| Sticky Revision: |
remember to typecast objects fetched from cache
Code Sync (use jsr107 / memcache for caching)
- Cache the XSLT template - Use a StringWriter for XML output instead of ByteArrayOutputStream (the stringwriter is giving correct charset on app engine)
Add a copy with partial support for google app engine
Just for fun use XSLT to produce html output
remember to put the arrival parameter in the output
Exprimental #3: update to changes in databeans
Experimental commit #2, move databeans to common
Experimental: use Simple (simple.sourceforge.net) for XML serialization
Seperate the metro-line identifiers with space
no need to check _all_ elements
Hopefully it works now ?!?
Re-add Debug log statements but with loglevel=fine
remove debug info
on www.bane.dk, dont get departures that already has left the station
URLencode the stationID used in request URL
Make metro plan extraction works in all cases ... and increase cache timeout to 1 minute
Add backend for fetching metro data
Sandbox: first try at making a parser for copenhagen metro
Also clean the timetable strings from www.bane.dk
Also make trainnumber extraction work with old site
Make jsoup act correct with old site
Write a log entry if no departures was found
also enable timetables to use both new azure and old www
Remove debug output
Enable usage of either the new azure parser or the old www.bane.dk parser
Bake bane.dk http timeout configurable
Lookup the stationName in the fetcher and put it in the departureBean That way we don't need db acccess on a cache hit and on a cache miss we saves one DB lookup operation
Don't show departures that already has departured
Not used anymore
Stationcode muste be URLencoded or bane.dk will use a lot of time on parsing the request
Experiment: remove the last bits of HtmlUnit
Experiment: switch to jsoup instead of htmlunit
Re-enable caching of depatures
adapt htmlview/jsp to new departurebean
Add code to extract notifications
Comment off-by-one error
refactor cancelledstring generation
Make cancelled strings from new site(trafikinfo.bane.dk) appear like they did on the old one
outcomment the temp site code - until we know whether it ever will be needed again
make extraction of s-tog number a little more elegant
Use enum instead of strings to determine the train-types (this saves some string comparisons)
remove debug log statements
Switch banedk fetchers to bane.dk's new site (http://trafikinfo.bane.dk/) ... this new site is some microsoft azure-based cloud thingie so hopefully it will have a better uptime/stability once we reach winter
include the arrival parameter in output xml
That implementation is not correct, so for now just throw an unsupported operation exception
Switch to servlet3.0 annotations
HtmlunitInvocation is a more correct name
Just send a plain error response instead of throwing exceptions
XML-Document size is reduced by ~13% just by limiting how many spaces should be used to indentation
Trim status field for white-space
Disable those XML header thingies
Show correct XML preamble
Add beginning of Rejseplanen fetcher code
Rename function
Re-enable the cache
Extract station ID's to timetable xml
remove debug output
Add backend support for reading arrivals
more correct impl
Remove old (pre-0.6.4) method of determining the train type.
in statistics, also keep track of how many request were unsuccessful
First take on some statistics
Better handling of train-types (regional vs s-train)
Move function getStationName() function do DAO class
Act as a Firefox 3 browser, and cleanup HtmlUnit WebClient instance when done
Rewrite to use a DAO class for DB access
Use JSTL in the jsp instead of scriptlets
Move some various parameters to web.xml
Forgot to remove debug log statement
Bug fix in temp site parser
Also make timetable degrade gracefully while bane.dk's temporary site is in place
Added a screenscraper for bane.dk's temporary site
Specify the exact return type of proceed
Updated with new circuitbreaker package names
Catch all IOExceptions
Nicer handling of incomplete requests
Degrade from throwable to exception
Fixed a bug, which caused a null pointer exception
Now it's (almost) a full map
unneeded import
Refactor TimeoutCache into a real Map
cleanup
Enabled the usage of Circuit breaker to guard agains bane.dk failures
Bugfix, better extraction of trainnumbers for S-trains
Don't log on cache miss - and loglevel=info should be sufficient for cache hit messages
Enable caching for departures and timetables
When bane.dk is pressed it needs some more time to respond
More graceful handling of timeout exceptions
Always save a time string, no matter whether bane.dk has one for us :)
Work arouund a bug in bane.dk (a departure at midnight sometimes occurs as a empty string)
No need for logging the requested bane.dk url
Added code for extracting a single train's timetable from bane.dk
Extract trainNumbers for S-train
Improve on error handling and logging
Make sure that the required table is present
work correctly when clock wraps around at midnight
Remove that frickin information 'i' from the note
Print a nice userfriendly string instead of the raw updated int code
Minor code tweaks
Finish new bane.dk screen-scraper
first version of new departure provider (this is supposed to replace the php mess)
| ViewVC Help | |
| Powered by ViewVC 1.1.20 |
Directory Listing