how can i achieve this 'start' in query ? i've already tried:
res = solr.query('library',{ 'q':'*:*', 'start':9, 'rows':15, })
but still after i used res.get_result_count() it returned 15 instead 6.

I have an i3 laptop with 8GB ram & SSD, and it was running very slow (taking up all ram & probably ). Any idea how we can:

1. make it faster
   or
2. separate, so don't run tests for everything ?
   or both ?

I have reviewed SolrClient and found that I does not have an option to update some part of document in Solr. It would be greate if this feature could be added in it.

example:
python run_tests.py -py 3.5 -solr 6.3.0 -test test_client.ClientTestIndexing.test_down_solr_exception

To put it somewhere in docs ? Maybe at SolrVagrant?

Hi,

How can i query for multiple facets? It looks like SolrClient doesn't know how to do that but it should be possible

Regards,
Hans

Been using SolrClient in my latest project and it is by far the cleanest SOLR api for python out there, so kudos for that! Hoping it will keep being developed and supported for the foreseeable future.

One thing I see that could be improved is that indexing multi-million set of documents using IndexQ and the finalize=True argument will lead to the todo/ directory being overwhelmed (impossible to 'ls' normally nor 'rm -rf' due to size of directory listing metadata). I was thinking about creating subfolders under todo/ and put data there instead, capping it at certain number (0.5M files, for ex.).

I understand that it is bad to proceed like this when you understand how it all works, but for first time users and non-experts this could actually prevent this unfortunate case.

• Grouped Queries
• Facets

Hi,

My application code depends on the latest commit#a75912f73730d3aaaf64ca5684ab035c2666eedc where integer conversion step is removed. But, i see the PyPi package of SolrClient doesn't have that issue fixed. Could you please update the PyPi package with the latest version.

Thanks,
Abhay

Currently all params (query string) must go on the "params={}" keyword argument.
But this sometimes works and sometimes doesn't. Example where it doesn't is the "query()". You can't set currently a "route='ay'" in there, unless you put it in the query, which sucks.

So, how to do that ?
My idea would be to have stuff like route,min_rf, etc in kwargs, that then can be included in the "params" right before making a request. Makes sense ?

And if there are kwargs-keys that are needed on other parts (ex: a router may need a "prefer_leader", it can pop it from the kwargs) ?

Makes sense ? Or maybe a nicer way ?

Hello I am using SolrClient in my project.
I want to make faceted fields multiple by using res.get_facet_keys_as_list, I mean I wanna see more fields.
However, it is not correlated with Solr syntax.
In Solr, it is used like this:
"fl":"genres,movie_title,title_year"
Is there any possibility to make it ?
Thanks

read title

Add method for easy cursor mark paging through collection.

Add a method in main SolrClient that easily pages through the data in Solr.

{
"responseHeader":{
"status":400,
"QTime":2},
"error":{
"msg":"undefined field text",
"code":400}}

Hello community, first thank you for creating this wonderful SolrClient application, lately I am experiencing that sometimes when I perform querys or high volume commits it stays in indeterminate standby mode, in which I think at least from the client I should be able to handle this problem. wait time. I see at the level of connection to solr it does it with requests and I think that right there you can add the timeout parameter. That is, you could send something like: data_solr = conectionSolr.query (collection, query, timeout = 80) and inside the _send function it would be something like: res = self.session.request (method, url, params = params, data = data , headers = headers, verify = False, timeout = param_timeOut). How feasible would it be to integrate this function? For my part, I still feel like a novice to make a change like this ..., thank you very much for everything, I will be waiting for your contributions

PD: In the screenshot I send the line where I think you can add the timeOut option

Some errors in the log. I stopped it after some time.

solrtestslog.txt

Can you please make a list so I can create pull requests and we get a new release on pypi ?

I need the .index() command to NOT return True/False but return the full json.

Should I create a new index_raw(?) or just edit the .index(if it's not in any release yet) ? Since this changes the return type.

Reason, is I need to know the "rf" parameter when I set "min_rf".

SolrClient is at version 0.2.1 on PyPi, but the GitHub repository is stuck at version 0.2.0. It would be nice to have a tagged version here that corresponded with what's available on PyPi. Thanks!

https://github.com/moonlitesolutions/SolrClient/blob/master/SolrClient/solrclient.py#L130-L142.

My guess is that the code is intended to be:

data = json.dumps(docs)
return self.index_json(collection, data, params, **kwargs)

Hello,
I have recently worked with your package, and noticed that it has a somewhat strange behaviour when encountering fields that are only present in some of the documents.
In my example, I had let's say 2500 documents with an unique "id" field, but only 2000 or so would have another field called "links".
Iterating over both the "id" list and "links" list is not possible, since they have different lengths, since get_field_values_as_list("links") would return only a list of length 2000, and I cannot discern which pages return empty.
For this, I simply altered the function to:
return [doc[field] if field in doc else [] for doc in self.docs]
which solved the problem in my case. It might not always be needed, but possibly this could be included as get_field_values_as_full_list, or something like this.

Best regards,
Dennis

Example, I was thinking of also adding:

1. method getting cluster state
2. keep track of hash-->shard-->host linking
3. when route is specified, route directly to correct node(with failover to replicas)
   3.1. route directly to leader for create,update,delete
4. prefer_leader=True will always contact leader of shard first and then replicas.
5. prefer_leader=False will contact leader last
6. prefer_leader=None will random.choice(replicas)
7. when multiple route are added "tenant1,tenant2,tenant3", pick random replica to contact

Makes sense ?

The whole point is to contact minimum amount of nodes necessary, reducing network hops/traffic, lowering cpu.

SolrClient 0.2.1 fails to run on Python 3.7.0:

$ python test.py 
Traceback (most recent call last):
  File "test.py", line 1, in <module>
    from SolrClient import SolrClient
  File "/home/aorth/.local/share/virtualenvs/solr-pipenv-nV7koRrN/lib/python3.7/site-packages/SolrClient/__init__.py", line 1, in <module>
    from .solrclient import SolrClient
  File "/home/aorth/.local/share/virtualenvs/solr-pipenv-nV7koRrN/lib/python3.7/site-packages/SolrClient/solrclient.py", line 10, in <module>
    from .zk import ZK
  File "/home/aorth/.local/share/virtualenvs/solr-pipenv-nV7koRrN/lib/python3.7/site-packages/SolrClient/zk.py", line 9, in <module>
    from kazoo.client import KazooClient
  File "/home/aorth/.local/share/virtualenvs/solr-pipenv-nV7koRrN/lib/python3.7/site-packages/kazoo/client.py", line 62, in <module>
    from kazoo.recipe.partitioner import SetPartitioner
  File "/home/aorth/.local/share/virtualenvs/solr-pipenv-nV7koRrN/lib/python3.7/site-packages/kazoo/recipe/partitioner.py", line 193
    self._child_watching(self._allocate_transition, async=True)
                                                        ^
SyntaxError: invalid syntax

Virtual environment was created with pipenv.

In the SolrResponse class, get_facets and get_facets_ranges return ordinary dict objects, so ordered results as returned by Solr get lost. Should just need to be returned as an OrderedDict.

This looks easy enough to fix, I'll see if I can create a pull request for it.

Hello,
I'm trying to submit this query to Solr Core with SolrClient, but the expected result object doesn't match the query since keys are not sorted by facet count.
Here the query object passed to SolrClient:

onto_suggest = solr.query('merged', {
    "q": "*:*",
    "rows": 0,
    "facet": True,
    "facet.field": "onto_suggest",
    "facet.sort": "count",
    "facet.limit": -1,
    "facet.mincount": 2
})

I've also tried to use "facet.count" to sort, same result.
Here an excerpt of the result

{
  "onto_suggest": {
    "Arthrobacter siderocapsulatus": 433,
    "Enterobacter aglomerans": 39,
    "Genetic endocrine tumor": 32,
    "ETOP": 15,
    "lymphoid lineage restricted progenitor cell": 47,
    "Escherichia coli O157:H7 EDL933": 127,
    "deferoxaminum": 2,
    "Bacillus francki": 472,
    "srf": 9,
    "Malignant Peripheral Nerve Sheath Tumors": 9,
    "Pancreas Cancer": 122,