I'm using FieldCollapsing to group the results.
Example: I search for : and group by names, like:
http://localhost:<port>/solr/select/?q=*:*
&group=true
&group.limit=200
&group.query=Jim
&group.query=Jon
&group.query=Frank Sinatra
It looks like, solr is running (internal) an separate query for every name. (whatever)
The point is, that i have to change the search parameter local in order to set an different search operator (from OR to AND).
To get valid results I need an query like this:
http://localhost:<port>/solr/select/?q=*:*
&group=true
&group.limit=200
&group.query={!q.op=AND defType=edismax}Jim
&group.query={!q.op=AND defType=edismax}Jon
&group.query={!q.op=AND defType=edismax}Frank Sinatra
This works very well. The Problem is, that solr returns the label of the group including the locale parameter!
<lst name="grouped">
<lst name="{!q.op=AND defType=edismax}Frank Sinatra"> <---- wrong label
<int name="matches">785</int><result name="doclist" numFound="10" start="0">
<doc>
[...]
An valide result is be:
<lst name="grouped">
<lst name="Frank Sinatra">
<int name="matches">785</int><result name="doclist" numFound="10" start="0">
<doc>
[...]
Is there a way to change to label to the real term where solr is searching for?
Related
We've got a multilingual search index with the "field-per-language" configuration with a lot of similar aliases in the search handler like this:
<str name="f.content_en.qf">Title_en^10 Text_en^1 ...</str>
<str name="f.content_de.qf">Title_de^10 Text_de^1 ...</str>
...
They are used in the q parameter:
<str name="q">{!edismax qf=$searchField pf=$searchField v=$searchText}</str>
The client knows, which language should be used and calls Solr like this, e.g.: /solr/core/search?searchText=TEXT&searchField=content_en
That works fine, but the configuration contains a lot of similar stuff.
I'd like to optimize the config to something like this:
<str name="df">content</str>
<str name="f.content.qf">Title_${lang}^10 Text_${lang}^1...</str>
Then the client would need to provide the lang parameter only.
I tried to use concat function like this:
paramLang=en
searchFields=concat("Title", "_", "${paramLang}", " ", "Text", "_", "${paramLang}")
and use it as the qf:
q={!edismax qf=$searchFields v=$searchText}
But it seems, the local params qf does not support Solr functions.
Is is possible with Solr at all?
Actually, the Parameter substitution / Macro Expansion works fine.
The issue was with those macros in the solrconfig.xml: there is a conflict with Solr system properties substitution. Solr could not resolve the query parameter macros.
I could not find a proper way, how to escape query parameters (macros) and used the following workaround:
<lst name="invariants">
<str name="defType">edismax</str>
<str name="searchFields">
Title_${lang:${lang}}^10
Text_${lang:${lang}}^1
...
<lst name="defaults">
<str name="q">*</str>
<str name="qf">${searchFields:${searchFields}}</str>
<str name="pf">${searchFields:${searchFields}}</str>
<str name="lang">en</str>
...
Query URL: /search?q=TEXT&lang=en
Update: proper way to deal with var substitution in solrconfig.xml - escape the dollar char by $$:
<str name="searchFields">
Title_$${lang}^10
Text_$${lang}^1
...
Update #2: do NOT define macros in the invariant or append sections when using a Solr Cloud! Otherwise, you'll a weird exception, e.g.:
"undefined field: \"Text_$\"
or
"msg": "Error from server at null: org.apache.solr.search.SyntaxError: Query Field '${searchFields}' is not a valid field name"
P.S. wt=json as "invariant" is also NOT compatible with Solr Cloud, giving "unexpected" content-type error.
So many "surprises" :(
This is newbie question, but trying understand solr search handler with defaults enabled.
As per my understanding df is used for Default search field, what will happen/how does it work if remove this part?
My scenario currently: I am pushing/adding all fields data (e.g title, department, etc.) to text. So when I search free text it can be searchable (using edismax) and it works as expected. But when tomorrow new fields come, I mean suppose it will be title, department , somefield, then I have to reindex all data again. Then it will be searchable. How can I achieve that? Can I have multiple fields in df?
example : <str name="df">title, department, etc,etc </str> like this ?
<requestHandler name="/select" class="solr.SearchHandler">
<!-- default values for query parameters can be specified, these
will be overridden by parameters in the request
-->
<lst name="defaults">
<str name="echoParams">explicit</str>
<int name="rows">10</int>
<!-- Default search field -->
<str name="df">text</str>
</lst>
Since you're using edismax, you can provide the qf argument with multiple fields instead - and weights between them. Another option is to have one common field, i.e. something like text and a copyField instruction that copies everything (source set to *) to text (dest set to text), then search against that field.
Using either won't require a reindex, since the content will either be copied from all documents indexed as they're being indexed, or the search will consider more fields (with qf).
You can set the default value for the qf in the same was a you've shown above in the config file under defaults.
I need to perform highlighting for multiple words into the same field but for each one using a specific formatter (prefix and postfix).
Let's say that I have the description field and for a document it has the value: Einstein always excelled at math and physics from a young age. How to highlight math with a pair of a specific prefix and postfix AND ALSO physicswith a different prefix-postfix pair? So, in the end I would like to obtain:
Einstein always excelled at <em class="hl-red">math</em> and <em class="hl-green">physics</em> from a young age
The reason is that in the frontend I have different CSS classes with background-color: red; for hl-red and background-color: green for hl-green for example.
However, I was managed to highlight multiple words into the same field but with the same prefix-postfix pair all over the places, which is not what I want actually. In addition, I tried to add multiple HtmlFormatter entries in solrconfig.xml:
<highlighting>
..............
<formatter name="html" default="true" class="solr.highlight.HtmlFormatter">
<lst name="defaults">
<str name="hl.simple.pre"><![CDATA[<em>]]></str>
<str name="hl.simple.post"><![CDATA[</em>]]></str>
</lst>
<lst name="hl-red">
<str name="hl.simple.pre"><![CDATA[<em class="hl-red">]]></str>
<str name="hl.simple.post"><![CDATA[</em>]]></str>
</lst>
<lst name="hl-green">
<str name="hl.simple.pre"><![CDATA[<em class="hl-green">]]></str>
<str name="hl.simple.post"><![CDATA[</em>]]></str>
</lst>
</formatter>
..............
</highlighting>
but I got HttpSolrClient$RemoteSolrException: Error from server at http://localhost:8983/solr: Unknown formatter: hl-green. Also, I didn't find a way to specify an array of prefixes in Solr Admin UI nor in spring-data-solr, just a simple query like this:
SimpleHighlightQuery query = new SimpleHighlightQuery(Objects.requireNonNull(criteria));
HighlightOptions highlightOptions = new HighlightOptions()
.addFields(fields)
.setSimplePrefix(prefix)
.setSimplePostfix(postfix);
query.setHighlightOptions(highlightOptions);
query.setPageRequest(pageable);
return solrTemplate.queryForHighlightPage(MY_CORE, query, MyModel.class);
My assumption is that it is a limitation of the Solr itself.
I was thinking about to write a custom fragmentsBuilder but I do not know exactly if it is the case nor how to do that. For another workaround I was thinking to execute for each word a highlight query, then to store the result, then to execute for the second word another highlight query, store the result and so on. But I don't think it is a good and elegant solution because I will have problems when executing the second query if the second word is: "em" or "class" or "red"/"green" (nested undesired highlighting will occur).
I am using spring-data-solr into a Spring Boot application and Solr 6.6.5 as a (http) service.
Does anyone know how to solve this? Please give me an advice! Any idea will be much appreciated!
To exclude some documents from the search result, I can use the not in or - negative sign to specify the ids like this through a query.
select/?q=:&fq=-id:86+-id:338
But i want to pre configure in solr that on any search the results of certain documents will never show up
You can add a parameter list to a definition for a requestHandler that appends a fq statement to all requests. The example from the wiki does something similar:
<lst name="appends">
<!-- no matter what other fq are also used, always remove these two documents -->
<str name="fq">-id:(86 338)</str>
</lst>
This fq will then always be appended to the request made.
We can also use following :
<lst name="appends">
<str name="excludeIds">86,338</str>
</lst>
I tried and it also gives expected result.
My Solr server seems only working with AND but not OR, e.g.
/solr/select?q=(marsden AND emma)&qt=yodl_handler works, but
/solr/select?q=(marsden OR mackey)&qt=yodl_handler doesn't.
For each individual query, it returns resutls, e.g.
/solr/select?q=marsden&qt=yodl_handler returns 2 results
/solr/select?q=mackey&qt=yodl_handler returns 3 results
Any suggestions are appreciated!
Here is the definition of yodl_handler:
<requestHandler name="yodl_handler" class="solr.DisMaxRequestHandler">
<lst name="defaults">
<str name="echoParams">explicit</str>
<float name="tie">0.01</float>
<str name="qf">
dc.title^1 dc3.title^1 dc.type^1 vra2.logical.creator^1 vra2.image.agent.name^1 vra2.image.agent.role^1 vra2.image.agent.role^1 vra2.image.location.name^1 vra2.image.rights.rightsholder^1 vra2.image.source.refid^1 vra2.image.title^1 vra2.image.worktype^1 vra2.work.agent.attribution^1 vra2.work.agent.name^1 vra2.work.agent.role^1 vra2.work.culturalContext^1 vra2.work.description^1 vra2.work.location-type^1 vra2.logical.location.place^1 vra2.logical.location.name^1 vra2.work.location.refid^1 vra2.work.material^1 vra2.work.rights.rightsHolder^1 vra2.work.StylePeriod^1 vra2.work.subject.term.name^1 vra2.work.subject.term.place^1 vra2.work.subject.term.keyword^1 vra2.work.technique^1 vra2.work.title^1 vra2.work.worktype^1 iris2.instrument.instrumentType^1 iris2.instrument.primaryInstrumentType^1 iris2.instrument.secondaryInstrumentType^1 iris2.instrument.instrumentType.alltypes^1 iris2.instrument.author^1 iris2.references.allauthors^1 iris2.instrument.researchArea^1 iris2.instrument.typeOfFile^1 iris2.instrument.software^1 iris2.instrument.dataType^1 iris2.instrument.linguisticTarget^1 iris2.instrument.sourceLanguage^1 iris2.instrument.funder^1 iris2.instrument.licence^1 iris2.participants.participantType^1 iris2.participants.firstLanguage^1 iris2.participants.targetLanguage^1 iris2.participants.gender^1 iris2.participants.proficiencyLearner^1 iris2.participants.proficiencyStudentsTaught^1 iris2.participants.yearsOfTeachingExperience^1 iris2.participants.domainOfUse^1 iris2.references.publicationType^1 iris2.references.author^1 iris2.references.author.lastnames^1 iris2.references.booktitle^1 iris2.references.journal^1 iris2.references.publicationDate^1 iris2.references.publicationLatestDate^1 iris2.references.publisher^1 iris2.references.placeOfPublication^1 iris2.references.editor^1 iris2.references.conferenceName^1
</str>
<int name="ps">100</int>
<str name="q.alt">*:*</str>
<str name="hl.fl">text features name</str>
<str name="f.name.hl.fragsize">0</str>
<str name="f.name.hl.alternateField">name</str>
<str name="f.text.hl.fragmenter">regex</str>
</lst>
</requestHandler>
The simple answer is, the Solr DisMax query parser does not support boolean logic in queries. The appearance of it working with the query involving "AND" is probably a side-effect of the way your fields are indexed (stop words?) or appears in the underlying data.
You can get a better idea what's happening under the hood if you send the debugQuery parameter, e.g.:
/solr/select?q=(marsden AND emma)&qt=yodl_handler&debugQuery=true
There's further documentation on the Solr wiki about the Dismax parser:
The Dismax query parser supports an extremely simplified subset of the Lucene QueryParser syntax. Quotes can be used to group phrases, and +/- can be used to denote mandatory and optional clauses ... but all other Lucene query parser special characters are escaped to simplify the user experience. (see DisMaxQParserPlugin)
The good news is, if you're using a modern version of Solr (3.1+), you have access to the new ExtendedDisMax parser, which DOES support boolean queries. `