Uniform access to HTTP resources

[simon04]

(This is meant as a broader follow-up of #12220.)

Currently, we have various ways to access HTTP resources. Especially, currently we have 3 implementations to follow HTTP redirects, which do not cover all possible HTTP calls.

src/org/openstreetmap/josm/tools/Utils.java
943:                    connection = openHttpConnection(new URL(connection.getHeaderField("Location")), keepAlive);

src/org/openstreetmap/josm/tools/HttpClient.java
85:                final String redirectLocation = connection.getHeaderField("Location");

src/org/openstreetmap/josm/data/cache/JCSCachedTileLoaderJob.java
327:                    urlConn = getURLConnection(new URL(urlConn.getHeaderField("Location")));
486:                urlConn = getURLConnection(new URL(urlConn.getHeaderField("Location")));

Despite having some utility functions, many classes perform operations directly on HttpURLConnections.

Inspired by ​https://github.com/google/google-http-java-client and our very handy ImageProvider, I implemented a HttpClient class.

My final goal is to replace all Utils.openURL* and HttpURLConnection usages with this new class.

[Don-vip]

Nice :) I think you can go ahead :)

[simon04]

In 9168/josm:

see #12231 - Uniform access to HTTP resources

[simon04]

In 9169/josm:

see #12231 - HttpClient: add support to uncompress streams

[simon04]

In 9170/josm:

see #12231 - Deprecate Utils.open* functions

[simon04]

In 9171/josm:

see #12231 - Use HttpClient instead of some Utils.openHttpConnection usages

[simon04]

In 9172/josm:

see #12231 - Use HttpClient for OSM API calls

This requires adaptors to the OAuth library: SignpostAdapters

[simon04]

In 9174/josm:

see #12231 - Update unit tests, fix regression

[simon04]

In 9175/josm:

see #12231 - Use HttpClient instead of remaining Utils.setupURLConnection

[simon04]

The DirectUpload plugin needs to be adapted in org.openstreetmap.josm.plugins.DirectUpload.UploadOsmConnection#addAuthHack – fixing a hack :( …

[Don-vip]

In 9177/josm:

see #12231 - javadoc

[Don-vip]

In 9178/josm:

see #12231 - add support for setFixedLengthStreamingMode (used in DirectUpload plugin)

[Don-vip]

In 9179/josm:

see #12231 - initialize API to use progress monitor during HTTP connection

[Don-vip]

Replying to simon04:

The DirectUpload plugin needs to be adapted in org.openstreetmap.josm.plugins.DirectUpload.UploadOsmConnection#addAuthHack – fixing a hack :( …

I have made some changes in r9178:9179 + [o31871:31872] in order to fix the build but this is not finished yet, the plugin now lacks the ability of displaying upload progress as before. Can you please look into that?

[Don-vip]

Also, two unit tests are now failing:

MultiFetchServerObjectReaderTest

OsmServerBackreferenceReaderTest

For both of them we have:

WARNING: Already here java.net.SocketTimeoutException: Read timed out
WARNING: Already here java.net.SocketTimeoutException: Read timed out
WARNING: Already here java.net.SocketTimeoutException: Read timed out
WARNING: Already here java.net.SocketTimeoutException: Read timed out
WARNING: Already here java.net.SocketTimeoutException: Read timed out

[simon04]

In 9182/josm:

see #12231 - No read timeout for OsmApi connections

Fixes regression from r9172

[simon04]

Replying to Don-vip:

I have made some changes in r9178:9179 + [o31871:31872] in order to fix the build but this is not finished yet, the plugin now lacks the ability of displaying upload progress as before. Can you please look into that?

Wow, thank you for taking care of the plugin! So what we would need is basically the analogue of ProgressInputStream when sending data to connection.getOutputStream?

[Don-vip]

It sounds nice, yes :)

[stoecker]

See #12236 for a possible regression.

[simon04]

In 9184/josm:

see #12231 - fix MultiFetchServerObjectReaderTest#multiGetWithNonExistingNode

[simon04]

Replying to comment:18

In 9185/josm:

see #12231 - HttpClient now reports status to ProgressMonitor

[Don-vip]

Replying to simon04:

Replying to comment:18

In 9185/josm:

see #12231 - HttpClient now reports status to ProgressMonitor

thanks! :)

[simon04]

In 9190/josm:

see #12231 - Fix logging of reasonForRequest

[Don-vip]

Replying to simon04:

In 9190/josm:

see #12231 - Fix logging of reasonForRequest

is there a reason to not use !reasonForRequest.isEmpty() ?

[simon04]

In 9200/josm:

see #12231 - Code style (reasonForRequest)

[simon04]

Replying to Don-vip:

is there a reason to not use !reasonForRequest.isEmpty() ?

No, I must have been confused while taking over this feature from https://josm.openstreetmap.de/browser/josm/trunk/src/org/openstreetmap/josm/io/OsmServerReader.java?rev=9078#L153

[simon04]

In 9314/josm:

fix #7094 see #12231 - HttpClient: set Content-Length for all POST/PUT/DELETE requests