def parseTimesResultRecursive(element, out, date): if "geo_facet_item" in element.tag: if element.text is not None: k = False parent = element.getparent().getparent() for c in parent: if "updated_date" in c.tag: publishdate = time.mktime(time.strptime(c.text[:-6], "%Y-%m-%dT%H:%M:%S")) publishdate = publishdate + int(c.text[-5:-3])*60*60 if publishdate > date: k = True lastupdate.append(publishdate) if k: out.add(('nyt', geoloc.getCoordinates(element.text.encode('utf-8')))) else: for child in element: parseTimesResultRecursive(child, out, date)
def parseTwitterResultRecursive(element, out, depth, date): if "location" in element.tag: if element.text is not None: k = False parent = element.getparent() for c in parent: if "published" in c.tag: publishdate = time.mktime(time.strptime(c.text, "%Y-%m-%dT%H:%M:%SZ")) if publishdate > date: k = True lastupdate.append(publishdate) if k: x = geoloc.getCoordinates(element.text.encode('utf-8')) if x is not None: out.add(('twitter', x)) else: if depth < 2: for child in element: parseTwitterResultRecursive(child, out, depth + 1, date)