mirrors/public-transport-enabler
1
0
Fork 0
mirror of https://gitlab.com/oeffi/public-transport-enabler.git synced 2025-07-17 18:29:51 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

split departures head parsing in coarse and fine

Browse source 
git-svn-id: https://public-transport-enabler.googlecode.com/svn/trunk@74 0924bc21-9374-b0fa-ee44-9ff1593b38f0
This commit is contained in:
andreas.schildbach 2010-08-12 19:02:37 +00:00
parent c0ecbe26f0
commit 187537ef7e
1 changed files with 47 additions and 37 deletions
Download patch file Download diff file Expand all files Collapse all files

84
src/de/schildbach/pte/MvvProvider.java
View file

@ -474,17 +474,19 @@ public class MvvProvider implements NetworkProvider
return DEPARTURE_URL + stationId; return DEPARTURE_URL + stationId;
} }
private static final Pattern P_DEPARTURES_HEAD = Pattern.compile(".*Von:[\\xa0\\s]*</b>(.*?)<br />.*?" // private static final Pattern P_DEPARTURES_HEAD_COARSE = Pattern.compile(".*?<body>(.*?Linie/Richtung.*?)</body>.*?", Pattern.DOTALL);
+ "Datum:[\\xa0\\s]*</b>\\w{2}\\.,\\s(\\d+)\\.\\s(\\w{3})\\.[\\xa0\\s]+(\\d{4})<br />.*", Pattern.DOTALL); private static final Pattern P_DEPARTURES_HEAD_FINE = Pattern.compile(".*?" //
+ "Von:[\\xa0\\s]*</b>(.*?)<br />.*?" // location
+ "Datum:[\\xa0\\s]*</b>\\w{2}\\.,\\s(\\d+)\\.\\s(\\w{3})\\.[\\xa0\\s]+(\\d{4})<br />.*?" // date
, Pattern.DOTALL);
private static final Pattern P_DEPARTURES_COARSE = Pattern.compile("<tr valign=\"top\" bgcolor=\"#\\w{6}\">(.+?)</tr>", Pattern.DOTALL); private static final Pattern P_DEPARTURES_COARSE = Pattern.compile("<tr valign=\"top\" bgcolor=\"#\\w{6}\">(.+?)</tr>", Pattern.DOTALL);
private static final Pattern P_DEPARTURES_FINE = Pattern.compile(".*?" // private static final Pattern P_DEPARTURES_FINE = Pattern.compile(".*?" //
+ "<td>" //
+ "(?:[\\xa0\\s]*<font color=\"red\">[\\xa0\\s]*(\\d+)\\.(\\d+)\\.[\\xa0\\s]*</font>)?" // date + "(?:[\\xa0\\s]*<font color=\"red\">[\\xa0\\s]*(\\d+)\\.(\\d+)\\.[\\xa0\\s]*</font>)?" // date
+ "(\\d+):(\\d+)</td>.*?" // time + "(\\d+):(\\d+)</td>.*?" // time
+ "(?:<img src=\"images/means.*?\" alt=\"(.*?)\" />.*?)?" // product + "(?:<img src=\"images/means.*?\" alt=\"(.*?)\" />.*?)?" // product
+ "<td width=\"100\">\\s*([^<]*?)[\\xa0\\s]*(?:<a .*?</a>.*?)?" // line + "<td width=\"100\">\\s*([^<]*?)[\\xa0\\s]*(?:<a .*?</a>.*?)?" // line
+ "<br />\\s*(.*?)\\s*<br />.*?" // destination + "<br />\\s*(.*?)\\s*<br />.*?" // destination
+ "</td>.*?", Pattern.DOTALL); , Pattern.DOTALL);
private static final Pattern P_DEPARTURES_URI_STATION_ID = Pattern.compile("nameInfo_dm=(\\d+)"); private static final Pattern P_DEPARTURES_URI_STATION_ID = Pattern.compile("nameInfo_dm=(\\d+)");
public QueryDeparturesResult queryDepartures(final String uri) throws IOException public QueryDeparturesResult queryDepartures(final String uri) throws IOException
@ -496,42 +498,50 @@ public class MvvProvider implements NetworkProvider
throw new IllegalStateException(uri); throw new IllegalStateException(uri);
final int stationId = Integer.parseInt(mStationId.group(1)); final int stationId = Integer.parseInt(mStationId.group(1));
final Matcher mHead = P_DEPARTURES_HEAD.matcher(page); final Matcher mHeadCoarse = P_DEPARTURES_HEAD_COARSE.matcher(page);
if (mHead.matches()) if (mHeadCoarse.matches())
{ {
final String location = ParserUtils.resolveEntities(mHead.group(1)); final Matcher mHeadFine = P_DEPARTURES_HEAD_FINE.matcher(mHeadCoarse.group(1));
final Date currentTime = parseDate(mHead.group(2), mHead.group(3), mHead.group(4)); if (mHeadFine.matches())
final List<Departure> departures = new ArrayList<Departure>(8);
final Calendar calendar = new GregorianCalendar();
final Matcher mDepCoarse = P_DEPARTURES_COARSE.matcher(page);
while (mDepCoarse.find())
{ {
final Matcher mDepFine = P_DEPARTURES_FINE.matcher(mDepCoarse.group(1)); final String location = ParserUtils.resolveEntities(mHeadFine.group(1));
if (mDepFine.matches()) final Date currentTime = parseDate(mHeadFine.group(2), mHeadFine.group(3), mHeadFine.group(4));
{ final List<Departure> departures = new ArrayList<Departure>(8);
calendar.setTime(currentTime);
final String day = mDepFine.group(1);
if (day != null)
calendar.set(Calendar.DAY_OF_MONTH, Integer.parseInt(day));
final String month = mDepFine.group(2);
if (month != null)
calendar.set(Calendar.MONTH, Integer.parseInt(month) - 1);
calendar.set(Calendar.HOUR_OF_DAY, Integer.parseInt(mDepFine.group(3)));
calendar.set(Calendar.MINUTE, Integer.parseInt(mDepFine.group(4)));
final String normalizedLine = normalizeLine(mDepFine.group(5), mDepFine.group(6));
final String destination = normalizeStationName(mDepFine.group(7));
final Departure departure = new Departure(calendar.getTime(), normalizedLine, LINES.get(normalizedLine), 0, destination);
departures.add(departure);
}
else
{
throw new IllegalArgumentException("cannot parse '" + mDepCoarse.group(1) + "' on " + uri);
}
}
return new QueryDeparturesResult(uri, stationId, location, currentTime, departures); final Calendar calendar = new GregorianCalendar();
final Matcher mDepCoarse = P_DEPARTURES_COARSE.matcher(page);
while (mDepCoarse.find())
{
final Matcher mDepFine = P_DEPARTURES_FINE.matcher(mDepCoarse.group(1));
if (mDepFine.matches())
{
calendar.setTime(currentTime);
final String day = mDepFine.group(1);
if (day != null)
calendar.set(Calendar.DAY_OF_MONTH, Integer.parseInt(day));
final String month = mDepFine.group(2);
if (month != null)
calendar.set(Calendar.MONTH, Integer.parseInt(month) - 1);
calendar.set(Calendar.HOUR_OF_DAY, Integer.parseInt(mDepFine.group(3)));
calendar.set(Calendar.MINUTE, Integer.parseInt(mDepFine.group(4)));
final String normalizedLine = normalizeLine(mDepFine.group(5), mDepFine.group(6));
final String destination = normalizeStationName(mDepFine.group(7));
final Departure departure = new Departure(calendar.getTime(), normalizedLine, LINES.get(normalizedLine), 0, destination);
departures.add(departure);
}
else
{
throw new IllegalArgumentException("cannot parse '" + mDepCoarse.group(1) + "' on " + uri);
}
}
return new QueryDeparturesResult(uri, stationId, location, currentTime, departures);
}
else
{
throw new IllegalArgumentException("cannot parse '" + mHeadCoarse.group(1) + "' on " + uri);
}
} }
else else
{ {