Request handler

[nuarhu]

This is a pull request that contains and replaces #112 , addressing #51 and #52 :

Additionally to specifying request handlers at the method level, it allows to specify a search_handler as instance variable. Also, a new instance variable use_qt_param is available that allows to change how request handlers are specified in the SOLR URL. The default SOLR behavior is to expect the handler as URL path variable (current #112 implementation). The pre 3.6 default behavior (and still available via configuration) specifies the request handler via qt parameter in the URL.

This pull request will allow to specify a custom request handler at construction time and thus allow to configure it as data source for Haystack like this:

HAYSTACK_CONNECTIONS = {
    'default': {
        'ENGINE': 'haystack.backends.solr_backend.SolrEngine',
        'URL': 'http://127.0.0.1:8983/solr/core0'
    },
    'autocomplete': {
        'ENGINE': 'haystack.backends.solr_backend.SolrEngine',
        'URL': 'http://127.0.0.1:8983/solr/core0',
        'KWARGS': {
            'search_handler': '/autocomplete',
            #'use_qt_param': False,  # False by default
        }
    },
}

Update README to include documentation of custom handlers.

I had to change the import statement in get-solr-download-url.py to make the tests run. (Using 3.4 but actually my fix should be for python3 in general.)

[nuarhu]

I've rebased this onto toastdriven/master.

[nuarhu]

Rebased onto current master.

[nuarhu]

The following test run fails:

Python 3.3.; SOLRCLOUD=False

During all of the tests in this run, no connection can be established to SOLR. This works in all the other test runs. It might be a one time thing?

Failed to establish a new connection: [Errno 111] Connection refused

I assume this means that my tests are fine, all over, seeing that all of the other runs finish just fine.

[acdha]

@nuarhu After restarting that build, it passed – we definitely see sporadic failures caused by how long it can take Solr to startup. I actually just made 05f0674 to reduce that, although it wouldn't have helped with this particular problem.

[nuarhu]

@acdha Thanks for checking! I hope everything is fine with my rebase and you are fine merging it.

[nuarhu]

rebased - again! Please consider merging this?