# HG changeset patch # User Brendan Cully # Date 1227311500 28800 # Node ID 526c40a74bd0a97330458a1754d88371ce176060 # Parent e2048f5c7bf5304a9756fac356864282d9835ddb templater: return data in increasing chunk sizes Currently hgweb is not streaming its output -- it accumulates the entire response before sending it. This patch restores streaming behaviour. To avoid having to synchronously write many tiny fragments, this patch also adds buffering to the template generator. Local testing of a fetch of a 100,000 line file with wget produces a slight slowdown overall (up from 6.5 seconds to 7.2 seconds), but instead of waiting 6 seconds for headers to arrive, output begins immediately. diff -r e2048f5c7bf5 -r 526c40a74bd0 mercurial/hgweb/hgweb_mod.py --- a/mercurial/hgweb/hgweb_mod.py Sat Nov 22 16:57:49 2008 +0100 +++ b/mercurial/hgweb/hgweb_mod.py Fri Nov 21 15:51:40 2008 -0800 @@ -182,20 +182,20 @@ content = getattr(webcommands, cmd)(self, req, tmpl) req.respond(HTTP_OK, ctype) - return ''.join(content), + return content except revlog.LookupError, err: req.respond(HTTP_NOT_FOUND, ctype) msg = str(err) if 'manifest' not in msg: msg = 'revision not found: %s' % err.name - return ''.join(tmpl('error', error=msg)), + return tmpl('error', error=msg) except (RepoError, revlog.RevlogError), inst: req.respond(HTTP_SERVER_ERROR, ctype) - return ''.join(tmpl('error', error=str(inst))), + return tmpl('error', error=str(inst)) except ErrorResponse, inst: req.respond(inst.code, ctype) - return ''.join(tmpl('error', error=inst.message)), + return tmpl('error', error=inst.message) def templater(self, req): diff -r e2048f5c7bf5 -r 526c40a74bd0 mercurial/hgweb/hgwebdir_mod.py --- a/mercurial/hgweb/hgwebdir_mod.py Sat Nov 22 16:57:49 2008 +0100 +++ b/mercurial/hgweb/hgwebdir_mod.py Fri Nov 21 15:51:40 2008 -0800 @@ -116,7 +116,7 @@ # top-level index elif not virtual: req.respond(HTTP_OK, ctype) - return ''.join(self.makeindex(req, tmpl)), + return self.makeindex(req, tmpl) # nested indexes and hgwebs @@ -138,7 +138,7 @@ subdir = virtual + '/' if [r for r in repos if r.startswith(subdir)]: req.respond(HTTP_OK, ctype) - return ''.join(self.makeindex(req, tmpl, subdir)), + return self.makeindex(req, tmpl, subdir) up = virtual.rfind('/') if up < 0: @@ -147,11 +147,11 @@ # prefixes not found req.respond(HTTP_NOT_FOUND, ctype) - return ''.join(tmpl("notfound", repo=virtual)), + return tmpl("notfound", repo=virtual) except ErrorResponse, err: req.respond(err.code, ctype) - return ''.join(tmpl('error', error=err.message or '')), + return tmpl('error', error=err.message or '') finally: tmpl = None diff -r e2048f5c7bf5 -r 526c40a74bd0 mercurial/templater.py --- a/mercurial/templater.py Sat Nov 22 16:57:49 2008 +0100 +++ b/mercurial/templater.py Fri Nov 21 15:51:40 2008 -0800 @@ -44,7 +44,8 @@ template_re = re.compile(r"(?:(?:#(?=[\w\|%]+#))|(?:{(?=[\w\|%]+})))" r"(\w+)(?:(?:%(\w+))|((?:\|\w+)*))[#}]") - def __init__(self, mapfile, filters={}, defaults={}, cache={}): + def __init__(self, mapfile, filters={}, defaults={}, cache={}, + minchunk=1024, maxchunk=65536): '''set up template engine. mapfile is name of file to read map definitions from. filters is dict of functions. each transforms a value into another. @@ -55,6 +56,7 @@ self.base = (mapfile and os.path.dirname(mapfile)) or '' self.filters = filters self.defaults = defaults + self.minchunk, self.maxchunk = minchunk, maxchunk if not mapfile: return @@ -130,6 +132,13 @@ yield v def __call__(self, t, **map): + stream = self.expand(t, **map) + if self.minchunk: + stream = util.increasingchunks(stream, min=self.minchunk, + max=self.maxchunk) + return stream + + def expand(self, t, **map): '''Perform expansion. t is name of map element to expand. map contains added elements for use during expansion. Is a generator.''' tmpl = self._template(t) diff -r e2048f5c7bf5 -r 526c40a74bd0 mercurial/util.py --- a/mercurial/util.py Sat Nov 22 16:57:49 2008 +0100 +++ b/mercurial/util.py Fri Nov 21 15:51:40 2008 -0800 @@ -290,6 +290,37 @@ l.sort() return l +def increasingchunks(source, min=1024, max=65536): + '''return no less than min bytes per chunk while data remains, + doubling min after each chunk until it reaches max''' + def log2(x): + if not x: + return 0 + i = 0 + while x: + x >>= 1 + i += 1 + return i - 1 + + buf = [] + blen = 0 + for chunk in source: + buf.append(chunk) + blen += len(chunk) + if blen >= min: + if min < max: + min = min << 1 + nmin = 1 << log2(blen) + if nmin > min: + min = nmin + if min > max: + min = max + yield ''.join(buf) + blen = 0 + buf = [] + if buf: + yield ''.join(buf) + class Abort(Exception): """Raised if a command needs to print an error and exit."""