pentabarf_to_prerole.py

#!/usr/bin/env python
# -*- coding:utf8 -*-

import urllib
import lxml.etree
from subprocess import Popen

rohdaten = urllib.urlopen("http://events.ccc.de/camp/2011/Fahrplan/schedule.en.xml").read()
#daten = unicode(rohdaten, "latin1")
daten = rohdaten
root = lxml.etree.fromstring(daten).getroottree()

SCRIPT_FULLPATH = "genPrerole.py"
EXTRA_ARGS = [
        #"-o", "/path/to/output",
        #"-w", "/tmp/workdir"
        ]

def call_genPrerole(info):
    command = ["python", SCRIPT_FULLPATH]

    arguments = dict(t="title",
                     s="subtitle",
                     p="speakers",
                     d="date",
                     b="start",
                     i="id")

    for flag, key in arguments.iteritems():
        flag = "-" + flag
        try:
            data = info[key]
        except KeyError:
            print "key %s does not exist for %s" % (key, info["id"])

        if data is None:
            print "value %s was None for %s" % (key, info["id"])
            data = ""
        if isinstance(data, list):
            command += [flag] + data
        elif isinstance(data, basestring):
            command += [flag, data]
        else:
            raise ValueError("unexpected datatype")

    command += EXTRA_ARGS

    process = Popen(command)
    returncode = process.wait()
    if returncode != 0:
        print "returncode %d != 0 for %r" % (returncode, command)

def daten_von_event(event, upper_info):
    info = upper_info.copy()
    info["id"] = event.get("id")
    things = "start room slug title subtitle".split(" ")
    for thing in things:
        thing_tag = event.find(thing)
        if thing_tag is None:
            print "ERROR: no tag %s in %r" % (thing, event)
        else:
            if event.find(thing).text is not None:
                info[thing] = event.find(thing).text

    info["speakers"] = []
    persons = event.find("persons")
    for person in list(persons):
        info["speakers"].append(person.text)

    call_genPrerole(info)

def daten_von_raum(raum, upper_info):
    info = upper_info.copy()
    info["name"] = raum.get("name")
    for event in list(raum):
        daten_von_event(event, info)

def daten_von_tag(tag):
    info = {}
    info["date"] = tag.get("date")
    info["index"] = tag.get("index")
    for raum in list(tag):
        daten_von_raum(raum, info)

def daten_von_schedule(schedule):
    for tag in schedule.xpath("day"):
        daten_von_tag(tag)

daten_von_schedule(root)