One thing I forgot to mention in my original post is that I wish to do this using the SolrJ client. I have my own rest server that presents a common API to our users, but the back-end can be anything I wish. I have been using "that other Lucene based product" :), but I wish to stick to a product that is more open and that perhaps I can contribute to.
I've searched for SolrJ examples for child documents and unfortunately there are far too many references to implementations based off of older versions of Solr. Specifically, I would like to insert beans with multiple child collections in them, but the latest I've read says this is not currently possible. Is that still true? In short, It isn't so important that REST based requests / responses from Solr are pure JSON so long as I can do what I want from the java client. Do you know if there have been recent additions / enhancements up through 6.5 that make this more straight-forward? Thanks ----- Original Message ----- From: "Mikhail Khludnev" <m...@apache.org> To: "solr-user" <solr-user@lucene.apache.org> Sent: Thursday, April 20, 2017 3:38:11 PM Subject: Re: Advice on how to work with pure JSON data. This is one of the features of the epic https://issues.apache.org/jira/browse/SOLR-10144. Until it's done the only way to achieve this is to properly set many params for https://cwiki.apache.org/confluence/display/solr/Transforming+Result+Documents#TransformingResultDocuments-[subquery] Note, here I assume that children mapping is static ie there is a limited list of optional scopes. Indexing and searching arbitrary JSON is esoteric (XML DB like) problem. Also, beware of https://issues.apache.org/jira/browse/SOLR-10500. I hope to fix it soon. On Thu, Apr 20, 2017 at 10:15 PM, <russell.lemas...@comcast.net> wrote: > > I have looked at many examples on how to do what I want, but they tend to > only show fragments or they > are based on older versions of Solr. I'm hoping there are new features > that make what I'm doing easier. > > I am running version 6.5 and am testing by running in cloud mode but only > on a single machine. > > Basically, I have a large number of documents stored as JSON in individual > files. I want to take that JSON > document and index it without having to do any pre-processing, etc. I also > need to be able to write newly indexed > JSON data back to individual files in the same format. > > For example, let's say I have a json document that looks like the > following: > > { > "id" : "bb903493-55b0-421f-a83e-2199ea11e136", > "productName_s" : "UsefulWidget", > "productCategory_s" : "tool", > "suppliers" : [ > { > "id" : " bb903493-55b0-421f-a83e-2199ea11e221", > "name_s" : "Acme Tools", > "productNumber_s" : "10342UW" > }, { > "id" : " bb903493-55b0-421a-a83e-2199ea11e445", > "name_s" : "Snappy Tools", > "productNumber_s" : "ST-X100023" > } > ], > "resellers" : [ > { > "id" : "cc 903493-55b0-421f-a83e-2199ea11e221", > "name_s" : "Target", > "productSKU_s" : "TA092310342UW" > }, { > "id" : "bc903493-55b0-421a-a83e-2199ea11e445", > "name_s" : "Wal-Mart", > "productSKU_s" : "029342ABLSWM" > } > ] > } > > I know I can use the /update/json/docs handler to insert the above but > from what I understand, I'd have to set up parameters > telling it how to split the children, etc. Though that is a bit of a pain, > I can make that happen. > > The problem is that, when I then try to query for the data, it comes back > with _childDocuments_ instead of the names of the > child document lists. So, how can I have Solr return the document as it > was originally indexed (I know it would be embedded > in the results structure, but I can deal with that)? > > I am running version 6.5 and I am hoping there is a method I haven't seen > documented that can do this. If not, can someone > point me to some examples of how to do this another way. > > If there is no easy way to do this with the current version, can someone > point me to a good resource for writing my own > handlers? > > Thank you. > > > > > > > > > -- Sincerely yours Mikhail Khludnev
