commands,
    extensions,
    help,
    minirst,
    ui as uimod,
)
from mercurial.i18n import (
    gettext,
    _,
)

table = commands.table
globalopts = commands.globalopts
helptable = help.helptable
loaddoc = help.loaddoc

def get_desc(docstr):
    if not docstr:
        return "", ""
    # sanitize
    docstr = docstr.strip("\n")
    docstr = docstr.rstrip()
    shortdesc = docstr.splitlines()[0].strip()

    i = docstr.find("\n")
    if i != -1:
        desc = docstr[i + 2:]
    else:
        desc = shortdesc

    desc = textwrap.dedent(desc)

    return (shortdesc, desc)

def get_opts(opts):
    for opt in opts:
        if len(opt) == 5:
            shortopt, longopt, default, desc, optlabel = opt
        else:
            shortopt, longopt, default, desc = opt
            optlabel = _("VALUE")
        allopts = []
        if shortopt:
            allopts.append("-%s" % shortopt)
        if longopt:
            allopts.append("--%s" % longopt)
        if isinstance(default, list):
            allopts[-1] += " <%s[+]>" % optlabel
        elif (default is not None) and not isinstance(default, bool):
            allopts[-1] += " <%s>" % optlabel
        if '\n' in desc:
            # only remove line breaks and indentation
            desc = ' '.join(l.lstrip() for l in desc.split('\n'))
        desc += default and _(" (default: %s)") % default or ""
        yield (", ".join(allopts), desc)

def get_cmd(cmd, cmdtable):
    d = {}
    attr = cmdtable[cmd]
    cmds = cmd.lstrip("^").split("|")

    d['cmd'] = cmds[0]
    d['aliases'] = cmd.split("|")[1:]
    d['desc'] = get_desc(gettext(attr[0].__doc__))
    d['opts'] = list(get_opts(attr[1]))

    s = 'hg ' + cmds[0]
    if len(attr) > 2:
        if not attr[2].startswith('hg'):
            s += ' ' + attr[2]
        else:
            s = attr[2]
    d['synopsis'] = s.strip()

    return d

def showdoc(ui):
    # print options
    ui.write(minirst.section(_("Options")))
    multioccur = False
    for optstr, desc in get_opts(globalopts):
        ui.write("%s\n    %s\n\n" % (optstr, desc))
        if optstr.endswith("[+]>"):
            multioccur = True
    if multioccur:
        ui.write(_("\n[+] marked option can be specified multiple times\n"))
        ui.write("\n")

    # print cmds
    ui.write(minirst.section(_("Commands")))
    commandprinter(ui, table, minirst.subsection)

    # print help topics
    # The config help topic is included in the hgrc.5 man page.
    helpprinter(ui, helptable, minirst.section, exclude=['config'])

    ui.write(minirst.section(_("Extensions")))
    ui.write(_("This section contains help for extensions that are "
               "distributed together with Mercurial. Help for other "
               "extensions is available in the help system."))
    ui.write(("\n\n"
             ".. contents::\n"
             "   :class: htmlonly\n"
             "   :local:\n"
             "   :depth: 1\n\n"))

    for extensionname in sorted(allextensionnames()):
        mod = extensions.load(ui, extensionname, None)
        ui.write(minirst.subsection(extensionname))
        ui.write("%s\n\n" % gettext(mod.__doc__))
        cmdtable = getattr(mod, 'cmdtable', None)
        if cmdtable:
            ui.write(minirst.subsubsection(_('Commands')))
            commandprinter(ui, cmdtable, minirst.subsubsubsection)

def showtopic(ui, topic):
    extrahelptable = [
        (["common"], '', loaddoc('common')),
        (["hg.1"], '', loaddoc('hg.1')),
        (["hg-ssh.8"], '', loaddoc('hg-ssh.8')),
        (["hgignore.5"], '', loaddoc('hgignore.5')),
        (["hgrc.5"], '', loaddoc('hgrc.5')),
        (["hgignore.5.gendoc"], '', loaddoc('hgignore')),
        (["hgrc.5.gendoc"], '', loaddoc('config')),
    ]
    helpprinter(ui, helptable + extrahelptable, None, include=[topic])

def helpprinter(ui, helptable, sectionfunc, include=[], exclude=[]):
    for names, sec, doc in helptable:
        if exclude and names[0] in exclude:
            continue
        if include and names[0] not in include:
            continue
        for name in names:
            ui.write(".. _%s:\n" % name)
        ui.write("\n")
        if sectionfunc:
            ui.write(sectionfunc(sec))
        if callable(doc):
            doc = doc(ui)
        ui.write(doc)
        ui.write("\n")

def commandprinter(ui, cmdtable, sectionfunc):
    h = {}
    for c, attr in cmdtable.items():
        f = c.split("|")[0]
        f = f.lstrip("^")
        h[f] = c
    cmds = h.keys()
    cmds.sort()

    for f in cmds:
        if f.startswith("debug"):
            continue
        d = get_cmd(h[f], cmdtable)
        ui.write(sectionfunc(d['cmd']))
        # short description
        ui.write(d['desc'][0])
        # synopsis
        ui.write("::\n\n")
        synopsislines = d['synopsis'].splitlines()
        for line in synopsislines:
            # some commands (such as rebase) have a multi-line
            # synopsis
            ui.write("   %s\n" % line)
        ui.write('\n')
        # description
        ui.write("%s\n\n" % d['desc'][1])
        # options
        opt_output = list(d['opts'])
        if opt_output:
            opts_len = max([len(line[0]) for line in opt_output])
            ui.write(_("Options:\n\n"))
            multioccur = False
            for optstr, desc in opt_output:
                if desc:
                    s = "%-*s  %s" % (opts_len, optstr, desc)
                else:
                    s = optstr
                ui.write("%s\n" % s)
                if optstr.endswith("[+]>"):
                    multioccur = True
            if multioccur:
                ui.write(_("\n[+] marked option can be specified"
                           " multiple times\n"))
            ui.write("\n")
        # aliases
        if d['aliases']:
            ui.write(_("    aliases: %s\n\n") % " ".join(d['aliases']))


def allextensionnames():
    return extensions.enabled().keys() + extensions.disabled().keys()

if __name__ == "__main__":
    doc = 'hg.1.gendoc'
    if len(sys.argv) > 1:
        doc = sys.argv[1]

    ui = uimod.ui()
    if doc == 'hg.1.gendoc':
        showdoc(ui)
    else:
        showtopic(ui, sys.argv[1])