Grouper Developers Forum

[Chris Hyzer <>]

Hey,

At Penn we have 500k people in our person source, and it is hard to find 
people in the UI with common names.

So I made a new jdbc source adapter that will split search strings by 
whitespace and "AND" all terms together in the query.

Let me know if you are interested in this and if it seems like it would fit 
your needs.  Note: we already edited the subject API for Grouper 1.5, and 
never branched, so this adapter is currently in grouper.  When I merge to the 
1.5 branch, I will put it in the subject.jar.

https://bugs.internet2.edu/jira/browse/GRP-280

Right now if I search for "john smith", and there is a user "john m. smith", 
it will not find the user.
Make a jdbc source adapter that can find the user

This is done. To implement:

1. provide a table or view which has the subject information including the 
search col (lowered). This will have better performance if you have a job 
which dumps the data to one table periodically, and not just joining a bunch 
of tables and calling to_lower()

2. configure in the sources.xml. Here is an example:

    <source 
adapterClass="edu.internet2.middleware.grouper.subj.GrouperJdbcSourceAdapter2">
    <!--
           CREATE TABLE `person_source_v` (
                   `penn_id` varchar(50) NOT NULL,
                   `name` varchar(50) default NULL,
                   `description` varchar(50) default NULL,
                   `pennname` varchar(50) default NULL,
                   `description_lower` varchar(50) default NULL,
                   PRIMARY KEY (`penn_id`)
                 );
insert into 
`person_source_v`(`penn_id`,`name`,`description`,`pennname`,`description_lower`)
 values ('10021368','Chris Hyzer','Chris Hyzer','mchyzer','chris hyzer, 
mchyzer, 10021368'),('10039438','John Smith','John Smith','jsmith','john 
smith, jsmith, 10039438');
           commit;
           -->
    <id>pennperson</id>
    <name>Penn person</name>
     <type>person</type>
     <init-param>
       <param-name>jdbcConnectionProvider</param-name>
       
<param-value>edu.internet2.middleware.grouper.subj.GrouperJdbcConnectionProvider</param-value>
     </init-param>

      <init-param>
       <param-name>dbTableOrView</param-name>
       <param-value>person_source_v</param-value>
     </init-param>
      <init-param>
       <param-name>subjectIdCol</param-name>
       <param-value>penn_id</param-value>
     </init-param>
     <init-param>
       <param-name>nameCol</param-name>
       <param-value>name</param-value>
     </init-param>
     <init-param>
       <param-name>descriptionCol</param-name>
       <param-value>description</param-value>
     </init-param>
     <init-param>
       <!-- search col where general searches take place, lower case -->
       <param-name>lowerSearchCol</param-name>
       <param-value>description_lower</param-value>
     </init-param>
     <init-param>
       <!-- optional col if you want the search results sorted in the API 
(note, UI might override) -->
       <param-name>defaultSortCol</param-name>
       <param-value>description</param-value>
     </init-param>
     <init-param>
       <!-- col which identifies the row, perhaps not subjectId -->
       <param-name>subjectIdentifierCol0</param-name>
       <param-value>pennname</param-value>
     </init-param>
     <init-param>
       <param-name>subjectIdentifierCol1</param-name>
       <param-value>penn_id</param-value>
     </init-param>
     <!-- now you can count up from 0 to N of attributes for various cols -->
     <init-param>
       <param-name>subjectAttributeCol0</param-name>
       <param-value>pennname</param-value>
     </init-param>
     <init-param>
       <param-name>subjectAttributeName0</param-name>
       <param-value>PENNNAME</param-value>
     </init-param>

   </source>

That is it, now you can search for "smith jo" and it will find john smith, 
but will not return any smiths without "jo" in there somewhere. Note at Penn 
we have the description field end in (active) for active employees, and 
(NOT_ACTIVE). So if I search for "John Smith (active)" it will only return 
active john smiths, including "john m. smith". We also have the data source 
in there. So I could search for "(active) John Smith staf" and it would find 
active john smiths who's primary affiliation with the university is 
'employee'. We could put alternate names in the search col, or alternate 
affiliations, etc. We load that table nightly with plsql so we can customize 
as needed.

Notes:
- Case insensitive column names (can be very confusing in the default source 
adapter)
- More efficient unmodifiable attribute maps and set values
- Supports multiple subjectIdentifier cols

Regards,
Chris