What happened to this solr range filter request?

Question

The following filter query returns null results (using *: * as the query):

-startDate:[* TO *] OR startDate:[* TO NOW/DAY+1DAY] 

But if I filter only:

 -startDate:[* TO *] 

I get 3 results.

If I filter only:

 startDate:[* TO NOW/DAY+1DAY] 

I get 161 results.

Why does combined FQ return null results? I want the filter to return any document whose start date is zero or the start date until today.

EDIT:

I am using Solr 4.2.1.2013.03.26.08.26.55

EDIT:

Well, strange, it can happen that a colleague suggested putting parentheses in two parts as follows:

 (-startDate:[* TO *]) OR (startDate:[* TO NOW/DAY+1DAY]) 

And somehow it worked. I'm still wondering why this made a difference. Hope someone can shed some light.

Thanks!

Answer 1

Solr supports purely negative requests. They do this essentially by expanding the net negative to something like:

 *:* -startDate:[* TO *] 

However, what you combine in BooleanQuery, I do not believe that it applies such logic. A negative query in lucene does not receive anything, but rather filters out matches caused by other, positive, query conditions. This is different from SQL queries, which in a sense begin with an implicit *:* or a full table of results and allow you to bypass it.

I believe that your OR effectively ignored, since, strictly speaking, it does not make sense in the context. Typically, OR is just syntactic sugar, I think ( field:this OR field:that equivalent to field:this field:that ).

So, actually your query is: startDate:[* TO NOW/DAY+1DAY] -startDate:[* TO *] , which makes the results more obvious. When you wrap it in parentheses, each terminological request is processed separately, and you get access to solr support for lone negative requests.

---

It is much better to remember storing the default value if you need to look for unset / null values. *:* , and as a result, pure negative queries like this should scan the entire index, and therefore run very poorly. Providing a default value will improve performance and prevent this confusing situation.

Answer 2

I used the femtoRgon answer and was able to build a query including a range and empty values.

Listed below are all documents with StartDate starting January 1, 2012, and all documents without StartDate.

 (StartDate:[2014-01-01T00:00:00Z TO *]) OR (-StartDate:([* TO *]) AND *:*) 

Magic (-StartDate:([* TO *]) AND *:*) . This will select documents without StartDate.

Answer 3

Pure negative queries do not work because they lose nothing.

Try:

: AND -startDate: [* TO *]

Answer 4

When you query with -startDate:[* TO *] , you get documents that don't have data for the startDate field.

When you request startDate:[* TO NOW/DAY+1DAY] , you receive documents whose values ​​are less than or equal to NOW/DAY+1DAY in the startDate field.

You can try -startDate:* OR startDate:[* TO NOW/DAY+1DAY] . The first part indicates documents that do not matter, and the second part says that the document has a value less than or equal to NOW/DAY+1DAY in the startDate field.