Solr query based on a string field's subset

I'd like to send a string to Solr and let it answer with all records which are a subset of that string.

The string I would send has intege开发者_如何学Cr numbers separated by spaces. I wanna make solr give me all records where a specific string field is a subset of the numbers I provide as the request string.

An example...

Imagine I have an string field indexed in Solr which is in reality a set of integers separated by space. For example, let's say I have the following record's field indexed in Solr:

• "888110"
• "888110 888120"
• "888110 888120 888130"
• "888110 888120 888130 888140"
• "888110 888130 888140"
• "888110 888140"
• "888140"
• "888120 888130"

I wanna Solr to receive a query with, for example, "888110 888140" and reply with the following records:

• "888110"
• "888110 888140"
• "888140"

If I query by "888110 888120 888130" the retrieved records would be...

• "888110"
• "888110 888120"
• "888110 888120 888130"
• "888120 888130"

The retrieved records must be exactly a subset of the numbers provided as a string.

Is it possible to make Solr behave like this?

I'm a bit confused why in the first example "888110" is not returned, but it is in the second example.

Anyways, if I understand generally what you are trying to do, I would be making a new field multi valued and use your boolean operators (AND ,OR) on the query.

eg in the schema

       <field name="code_string" ... />
       <field name="codes" ... multiValued="true"/>

so you have a document like

<doc>
    <arr name="codes">
       <str>811001</str>
       <str>811002</str>
    </arr>

and in your query

?=codes=811001 OR codes=811002 OR ....

In my experience with solr it is generally cleaner / more maintainable to sacrifice a little memory rather than creating debilitatingly complex chains of filters etc