private static final void parseDateTime(final Calendar calendar, final CharSequence str) { final Matcher m = P_DATE_TIME.matcher(str); if (!m.matches()) throw new RuntimeException("cannot parse: '" + str + "'"); ParserUtils.parseGermanDate(calendar, m.group(1)); ParserUtils.parseEuropeanTime(calendar, m.group(2)); }
@Override public GetConnectionDetailsResult getConnectionDetails(final String uri) throws IOException { final CharSequence page = ParserUtils.scrape(uri); final Matcher mError = P_CONNECTION_DETAILS_ERROR.matcher(page); if (mError.find()) { if (mError.group(1) != null) throw new SessionExpiredException(); } final Matcher mHead = P_CONNECTION_DETAILS_HEAD.matcher(page); if (mHead.matches()) { final List<Connection.Part> parts = new ArrayList<Connection.Part>(4); Date firstDepartureTime = null; Location firstDeparture = null; Date lastArrivalTime = null; Location lastArrival = null; final Matcher mDetCoarse = P_CONNECTION_DETAILS_COARSE.matcher(mHead.group(1)); while (mDetCoarse.find()) { final String section = mDetCoarse.group(1); if (P_CONNECTION_DETAILS_MESSAGES.matcher(section).find()) { // ignore message for now } else { final Matcher mDetFine = P_CONNECTION_DETAILS_FINE.matcher(section); if (mDetFine.matches()) { final Location departure = new Location( LocationType.ANY, 0, null, ParserUtils.resolveEntities(mDetFine.group(1))); if (departure != null && firstDeparture == null) firstDeparture = departure; if (mDetFine.group(2) != null) { final String lineStr = normalizeLine(ParserUtils.resolveEntities(mDetFine.group(2))); final Line line = new Line(null, lineStr, lineColors(lineStr)); final Calendar departureTime = new GregorianCalendar(timeZone()); departureTime.clear(); ParserUtils.parseEuropeanTime(departureTime, mDetFine.group(3)); ParserUtils.parseGermanDate(departureTime, mDetFine.group(5)); final String departurePosition = ParserUtils.resolveEntities(mDetFine.group(4)); final Location arrival = new Location( LocationType.ANY, 0, null, ParserUtils.resolveEntities(mDetFine.group(6))); final Calendar arrivalTime = new GregorianCalendar(timeZone()); arrivalTime.clear(); ParserUtils.parseEuropeanTime(arrivalTime, mDetFine.group(7)); ParserUtils.parseGermanDate(arrivalTime, mDetFine.group(9)); final String arrivalPosition = ParserUtils.resolveEntities(mDetFine.group(8)); parts.add( new Connection.Trip( line, null, departureTime.getTime(), departurePosition, departure, arrivalTime.getTime(), arrivalPosition, arrival, null, null)); if (firstDepartureTime == null) firstDepartureTime = departureTime.getTime(); lastArrival = arrival; lastArrivalTime = arrivalTime.getTime(); } else if (mDetFine.group(10) != null) { final String min = mDetFine.group(10); final Location arrival = new Location( LocationType.ANY, 0, null, ParserUtils.resolveEntities(mDetFine.group(11))); if (parts.size() > 0 && parts.get(parts.size() - 1) instanceof Connection.Footway) { final Connection.Footway lastFootway = (Connection.Footway) parts.remove(parts.size() - 1); parts.add( new Connection.Footway( lastFootway.min + Integer.parseInt(min), lastFootway.departure, arrival, null)); } else { parts.add(new Connection.Footway(Integer.parseInt(min), departure, arrival, null)); } lastArrival = arrival; } else { final Location arrival = new Location( LocationType.ANY, 0, null, ParserUtils.resolveEntities(mDetFine.group(12))); parts.add(new Connection.Footway(0, departure, arrival, null)); } } else { throw new IllegalArgumentException("cannot parse '" + section + "' on " + uri); } } } // verify if (firstDepartureTime == null || lastArrivalTime == null) throw new IllegalStateException( "could not parse all parts of:\n" + mHead.group(1) + "\n" + parts); return new GetConnectionDetailsResult( new GregorianCalendar(timeZone()).getTime(), new Connection( AbstractHafasProvider.extractConnectionId(uri), uri, firstDepartureTime, lastArrivalTime, firstDeparture, lastArrival, parts, null, null)); } else { throw new IOException(page.toString()); } }
public QueryDeparturesResult queryDepartures( final int stationId, final int maxDepartures, final boolean equivs) throws IOException { final ResultHeader header = new ResultHeader(SERVER_PRODUCT); final QueryDeparturesResult result = new QueryDeparturesResult(header); if (stationId < 1000000) // live { // scrape page final String uri = departuresQueryLiveUri(stationId); final CharSequence page = ParserUtils.scrape(uri); final Matcher mError = P_DEPARTURES_LIVE_ERRORS.matcher(page); if (mError.find()) { if (mError.group(1) != null) return new QueryDeparturesResult(header, QueryDeparturesResult.Status.INVALID_STATION); if (mError.group(2) != null) return new QueryDeparturesResult(header, QueryDeparturesResult.Status.SERVICE_DOWN); if (mError.group(3) != null) throw new UnexpectedRedirectException(); } // parse page final Matcher mHead = P_DEPARTURES_LIVE_HEAD.matcher(page); if (mHead.matches()) { final String[] placeAndName = splitPlaceAndName(ParserUtils.resolveEntities(mHead.group(1))); final Calendar currentTime = new GregorianCalendar(timeZone()); currentTime.clear(); parseDateTime(currentTime, mHead.group(2)); final Map<String, String> messages = new HashMap<String, String>(); final Matcher mMsgsCoarse = P_DEPARTURES_LIVE_MSGS_COARSE.matcher(page); while (mMsgsCoarse.find()) { final Matcher mMsgsFine = P_DEPARTURES_LIVE_MSGS_FINE.matcher(mMsgsCoarse.group(1)); if (mMsgsFine.matches()) { final String lineName = ParserUtils.resolveEntities(mMsgsFine.group(1)); final char linePproduct = normalizeType(categoryFromName(lineName)); final Line line = newLine(linePproduct, normalizeLineName(lineName), null); final String message = ParserUtils.resolveEntities(mMsgsFine.group(3)).replace('\n', ' '); messages.put(line.label, message); } else { throw new IllegalArgumentException( "cannot parse '" + mMsgsCoarse.group(1) + "' on " + uri); } } final List<Departure> departures = new ArrayList<Departure>(8); final Matcher mDepCoarse = P_DEPARTURES_LIVE_COARSE.matcher(page); while (mDepCoarse.find()) { final Matcher mDepFine = P_DEPARTURES_LIVE_FINE.matcher(mDepCoarse.group(1)); if (mDepFine.matches()) { final Calendar parsedTime = new GregorianCalendar(timeZone()); parsedTime.setTimeInMillis(currentTime.getTimeInMillis()); ParserUtils.parseEuropeanTime(parsedTime, mDepFine.group(1)); if (parsedTime.getTimeInMillis() - currentTime.getTimeInMillis() < -PARSER_DAY_ROLLOVER_THRESHOLD_MS) parsedTime.add(Calendar.DAY_OF_MONTH, 1); boolean isPlanned = mDepFine.group(2) != null; Date plannedTime = null; Date predictedTime = null; if (!isPlanned) predictedTime = parsedTime.getTime(); else plannedTime = parsedTime.getTime(); final String lineName = ParserUtils.resolveEntities(mDepFine.group(3)); final char lineProduct = normalizeType(categoryFromName(lineName)); final Line line = newLine(lineProduct, normalizeLineName(lineName), null); final Position position = null; final String[] destinationPlaceAndName = splitPlaceAndName(ParserUtils.resolveEntities(mDepFine.group(4))); final Location destination = new Location( LocationType.ANY, 0, destinationPlaceAndName[0], destinationPlaceAndName[1]); final String message = messages.get(line.label); final Departure dep = new Departure( plannedTime, predictedTime, line, position, destination, null, message); if (!departures.contains(dep)) departures.add(dep); } else { throw new IllegalArgumentException( "cannot parse '" + mDepCoarse.group(1) + "' on " + uri); } } result.stationDepartures.add( new StationDepartures( new Location(LocationType.STATION, stationId, placeAndName[0], placeAndName[1]), departures, null)); return result; } else { throw new IllegalArgumentException("cannot parse '" + page + "' on " + uri); } } else { // scrape page final String uri = departuresQueryPlanUri(stationId, maxDepartures); final CharSequence page = ParserUtils.scrape(uri); final Matcher mError = P_DEPARTURES_PLAN_ERRORS.matcher(page); if (mError.find()) { if (mError.group(1) != null) return new QueryDeparturesResult(header, QueryDeparturesResult.Status.INVALID_STATION); if (mError.group(2) != null) return new QueryDeparturesResult(header, QueryDeparturesResult.Status.SERVICE_DOWN); if (mError.group(3) != null) throw new UnexpectedRedirectException(); } // parse page final Matcher mHead = P_DEPARTURES_PLAN_HEAD.matcher(page); if (mHead.matches()) { final String[] placeAndName = splitPlaceAndName(ParserUtils.resolveEntities(mHead.group(1))); final Calendar currentTime = new GregorianCalendar(timeZone()); currentTime.clear(); ParserUtils.parseGermanDate(currentTime, mHead.group(2)); final List<Departure> departures = new ArrayList<Departure>(8); final Matcher mDepCoarse = P_DEPARTURES_PLAN_COARSE.matcher(page); while (mDepCoarse.find()) { final Matcher mDepFine = P_DEPARTURES_PLAN_FINE.matcher(mDepCoarse.group(1)); if (mDepFine.matches()) { final Calendar parsedTime = new GregorianCalendar(timeZone()); parsedTime.setTimeInMillis(currentTime.getTimeInMillis()); ParserUtils.parseEuropeanTime(parsedTime, mDepFine.group(1)); if (parsedTime.getTimeInMillis() - currentTime.getTimeInMillis() < -PARSER_DAY_ROLLOVER_THRESHOLD_MS) parsedTime.add(Calendar.DAY_OF_MONTH, 1); final Date plannedTime = parsedTime.getTime(); final String lineName = ParserUtils.resolveEntities(mDepFine.group(2)); final char lineProduct = normalizeType(categoryFromName(lineName)); final Line line = newLine(lineProduct, normalizeLineName(lineName), null); final Position position = new Position(ParserUtils.resolveEntities(mDepFine.group(3))); final int destinationId = Integer.parseInt(mDepFine.group(4)); final String[] destinationPlaceAndName = splitPlaceAndName(ParserUtils.resolveEntities(mDepFine.group(5))); final Location destination = new Location( destinationId > 0 ? LocationType.STATION : LocationType.ANY, destinationId, destinationPlaceAndName[0], destinationPlaceAndName[1]); final Departure dep = new Departure(plannedTime, null, line, position, destination, null, null); if (!departures.contains(dep)) departures.add(dep); } else { throw new IllegalArgumentException( "cannot parse '" + mDepCoarse.group(1) + "' on " + uri); } } result.stationDepartures.add( new StationDepartures( new Location(LocationType.STATION, stationId, placeAndName[0], placeAndName[1]), departures, null)); return result; } else { throw new IllegalArgumentException("cannot parse '" + page + "' on " + uri); } } }
private QueryConnectionsResult queryConnections(final String uri, final CharSequence page) throws IOException { final Matcher mError = P_CHECK_CONNECTIONS_ERROR.matcher(page); if (mError.find()) { if (mError.group(1) != null) return new QueryConnectionsResult(null, QueryConnectionsResult.Status.TOO_CLOSE); if (mError.group(2) != null) return new QueryConnectionsResult(null, QueryConnectionsResult.Status.UNRESOLVABLE_ADDRESS); if (mError.group(3) != null) return new QueryConnectionsResult(null, QueryConnectionsResult.Status.NO_CONNECTIONS); if (mError.group(4) != null) return new QueryConnectionsResult(null, QueryConnectionsResult.Status.INVALID_DATE); if (mError.group(5) != null) throw new SessionExpiredException(); } final Matcher mHead = P_CONNECTIONS_HEAD.matcher(page); if (mHead.matches()) { final Location from = new Location(LocationType.ANY, 0, null, ParserUtils.resolveEntities(mHead.group(1))); final Location to = new Location(LocationType.ANY, 0, null, ParserUtils.resolveEntities(mHead.group(2))); final Calendar currentDate = new GregorianCalendar(timeZone()); currentDate.clear(); ParserUtils.parseGermanDate(currentDate, mHead.group(3)); // final String linkEarlier = mHead.group(4) != null ? // ParserUtils.resolveEntities(mHead.group(4)) : null; final String linkLater = mHead.group(5) != null ? ParserUtils.resolveEntities(mHead.group(5)) : null; final List<Connection> connections = new ArrayList<Connection>(); final Matcher mConCoarse = P_CONNECTIONS_COARSE.matcher(page); while (mConCoarse.find()) { final Matcher mConFine = P_CONNECTIONS_FINE.matcher(mConCoarse.group(1)); if (mConFine.matches()) { final String link = ParserUtils.resolveEntities(mConFine.group(1)); final Calendar departureTime = new GregorianCalendar(timeZone()); departureTime.setTimeInMillis(currentDate.getTimeInMillis()); ParserUtils.parseEuropeanTime(departureTime, mConFine.group(2)); if (!connections.isEmpty()) { final long diff = departureTime.getTimeInMillis() - connections.get(connections.size() - 1).departureTime.getTime(); if (diff > PARSER_DAY_ROLLOVER_THRESHOLD_MS) departureTime.add(Calendar.DAY_OF_YEAR, -1); else if (diff < -PARSER_DAY_ROLLOVER_THRESHOLD_MS) departureTime.add(Calendar.DAY_OF_YEAR, 1); } final Calendar arrivalTime = new GregorianCalendar(timeZone()); arrivalTime.setTimeInMillis(currentDate.getTimeInMillis()); ParserUtils.parseEuropeanTime(arrivalTime, mConFine.group(3)); if (departureTime.after(arrivalTime)) arrivalTime.add(Calendar.DAY_OF_YEAR, 1); final Connection connection = new Connection( AbstractHafasProvider.extractConnectionId(link), link, departureTime.getTime(), arrivalTime.getTime(), from, to, null, null, null); connections.add(connection); } else { throw new IllegalArgumentException( "cannot parse '" + mConCoarse.group(1) + "' on " + uri); } } return new QueryConnectionsResult( new ResultHeader(SERVER_PRODUCT), uri, from, null, to, linkLater, connections); } else { throw new IOException(page.toString()); } }