def get_taxbot_knowledge():
global taxbot_knowledge
if taxbot_knowledge is not None:
return taxbot_knowledge
sink = TaxonomySink()
try:
run(file("/home/sioclog/taxonomy.log"), sink)
except:
print_exc()
taxbot_knowledge = sink.taxonomy
return taxbot_knowledge
def runcgi(datarooturi, logfiles):
HTTP_HOST = os.environ.get('HTTP_HOST', "")
SERVER_PORT = os.environ.get('SERVER_PORT', "")
REQUEST_URI = os.environ.get('REQUEST_URI', "")
HTTP_ACCEPT = os.environ.get('HTTP_ACCEPT', "")
PATH_INFO = os.environ.get('PATH_INFO', "")
query = cgi.FieldStorage()
up_to = query.getfirst("up_to", None)
# channel = query.getfirst("channel", "")
# timeprefix = query.getfirst("time", "")
if PATH_INFO == "/styles.css":
print "Content-type: text/css"
print
css_stylesheet()
return
elif PATH_INFO == "/sitemap.xml":
print "Content-type: text/xml"
sink = ChannelsAndDaysSink()
run(logfiles, sink)
sitemap_index(sink, datarooturi)
return
if REQUEST_URI.endswith(".html"):
extension = ".html"
format = "html"
elif REQUEST_URI.endswith(".turtle"):
extension = ".turtle"
format = "turtle"
elif REQUEST_URI.endswith(".ttl"):
extension = ".ttl"
format = "turtle"
elif REQUEST_URI.endswith(".txt"):
extension = ".txt"
format = "raw"
else:
# XXX do real content negotiation, e.g. mimeparse.py
extension = ""
if "turtle" in HTTP_ACCEPT:
format = "turtle"
elif "html" in HTTP_ACCEPT:
format = "html"
elif "text" in HTTP_ACCEPT:
format = "raw"
else:
if "Googlebot" in os.environ.get('HTTP_USER_AGENT', ""):
format = "html" # Accept: */* isn't the full truth...
else:
format = "turtle" # default
parts = PATH_INFO.split('/')
# remove extension if any:
if parts[-1].endswith(extension):
parts[-1] = parts[-1][:-len(extension) or None]
if len(parts) > 1 and parts[1] not in ["channels", "users", "backlog"]:
parts.insert(1, "channels") # XXX default type for now
if len(parts) > 1:
restype = parts[1]
else:
restype = ""
if len(parts) > 2:
channel = parts[2]
else:
channel = ""
if len(parts) > 3:
timeprefix = parts[3]
else:
timeprefix = ""
# XXX the following assumes http over port 80, no QUERY_STRING
requesturi = "http://"+HTTP_HOST+REQUEST_URI
datauri = requesturi.split("?")[0] # exclude QUERY_STRING
# remove extension if any, to reset content negotiation in datauri:
if datauri.endswith(extension):
datauri = datauri[:-len(extension) or None]
crumbs = list(create_index_crumbs(datarooturi, datauri, restype, channel,
timeprefix))
if format == "html":
print "Content-type: text/html; charset=utf-8"
print
elif format == "turtle":
print "Content-type: application/x-turtle; charset=utf-8"
print
elif format == "raw":
print "Content-type: text/plain; charset=utf-8"
print
if restype == "users" and channel:
sink = ChannelsAndDaysSink()
run(logfiles, sink)
latestsink = EventSink(datarooturi, None, None, datauri)
latestpipeline = OffFilter(UserFilter(channel, ChannelMessageTailFilter(1, AddLinksFilter(latestsink))))
run(logfiles, latestpipeline)
render_user(sink, format, crumbs, datarooturi, channel, datauri, latestsink)
elif restype == "users":
sink = ChannelsAndDaysSink()
run(logfiles, sink)
render_user_index(sink, format, crumbs, datarooturi, datauri)
elif channel and timeprefix:
# show log
if format == "html":
if restype == "backlog":
# FIXME temporary hack to get the params right:
nick = channel
channel = timeprefix
timeprefix = nick
sink = AddLinksFilter(BackLogHtmlSink(nick, up_to, crumbs, datarooturi, channel, timeprefix, datauri))
timeprefix = ""
else:
sink = AddLinksFilter(HtmlSink(crumbs, datarooturi, channel, timeprefix, datauri))
elif format == "turtle":
sink = AddLinksFilter(TurtleSink(datarooturi, channel, timeprefix))
elif format == "raw":
sink = RawSink()
pipeline = OffFilter(ChannelFilter('#'+channel,
TimeFilter(timeprefix,
sink
)
))
run(logfiles, pipeline)
else:
# show index
sink = ChannelsAndDaysSink()
if channel:
pipeline = ChannelFilter('#'+channel, sink)
elif timeprefix:
pipeline = TimeFilter(timeprefix, sink)
else:
pipeline = sink
run(logfiles, pipeline)
if format == "html":
html_index(sink, crumbs, datarooturi, datauri, channel)
elif format == "turtle":
turtle_index(sink, datarooturi, datauri, channel)
