Mon Dec 19 14:42:02 ART 2005 Alberto Bertogli tagged 0.13 Mon Dec 19 14:30:50 ART 2005 Alberto Bertogli * Shorten the author string in shortlog. Tue Dec 13 11:12:44 ART 2005 Alberto Bertogli * Small changes to do_atom(), make it validate. Thu Dec 1 19:07:08 ART 2005 stephane@sources.org * ATOM support for syndication Tue Dec 13 10:28:49 ART 2005 Alberto Bertogli * Use darcsweb.cgi's modification time to invalidate cache entries. Besides the repository inventory modification time, we also want to check the script's modification time, so if we make changes to darcsweb (updating it, for instance), there's no possibility of serving stale pages if the user forgets to clean the cache. You still need to clean it for configuration changes. Wed Nov 30 08:36:26 ART 2005 Kirill Smelkov * fixu8: honour config.repoencoding when decoding characters like [_\e3] Sat Nov 12 20:59:16 ART 2005 Alexandre Rossi * another basic validation issue The issue was that there was an empty line before the XML declaration, which is not valid. Thu Nov 17 11:41:33 ART 2005 Alberto Bertogli * Make how_old() return a fixed date when caching is enabled. how_old() becomes a problem when we have a cache, because the relative dates will get stalled in the cache. This is makes a workaround by making how_old() return a string containing the fixed date, which isn't quite nice but it will do the trick until a better solution comes up. Thu Nov 17 11:40:27 ART 2005 Alberto Bertogli * Don't close the file in a cache miss. Thu Nov 17 00:27:57 ART 2005 Alberto Bertogli * Call cache.cancel() only if we have a cache. Thu Nov 17 00:10:28 ART 2005 Alberto Bertogli * Implement a simple cache. This patch implements a very simple but effective cache, so darcsweb can avoid regenerating everything all the time. Based on an idea from Alexandre Rossi. Thu Nov 10 12:22:39 ART 2005 Leandro Lucarella * Use a more human-friendly format por list configuration variables. Wed Nov 9 21:34:34 ART 2005 Alberto Bertogli * Highlight tags in the shortlog/summary. Wed Nov 9 20:59:47 ART 2005 Alberto Bertogli * Add exclusion lists to multidir configuration. Tue Nov 8 22:10:31 ART 2005 Alberto Bertogli tagged 0.12 diff -rN -u old-darcsweb/config.py.sample new-darcsweb/config.py.sample --- old-darcsweb/config.py.sample 2005-12-19 20:33:22.000000000 -0300 +++ new-darcsweb/config.py.sample 2005-12-19 20:33:22.000000000 -0300 @@ -29,6 +29,17 @@ # alternative footer here; it's optional, of course #footer = "I don't like shoes" + # It is possible to have a cache where darcsweb will store the pages + # it generates; entries are automatically updated when the repository + # changes. This will speed things up significatively, specially for + # popular sites. + # It's recommended that you clean the directory with some regularity, + # to avoid having too many unused files. A simple rm will do just + # fine. + # If you leave the entry commented, no cache will be ever used; + # otherwise the directory is assumed to exist and be writeable. + #cachedir = '/tmp/darcsweb-cache' + # # From now on, every class is a repo configuration, with the same format @@ -89,4 +100,8 @@ repourl = 'http://auriga.wearlab.de/~alb/repos/%(name)s' repoencoding = 'latin1' + # if you want to exclude some directories, add them to this list (note + # they're relative to multidir, not absolute) + #exclude = 'dir1', 'dir2' + diff -rN -u old-darcsweb/darcsweb.cgi new-darcsweb/darcsweb.cgi --- old-darcsweb/darcsweb.cgi 2005-12-19 20:33:22.000000000 -0300 +++ new-darcsweb/darcsweb.cgi 2005-12-19 20:33:22.000000000 -0300 @@ -13,12 +13,15 @@ import string import time import stat +import sha import cgi import cgitb; cgitb.enable() import urllib import xml.sax from xml.sax.saxutils import escape as xml_escape +import email.Utils +iso_datetime = '%Y-%m-%dT%H:%M:%SZ' # empty configuration class, we will fill it in later depending on the repo class config: @@ -95,7 +98,7 @@ # finally, replace s with our new improved string, and # repeat the ugly procedure - char = char.decode('raw_unicode_escape') + char = char.decode(config.repoencoding) mn = '[_\\' + middle + '_]' s = s.replace(mn, char, 1) openpos = s.find('[_', openpos + 1) @@ -112,6 +115,13 @@ return s def how_old(epoch): + if config.cachedir: + # when we have a cache, the how_old() becomes a problem since + # the cached entries will have old data; so in this case just + # return a nice string + t = time.localtime(epoch) + s = time.strftime("%d %b %H:%M", t) + return s age = int(time.time()) - int(epoch) if age > 60*60*24*365*2: s = str(age/60/60/24/365) @@ -205,7 +215,7 @@ # def print_header(): - print "Content-type: text/html; charset=utf-8\n" + print "Content-type: text/html; charset=utf-8" print """ + @@ -384,6 +396,91 @@ print "Content-type: text/plain; charset=utf-8\n" + +# +# basic caching +# + +class Cache: + def __init__(self, basedir, url): + self.basedir = basedir + self.url = url + self.fname = sha.sha(url).hexdigest() + self.file = None + self.mode = None + self.real_stdout = sys.stdout + + def open(self): + "Returns 1 on hit, 0 on miss" + fname = self.basedir + '/' + self.fname + + if not os.access(fname, os.R_OK): + # the file doesn't exist, direct miss + pid = str(os.getpid()) + fname = self.basedir + '/.' + self.fname + '-' + pid + self.file = open(fname, 'w') + self.mode = 'w' + + # step over stdout so when "print" tries to write + # output, we get it first + sys.stdout = self + return 0 + + inv = config.repodir + '/_darcs/inventory' + cache_lastmod = os.stat(fname).st_mtime + repo_lastmod = os.stat(inv).st_mtime + dw_lastmod = os.stat(sys.argv[0]).st_mtime + + if repo_lastmod > cache_lastmod or dw_lastmod > cache_lastmod: + # the entry is too old, remove it and return a miss + os.unlink(fname) + + pid = str(os.getpid()) + fname = self.basedir + '/.' + self.fname + '-' + pid + self.file = open(fname, 'w') + self.mode = 'w' + sys.stdout = self + return 0 + + # the entry is still valid, hit! + self.file = open(fname, 'r') + self.mode = 'r' + return 1 + + + def dump(self): + for l in self.file: + self.real_stdout.write(l) + + def write(self, s): + # this gets called from print, because we replaced stdout with + # ourselves + self.file.write(s) + self.real_stdout.write(s) + + def close(self): + if self.file: + self.file.close() + sys.stdout = self.real_stdout + if self.mode == 'w': + pid = str(os.getpid()) + fname1 = self.basedir + '/.' + self.fname + '-' + pid + fname2 = self.basedir + '/' + self.fname + os.rename(fname1, fname2) + self.mode = 'c' + + def cancel(self): + "Like close() but don't save the entry." + if self.file: + self.file.close() + sys.stdout = self.real_stdout + if self.mode == 'w': + pid = str(os.getpid()) + fname = self.basedir + '/.' + self.fname + '-' + pid + os.unlink(fname) + self.mode = 'c' + + # # darcs repo manipulation # @@ -870,7 +967,9 @@ alt = False for p in ps: - if alt: + if p.name.startswith("TAG "): + print '' + elif alt: print '' else: print '' @@ -890,7 +989,7 @@ """ % { 'age': how_old(p.local_date), - 'author': p.shortauthor, + 'author': shorten_str(p.shortauthor, 26), 'myrname': config.myreponame, 'hash': p.hash, 'name': shorten_str(p.name), @@ -1620,6 +1719,86 @@ print_log(topi = topi) print_footer() +def do_atom(): + print "Content-type: application/atom+xml; charset=utf-8\n" + print '' + inv = config.repodir + '/_darcs/inventory' + repo_lastmod = os.stat(inv).st_mtime + str_lastmod = time.strftime(iso_datetime, + time.localtime(repo_lastmod)) + + print """ + + %(reponame)s darcs repository + + + %(url)s + darcs repository (several authors) + darcsweb.cgi + %(lastmod)s + %(desc)s + """ % { + 'reponame': config.reponame, + 'url': config.myurl + '/' + config.myreponame, + 'desc': config.repodesc, + 'lastmod': str_lastmod, + } + + ps = get_last_patches(20) + for p in ps: + title = time.strftime('%d %b %H:%M', time.localtime(p.date)) + title += ' - ' + p.name + pdate = time.strftime(iso_datetime, + time.localtime(p.date)) + link = '%s/%s;a=commit;h=%s' % (config.myurl, + config.myreponame, p.hash) + + addr, author = email.Utils.parseaddr(p.author) + if not addr: + addr = "unknown_email@example.com" + if not author: + author = addr + + print """ + + %(title)s + + %(author)s + %(email)s + + %(pdate)s + %(link)s + + %(desc)s +

+ """ % { + 'title': escape(title), + 'author': author, + 'email': addr, + 'url': config.myurl + '/' + config.myreponame, + 'pdate': pdate, + 'myrname': config.myreponame, + 'hash': p.hash, + 'pname': escape(p.name), + 'link': link, + 'desc': escape(p.name), + } + + # TODO: allow to get plain text, not HTML? + print escape(p.name) + '
' + if p.comment: + print '
' + print escape(p.comment).replace('\n', '
\n') + print '
' + print '
' + changed = p.adds + p.removes + p.modifies.keys() + \ + p.moves.keys() + p.diradds + p.dirremoves + \ + p.replaces.keys() + for i in changed: # TODO: link to the file + print '%s
' % i + print '

' + print '' + print '' def do_rss(): print "Content-type: text/xml; charset=utf-8\n" @@ -1691,7 +1870,6 @@ print_header() print "

Error! Malformed query

" print_footer() - sys.exit(1) def do_listrepos(): @@ -1783,6 +1961,8 @@ if 'multidir' not in dir(c): continue + if 'exclude' not in dir(c): + c.exclude = [] entries = os.listdir(c.multidir) entries.sort() for name in entries: @@ -1791,6 +1971,8 @@ fulldir = c.multidir + '/' + name if not os.path.isdir(fulldir + '/_darcs'): continue + if name in c.exclude: + continue rdir = fulldir desc = c.repodesc % { 'name': name } @@ -1854,6 +2036,11 @@ These are all the available repositories.
""" + if "cachedir" in dir(base): + config.cachedir = base.cachedir + else: + config.cachedir = None + if name and "footer" in dir(c): config.footer = c.footer elif "footer" in dir(base): @@ -1863,6 +2050,7 @@ + "crece desde el pie" + # # main # @@ -1886,6 +2074,18 @@ else: action = filter_act(form["a"].value) +# check if we have the page in the cache +if config.cachedir: + url_request = os.environ['QUERY_STRING'] + cache = Cache(config.cachedir, url_request) + if cache.open(): + # we have a hit, dump and run + cache.dump() + cache.close() + sys.exit(0) + # if there is a miss, the cache will step over stdout, intercepting + # all "print"s and writing them to the cache file automatically + # see what should we do according to the received action if action == "summary": @@ -2012,9 +2212,16 @@ elif action == 'rss': do_rss() +elif action == 'atom': + do_atom() + else: action = "invalid query" do_die() + if config.cachedir: + cache.cancel() +if config.cachedir: + cache.close() diff -rN -u old-darcsweb/style.css new-darcsweb/style.css --- old-darcsweb/style.css 2005-12-19 20:33:22.000000000 -0300 +++ new-darcsweb/style.css 2005-12-19 20:33:22.000000000 -0300 @@ -158,6 +158,14 @@ background-color:#edece6; } +.tag { + background-color:#f0f0ff; +} + +.tag:hover { + background-color:#e0e0ff; +} + td { padding:2px 5px; font-size:12px;