hgweb: skip body creation of HEAD for most requests

The body is thrown away anyway, so this just wastes a lot of CPU time.
In the case of /archive/, this skips manifest processing and the actual
file archiving, resulting in a huge difference.

The most tricky part here is skipping the Content-Length creation as it
would indicate the output size for the corresponding GET request.

mercurial/hgweb/hgweb_mod.py

@@ -1,515 +1,516 b''
 # hgweb/hgweb_mod.py - Web interface for a repository.
 #
 # Copyright 21 May 2005 - (c) 2005 Jake Edge <[e-mail]>
 # Copyright 2005-2007 Olivia Mackall <[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.
 
 
 import contextlib
 import os
 
 from .common import (
     ErrorResponse,
     HTTP_BAD_REQUEST,
     cspvalues,
     permhooks,
     statusmessage,
 )
 from ..pycompat import getattr
 
 from .. import (
     encoding,
     error,
     extensions,
     formatter,
     hg,
     hook,
     profiling,
     pycompat,
     registrar,
     repoview,
     templatefilters,
     templater,
     templateutil,
     ui as uimod,
     util,
     wireprotoserver,
 )
 
 from . import (
     request as requestmod,
     webcommands,
     webutil,
     wsgicgi,
 )
 
 
 def getstyle(req, configfn, templatepath):
     styles = (
         req.qsparams.get(b'style', None),
         configfn(b'web', b'style'),
         b'paper',
     )
     return styles, _stylemap(styles, templatepath)
 
 
 def _stylemap(styles, path=None):
     """Return path to mapfile for a given style.
 
     Searches mapfile in the following locations:
     1. templatepath/style/map
     2. templatepath/map-style
     3. templatepath/map
     """
 
     for style in styles:
         # only plain name is allowed to honor template paths
         if (
             not style
             or style in (pycompat.oscurdir, pycompat.ospardir)
             or pycompat.ossep in style
             or pycompat.osaltsep
             and pycompat.osaltsep in style
         ):
             continue
         locations = (os.path.join(style, b'map'), b'map-' + style, b'map')
 
         for location in locations:
             mapfile, fp = templater.try_open_template(location, path)
             if mapfile:
                 return style, mapfile, fp
 
     raise RuntimeError(b"No hgweb templates found in %r" % path)
 
 
 def makebreadcrumb(url, prefix=b''):
     """Return a 'URL breadcrumb' list
 
     A 'URL breadcrumb' is a list of URL-name pairs,
     corresponding to each of the path items on a URL.
     This can be used to create path navigation entries.
     """
     if url.endswith(b'/'):
         url = url[:-1]
     if prefix:
         url = b'/' + prefix + url
     relpath = url
     if relpath.startswith(b'/'):
         relpath = relpath[1:]
 
     breadcrumb = []
     urlel = url
     pathitems = [b''] + relpath.split(b'/')
     for pathel in reversed(pathitems):
         if not pathel or not urlel:
             break
         breadcrumb.append({b'url': urlel, b'name': pathel})
         urlel = os.path.dirname(urlel)
     return templateutil.mappinglist(reversed(breadcrumb))
 
 
 class requestcontext:
     """Holds state/context for an individual request.
 
     Servers can be multi-threaded. Holding state on the WSGI application
     is prone to race conditions. Instances of this class exist to hold
     mutable and race-free state for requests.
     """
 
     def __init__(self, app, repo, req, res):
         self.repo = repo
         self.reponame = app.reponame
         self.req = req
         self.res = res
 
         self.maxchanges = self.configint(b'web', b'maxchanges')
         self.stripecount = self.configint(b'web', b'stripes')
         self.maxshortchanges = self.configint(b'web', b'maxshortchanges')
         self.maxfiles = self.configint(b'web', b'maxfiles')
         self.allowpull = self.configbool(b'web', b'allow-pull')
 
         # we use untrusted=False to prevent a repo owner from using
         # web.templates in .hg/hgrc to get access to any file readable
         # by the user running the CGI script
         self.templatepath = self.config(b'web', b'templates', untrusted=False)
 
         # This object is more expensive to build than simple config values.
         # It is shared across requests. The app will replace the object
         # if it is updated. Since this is a reference and nothing should
         # modify the underlying object, it should be constant for the lifetime
         # of the request.
         self.websubtable = app.websubtable
 
         self.csp, self.nonce = cspvalues(self.repo.ui)
 
     # Trust the settings from the .hg/hgrc files by default.
     def config(self, *args, **kwargs):
         kwargs.setdefault('untrusted', True)
         return self.repo.ui.config(*args, **kwargs)
 
     def configbool(self, *args, **kwargs):
         kwargs.setdefault('untrusted', True)
         return self.repo.ui.configbool(*args, **kwargs)
 
     def configint(self, *args, **kwargs):
         kwargs.setdefault('untrusted', True)
         return self.repo.ui.configint(*args, **kwargs)
 
     def configlist(self, *args, **kwargs):
         kwargs.setdefault('untrusted', True)
         return self.repo.ui.configlist(*args, **kwargs)
 
     def archivelist(self, nodeid):
         return webutil.archivelist(self.repo.ui, nodeid)
 
     def templater(self, req):
         # determine scheme, port and server name
         # this is needed to create absolute urls
         logourl = self.config(b'web', b'logourl')
         logoimg = self.config(b'web', b'logoimg')
         staticurl = (
             self.config(b'web', b'staticurl')
             or req.apppath.rstrip(b'/') + b'/static/'
         )
         if not staticurl.endswith(b'/'):
             staticurl += b'/'
 
         # figure out which style to use
 
         vars = {}
         styles, (style, mapfile, fp) = getstyle(
             req, self.config, self.templatepath
         )
         if style == styles[0]:
             vars[b'style'] = style
 
         sessionvars = webutil.sessionvars(vars, b'?')
 
         if not self.reponame:
             self.reponame = (
                 self.config(b'web', b'name', b'')
                 or req.reponame
                 or req.apppath
                 or self.repo.root
             )
 
         filters = {}
         templatefilter = registrar.templatefilter(filters)
 
         @templatefilter(b'websub', intype=bytes)
         def websubfilter(text):
             return templatefilters.websub(text, self.websubtable)
 
         # create the templater
         # TODO: export all keywords: defaults = templatekw.keywords.copy()
         defaults = {
             b'url': req.apppath + b'/',
             b'logourl': logourl,
             b'logoimg': logoimg,
             b'staticurl': staticurl,
             b'urlbase': req.advertisedbaseurl,
             b'repo': self.reponame,
             b'encoding': encoding.encoding,
             b'sessionvars': sessionvars,
             b'pathdef': makebreadcrumb(req.apppath),
             b'style': style,
             b'nonce': self.nonce,
         }
         templatekeyword = registrar.templatekeyword(defaults)
 
         @templatekeyword(b'motd', requires=())
         def motd(context, mapping):
             yield self.config(b'web', b'motd')
 
         tres = formatter.templateresources(self.repo.ui, self.repo)
         return templater.templater.frommapfile(
             mapfile, fp=fp, filters=filters, defaults=defaults, resources=tres
         )
 
     def sendtemplate(self, name, **kwargs):
         """Helper function to send a response generated from a template."""
-        kwargs = pycompat.byteskwargs(kwargs)
-        self.res.setbodygen(self.tmpl.generate(name, kwargs))
+        if self.req.method != b'HEAD':
+            kwargs = pycompat.byteskwargs(kwargs)
+            self.res.setbodygen(self.tmpl.generate(name, kwargs))
         return self.res.sendresponse()
 
 
 class hgweb:
     """HTTP server for individual repositories.
 
     Instances of this class serve HTTP responses for a particular
     repository.
 
     Instances are typically used as WSGI applications.
 
     Some servers are multi-threaded. On these servers, there may
     be multiple active threads inside __call__.
     """
 
     def __init__(self, repo, name=None, baseui=None):
         if isinstance(repo, bytes):
             if baseui:
                 u = baseui.copy()
             else:
                 u = uimod.ui.load()
                 extensions.loadall(u)
                 extensions.populateui(u)
             r = hg.repository(u, repo)
         else:
             # we trust caller to give us a private copy
             r = repo
 
         r.ui.setconfig(b'ui', b'report_untrusted', b'off', b'hgweb')
         r.baseui.setconfig(b'ui', b'report_untrusted', b'off', b'hgweb')
         r.ui.setconfig(b'ui', b'nontty', b'true', b'hgweb')
         r.baseui.setconfig(b'ui', b'nontty', b'true', b'hgweb')
         # resolve file patterns relative to repo root
         r.ui.setconfig(b'ui', b'forcecwd', r.root, b'hgweb')
         r.baseui.setconfig(b'ui', b'forcecwd', r.root, b'hgweb')
         # it's unlikely that we can replace signal handlers in WSGI server,
         # and mod_wsgi issues a big warning. a plain hgweb process (with no
         # threading) could replace signal handlers, but we don't bother
         # conditionally enabling it.
         r.ui.setconfig(b'ui', b'signal-safe-lock', b'false', b'hgweb')
         r.baseui.setconfig(b'ui', b'signal-safe-lock', b'false', b'hgweb')
         # displaying bundling progress bar while serving feel wrong and may
         # break some wsgi implementation.
         r.ui.setconfig(b'progress', b'disable', b'true', b'hgweb')
         r.baseui.setconfig(b'progress', b'disable', b'true', b'hgweb')
         self._repos = [hg.cachedlocalrepo(self._webifyrepo(r))]
         self._lastrepo = self._repos[0]
         hook.redirect(True)
         self.reponame = name
 
     def _webifyrepo(self, repo):
         repo = getwebview(repo)
         self.websubtable = webutil.getwebsubs(repo)
         return repo
 
     @contextlib.contextmanager
     def _obtainrepo(self):
         """Obtain a repo unique to the caller.
 
         Internally we maintain a stack of cachedlocalrepo instances
         to be handed out. If one is available, we pop it and return it,
         ensuring it is up to date in the process. If one is not available,
         we clone the most recently used repo instance and return it.
 
         It is currently possible for the stack to grow without bounds
         if the server allows infinite threads. However, servers should
         have a thread limit, thus establishing our limit.
         """
         if self._repos:
             cached = self._repos.pop()
             r, created = cached.fetch()
         else:
             cached = self._lastrepo.copy()
             r, created = cached.fetch()
         if created:
             r = self._webifyrepo(r)
 
         self._lastrepo = cached
         self.mtime = cached.mtime
         try:
             yield r
         finally:
             self._repos.append(cached)
 
     def run(self):
         """Start a server from CGI environment.
 
         Modern servers should be using WSGI and should avoid this
         method, if possible.
         """
         if not encoding.environ.get(b'GATEWAY_INTERFACE', b'').startswith(
             b"CGI/1."
         ):
             raise RuntimeError(
                 b"This function is only intended to be "
                 b"called while running as a CGI script."
             )
         wsgicgi.launch(self)
 
     def __call__(self, env, respond):
         """Run the WSGI application.
 
         This may be called by multiple threads.
         """
         req = requestmod.parserequestfromenv(env)
         res = requestmod.wsgiresponse(req, respond)
 
         return self.run_wsgi(req, res)
 
     def run_wsgi(self, req, res):
         """Internal method to run the WSGI application.
 
         This is typically only called by Mercurial. External consumers
         should be using instances of this class as the WSGI application.
         """
         with self._obtainrepo() as repo:
             profile = repo.ui.configbool(b'profiling', b'enabled')
             with profiling.profile(repo.ui, enabled=profile):
                 for r in self._runwsgi(req, res, repo):
                     yield r
 
     def _runwsgi(self, req, res, repo):
         rctx = requestcontext(self, repo, req, res)
 
         # This state is global across all threads.
         encoding.encoding = rctx.config(b'web', b'encoding')
         rctx.repo.ui.environ = req.rawenv
 
         if rctx.csp:
             # hgwebdir may have added CSP header. Since we generate our own,
             # replace it.
             res.headers[b'Content-Security-Policy'] = rctx.csp
 
         handled = wireprotoserver.handlewsgirequest(
             rctx, req, res, self.check_perm
         )
         if handled:
             return res.sendresponse()
 
         # Old implementations of hgweb supported dispatching the request via
         # the initial query string parameter instead of using PATH_INFO.
         # If PATH_INFO is present (signaled by ``req.dispatchpath`` having
         # a value), we use it. Otherwise fall back to the query string.
         if req.dispatchpath is not None:
             query = req.dispatchpath
         else:
             query = req.querystring.partition(b'&')[0].partition(b';')[0]
 
         # translate user-visible url structure to internal structure
 
         args = query.split(b'/', 2)
         if b'cmd' not in req.qsparams and args and args[0]:
             cmd = args.pop(0)
             style = cmd.rfind(b'-')
             if style != -1:
                 req.qsparams[b'style'] = cmd[:style]
                 cmd = cmd[style + 1 :]
 
             # avoid accepting e.g. style parameter as command
             if util.safehasattr(webcommands, cmd):
                 req.qsparams[b'cmd'] = cmd
 
             if cmd == b'static':
                 req.qsparams[b'file'] = b'/'.join(args)
             else:
                 if args and args[0]:
                     node = args.pop(0).replace(b'%2F', b'/')
                     req.qsparams[b'node'] = node
                 if args:
                     if b'file' in req.qsparams:
                         del req.qsparams[b'file']
                     for a in args:
                         req.qsparams.add(b'file', a)
 
             ua = req.headers.get(b'User-Agent', b'')
             if cmd == b'rev' and b'mercurial' in ua:
                 req.qsparams[b'style'] = b'raw'
 
             if cmd == b'archive':
                 fn = req.qsparams[b'node']
                 for type_, spec in webutil.archivespecs.items():
                     ext = spec[2]
                     if fn.endswith(ext):
                         req.qsparams[b'node'] = fn[: -len(ext)]
                         req.qsparams[b'type'] = type_
         else:
             cmd = req.qsparams.get(b'cmd', b'')
 
         # process the web interface request
 
         try:
             rctx.tmpl = rctx.templater(req)
             ctype = rctx.tmpl.render(
                 b'mimetype', {b'encoding': encoding.encoding}
             )
 
             # check read permissions non-static content
             if cmd != b'static':
                 self.check_perm(rctx, req, None)
 
             if cmd == b'':
                 req.qsparams[b'cmd'] = rctx.tmpl.render(b'default', {})
                 cmd = req.qsparams[b'cmd']
 
             # Don't enable caching if using a CSP nonce because then it wouldn't
             # be a nonce.
             if rctx.configbool(b'web', b'cache') and not rctx.nonce:
                 tag = b'W/"%d"' % self.mtime
                 if req.headers.get(b'If-None-Match') == tag:
                     res.status = b'304 Not Modified'
                     # Content-Type may be defined globally. It isn't valid on a
                     # 304, so discard it.
                     try:
                         del res.headers[b'Content-Type']
                     except KeyError:
                         pass
                     # Response body not allowed on 304.
                     res.setbodybytes(b'')
                     return res.sendresponse()
 
                 res.headers[b'ETag'] = tag
 
             if cmd not in webcommands.__all__:
                 msg = b'no such method: %s' % cmd
                 raise ErrorResponse(HTTP_BAD_REQUEST, msg)
             else:
                 # Set some globals appropriate for web handlers. Commands can
                 # override easily enough.
                 res.status = b'200 Script output follows'
                 res.headers[b'Content-Type'] = ctype
                 return getattr(webcommands, cmd)(rctx)
 
         except (error.LookupError, error.RepoLookupError) as err:
             msg = pycompat.bytestr(err)
             if util.safehasattr(err, b'name') and not isinstance(
                 err, error.ManifestLookupError
             ):
                 msg = b'revision not found: %s' % err.name
 
             res.status = b'404 Not Found'
             res.headers[b'Content-Type'] = ctype
             return rctx.sendtemplate(b'error', error=msg)
         except (error.RepoError, error.StorageError) as e:
             res.status = b'500 Internal Server Error'
             res.headers[b'Content-Type'] = ctype
             return rctx.sendtemplate(b'error', error=pycompat.bytestr(e))
         except error.Abort as e:
             res.status = b'403 Forbidden'
             res.headers[b'Content-Type'] = ctype
             return rctx.sendtemplate(b'error', error=e.message)
         except ErrorResponse as e:
             for k, v in e.headers:
                 res.headers[k] = v
             res.status = statusmessage(e.code, pycompat.bytestr(e))
             res.headers[b'Content-Type'] = ctype
             return rctx.sendtemplate(b'error', error=pycompat.bytestr(e))
 
     def check_perm(self, rctx, req, op):
         for permhook in permhooks:
             permhook(rctx, req, op)
 
 
 def getwebview(repo):
     """The 'web.view' config controls changeset filter to hgweb. Possible
     values are ``served``, ``visible`` and ``all``. Default is ``served``.
     The ``served`` filter only shows changesets that can be pulled from the
     hgweb instance.  The``visible`` filter includes secret changesets but
     still excludes "hidden" one.
 
     See the repoview module for details.
 
     The option has been around undocumented since Mercurial 2.5, but no
     user ever asked about it. So we better keep it undocumented for now."""
     # experimental config: web.view
     viewconfig = repo.ui.config(b'web', b'view', untrusted=True)
     if viewconfig == b'all':
         return repo.unfiltered()
     elif viewconfig in repoview.filtertable:
         return repo.filtered(viewconfig)
     else:
         return repo.filtered(b'served')

mercurial/hgweb/request.py

@@ -1,632 +1,635 b''
 # hgweb/request.py - An http request from either CGI or the standalone server.
 #
 # Copyright 21 May 2005 - (c) 2005 Jake Edge <[e-mail]>
 # Copyright 2005, 2006 Olivia Mackall <[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.
 
 
 # import wsgiref.validate
 
 from ..thirdparty import attr
 from .. import (
     encoding,
     error,
     pycompat,
     util,
 )
 from ..utils import (
     urlutil,
 )
 
 
 class multidict:
     """A dict like object that can store multiple values for a key.
 
     Used to store parsed request parameters.
 
     This is inspired by WebOb's class of the same name.
     """
 
     def __init__(self):
         self._items = {}
 
     def __getitem__(self, key):
         """Returns the last set value for a key."""
         return self._items[key][-1]
 
     def __setitem__(self, key, value):
         """Replace a values for a key with a new value."""
         self._items[key] = [value]
 
     def __delitem__(self, key):
         """Delete all values for a key."""
         del self._items[key]
 
     def __contains__(self, key):
         return key in self._items
 
     def __len__(self):
         return len(self._items)
 
     def get(self, key, default=None):
         try:
             return self.__getitem__(key)
         except KeyError:
             return default
 
     def add(self, key, value):
         """Add a new value for a key. Does not replace existing values."""
         self._items.setdefault(key, []).append(value)
 
     def getall(self, key):
         """Obtains all values for a key."""
         return self._items.get(key, [])
 
     def getone(self, key):
         """Obtain a single value for a key.
 
         Raises KeyError if key not defined or it has multiple values set.
         """
         vals = self._items[key]
 
         if len(vals) > 1:
             raise KeyError(b'multiple values for %r' % key)
 
         return vals[0]
 
     def asdictoflists(self):
         return {k: list(v) for k, v in self._items.items()}
 
 
 @attr.s(frozen=True)
 class parsedrequest:
     """Represents a parsed WSGI request.
 
     Contains both parsed parameters as well as a handle on the input stream.
     """
 
     # Request method.
     method = attr.ib()
     # Full URL for this request.
     url = attr.ib()
     # URL without any path components. Just <proto>://<host><port>.
     baseurl = attr.ib()
     # Advertised URL. Like ``url`` and ``baseurl`` but uses SERVER_NAME instead
     # of HTTP: Host header for hostname. This is likely what clients used.
     advertisedurl = attr.ib()
     advertisedbaseurl = attr.ib()
     # URL scheme (part before ``://``). e.g. ``http`` or ``https``.
     urlscheme = attr.ib()
     # Value of REMOTE_USER, if set, or None.
     remoteuser = attr.ib()
     # Value of REMOTE_HOST, if set, or None.
     remotehost = attr.ib()
     # Relative WSGI application path. If defined, will begin with a
     # ``/``.
     apppath = attr.ib()
     # List of path parts to be used for dispatch.
     dispatchparts = attr.ib()
     # URL path component (no query string) used for dispatch. Can be
     # ``None`` to signal no path component given to the request, an
     # empty string to signal a request to the application's root URL,
     # or a string not beginning with ``/`` containing the requested
     # path under the application.
     dispatchpath = attr.ib()
     # The name of the repository being accessed.
     reponame = attr.ib()
     # Raw query string (part after "?" in URL).
     querystring = attr.ib()
     # multidict of query string parameters.
     qsparams = attr.ib()
     # wsgiref.headers.Headers instance. Operates like a dict with case
     # insensitive keys.
     headers = attr.ib()
     # Request body input stream.
     bodyfh = attr.ib()
     # WSGI environment dict, unmodified.
     rawenv = attr.ib()
 
 
 def parserequestfromenv(env, reponame=None, altbaseurl=None, bodyfh=None):
     """Parse URL components from environment variables.
 
     WSGI defines request attributes via environment variables. This function
     parses the environment variables into a data structure.
 
     If ``reponame`` is defined, the leading path components matching that
     string are effectively shifted from ``PATH_INFO`` to ``SCRIPT_NAME``.
     This simulates the world view of a WSGI application that processes
     requests from the base URL of a repo.
 
     If ``altbaseurl`` (typically comes from ``web.baseurl`` config option)
     is defined, it is used - instead of the WSGI environment variables - for
     constructing URL components up to and including the WSGI application path.
     For example, if the current WSGI application is at ``/repo`` and a request
     is made to ``/rev/@`` with this argument set to
     ``http://myserver:9000/prefix``, the URL and path components will resolve as
     if the request were to ``http://myserver:9000/prefix/rev/@``. In other
     words, ``wsgi.url_scheme``, ``SERVER_NAME``, ``SERVER_PORT``, and
     ``SCRIPT_NAME`` are all effectively replaced by components from this URL.
 
     ``bodyfh`` can be used to specify a file object to read the request body
     from. If not defined, ``wsgi.input`` from the environment dict is used.
     """
     # PEP 3333 defines the WSGI spec and is a useful reference for this code.
 
     # We first validate that the incoming object conforms with the WSGI spec.
     # We only want to be dealing with spec-conforming WSGI implementations.
     # TODO enable this once we fix internal violations.
     # wsgiref.validate.check_environ(env)
 
     # PEP-0333 states that environment keys and values are native strings.
     # The code points for the Unicode strings on Python 3 must be between
     # \00000-\000FF. We deal with bytes in Mercurial, so mass convert string
     # keys and values to bytes.
     def tobytes(s):
         if not isinstance(s, str):
             return s
         if pycompat.iswindows:
             # This is what mercurial.encoding does for os.environ on
             # Windows.
             return encoding.strtolocal(s)
         else:
             # This is what is documented to be used for os.environ on Unix.
             return pycompat.fsencode(s)
 
     env = {tobytes(k): tobytes(v) for k, v in env.items()}
 
     # Some hosting solutions are emulating hgwebdir, and dispatching directly
     # to an hgweb instance using this environment variable.  This was always
     # checked prior to d7fd203e36cc; keep doing so to avoid breaking them.
     if not reponame:
         reponame = env.get(b'REPO_NAME')
 
     if altbaseurl:
         altbaseurl = urlutil.url(altbaseurl)
 
     # https://www.python.org/dev/peps/pep-0333/#environ-variables defines
     # the environment variables.
     # https://www.python.org/dev/peps/pep-0333/#url-reconstruction defines
     # how URLs are reconstructed.
     fullurl = env[b'wsgi.url_scheme'] + b'://'
 
     if altbaseurl and altbaseurl.scheme:
         advertisedfullurl = altbaseurl.scheme + b'://'
     else:
         advertisedfullurl = fullurl
 
     def addport(s, port):
         if s.startswith(b'https://'):
             if port != b'443':
                 s += b':' + port
         else:
             if port != b'80':
                 s += b':' + port
 
         return s
 
     if env.get(b'HTTP_HOST'):
         fullurl += env[b'HTTP_HOST']
     else:
         fullurl += env[b'SERVER_NAME']
         fullurl = addport(fullurl, env[b'SERVER_PORT'])
 
     if altbaseurl and altbaseurl.host:
         advertisedfullurl += altbaseurl.host
 
         if altbaseurl.port:
             port = altbaseurl.port
         elif altbaseurl.scheme == b'http' and not altbaseurl.port:
             port = b'80'
         elif altbaseurl.scheme == b'https' and not altbaseurl.port:
             port = b'443'
         else:
             port = env[b'SERVER_PORT']
 
         advertisedfullurl = addport(advertisedfullurl, port)
     else:
         advertisedfullurl += env[b'SERVER_NAME']
         advertisedfullurl = addport(advertisedfullurl, env[b'SERVER_PORT'])
 
     baseurl = fullurl
     advertisedbaseurl = advertisedfullurl
 
     fullurl += util.urlreq.quote(env.get(b'SCRIPT_NAME', b''))
     fullurl += util.urlreq.quote(env.get(b'PATH_INFO', b''))
 
     if altbaseurl:
         path = altbaseurl.path or b''
         if path and not path.startswith(b'/'):
             path = b'/' + path
         advertisedfullurl += util.urlreq.quote(path)
     else:
         advertisedfullurl += util.urlreq.quote(env.get(b'SCRIPT_NAME', b''))
 
     advertisedfullurl += util.urlreq.quote(env.get(b'PATH_INFO', b''))
 
     if env.get(b'QUERY_STRING'):
         fullurl += b'?' + env[b'QUERY_STRING']
         advertisedfullurl += b'?' + env[b'QUERY_STRING']
 
     # If ``reponame`` is defined, that must be a prefix on PATH_INFO
     # that represents the repository being dispatched to. When computing
     # the dispatch info, we ignore these leading path components.
 
     if altbaseurl:
         apppath = altbaseurl.path or b''
         if apppath and not apppath.startswith(b'/'):
             apppath = b'/' + apppath
     else:
         apppath = env.get(b'SCRIPT_NAME', b'')
 
     if reponame:
         repoprefix = b'/' + reponame.strip(b'/')
 
         if not env.get(b'PATH_INFO'):
             raise error.ProgrammingError(b'reponame requires PATH_INFO')
 
         if not env[b'PATH_INFO'].startswith(repoprefix):
             raise error.ProgrammingError(
                 b'PATH_INFO does not begin with repo '
                 b'name: %s (%s)' % (env[b'PATH_INFO'], reponame)
             )
 
         dispatchpath = env[b'PATH_INFO'][len(repoprefix) :]
 
         if dispatchpath and not dispatchpath.startswith(b'/'):
             raise error.ProgrammingError(
                 b'reponame prefix of PATH_INFO does '
                 b'not end at path delimiter: %s (%s)'
                 % (env[b'PATH_INFO'], reponame)
             )
 
         apppath = apppath.rstrip(b'/') + repoprefix
         dispatchparts = dispatchpath.strip(b'/').split(b'/')
         dispatchpath = b'/'.join(dispatchparts)
 
     elif b'PATH_INFO' in env:
         if env[b'PATH_INFO'].strip(b'/'):
             dispatchparts = env[b'PATH_INFO'].strip(b'/').split(b'/')
             dispatchpath = b'/'.join(dispatchparts)
         else:
             dispatchparts = []
             dispatchpath = b''
     else:
         dispatchparts = []
         dispatchpath = None
 
     querystring = env.get(b'QUERY_STRING', b'')
 
     # We store as a list so we have ordering information. We also store as
     # a dict to facilitate fast lookup.
     qsparams = multidict()
     for k, v in util.urlreq.parseqsl(querystring, keep_blank_values=True):
         qsparams.add(k, v)
 
     # HTTP_* keys contain HTTP request headers. The Headers structure should
     # perform case normalization for us. We just rewrite underscore to dash
     # so keys match what likely went over the wire.
     headers = []
     for k, v in env.items():
         if k.startswith(b'HTTP_'):
             headers.append((k[len(b'HTTP_') :].replace(b'_', b'-'), v))
 
     from . import wsgiheaders  # avoid cycle
 
     headers = wsgiheaders.Headers(headers)
 
     # This is kind of a lie because the HTTP header wasn't explicitly
     # sent. But for all intents and purposes it should be OK to lie about
     # this, since a consumer will either either value to determine how many
     # bytes are available to read.
     if b'CONTENT_LENGTH' in env and b'HTTP_CONTENT_LENGTH' not in env:
         headers[b'Content-Length'] = env[b'CONTENT_LENGTH']
 
     if b'CONTENT_TYPE' in env and b'HTTP_CONTENT_TYPE' not in env:
         headers[b'Content-Type'] = env[b'CONTENT_TYPE']
 
     if bodyfh is None:
         bodyfh = env[b'wsgi.input']
         if b'Content-Length' in headers:
             bodyfh = util.cappedreader(
                 bodyfh, int(headers[b'Content-Length'] or b'0')
             )
 
     return parsedrequest(
         method=env[b'REQUEST_METHOD'],
         url=fullurl,
         baseurl=baseurl,
         advertisedurl=advertisedfullurl,
         advertisedbaseurl=advertisedbaseurl,
         urlscheme=env[b'wsgi.url_scheme'],
         remoteuser=env.get(b'REMOTE_USER'),
         remotehost=env.get(b'REMOTE_HOST'),
         apppath=apppath,
         dispatchparts=dispatchparts,
         dispatchpath=dispatchpath,
         reponame=reponame,
         querystring=querystring,
         qsparams=qsparams,
         headers=headers,
         bodyfh=bodyfh,
         rawenv=env,
     )
 
 
 class offsettrackingwriter:
     """A file object like object that is append only and tracks write count.
 
     Instances are bound to a callable. This callable is called with data
     whenever a ``write()`` is attempted.
 
     Instances track the amount of written data so they can answer ``tell()``
     requests.
 
     The intent of this class is to wrap the ``write()`` function returned by
     a WSGI ``start_response()`` function. Since ``write()`` is a callable and
     not a file object, it doesn't implement other file object methods.
     """
 
     def __init__(self, writefn):
         self._write = writefn
         self._offset = 0
 
     def write(self, s):
         res = self._write(s)
         # Some Python objects don't report the number of bytes written.
         if res is None:
             self._offset += len(s)
         else:
             self._offset += res
 
     def flush(self):
         pass
 
     def tell(self):
         return self._offset
 
 
 class wsgiresponse:
     """Represents a response to a WSGI request.
 
     A response consists of a status line, headers, and a body.
 
     Consumers must populate the ``status`` and ``headers`` fields and
     make a call to a ``setbody*()`` method before the response can be
     issued.
 
     When it is time to start sending the response over the wire,
     ``sendresponse()`` is called. It handles emitting the header portion
     of the response message. It then yields chunks of body data to be
     written to the peer. Typically, the WSGI application itself calls
     and returns the value from ``sendresponse()``.
     """
 
     def __init__(self, req, startresponse):
         """Create an empty response tied to a specific request.
 
         ``req`` is a ``parsedrequest``. ``startresponse`` is the
         ``start_response`` function passed to the WSGI application.
         """
         self._req = req
         self._startresponse = startresponse
 
         self.status = None
         from . import wsgiheaders  # avoid cycle
 
         self.headers = wsgiheaders.Headers([])
 
         self._bodybytes = None
         self._bodygen = None
         self._bodywillwrite = False
         self._started = False
         self._bodywritefn = None
 
     def _verifybody(self):
         if (
             self._bodybytes is not None
             or self._bodygen is not None
             or self._bodywillwrite
         ):
             raise error.ProgrammingError(b'cannot define body multiple times')
 
     def setbodybytes(self, b):
         """Define the response body as static bytes.
 
         The empty string signals that there is no response body.
         """
         self._verifybody()
         self._bodybytes = b
         self.headers[b'Content-Length'] = b'%d' % len(b)
 
     def setbodygen(self, gen):
         """Define the response body as a generator of bytes."""
         self._verifybody()
         self._bodygen = gen
 
     def setbodywillwrite(self):
         """Signal an intent to use write() to emit the response body.
 
         **This is the least preferred way to send a body.**
 
         It is preferred for WSGI applications to emit a generator of chunks
         constituting the response body. However, some consumers can't emit
         data this way. So, WSGI provides a way to obtain a ``write(data)``
         function that can be used to synchronously perform an unbuffered
         write.
 
         Calling this function signals an intent to produce the body in this
         manner.
         """
         self._verifybody()
         self._bodywillwrite = True
 
     def sendresponse(self):
         """Send the generated response to the client.
 
         Before this is called, ``status`` must be set and one of
         ``setbodybytes()`` or ``setbodygen()`` must be called.
 
         Calling this method multiple times is not allowed.
         """
         if self._started:
             raise error.ProgrammingError(
                 b'sendresponse() called multiple times'
             )
 
         self._started = True
 
         if not self.status:
             raise error.ProgrammingError(b'status line not defined')
 
         if (
             self._bodybytes is None
             and self._bodygen is None
             and not self._bodywillwrite
+            and self._req.method != b'HEAD'
         ):
             raise error.ProgrammingError(b'response body not defined')
 
         # RFC 7232 Section 4.1 states that a 304 MUST generate one of
         # {Cache-Control, Content-Location, Date, ETag, Expires, Vary}
         # and SHOULD NOT generate other headers unless they could be used
         # to guide cache updates. Furthermore, RFC 7230 Section 3.3.2
         # states that no response body can be issued. Content-Length can
         # be sent. But if it is present, it should be the size of the response
         # that wasn't transferred.
         if self.status.startswith(b'304 '):
             # setbodybytes('') will set C-L to 0. This doesn't conform with the
             # spec. So remove it.
             if self.headers.get(b'Content-Length') == b'0':
                 del self.headers[b'Content-Length']
 
             # Strictly speaking, this is too strict. But until it causes
             # problems, let's be strict.
             badheaders = {
                 k
                 for k in self.headers.keys()
                 if k.lower()
                 not in (
                     b'date',
                     b'etag',
                     b'expires',
                     b'cache-control',
                     b'content-location',
                     b'content-security-policy',
                     b'vary',
                 )
             }
             if badheaders:
                 raise error.ProgrammingError(
                     b'illegal header on 304 response: %s'
                     % b', '.join(sorted(badheaders))
                 )
 
             if self._bodygen is not None or self._bodywillwrite:
                 raise error.ProgrammingError(
                     b"must use setbodybytes('') with 304 responses"
                 )
 
         # Various HTTP clients (notably httplib) won't read the HTTP response
         # until the HTTP request has been sent in full. If servers (us) send a
         # response before the HTTP request has been fully sent, the connection
         # may deadlock because neither end is reading.
         #
         # We work around this by "draining" the request data before
         # sending any response in some conditions.
         drain = False
         close = False
 
         # If the client sent Expect: 100-continue, we assume it is smart enough
         # to deal with the server sending a response before reading the request.
         # (httplib doesn't do this.)
         if self._req.headers.get(b'Expect', b'').lower() == b'100-continue':
             pass
         # Only tend to request methods that have bodies. Strictly speaking,
         # we should sniff for a body. But this is fine for our existing
         # WSGI applications.
         elif self._req.method not in (b'POST', b'PUT'):
             pass
         else:
             # If we don't know how much data to read, there's no guarantee
             # that we can drain the request responsibly. The WSGI
             # specification only says that servers *should* ensure the
             # input stream doesn't overrun the actual request. So there's
             # no guarantee that reading until EOF won't corrupt the stream
             # state.
             if not isinstance(self._req.bodyfh, util.cappedreader):
                 close = True
             else:
                 # We /could/ only drain certain HTTP response codes. But 200 and
                 # non-200 wire protocol responses both require draining. Since
                 # we have a capped reader in place for all situations where we
                 # drain, it is safe to read from that stream. We'll either do
                 # a drain or no-op if we're already at EOF.
                 drain = True
 
         if close:
             self.headers[b'Connection'] = b'Close'
 
         if drain:
             assert isinstance(self._req.bodyfh, util.cappedreader)
             while True:
                 chunk = self._req.bodyfh.read(32768)
                 if not chunk:
                     break
 
         strheaders = [
             (pycompat.strurl(k), pycompat.strurl(v))
             for k, v in self.headers.items()
         ]
         write = self._startresponse(pycompat.sysstr(self.status), strheaders)
 
         if self._bodybytes:
             yield self._bodybytes
         elif self._bodygen:
             for chunk in self._bodygen:
                 # PEP-3333 says that output must be bytes. And some WSGI
                 # implementations enforce this. We cast bytes-like types here
                 # for convenience.
                 if isinstance(chunk, bytearray):
                     chunk = bytes(chunk)
 
                 yield chunk
         elif self._bodywillwrite:
             self._bodywritefn = write
+        elif self._req.method == b'HEAD':
+            pass
         else:
             error.ProgrammingError(b'do not know how to send body')
 
     def getbodyfile(self):
         """Obtain a file object like object representing the response body.
 
         For this to work, you must call ``setbodywillwrite()`` and then
         ``sendresponse()`` first. ``sendresponse()`` is a generator and the
         function won't run to completion unless the generator is advanced. The
         generator yields not items. The easiest way to consume it is with
         ``list(res.sendresponse())``, which should resolve to an empty list -
         ``[]``.
         """
         if not self._bodywillwrite:
             raise error.ProgrammingError(b'must call setbodywillwrite() first')
 
         if not self._started:
             raise error.ProgrammingError(
                 b'must call sendresponse() first; did '
                 b'you remember to consume it since it '
                 b'is a generator?'
             )
 
         assert self._bodywritefn
         return offsettrackingwriter(self._bodywritefn)
 
 
 def wsgiapplication(app_maker):
     """For compatibility with old CGI scripts. A plain hgweb() or hgwebdir()
     can and should now be used as a WSGI application."""
     application = app_maker()
 
     def run_wsgi(env, respond):
         return application(env, respond)
 
     return run_wsgi

mercurial/hgweb/server.py

@@ -1,417 +1,424 b''
 # hgweb/server.py - The standalone hg web server.
 #
 # Copyright 21 May 2005 - (c) 2005 Jake Edge <[e-mail]>
 # Copyright 2005-2007 Olivia Mackall <[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.
 
 
 import errno
 import os
 import socket
 import sys
 import traceback
 import wsgiref.validate
 
 from ..i18n import _
 from ..pycompat import (
     getattr,
     open,
 )
 
 from .. import (
     encoding,
     error,
     pycompat,
     util,
 )
 from ..utils import (
     urlutil,
 )
 
 httpservermod = util.httpserver
 socketserver = util.socketserver
 urlerr = util.urlerr
 urlreq = util.urlreq
 
 from . import common
 
 
 def _splitURI(uri):
     """Return path and query that has been split from uri
 
     Just like CGI environment, the path is unquoted, the query is
     not.
     """
     if '?' in uri:
         path, query = uri.split('?', 1)
     else:
         path, query = uri, r''
     return urlreq.unquote(path), query
 
 
 class _error_logger:
     def __init__(self, handler):
         self.handler = handler
 
     def flush(self):
         pass
 
     def write(self, str):
         self.writelines(str.split(b'\n'))
 
     def writelines(self, seq):
         for msg in seq:
             self.handler.log_error("HG error:  %s", encoding.strfromlocal(msg))
 
 
 class _httprequesthandler(httpservermod.basehttprequesthandler):
 
     url_scheme = b'http'
 
     @staticmethod
     def preparehttpserver(httpserver, ui):
         """Prepare .socket of new HTTPServer instance"""
 
     def __init__(self, *args, **kargs):
         self.protocol_version = r'HTTP/1.1'
         httpservermod.basehttprequesthandler.__init__(self, *args, **kargs)
 
     def _log_any(self, fp, format, *args):
         fp.write(
             pycompat.sysbytes(
                 r"%s - - [%s] %s"
                 % (
                     self.client_address[0],
                     self.log_date_time_string(),
                     format % args,
                 )
             )
             + b'\n'
         )
         fp.flush()
 
     def log_error(self, format, *args):
         self._log_any(self.server.errorlog, format, *args)
 
     def log_message(self, format, *args):
         self._log_any(self.server.accesslog, format, *args)
 
     def log_request(self, code='-', size='-'):
         xheaders = []
         if util.safehasattr(self, b'headers'):
             xheaders = [
                 h for h in self.headers.items() if h[0].startswith('x-')
             ]
         self.log_message(
             '"%s" %s %s%s',
             self.requestline,
             str(code),
             str(size),
             ''.join([' %s:%s' % h for h in sorted(xheaders)]),
         )
 
     def do_write(self):
         try:
             self.do_hgweb()
         except BrokenPipeError:
             pass
 
     def do_POST(self):
         try:
             self.do_write()
         except Exception as e:
             # I/O below could raise another exception. So log the original
             # exception first to ensure it is recorded.
             if not (
                 isinstance(e, (OSError, socket.error))
                 and e.errno == errno.ECONNRESET
             ):
                 tb = "".join(traceback.format_exception(*sys.exc_info()))
                 # We need a native-string newline to poke in the log
                 # message, because we won't get a newline when using an
                 # r-string. This is the easy way out.
                 newline = chr(10)
                 self.log_error(
                     r"Exception happened during processing "
                     "request '%s':%s%s",
                     self.path,
                     newline,
                     tb,
                 )
 
             self._start_response("500 Internal Server Error", [])
             self._write(b"Internal Server Error")
             self._done()
 
     def do_PUT(self):
         self.do_POST()
 
     def do_GET(self):
         self.do_POST()
 
+    def do_HEAD(self):
+        self.do_POST()
+
     def do_hgweb(self):
         self.sent_headers = False
         path, query = _splitURI(self.path)
 
         # Ensure the slicing of path below is valid
         if path != self.server.prefix and not path.startswith(
             self.server.prefix + b'/'
         ):
             self._start_response(pycompat.strurl(common.statusmessage(404)), [])
             if self.command == 'POST':
                 # Paranoia: tell the client we're going to close the
                 # socket so they don't try and reuse a socket that
                 # might have a POST body waiting to confuse us. We do
                 # this by directly munging self.saved_headers because
                 # self._start_response ignores Connection headers.
                 self.saved_headers = [('Connection', 'Close')]
             self._write(b"Not Found")
             self._done()
             return
 
         env = {}
         env['GATEWAY_INTERFACE'] = 'CGI/1.1'
         env['REQUEST_METHOD'] = self.command
         env['SERVER_NAME'] = self.server.server_name
         env['SERVER_PORT'] = str(self.server.server_port)
         env['REQUEST_URI'] = self.path
         env['SCRIPT_NAME'] = pycompat.sysstr(self.server.prefix)
         env['PATH_INFO'] = pycompat.sysstr(path[len(self.server.prefix) :])
         env['REMOTE_HOST'] = self.client_address[0]
         env['REMOTE_ADDR'] = self.client_address[0]
         env['QUERY_STRING'] = query or ''
 
         if self.headers.get_content_type() is None:
             env['CONTENT_TYPE'] = self.headers.get_default_type()
         else:
             env['CONTENT_TYPE'] = self.headers.get_content_type()
         length = self.headers.get('content-length')
         if length:
             env['CONTENT_LENGTH'] = length
         for header in [
             h
             for h in self.headers.keys()
             if h.lower() not in ('content-type', 'content-length')
         ]:
             hkey = 'HTTP_' + header.replace('-', '_').upper()
             hval = self.headers.get(header)
             hval = hval.replace('\n', '').strip()
             if hval:
                 env[hkey] = hval
         env['SERVER_PROTOCOL'] = self.request_version
         env['wsgi.version'] = (1, 0)
         env['wsgi.url_scheme'] = pycompat.sysstr(self.url_scheme)
         if env.get('HTTP_EXPECT', b'').lower() == b'100-continue':
             self.rfile = common.continuereader(self.rfile, self.wfile.write)
 
         env['wsgi.input'] = self.rfile
         env['wsgi.errors'] = _error_logger(self)
         env['wsgi.multithread'] = isinstance(
             self.server, socketserver.ThreadingMixIn
         )
         if util.safehasattr(socketserver, b'ForkingMixIn'):
             env['wsgi.multiprocess'] = isinstance(
                 self.server, socketserver.ForkingMixIn
             )
         else:
             env['wsgi.multiprocess'] = False
 
         env['wsgi.run_once'] = 0
 
         wsgiref.validate.check_environ(env)
 
         self.saved_status = None
         self.saved_headers = []
         self.length = None
         self._chunked = None
         for chunk in self.server.application(env, self._start_response):
             self._write(chunk)
         if not self.sent_headers:
             self.send_headers()
         self._done()
 
     def send_headers(self):
         if not self.saved_status:
             raise AssertionError(
                 b"Sending headers before start_response() called"
             )
         saved_status = self.saved_status.split(None, 1)
         saved_status[0] = int(saved_status[0])
         self.send_response(*saved_status)
         self.length = None
         self._chunked = False
         for h in self.saved_headers:
             self.send_header(*h)
             if h[0].lower() == 'content-length':
                 self.length = int(h[1])
-        if self.length is None and saved_status[0] != common.HTTP_NOT_MODIFIED:
+        if (
+            self.length is None
+            and saved_status[0] != common.HTTP_NOT_MODIFIED
+            and self.command != 'HEAD'
+        ):
             self._chunked = (
                 not self.close_connection and self.request_version == 'HTTP/1.1'
             )
             if self._chunked:
                 self.send_header('Transfer-Encoding', 'chunked')
             else:
                 self.send_header('Connection', 'close')
         self.end_headers()
         self.sent_headers = True
 
     def _start_response(self, http_status, headers, exc_info=None):
         assert isinstance(http_status, str)
         code, msg = http_status.split(None, 1)
         code = int(code)
         self.saved_status = http_status
         bad_headers = ('connection', 'transfer-encoding')
         self.saved_headers = [
             h for h in headers if h[0].lower() not in bad_headers
         ]
         return self._write
 
     def _write(self, data):
         if not self.saved_status:
             raise AssertionError(b"data written before start_response() called")
         elif not self.sent_headers:
             self.send_headers()
         if self.length is not None:
             if len(data) > self.length:
                 raise AssertionError(
                     b"Content-length header sent, but more "
                     b"bytes than specified are being written."
                 )
             self.length = self.length - len(data)
         elif self._chunked and data:
             data = b'%x\r\n%s\r\n' % (len(data), data)
         self.wfile.write(data)
         self.wfile.flush()
 
     def _done(self):
         if self._chunked:
             self.wfile.write(b'0\r\n\r\n')
             self.wfile.flush()
 
     def version_string(self):
         if self.server.serverheader:
             return encoding.strfromlocal(self.server.serverheader)
         return httpservermod.basehttprequesthandler.version_string(self)
 
 
 class _httprequesthandlerssl(_httprequesthandler):
     """HTTPS handler based on Python's ssl module"""
 
     url_scheme = b'https'
 
     @staticmethod
     def preparehttpserver(httpserver, ui):
         try:
             from .. import sslutil
 
             sslutil.wrapserversocket
         except ImportError:
             raise error.Abort(_(b"SSL support is unavailable"))
 
         certfile = ui.config(b'web', b'certificate')
 
         # These config options are currently only meant for testing. Use
         # at your own risk.
         cafile = ui.config(b'devel', b'servercafile')
         reqcert = ui.configbool(b'devel', b'serverrequirecert')
 
         httpserver.socket = sslutil.wrapserversocket(
             httpserver.socket,
             ui,
             certfile=certfile,
             cafile=cafile,
             requireclientcert=reqcert,
         )
 
     def setup(self):
         self.connection = self.request
         self.rfile = self.request.makefile("rb", self.rbufsize)
         self.wfile = self.request.makefile("wb", self.wbufsize)
 
 
 try:
     import threading
 
     threading.active_count()  # silence pyflakes and bypass demandimport
     _mixin = socketserver.ThreadingMixIn
 except ImportError:
     if util.safehasattr(os, b"fork"):
         _mixin = socketserver.ForkingMixIn
     else:
 
         class _mixin:
             pass
 
 
 def openlog(opt, default):
     if opt and opt != b'-':
         return open(opt, b'ab')
     return default
 
 
 class MercurialHTTPServer(_mixin, httpservermod.httpserver, object):
 
     # SO_REUSEADDR has broken semantics on windows
     if pycompat.iswindows:
         allow_reuse_address = 0
 
     def __init__(self, ui, app, addr, handler, **kwargs):
         httpservermod.httpserver.__init__(self, addr, handler, **kwargs)
         self.daemon_threads = True
         self.application = app
 
         handler.preparehttpserver(self, ui)
 
         prefix = ui.config(b'web', b'prefix')
         if prefix:
             prefix = b'/' + prefix.strip(b'/')
         self.prefix = prefix
 
         alog = openlog(ui.config(b'web', b'accesslog'), ui.fout)
         elog = openlog(ui.config(b'web', b'errorlog'), ui.ferr)
         self.accesslog = alog
         self.errorlog = elog
 
         self.addr, self.port = self.socket.getsockname()[0:2]
         self.fqaddr = self.server_name
 
         self.serverheader = ui.config(b'web', b'server-header')
 
 
 class IPv6HTTPServer(MercurialHTTPServer):
     address_family = getattr(socket, 'AF_INET6', None)
 
     def __init__(self, *args, **kwargs):
         if self.address_family is None:
             raise error.RepoError(_(b'IPv6 is not available on this system'))
         super(IPv6HTTPServer, self).__init__(*args, **kwargs)
 
 
 def create_server(ui, app):
 
     if ui.config(b'web', b'certificate'):
         handler = _httprequesthandlerssl
     else:
         handler = _httprequesthandler
 
     if ui.configbool(b'web', b'ipv6'):
         cls = IPv6HTTPServer
     else:
         cls = MercurialHTTPServer
 
     # ugly hack due to python issue5853 (for threaded use)
     import mimetypes
 
     mimetypes.init()
 
     address = ui.config(b'web', b'address')
     port = urlutil.getport(ui.config(b'web', b'port'))
     try:
         return cls(ui, app, (address, port), handler)
     except socket.error as inst:
         raise error.Abort(
             _(b"cannot start server at '%s:%d': %s")
             % (address, port, encoding.strtolocal(inst.args[1]))
         )

mercurial/hgweb/webcommands.py

@@ -1,1594 +1,1597 b''
 #
 # Copyright 21 May 2005 - (c) 2005 Jake Edge <[e-mail]>
 # Copyright 2005-2007 Olivia Mackall <[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.
 
 
 import copy
 import mimetypes
 import os
 import re
 
 from ..i18n import _
 from ..node import hex, short
 from ..pycompat import getattr
 
 from .common import (
     ErrorResponse,
     HTTP_FORBIDDEN,
     HTTP_NOT_FOUND,
     get_contact,
     paritygen,
     staticfile,
 )
 
 from .. import (
     archival,
     dagop,
     encoding,
     error,
     graphmod,
     pycompat,
     revset,
     revsetlang,
     scmutil,
     smartset,
     templateutil,
 )
 
 from ..utils import stringutil
 
 from . import webutil
 
 __all__ = []
 commands = {}
 
 
 class webcommand:
     """Decorator used to register a web command handler.
 
     The decorator takes as its positional arguments the name/path the
     command should be accessible under.
 
     When called, functions receive as arguments a ``requestcontext``,
     ``wsgirequest``, and a templater instance for generatoring output.
     The functions should populate the ``rctx.res`` object with details
     about the HTTP response.
 
     The function returns a generator to be consumed by the WSGI application.
     For most commands, this should be the result from
     ``web.res.sendresponse()``. Many commands will call ``web.sendtemplate()``
     to render a template.
 
     Usage:
 
     @webcommand('mycommand')
     def mycommand(web):
         pass
     """
 
     def __init__(self, name):
         self.name = name
 
     def __call__(self, func):
         __all__.append(self.name)
         commands[self.name] = func
         return func
 
 
 @webcommand(b'log')
 def log(web):
     """
     /log[/{revision}[/{path}]]
     --------------------------
 
     Show repository or file history.
 
     For URLs of the form ``/log/{revision}``, a list of changesets starting at
     the specified changeset identifier is shown. If ``{revision}`` is not
     defined, the default is ``tip``. This form is equivalent to the
     ``changelog`` handler.
 
     For URLs of the form ``/log/{revision}/{file}``, the history for a specific
     file will be shown. This form is equivalent to the ``filelog`` handler.
     """
 
     if web.req.qsparams.get(b'file'):
         return filelog(web)
     else:
         return changelog(web)
 
 
 @webcommand(b'rawfile')
 def rawfile(web):
     guessmime = web.configbool(b'web', b'guessmime')
 
     path = webutil.cleanpath(web.repo, web.req.qsparams.get(b'file', b''))
     if not path:
         return manifest(web)
 
     try:
         fctx = webutil.filectx(web.repo, web.req)
     except error.LookupError as inst:
         try:
             return manifest(web)
         except ErrorResponse:
             raise inst
 
     path = fctx.path()
     text = fctx.data()
     mt = b'application/binary'
     if guessmime:
         mt = mimetypes.guess_type(pycompat.fsdecode(path))[0]
         if mt is None:
             if stringutil.binary(text):
                 mt = b'application/binary'
             else:
                 mt = b'text/plain'
         else:
             mt = pycompat.sysbytes(mt)
 
     if mt.startswith(b'text/'):
         mt += b'; charset="%s"' % encoding.encoding
 
     web.res.headers[b'Content-Type'] = mt
     filename = (
         path.rpartition(b'/')[-1].replace(b'\\', b'\\\\').replace(b'"', b'\\"')
     )
     web.res.headers[b'Content-Disposition'] = (
         b'inline; filename="%s"' % filename
     )
     web.res.setbodybytes(text)
     return web.res.sendresponse()
 
 
 def _filerevision(web, fctx):
     f = fctx.path()
     text = fctx.data()
     parity = paritygen(web.stripecount)
     ishead = fctx.filenode() in fctx.filelog().heads()
 
     if stringutil.binary(text):
         mt = pycompat.sysbytes(
             mimetypes.guess_type(pycompat.fsdecode(f))[0]
             or r'application/octet-stream'
         )
         text = b'(binary:%s)' % mt
 
     def lines(context):
         for lineno, t in enumerate(text.splitlines(True)):
             yield {
                 b"line": t,
                 b"lineid": b"l%d" % (lineno + 1),
                 b"linenumber": b"% 6d" % (lineno + 1),
                 b"parity": next(parity),
             }
 
     return web.sendtemplate(
         b'filerevision',
         file=f,
         path=webutil.up(f),
         text=templateutil.mappinggenerator(lines),
         symrev=webutil.symrevorshortnode(web.req, fctx),
         rename=webutil.renamelink(fctx),
         permissions=fctx.manifest().flags(f),
         ishead=int(ishead),
         **pycompat.strkwargs(webutil.commonentry(web.repo, fctx))
     )
 
 
 @webcommand(b'file')
 def file(web):
     """
     /file/{revision}[/{path}]
     -------------------------
 
     Show information about a directory or file in the repository.
 
     Info about the ``path`` given as a URL parameter will be rendered.
 
     If ``path`` is a directory, information about the entries in that
     directory will be rendered. This form is equivalent to the ``manifest``
     handler.
 
     If ``path`` is a file, information about that file will be shown via
     the ``filerevision`` template.
 
     If ``path`` is not defined, information about the root directory will
     be rendered.
     """
     if web.req.qsparams.get(b'style') == b'raw':
         return rawfile(web)
 
     path = webutil.cleanpath(web.repo, web.req.qsparams.get(b'file', b''))
     if not path:
         return manifest(web)
     try:
         return _filerevision(web, webutil.filectx(web.repo, web.req))
     except error.LookupError as inst:
         try:
             return manifest(web)
         except ErrorResponse:
             raise inst
 
 
 def _search(web):
     MODE_REVISION = b'rev'
     MODE_KEYWORD = b'keyword'
     MODE_REVSET = b'revset'
 
     def revsearch(ctx):
         yield ctx
 
     def keywordsearch(query):
         lower = encoding.lower
         qw = lower(query).split()
 
         def revgen():
             cl = web.repo.changelog
             for i in range(len(web.repo) - 1, 0, -100):
                 l = []
                 for j in cl.revs(max(0, i - 99), i):
                     ctx = web.repo[j]
                     l.append(ctx)
                 l.reverse()
                 for e in l:
                     yield e
 
         for ctx in revgen():
             miss = 0
             for q in qw:
                 if not (
                     q in lower(ctx.user())
                     or q in lower(ctx.description())
                     or q in lower(b" ".join(ctx.files()))
                 ):
                     miss = 1
                     break
             if miss:
                 continue
 
             yield ctx
 
     def revsetsearch(revs):
         for r in revs:
             yield web.repo[r]
 
     searchfuncs = {
         MODE_REVISION: (revsearch, b'exact revision search'),
         MODE_KEYWORD: (keywordsearch, b'literal keyword search'),
         MODE_REVSET: (revsetsearch, b'revset expression search'),
     }
 
     def getsearchmode(query):
         try:
             ctx = scmutil.revsymbol(web.repo, query)
         except (error.RepoError, error.LookupError):
             # query is not an exact revision pointer, need to
             # decide if it's a revset expression or keywords
             pass
         else:
             return MODE_REVISION, ctx
 
         revdef = b'reverse(%s)' % query
         try:
             tree = revsetlang.parse(revdef)
         except error.ParseError:
             # can't parse to a revset tree
             return MODE_KEYWORD, query
 
         if revsetlang.depth(tree) <= 2:
             # no revset syntax used
             return MODE_KEYWORD, query
 
         if any(
             (token, (value or b'')[:3]) == (b'string', b're:')
             for token, value, pos in revsetlang.tokenize(revdef)
         ):
             return MODE_KEYWORD, query
 
         funcsused = revsetlang.funcsused(tree)
         if not funcsused.issubset(revset.safesymbols):
             return MODE_KEYWORD, query
 
         try:
             mfunc = revset.match(
                 web.repo.ui, revdef, lookup=revset.lookupfn(web.repo)
             )
             revs = mfunc(web.repo)
             return MODE_REVSET, revs
             # ParseError: wrongly placed tokens, wrongs arguments, etc
             # RepoLookupError: no such revision, e.g. in 'revision:'
             # Abort: bookmark/tag not exists
             # LookupError: ambiguous identifier, e.g. in '(bc)' on a large repo
         except (
             error.ParseError,
             error.RepoLookupError,
             error.Abort,
             LookupError,
         ):
             return MODE_KEYWORD, query
 
     def changelist(context):
         count = 0
 
         for ctx in searchfunc[0](funcarg):
             count += 1
             n = scmutil.binnode(ctx)
             showtags = webutil.showtag(web.repo, b'changelogtag', n)
             files = webutil.listfilediffs(ctx.files(), n, web.maxfiles)
 
             lm = webutil.commonentry(web.repo, ctx)
             lm.update(
                 {
                     b'parity': next(parity),
                     b'changelogtag': showtags,
                     b'files': files,
                 }
             )
             yield lm
 
             if count >= revcount:
                 break
 
     query = web.req.qsparams[b'rev']
     revcount = web.maxchanges
     if b'revcount' in web.req.qsparams:
         try:
             revcount = int(web.req.qsparams.get(b'revcount', revcount))
             revcount = max(revcount, 1)
             web.tmpl.defaults[b'sessionvars'][b'revcount'] = revcount
         except ValueError:
             pass
 
     lessvars = copy.copy(web.tmpl.defaults[b'sessionvars'])
     lessvars[b'revcount'] = max(revcount // 2, 1)
     lessvars[b'rev'] = query
     morevars = copy.copy(web.tmpl.defaults[b'sessionvars'])
     morevars[b'revcount'] = revcount * 2
     morevars[b'rev'] = query
 
     mode, funcarg = getsearchmode(query)
 
     if b'forcekw' in web.req.qsparams:
         showforcekw = b''
         showunforcekw = searchfuncs[mode][1]
         mode = MODE_KEYWORD
         funcarg = query
     else:
         if mode != MODE_KEYWORD:
             showforcekw = searchfuncs[MODE_KEYWORD][1]
         else:
             showforcekw = b''
         showunforcekw = b''
 
     searchfunc = searchfuncs[mode]
 
     tip = web.repo[b'tip']
     parity = paritygen(web.stripecount)
 
     return web.sendtemplate(
         b'search',
         query=query,
         node=tip.hex(),
         symrev=b'tip',
         entries=templateutil.mappinggenerator(changelist, name=b'searchentry'),
         archives=web.archivelist(b'tip'),
         morevars=morevars,
         lessvars=lessvars,
         modedesc=searchfunc[1],
         showforcekw=showforcekw,
         showunforcekw=showunforcekw,
     )
 
 
 @webcommand(b'changelog')
 def changelog(web, shortlog=False):
     """
     /changelog[/{revision}]
     -----------------------
 
     Show information about multiple changesets.
 
     If the optional ``revision`` URL argument is absent, information about
     all changesets starting at ``tip`` will be rendered. If the ``revision``
     argument is present, changesets will be shown starting from the specified
     revision.
 
     If ``revision`` is absent, the ``rev`` query string argument may be
     defined. This will perform a search for changesets.
 
     The argument for ``rev`` can be a single revision, a revision set,
     or a literal keyword to search for in changeset data (equivalent to
     :hg:`log -k`).
 
     The ``revcount`` query string argument defines the maximum numbers of
     changesets to render.
 
     For non-searches, the ``changelog`` template will be rendered.
     """
 
     query = b''
     if b'node' in web.req.qsparams:
         ctx = webutil.changectx(web.repo, web.req)
         symrev = webutil.symrevorshortnode(web.req, ctx)
     elif b'rev' in web.req.qsparams:
         return _search(web)
     else:
         ctx = web.repo[b'tip']
         symrev = b'tip'
 
     def changelist(maxcount):
         revs = []
         if pos != -1:
             revs = web.repo.changelog.revs(pos, 0)
 
         for entry in webutil.changelistentries(web, revs, maxcount, parity):
             yield entry
 
     if shortlog:
         revcount = web.maxshortchanges
     else:
         revcount = web.maxchanges
 
     if b'revcount' in web.req.qsparams:
         try:
             revcount = int(web.req.qsparams.get(b'revcount', revcount))
             revcount = max(revcount, 1)
             web.tmpl.defaults[b'sessionvars'][b'revcount'] = revcount
         except ValueError:
             pass
 
     lessvars = copy.copy(web.tmpl.defaults[b'sessionvars'])
     lessvars[b'revcount'] = max(revcount // 2, 1)
     morevars = copy.copy(web.tmpl.defaults[b'sessionvars'])
     morevars[b'revcount'] = revcount * 2
 
     count = len(web.repo)
     pos = ctx.rev()
     parity = paritygen(web.stripecount)
 
     changenav = webutil.revnav(web.repo).gen(pos, revcount, count)
 
     entries = list(changelist(revcount + 1))
     latestentry = entries[:1]
     if len(entries) > revcount:
         nextentry = entries[-1:]
         entries = entries[:-1]
     else:
         nextentry = []
 
     return web.sendtemplate(
         b'shortlog' if shortlog else b'changelog',
         changenav=changenav,
         node=ctx.hex(),
         rev=pos,
         symrev=symrev,
         changesets=count,
         entries=templateutil.mappinglist(entries),
         latestentry=templateutil.mappinglist(latestentry),
         nextentry=templateutil.mappinglist(nextentry),
         archives=web.archivelist(b'tip'),
         revcount=revcount,
         morevars=morevars,
         lessvars=lessvars,
         query=query,
     )
 
 
 @webcommand(b'shortlog')
 def shortlog(web):
     """
     /shortlog
     ---------
 
     Show basic information about a set of changesets.
 
     This accepts the same parameters as the ``changelog`` handler. The only
     difference is the ``shortlog`` template will be rendered instead of the
     ``changelog`` template.
     """
     return changelog(web, shortlog=True)
 
 
 @webcommand(b'changeset')
 def changeset(web):
     """
     /changeset[/{revision}]
     -----------------------
 
     Show information about a single changeset.
 
     A URL path argument is the changeset identifier to show. See ``hg help
     revisions`` for possible values. If not defined, the ``tip`` changeset
     will be shown.
 
     The ``changeset`` template is rendered. Contents of the ``changesettag``,
     ``changesetbookmark``, ``filenodelink``, ``filenolink``, and the many
     templates related to diffs may all be used to produce the output.
     """
     ctx = webutil.changectx(web.repo, web.req)
 
     return web.sendtemplate(b'changeset', **webutil.changesetentry(web, ctx))
 
 
 rev = webcommand(b'rev')(changeset)
 
 
 def decodepath(path):
     # type: (bytes) -> bytes
     """Hook for mapping a path in the repository to a path in the
     working copy.
 
     Extensions (e.g., largefiles) can override this to remap files in
     the virtual file system presented by the manifest command below."""
     return path
 
 
 @webcommand(b'manifest')
 def manifest(web):
     """
     /manifest[/{revision}[/{path}]]
     -------------------------------
 
     Show information about a directory.
 
     If the URL path arguments are omitted, information about the root
     directory for the ``tip`` changeset will be shown.
 
     Because this handler can only show information for directories, it
     is recommended to use the ``file`` handler instead, as it can handle both
     directories and files.
 
     The ``manifest`` template will be rendered for this handler.
     """
     if b'node' in web.req.qsparams:
         ctx = webutil.changectx(web.repo, web.req)
         symrev = webutil.symrevorshortnode(web.req, ctx)
     else:
         ctx = web.repo[b'tip']
         symrev = b'tip'
     path = webutil.cleanpath(web.repo, web.req.qsparams.get(b'file', b''))
     mf = ctx.manifest()
     node = scmutil.binnode(ctx)
 
     files = {}
     dirs = {}
     parity = paritygen(web.stripecount)
 
     if path and path[-1:] != b"/":
         path += b"/"
     l = len(path)
     abspath = b"/" + path
 
     for full, n in mf.items():
         # the virtual path (working copy path) used for the full
         # (repository) path
         f = decodepath(full)
 
         if f[:l] != path:
             continue
         remain = f[l:]
         elements = remain.split(b'/')
         if len(elements) == 1:
             files[remain] = full
         else:
             h = dirs  # need to retain ref to dirs (root)
             for elem in elements[0:-1]:
                 if elem not in h:
                     h[elem] = {}
                 h = h[elem]
                 if len(h) > 1:
                     break
             h[None] = None  # denotes files present
 
     if mf and not files and not dirs:
         raise ErrorResponse(HTTP_NOT_FOUND, b'path not found: ' + path)
 
     def filelist(context):
         for f in sorted(files):
             full = files[f]
 
             fctx = ctx.filectx(full)
             yield {
                 b"file": full,
                 b"parity": next(parity),
                 b"basename": f,
                 b"date": fctx.date(),
                 b"size": fctx.size(),
                 b"permissions": mf.flags(full),
             }
 
     def dirlist(context):
         for d in sorted(dirs):
 
             emptydirs = []
             h = dirs[d]
             while isinstance(h, dict) and len(h) == 1:
                 k, v = next(iter(h.items()))
                 if v:
                     emptydirs.append(k)
                 h = v
 
             path = b"%s%s" % (abspath, d)
             yield {
                 b"parity": next(parity),
                 b"path": path,
                 # pytype: disable=wrong-arg-types
                 b"emptydirs": b"/".join(emptydirs),
                 # pytype: enable=wrong-arg-types
                 b"basename": d,
             }
 
     return web.sendtemplate(
         b'manifest',
         symrev=symrev,
         path=abspath,
         up=webutil.up(abspath),
         upparity=next(parity),
         fentries=templateutil.mappinggenerator(filelist),
         dentries=templateutil.mappinggenerator(dirlist),
         archives=web.archivelist(hex(node)),
         **pycompat.strkwargs(webutil.commonentry(web.repo, ctx))
     )
 
 
 @webcommand(b'tags')
 def tags(web):
     """
     /tags
     -----
 
     Show information about tags.
 
     No arguments are accepted.
 
     The ``tags`` template is rendered.
     """
     i = list(reversed(web.repo.tagslist()))
     parity = paritygen(web.stripecount)
 
     def entries(context, notip, latestonly):
         t = i
         if notip:
             t = [(k, n) for k, n in i if k != b"tip"]
         if latestonly:
             t = t[:1]
         for k, n in t:
             yield {
                 b"parity": next(parity),
                 b"tag": k,
                 b"date": web.repo[n].date(),
                 b"node": hex(n),
             }
 
     return web.sendtemplate(
         b'tags',
         node=hex(web.repo.changelog.tip()),
         entries=templateutil.mappinggenerator(entries, args=(False, False)),
         entriesnotip=templateutil.mappinggenerator(entries, args=(True, False)),
         latestentry=templateutil.mappinggenerator(entries, args=(True, True)),
     )
 
 
 @webcommand(b'bookmarks')
 def bookmarks(web):
     """
     /bookmarks
     ----------
 
     Show information about bookmarks.
 
     No arguments are accepted.
 
     The ``bookmarks`` template is rendered.
     """
     i = [b for b in web.repo._bookmarks.items() if b[1] in web.repo]
     sortkey = lambda b: (web.repo[b[1]].rev(), b[0])
     i = sorted(i, key=sortkey, reverse=True)
     parity = paritygen(web.stripecount)
 
     def entries(context, latestonly):
         t = i
         if latestonly:
             t = i[:1]
         for k, n in t:
             yield {
                 b"parity": next(parity),
                 b"bookmark": k,
                 b"date": web.repo[n].date(),
                 b"node": hex(n),
             }
 
     if i:
         latestrev = i[0][1]
     else:
         latestrev = -1
     lastdate = web.repo[latestrev].date()
 
     return web.sendtemplate(
         b'bookmarks',
         node=hex(web.repo.changelog.tip()),
         lastchange=templateutil.mappinglist([{b'date': lastdate}]),
         entries=templateutil.mappinggenerator(entries, args=(False,)),
         latestentry=templateutil.mappinggenerator(entries, args=(True,)),
     )
 
 
 @webcommand(b'branches')
 def branches(web):
     """
     /branches
     ---------
 
     Show information about branches.
 
     All known branches are contained in the output, even closed branches.
 
     No arguments are accepted.
 
     The ``branches`` template is rendered.
     """
     entries = webutil.branchentries(web.repo, web.stripecount)
     latestentry = webutil.branchentries(web.repo, web.stripecount, 1)
 
     return web.sendtemplate(
         b'branches',
         node=hex(web.repo.changelog.tip()),
         entries=entries,
         latestentry=latestentry,
     )
 
 
 @webcommand(b'summary')
 def summary(web):
     """
     /summary
     --------
 
     Show a summary of repository state.
 
     Information about the latest changesets, bookmarks, tags, and branches
     is captured by this handler.
 
     The ``summary`` template is rendered.
     """
     i = reversed(web.repo.tagslist())
 
     def tagentries(context):
         parity = paritygen(web.stripecount)
         count = 0
         for k, n in i:
             if k == b"tip":  # skip tip
                 continue
 
             count += 1
             if count > 10:  # limit to 10 tags
                 break
 
             yield {
                 b'parity': next(parity),
                 b'tag': k,
                 b'node': hex(n),
                 b'date': web.repo[n].date(),
             }
 
     def bookmarks(context):
         parity = paritygen(web.stripecount)
         marks = [b for b in web.repo._bookmarks.items() if b[1] in web.repo]
         sortkey = lambda b: (web.repo[b[1]].rev(), b[0])
         marks = sorted(marks, key=sortkey, reverse=True)
         for k, n in marks[:10]:  # limit to 10 bookmarks
             yield {
                 b'parity': next(parity),
                 b'bookmark': k,
                 b'date': web.repo[n].date(),
                 b'node': hex(n),
             }
 
     def changelist(context):
         parity = paritygen(web.stripecount, offset=start - end)
         l = []  # build a list in forward order for efficiency
         revs = []
         if start < end:
             revs = web.repo.changelog.revs(start, end - 1)
         for i in revs:
             ctx = web.repo[i]
             lm = webutil.commonentry(web.repo, ctx)
             lm[b'parity'] = next(parity)
             l.append(lm)
 
         for entry in reversed(l):
             yield entry
 
     tip = web.repo[b'tip']
     count = len(web.repo)
     start = max(0, count - web.maxchanges)
     end = min(count, start + web.maxchanges)
 
     desc = web.config(b"web", b"description")
     if not desc:
         desc = b'unknown'
     labels = web.configlist(b'web', b'labels')
 
     return web.sendtemplate(
         b'summary',
         desc=desc,
         owner=get_contact(web.config) or b'unknown',
         lastchange=tip.date(),
         tags=templateutil.mappinggenerator(tagentries, name=b'tagentry'),
         bookmarks=templateutil.mappinggenerator(bookmarks),
         branches=webutil.branchentries(web.repo, web.stripecount, 10),
         shortlog=templateutil.mappinggenerator(
             changelist, name=b'shortlogentry'
         ),
         node=tip.hex(),
         symrev=b'tip',
         archives=web.archivelist(b'tip'),
         labels=templateutil.hybridlist(labels, name=b'label'),
     )
 
 
 @webcommand(b'filediff')
 def filediff(web):
     """
     /diff/{revision}/{path}
     -----------------------
 
     Show how a file changed in a particular commit.
 
     The ``filediff`` template is rendered.
 
     This handler is registered under both the ``/diff`` and ``/filediff``
     paths. ``/diff`` is used in modern code.
     """
     fctx, ctx = None, None
     try:
         fctx = webutil.filectx(web.repo, web.req)
     except LookupError:
         ctx = webutil.changectx(web.repo, web.req)
         path = webutil.cleanpath(web.repo, web.req.qsparams[b'file'])
         if path not in ctx.files():
             raise
 
     if fctx is not None:
         path = fctx.path()
         ctx = fctx.changectx()
     basectx = ctx.p1()
 
     style = web.config(b'web', b'style')
     if b'style' in web.req.qsparams:
         style = web.req.qsparams[b'style']
 
     diffs = webutil.diffs(web, ctx, basectx, [path], style)
     if fctx is not None:
         rename = webutil.renamelink(fctx)
         ctx = fctx
     else:
         rename = templateutil.mappinglist([])
         ctx = ctx
 
     return web.sendtemplate(
         b'filediff',
         file=path,
         symrev=webutil.symrevorshortnode(web.req, ctx),
         rename=rename,
         diff=diffs,
         **pycompat.strkwargs(webutil.commonentry(web.repo, ctx))
     )
 
 
 diff = webcommand(b'diff')(filediff)
 
 
 @webcommand(b'comparison')
 def comparison(web):
     """
     /comparison/{revision}/{path}
     -----------------------------
 
     Show a comparison between the old and new versions of a file from changes
     made on a particular revision.
 
     This is similar to the ``diff`` handler. However, this form features
     a split or side-by-side diff rather than a unified diff.
 
     The ``context`` query string argument can be used to control the lines of
     context in the diff.
 
     The ``filecomparison`` template is rendered.
     """
     ctx = webutil.changectx(web.repo, web.req)
     if b'file' not in web.req.qsparams:
         raise ErrorResponse(HTTP_NOT_FOUND, b'file not given')
     path = webutil.cleanpath(web.repo, web.req.qsparams[b'file'])
 
     parsecontext = lambda v: v == b'full' and -1 or int(v)
     if b'context' in web.req.qsparams:
         context = parsecontext(web.req.qsparams[b'context'])
     else:
         context = parsecontext(web.config(b'web', b'comparisoncontext'))
 
     def filelines(f):
         if f.isbinary():
             mt = pycompat.sysbytes(
                 mimetypes.guess_type(pycompat.fsdecode(f.path()))[0]
                 or r'application/octet-stream'
             )
             return [_(b'(binary file %s, hash: %s)') % (mt, hex(f.filenode()))]
         return f.data().splitlines()
 
     fctx = None
     parent = ctx.p1()
     leftrev = parent.rev()
     leftnode = parent.node()
     rightrev = ctx.rev()
     rightnode = scmutil.binnode(ctx)
     if path in ctx:
         fctx = ctx[path]
         rightlines = filelines(fctx)
         if path not in parent:
             leftlines = ()
         else:
             pfctx = parent[path]
             leftlines = filelines(pfctx)
     else:
         rightlines = ()
         pfctx = ctx.p1()[path]
         leftlines = filelines(pfctx)
 
     comparison = webutil.compare(context, leftlines, rightlines)
     if fctx is not None:
         rename = webutil.renamelink(fctx)
         ctx = fctx
     else:
         rename = templateutil.mappinglist([])
         ctx = ctx
 
     return web.sendtemplate(
         b'filecomparison',
         file=path,
         symrev=webutil.symrevorshortnode(web.req, ctx),
         rename=rename,
         leftrev=leftrev,
         leftnode=hex(leftnode),
         rightrev=rightrev,
         rightnode=hex(rightnode),
         comparison=comparison,
         **pycompat.strkwargs(webutil.commonentry(web.repo, ctx))
     )
 
 
 @webcommand(b'annotate')
 def annotate(web):
     """
     /annotate/{revision}/{path}
     ---------------------------
 
     Show changeset information for each line in a file.
 
     The ``ignorews``, ``ignorewsamount``, ``ignorewseol``, and
     ``ignoreblanklines`` query string arguments have the same meaning as
     their ``[annotate]`` config equivalents. It uses the hgrc boolean
     parsing logic to interpret the value. e.g. ``0`` and ``false`` are
     false and ``1`` and ``true`` are true. If not defined, the server
     default settings are used.
 
     The ``fileannotate`` template is rendered.
     """
     fctx = webutil.filectx(web.repo, web.req)
     f = fctx.path()
     parity = paritygen(web.stripecount)
     ishead = fctx.filenode() in fctx.filelog().heads()
 
     # parents() is called once per line and several lines likely belong to
     # same revision. So it is worth caching.
     # TODO there are still redundant operations within basefilectx.parents()
     # and from the fctx.annotate() call itself that could be cached.
     parentscache = {}
 
     def parents(context, f):
         rev = f.rev()
         if rev not in parentscache:
             parentscache[rev] = []
             for p in f.parents():
                 entry = {
                     b'node': p.hex(),
                     b'rev': p.rev(),
                 }
                 parentscache[rev].append(entry)
 
         for p in parentscache[rev]:
             yield p
 
     def annotate(context):
         if fctx.isbinary():
             mt = pycompat.sysbytes(
                 mimetypes.guess_type(pycompat.fsdecode(fctx.path()))[0]
                 or r'application/octet-stream'
             )
             lines = [
                 dagop.annotateline(
                     fctx=fctx.filectx(fctx.filerev()),
                     lineno=1,
                     text=b'(binary:%s)' % mt,
                 )
             ]
         else:
             lines = webutil.annotate(web.req, fctx, web.repo.ui)
 
         previousrev = None
         blockparitygen = paritygen(1)
         for lineno, aline in enumerate(lines):
             f = aline.fctx
             rev = f.rev()
             if rev != previousrev:
                 blockhead = True
                 blockparity = next(blockparitygen)
             else:
                 blockhead = None
             previousrev = rev
             yield {
                 b"parity": next(parity),
                 b"node": f.hex(),
                 b"rev": rev,
                 b"author": f.user(),
                 b"parents": templateutil.mappinggenerator(parents, args=(f,)),
                 b"desc": f.description(),
                 b"extra": f.extra(),
                 b"file": f.path(),
                 b"blockhead": blockhead,
                 b"blockparity": blockparity,
                 b"targetline": aline.lineno,
                 b"line": aline.text,
                 b"lineno": lineno + 1,
                 b"lineid": b"l%d" % (lineno + 1),
                 b"linenumber": b"% 6d" % (lineno + 1),
                 b"revdate": f.date(),
             }
 
     diffopts = webutil.difffeatureopts(web.req, web.repo.ui, b'annotate')
     diffopts = {k: getattr(diffopts, k) for k in diffopts.defaults}
 
     return web.sendtemplate(
         b'fileannotate',
         file=f,
         annotate=templateutil.mappinggenerator(annotate),
         path=webutil.up(f),
         symrev=webutil.symrevorshortnode(web.req, fctx),
         rename=webutil.renamelink(fctx),
         permissions=fctx.manifest().flags(f),
         ishead=int(ishead),
         diffopts=templateutil.hybriddict(diffopts),
         **pycompat.strkwargs(webutil.commonentry(web.repo, fctx))
     )
 
 
 @webcommand(b'filelog')
 def filelog(web):
     """
     /filelog/{revision}/{path}
     --------------------------
 
     Show information about the history of a file in the repository.
 
     The ``revcount`` query string argument can be defined to control the
     maximum number of entries to show.
 
     The ``filelog`` template will be rendered.
     """
 
     try:
         fctx = webutil.filectx(web.repo, web.req)
         f = fctx.path()
         fl = fctx.filelog()
     except error.LookupError:
         f = webutil.cleanpath(web.repo, web.req.qsparams[b'file'])
         fl = web.repo.file(f)
         numrevs = len(fl)
         if not numrevs:  # file doesn't exist at all
             raise
         rev = webutil.changectx(web.repo, web.req).rev()
         first = fl.linkrev(0)
         if rev < first:  # current rev is from before file existed
             raise
         frev = numrevs - 1
         while fl.linkrev(frev) > rev:
             frev -= 1
         fctx = web.repo.filectx(f, fl.linkrev(frev))
 
     revcount = web.maxshortchanges
     if b'revcount' in web.req.qsparams:
         try:
             revcount = int(web.req.qsparams.get(b'revcount', revcount))
             revcount = max(revcount, 1)
             web.tmpl.defaults[b'sessionvars'][b'revcount'] = revcount
         except ValueError:
             pass
 
     lrange = webutil.linerange(web.req)
 
     lessvars = copy.copy(web.tmpl.defaults[b'sessionvars'])
     lessvars[b'revcount'] = max(revcount // 2, 1)
     morevars = copy.copy(web.tmpl.defaults[b'sessionvars'])
     morevars[b'revcount'] = revcount * 2
 
     patch = b'patch' in web.req.qsparams
     if patch:
         lessvars[b'patch'] = morevars[b'patch'] = web.req.qsparams[b'patch']
     descend = b'descend' in web.req.qsparams
     if descend:
         lessvars[b'descend'] = morevars[b'descend'] = web.req.qsparams[
             b'descend'
         ]
 
     count = fctx.filerev() + 1
     start = max(0, count - revcount)  # first rev on this page
     end = min(count, start + revcount)  # last rev on this page
     parity = paritygen(web.stripecount, offset=start - end)
 
     repo = web.repo
     filelog = fctx.filelog()
     revs = [
         filerev
         for filerev in filelog.revs(start, end - 1)
         if filelog.linkrev(filerev) in repo
     ]
     entries = []
 
     diffstyle = web.config(b'web', b'style')
     if b'style' in web.req.qsparams:
         diffstyle = web.req.qsparams[b'style']
 
     def diff(fctx, linerange=None):
         ctx = fctx.changectx()
         basectx = ctx.p1()
         path = fctx.path()
         return webutil.diffs(
             web,
             ctx,
             basectx,
             [path],
             diffstyle,
             linerange=linerange,
             lineidprefix=b'%s-' % ctx.hex()[:12],
         )
 
     linerange = None
     if lrange is not None:
         assert lrange is not None  # help pytype (!?)
         linerange = webutil.formatlinerange(*lrange)
         # deactivate numeric nav links when linerange is specified as this
         # would required a dedicated "revnav" class
         nav = templateutil.mappinglist([])
         if descend:
             it = dagop.blockdescendants(fctx, *lrange)
         else:
             it = dagop.blockancestors(fctx, *lrange)
         for i, (c, lr) in enumerate(it, 1):
             diffs = None
             if patch:
                 diffs = diff(c, linerange=lr)
             # follow renames accross filtered (not in range) revisions
             path = c.path()
             lm = webutil.commonentry(repo, c)
             lm.update(
                 {
                     b'parity': next(parity),
                     b'filerev': c.rev(),
                     b'file': path,
                     b'diff': diffs,
                     b'linerange': webutil.formatlinerange(*lr),
                     b'rename': templateutil.mappinglist([]),
                 }
             )
             entries.append(lm)
             if i == revcount:
                 break
         lessvars[b'linerange'] = webutil.formatlinerange(*lrange)
         morevars[b'linerange'] = lessvars[b'linerange']
     else:
         for i in revs:
             iterfctx = fctx.filectx(i)
             diffs = None
             if patch:
                 diffs = diff(iterfctx)
             lm = webutil.commonentry(repo, iterfctx)
             lm.update(
                 {
                     b'parity': next(parity),
                     b'filerev': i,
                     b'file': f,
                     b'diff': diffs,
                     b'rename': webutil.renamelink(iterfctx),
                 }
             )
             entries.append(lm)
         entries.reverse()
         revnav = webutil.filerevnav(web.repo, fctx.path())
         nav = revnav.gen(end - 1, revcount, count)
 
     latestentry = entries[:1]
 
     return web.sendtemplate(
         b'filelog',
         file=f,
         nav=nav,
         symrev=webutil.symrevorshortnode(web.req, fctx),
         entries=templateutil.mappinglist(entries),
         descend=descend,
         patch=patch,
         latestentry=templateutil.mappinglist(latestentry),
         linerange=linerange,
         revcount=revcount,
         morevars=morevars,
         lessvars=lessvars,
         **pycompat.strkwargs(webutil.commonentry(web.repo, fctx))
     )
 
 
 @webcommand(b'archive')
 def archive(web):
     """
     /archive/{revision}.{format}[/{path}]
     -------------------------------------
 
     Obtain an archive of repository content.
 
     The content and type of the archive is defined by a URL path parameter.
     ``format`` is the file extension of the archive type to be generated. e.g.
     ``zip`` or ``tar.bz2``. Not all archive types may be allowed by your
     server configuration.
 
     The optional ``path`` URL parameter controls content to include in the
     archive. If omitted, every file in the specified revision is present in the
     archive. If included, only the specified file or contents of the specified
     directory will be included in the archive.
 
     No template is used for this handler. Raw, binary content is generated.
     """
 
     type_ = web.req.qsparams.get(b'type')
     allowed = web.configlist(b"web", b"allow-archive")
     key = web.req.qsparams[b'node']
 
     if type_ not in webutil.archivespecs:
         msg = b'Unsupported archive type: %s' % stringutil.pprint(type_)
         raise ErrorResponse(HTTP_NOT_FOUND, msg)
 
     if not ((type_ in allowed or web.configbool(b"web", b"allow" + type_))):
         msg = b'Archive type not allowed: %s' % type_
         raise ErrorResponse(HTTP_FORBIDDEN, msg)
 
     reponame = re.sub(br"\W+", b"-", os.path.basename(web.reponame))
     cnode = web.repo.lookup(key)
     arch_version = key
     if cnode == key or key == b'tip':
         arch_version = short(cnode)
     name = b"%s-%s" % (reponame, arch_version)
 
     ctx = webutil.changectx(web.repo, web.req)
     match = scmutil.match(ctx, [])
     file = web.req.qsparams.get(b'file')
     if file:
         pats = [b'path:' + file]
         match = scmutil.match(ctx, pats, default=b'path')
         if pats:
             files = [f for f in ctx.manifest().keys() if match(f)]
             if not files:
                 raise ErrorResponse(
                     HTTP_NOT_FOUND, b'file(s) not found: %s' % file
                 )
 
     mimetype, artype, extension, encoding = webutil.archivespecs[type_]
 
     web.res.headers[b'Content-Type'] = mimetype
     web.res.headers[b'Content-Disposition'] = b'attachment; filename=%s%s' % (
         name,
         extension,
     )
 
     if encoding:
         web.res.headers[b'Content-Encoding'] = encoding
 
     web.res.setbodywillwrite()
     if list(web.res.sendresponse()):
         raise error.ProgrammingError(
             b'sendresponse() should not emit data if writing later'
         )
 
+    if web.req.method == b'HEAD':
+        return []
+
     bodyfh = web.res.getbodyfile()
 
     archival.archive(
         web.repo,
         bodyfh,
         cnode,
         artype,
         prefix=name,
         match=match,
         subrepos=web.configbool(b"web", b"archivesubrepos"),
     )
 
     return []
 
 
 @webcommand(b'static')
 def static(web):
     fname = web.req.qsparams[b'file']
     # a repo owner may set web.static in .hg/hgrc to get any file
     # readable by the user running the CGI script
     static = web.config(b"web", b"static", untrusted=False)
     staticfile(web.templatepath, static, fname, web.res)
     return web.res.sendresponse()
 
 
 @webcommand(b'graph')
 def graph(web):
     """
     /graph[/{revision}]
     -------------------
 
     Show information about the graphical topology of the repository.
 
     Information rendered by this handler can be used to create visual
     representations of repository topology.
 
     The ``revision`` URL parameter controls the starting changeset. If it's
     absent, the default is ``tip``.
 
     The ``revcount`` query string argument can define the number of changesets
     to show information for.
 
     The ``graphtop`` query string argument can specify the starting changeset
     for producing ``jsdata`` variable that is used for rendering graph in
     JavaScript. By default it has the same value as ``revision``.
 
     This handler will render the ``graph`` template.
     """
 
     if b'node' in web.req.qsparams:
         ctx = webutil.changectx(web.repo, web.req)
         symrev = webutil.symrevorshortnode(web.req, ctx)
     else:
         ctx = web.repo[b'tip']
         symrev = b'tip'
     rev = ctx.rev()
 
     bg_height = 39
     revcount = web.maxshortchanges
     if b'revcount' in web.req.qsparams:
         try:
             revcount = int(web.req.qsparams.get(b'revcount', revcount))
             revcount = max(revcount, 1)
             web.tmpl.defaults[b'sessionvars'][b'revcount'] = revcount
         except ValueError:
             pass
 
     lessvars = copy.copy(web.tmpl.defaults[b'sessionvars'])
     lessvars[b'revcount'] = max(revcount // 2, 1)
     morevars = copy.copy(web.tmpl.defaults[b'sessionvars'])
     morevars[b'revcount'] = revcount * 2
 
     graphtop = web.req.qsparams.get(b'graphtop', ctx.hex())
     graphvars = copy.copy(web.tmpl.defaults[b'sessionvars'])
     graphvars[b'graphtop'] = graphtop
 
     count = len(web.repo)
     pos = rev
 
     uprev = min(max(0, count - 1), rev + revcount)
     downrev = max(0, rev - revcount)
     changenav = webutil.revnav(web.repo).gen(pos, revcount, count)
 
     tree = []
     nextentry = []
     lastrev = 0
     if pos != -1:
         allrevs = web.repo.changelog.revs(pos, 0)
         revs = []
         for i in allrevs:
             revs.append(i)
             if len(revs) >= revcount + 1:
                 break
 
         if len(revs) > revcount:
             nextentry = [webutil.commonentry(web.repo, web.repo[revs[-1]])]
             revs = revs[:-1]
 
         lastrev = revs[-1]
 
         # We have to feed a baseset to dagwalker as it is expecting smartset
         # object. This does not have a big impact on hgweb performance itself
         # since hgweb graphing code is not itself lazy yet.
         dag = graphmod.dagwalker(web.repo, smartset.baseset(revs))
         # As we said one line above... not lazy.
         tree = list(
             item
             for item in graphmod.colored(dag, web.repo)
             if item[1] == graphmod.CHANGESET
         )
 
     def fulltree():
         pos = web.repo[graphtop].rev()
         tree = []
         if pos != -1:
             revs = web.repo.changelog.revs(pos, lastrev)
             dag = graphmod.dagwalker(web.repo, smartset.baseset(revs))
             tree = list(
                 item
                 for item in graphmod.colored(dag, web.repo)
                 if item[1] == graphmod.CHANGESET
             )
         return tree
 
     def jsdata(context):
         for (id, type, ctx, vtx, edges) in fulltree():
             yield {
                 b'node': pycompat.bytestr(ctx),
                 b'graphnode': webutil.getgraphnode(web.repo, ctx),
                 b'vertex': vtx,
                 b'edges': edges,
             }
 
     def nodes(context):
         parity = paritygen(web.stripecount)
         for row, (id, type, ctx, vtx, edges) in enumerate(tree):
             entry = webutil.commonentry(web.repo, ctx)
             edgedata = [
                 {
                     b'col': edge[0],
                     b'nextcol': edge[1],
                     b'color': (edge[2] - 1) % 6 + 1,
                     b'width': edge[3],
                     b'bcolor': edge[4],
                 }
                 for edge in edges
             ]
 
             entry.update(
                 {
                     b'col': vtx[0],
                     b'color': (vtx[1] - 1) % 6 + 1,
                     b'parity': next(parity),
                     b'edges': templateutil.mappinglist(edgedata),
                     b'row': row,
                     b'nextrow': row + 1,
                 }
             )
 
             yield entry
 
     rows = len(tree)
 
     return web.sendtemplate(
         b'graph',
         rev=rev,
         symrev=symrev,
         revcount=revcount,
         uprev=uprev,
         lessvars=lessvars,
         morevars=morevars,
         downrev=downrev,
         graphvars=graphvars,
         rows=rows,
         bg_height=bg_height,
         changesets=count,
         nextentry=templateutil.mappinglist(nextentry),
         jsdata=templateutil.mappinggenerator(jsdata),
         nodes=templateutil.mappinggenerator(nodes),
         node=ctx.hex(),
         archives=web.archivelist(b'tip'),
         changenav=changenav,
     )
 
 
 def _getdoc(e):
     doc = e[0].__doc__
     if doc:
         doc = _(doc).partition(b'\n')[0]
     else:
         doc = _(b'(no help text available)')
     return doc
 
 
 @webcommand(b'help')
 def help(web):
     """
     /help[/{topic}]
     ---------------
 
     Render help documentation.
 
     This web command is roughly equivalent to :hg:`help`. If a ``topic``
     is defined, that help topic will be rendered. If not, an index of
     available help topics will be rendered.
 
     The ``help`` template will be rendered when requesting help for a topic.
     ``helptopics`` will be rendered for the index of help topics.
     """
     from .. import commands, help as helpmod  # avoid cycle
 
     topicname = web.req.qsparams.get(b'node')
     if not topicname:
 
         def topics(context):
             for h in helpmod.helptable:
                 entries, summary, _doc = h[0:3]
                 yield {b'topic': entries[0], b'summary': summary}
 
         early, other = [], []
         primary = lambda s: s.partition(b'|')[0]
         for c, e in commands.table.items():
             doc = _getdoc(e)
             if b'DEPRECATED' in doc or c.startswith(b'debug'):
                 continue
             cmd = primary(c)
             if getattr(e[0], 'helpbasic', False):
                 early.append((cmd, doc))
             else:
                 other.append((cmd, doc))
 
         early.sort()
         other.sort()
 
         def earlycommands(context):
             for c, doc in early:
                 yield {b'topic': c, b'summary': doc}
 
         def othercommands(context):
             for c, doc in other:
                 yield {b'topic': c, b'summary': doc}
 
         return web.sendtemplate(
             b'helptopics',
             topics=templateutil.mappinggenerator(topics),
             earlycommands=templateutil.mappinggenerator(earlycommands),
             othercommands=templateutil.mappinggenerator(othercommands),
             title=b'Index',
         )
 
     # Render an index of sub-topics.
     if topicname in helpmod.subtopics:
         topics = []
         for entries, summary, _doc in helpmod.subtopics[topicname]:
             topics.append(
                 {
                     b'topic': b'%s.%s' % (topicname, entries[0]),
                     b'basename': entries[0],
                     b'summary': summary,
                 }
             )
 
         return web.sendtemplate(
             b'helptopics',
             topics=templateutil.mappinglist(topics),
             title=topicname,
             subindex=True,
         )
 
     u = webutil.wsgiui.load()
     u.verbose = True
 
     # Render a page from a sub-topic.
     if b'.' in topicname:
         # TODO implement support for rendering sections, like
         # `hg help` works.
         topic, subtopic = topicname.split(b'.', 1)
         if topic not in helpmod.subtopics:
             raise ErrorResponse(HTTP_NOT_FOUND)
     else:
         topic = topicname
         subtopic = None
 
     try:
         doc = helpmod.help_(u, commands, topic, subtopic=subtopic)
     except error.Abort:
         raise ErrorResponse(HTTP_NOT_FOUND)
 
     return web.sendtemplate(b'help', topic=topicname, doc=doc)
 
 
 # tell hggettext to extract docstrings from these functions:
 i18nfunctions = commands.values()