Apache Solr extract, highlight HTML elements based on query, filter query terms

Question

Update. (+18d) edited title and provided answer addressing original question.

---

tl/dr

I am indexing HTML pages and dumping the <p>...</p> content as a snippet for search query returns. However, I don't want / need all that content (just the context around the query matched text).

Background

With these in my [classic] schema,

<fieldType name="text_general" class="solr.TextField" positionIncrementGap="100" 
autoGeneratePhraseQueries="true" multiValued="true">

<field name="p" type="text_general" indexed="true" stored="true" multiValued="true" 
omitNorms="true" termVectors="true" />

and these in my solrconfig.xml

<str name="queryAnalyzerFieldType">text_general</str>

<updateProcessor class="solr.AddSchemaFieldsUpdateProcessorFactory" name="add-schema-fields">
  <lst name="typeMapping">
    <str name="valueClass">java.lang.String</str>
    <str name="fieldType">text_general</str>
    <lst name="copyField">
      <str name="dest">*_str</str>
      <int name="maxChars">256</int>
    </lst>
    ...

<initParams path="/update/**,/query,/select,/spell">
  <lst name="defaults">
    <str name="df">_text_</str>
  </lst>
</initParams>

<requestHandler name="/update/extract"
class="org.apache.solr.handler.extraction.ExtractingRequestHandler">
  <lst name="defaults">
    <str name="lowernames">true</str>
    <str name="uprefix">ignored_</str>
    <str name="capture">div</str>
    <str name="fmap.div">div</str>
    <str name="capture">p</str>
    <str name="fmap.p">p</str>
    <str name="processor">uuid,remove-blank,field-name-mutating,parse-boolean,
               parse-long,parse-double,parse-date</str>
  </lst>
</requestHandler>

<requestHandler name="/query" class="solr.SearchHandler">
  <lst name="defaults">
    <str name="echoParams">explicit</str>
    <str name="wt">json</str>
    <str name="indent">true</str>
  </lst>
</requestHandler>

<queryResponseWriter name="json" class="solr.JSONResponseWriter">
  <!-- For the purposes of the tutorial, JSON responses are written as
   plain text so that they are easy to read in *any* browser.
   If you expect a MIME type of "application/json" just remove this override.
  -->
  <str name="content-type">text/plain; charset=UTF-8</str>
</queryResponseWriter>

I get this result [Solr Admin UI; facsimile shown here],

"p":["Sentence 1. Sentence 2. Sentence 3. Sentence 4. ..."]

In the source HTML document those sentences occur singly in p-tags, e.g. <p>Sentence 1.</p>, <p>Sentence 1.</p>, ...

Questions

1. How can I index them, singly? My rationale is that I want to display a snippet of the context around the search result target (not the entire p-tagged content).

2. Additionally, in the Linux grep command we can, e.g., return a line before and after the matched line (-C1, context, argument). Can we do something similar, here?

   i.e., if the Solr query match is in Sentence 2, the snippet would contain Sentences 1-3?

I tried assigning unique id's to the p-elements (<p id="a">...</p> <p id="b">...</p> but I just got this in Solr,

"p":["a Sentence 1. b Sentence 2. Sentence d 3. Sentence d 4. ..."]

Answer 1

Update [2020-12-31]

• Please overlook the answering of my own question, as 18 days have passed with one comment and no answers.

I am building a search page with Solr as the backend, inspired by the following Ajax Solr tutorial. https://github.com/evolvingweb/ajax-solr

Ultimately, I decided to forgo Solr highlighting in favor of a more flexible, bespoke JavaScript (JS) solution.

Basically, I:

• collect the Solr query (q) and filter query (fq) values (terms) in an array (simplified example shown below; more complete JS code appended)

  for (var i = 0, l = this.manager.response.response.docs.length; i < l; i++) {
      var doc = this.manager.response.response.docs[i];
  }
  

• extract sentences matching those terms (words) via a JS regex expression

  var mySentences = doc_p.replace(/([.?!])\s*(?=['"A-Z])/g, "$1|").split("|");
  

  where doc.p is a Solr field (defined in schema.xml) corresponding to indexed HTML p-element (<p>...</p>) text.

  • details: see Split string into sentences in javascript

• highlight those query terms

  var query = this.manager.store.get('q').value;  /* or loop over array */
  
  const replacer = (str, replace) => {
      const re = new RegExp(`(${replace})`, 'gi')
      return str.replaceAll(re, '<font style="background:#FFFF99">$1</font>')
  }
  var doc_p_hl = replacer(doc.p.toString(), query);
  

  • details: see JavaScript replaceAll case-insensitive search using variable rather than a string

• use those term-highlighted strings as snippets on the frontend

• apply a similar approach to the highighting of query terms in the full documents, doc.p.toString() ...

---

Addendum

Here is the JS code I wrote to collect Solr "q" and "fq" terms in an array. Note that Solr returns single fq as a string, and multiple fq terms as an array.

var q_arr = [];
var fq_arr = [];
var highlight_arr = [];
var snippets_arr = [];
var fq_vals = [];

if ((this.manager.store.get('q').value !== undefined) &&
    (this.manager.store.get('q').value !== '*:*')) {
    query = this.manager.store.get('q').value;
    q_arr.push(query);
    highlight_arr.push(query);
    console.log('q_arr:', q_arr, '| type:', typeof q_arr, '| length:', q_arr.length)
}

var doc_responseHeader = this.manager.response.responseHeader;
if (doc_responseHeader.params.fq !== undefined) {

    /* ONE "fq" (FILTER QUERY) TERM: */
    if (typeof doc_responseHeader.params.fq === 'string' ||
        doc_responseHeader.params.fq instanceof String) {
        fq_arr.push(doc_responseHeader.params.fq);
    }

    /* MORE THAN ONE "fq" (FILTER QUERY) TERM: */
    if  (typeof doc_responseHeader.params.fq === 'object' ||
        doc_responseHeader.params.fq instanceof Object) {

        for (var i = 0, l = doc_responseHeader.params.fq.length; i < l; i++) {
            fq_arr.push(doc_responseHeader.params.fq[i].toString());
        }
    }

    fq_vals = fq_arr.map(function(x){return x.replace(/keywords:/g, '');})
    console.log('fq_vals', fq_vals, '| type:', typeof fq_vals, '| length:', fq_vals.length)

    for (var i = 0, l = fq_vals.length; i < l; i++) {
        highlight_arr.push(fq_vals[i].toString());
    }
}