Solr ExtractingRequestHandler extracting “rect” in links

问题

I am utilizing solr ExtractingRequestHandler to extract and index HTML content. My issue comes to the extracted links section that it produces. The extracted content returned has "rect" inserted where they do not exist in the HTML source.

I have my solrconfig cell configuration as follows:

  <requestHandler name="/upate/extract" 
              startup="lazy"
              class="solr.extraction.ExtractingRequestHandler" >
<lst name="defaults">
  <str name="lowernames">true</str>
  <!-- capture link hrefs but ignore div attributes -->
  <str name="captureAttr">true</str>
  <str name="fmap.div">ignored_</str>
</lst>

And my solr schema.xml with the following etnries:

   <field name="content_type" type="string" indexed="true" stored="true" multiValued="true"/>
   <field name="links" type="string" indexed="true" stored="true" multiValued="true"/>
   <field name="meta" type="string" indexed="true" stored="true" multiValued="true"/>
   <field name="content_encoding" type="string" indexed="false" stored="true" multiValued="false"/>
   <field name="content" type="text_general" indexed="false" stored="true" multiValued="true"/>

I post the following HTML to sorl cell:

<!DOCTYPE html>
<html>
<body>
  <h1>Heading1</h1><a href="http://www.google.com">Link to Google</a><a href=
  "http://www.google.com">Link to Google2</a><a href="http://www.google.com">Link to
  Google3</a><a href="http://www.google.com">Link to Google</a>

  <p>Paragraph1</p>
</body>
</html>

Solr has the following indexed:

      {
    "meta": [
      "Content-Encoding",
      "ISO-8859-1",
      "ignored_hbaseindexer_mime_type",
      "text/html",
      "Content-Type",
      "text/html; charset=ISO-8859-1"
    ],
    "links": [
      "rect",
      "http://www.google.com",
      "rect",
      "http://www.google.com",
      "rect",
      "http://www.google.com",
      "rect",
      "http://www.google.com"
    ],
    "content_encoding": "ISO-8859-1",
    "content_type": [
      "text/html; charset=ISO-8859-1"
    ],
    "content": [
      "             Heading1  Link to Google  Link to Google2  Link to Google3  Link to Google  Paragraph1   "
    ],
    "id": "row69",
    "_version_": 1461665607851180000
  }

Notice the "rect" between every link. Why is solr cell or tika inserting these? I am not defining a tika config file to use. Do i need to configure tika?

回答1:

Although an old Question, I also encountered this issue while indexing HTML documents via Solr 8.7.0.

<requestHandler name="/update/extract" 
    class="org.apache.solr.handler.extraction.ExtractingRequestHandler">
    ....

HTML:

<p>My website is <a href="https://buriedtruth.com/">BuriedTruth.com</a>.</p>

Result:

My website is rect https://buriedtruth.com/ BuriedTruth.com .

[ I am posting/indexing on the Linux command-line: solr restart; sleep 1; post -c gettingstarted /mnt/Vancouver/programming/datasci/solr/test/solr_test9.html; ]

I grepped (ripgrep: rg --color=always -w -e 'rect' . |less) the Solr code for that word, but found nothing, so the source of rect http... in indexed URLs eludes me.

My solution was to add a regex processor to mysolrconfig.xml:

  <updateRequestProcessorChain name="add-unknown-fields-to-the-schema" default="${update.autoCreateFields:true}"
           processor="uuid,remove-blank,field-name-mutating,parse-boolean,parse-long,parse-double,parse-date,add-schema-fields">
    <processor class="solr.LogUpdateProcessorFactory"/>
    <processor class="solr.DistributedUpdateProcessorFactory"/>
    <!-- ======================================== -->
    <!-- https://lucene.apache.org/solr/7_4_0/solr-core/org/apache/solr/update/processor/RegexReplaceProcessorFactory.html -->
    <!-- Solr bug? URLs parse as "rect https..."  Managed-schema (Admin UI): defined p as text_general -->
    <!-- but did not parse. Looking at content | title: text_general copied to string, so added  -->
    <!-- copyfield of p (text_general) as p_str ... regex below now works! -->
    <processor class="solr.RegexReplaceProcessorFactory">
      <str name="fieldName">content</str>
      <str name="fieldName">title</str>
      <str name="fieldName">p</str>
      <!-- Case-sensitive (and only one pattern:replacement allowed, so use as many copies): -->
      <!-- of this processor as needed: -->
      <str name="pattern">rect http</str>
      <str name="replacement">http</str>
      <bool name="literalReplacement">true</bool>
    </processor>
    <!-- ======================================== -->
    <!-- This needs to be last (may need to clear documents and reindex to see changes, e.g. Solr Admin UI): -->
    <processor class="solr.RunUpdateProcessorFactory"/>
  </updateRequestProcessorChain>

As alluded in my comments in that processor, I am extracting <p />-formatted HTML content to a p field (field: p | type: text_general).

That content did not parse with the RegexReplaceProcessorFactory processor.

In the Solr Admin UI I noted that title and content were copied as strings (e.g.: field: content | type: text_general | copied to: content_str), so I made copy field (p >> p_str) that resolved the regex issue.

For completeness, here are the relevant parts of my solrconfig.xml related to HTML document indexing,

  <lib dir="${solr.install.dir:../../..}/contrib/extraction/lib" regex=".*\.jar" />
  <lib dir="${solr.install.dir:../../..}/dist/" regex="solr-cell-\d.*\.jar" />

  <!-- https://lucene.472066.n3.nabble.com/Prons-an-Cons-of-Startup-Lazy-a-Handler-td4059111.html -->
                  <!-- startup="lazy" -->

  <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>
    </lst>
  </requestHandler>

... noting again that I added fields to the managed-schema via the Solr Admin UI.

Result:

My website is https://buriedtruth.com/ BuriedTruth.com .

  <field name="p" type="text_general" uninvertible="true" indexed="true" stored="true"/>
  <copyField source="p" dest="p_str"/>