Metadata Troubleshooting


This page gives some details on troubleshooting metadata search.

Metadata Search not Returning Results

Check for the following possible causes of error:

  • Check the metamap.cfg and/or xml.cfg files for omissions, syntactical errors (missing commas, backslashes in XPaths, etc) and overriding of reserved metadata classes (i.e. d,h,i,k,m,t,u,v).
  • If metadata summaries are not displaying correctly then check that metadata summary mode is enabled (-SM=meta) and that one or more fields from the xml.cfg and/or metamap.cfg are included in the summary fields option (-SF=[fields]). For example, the collection.cfg file should contain a line similar to:
    query_processor_options=-SM=meta -SF=[c,d,s,t,longexample]
  • If the search interface is generated by search.html (i.e. not using direct XML or JSON) then check that there are metadata summary directives in the relevant form files (e.g. simple.form) for each of the metadata fields that are to appear in the summary. For example, to display author metadata as field a, the following line should be in the results section of the form file:
      <#if s.result.metaData["a"]??>
        Author(s): ${s.result.metaData["a"]!?replace("|", ", ")}
  • Check that the xml.cfg and metamap.cfg files were correctly loaded by the indexer. The top of the Step-Index.log file displays a formatted record for every line in both of these files with no error messages if they were loaded successfully. For example,
     Loading XML mappings from: /opt/funnelback/conf/fctest/xml.cfg
       XML: "/title" -> 't'
       XML: "/Title" -> 't'
       XML: "/TITLE" -> 't'
  • Check that the indexer found metadata in the documents by searching the Step-Index.log file for the name of a document that is known to contain metadata. The line just below the filename will also contain a list of metadata fields that were found in the document. The list of fields is just before the date, as in the example...
    10 [default] modernui-datamodel/com/funnelback/publicui/search/model/curator/action/RemoveUrls.html 10 
       {0 UTF-8 HTML 23976 0 "c d s t longexample " 2016-11-09  38 [ T=11.000 W=577.000 Z=2.768 H=6.917 F=0.258 C=0.164 L=5.742 D=173.000 M=3.986 ] html}

This record contained a description (c), a date (d), keywords (s), a title (t) and some 'longexample' metadata.

  • Check that the documents contain metadata in standard HTML notation such as <meta name="date" content="2003-03-04"> or <meta http-equiv='' content='Christopher Marlowe'>

See also