Mowracer

1. A user logs into a Quickr Place and navigates to the Library
2. The user then chooses “Upload New Policy” which is a custom form
   This form contains additional fields to the OOB file upload. These fields are c_AvailableInPortalSearch, c_SerialNumber, c_Description, c_Department, c_Keywords and c_Action. Additionally, there are standard fields such as author, file attachment and modified date.
3. Finally, the user attaches a binary document such as Word, Power Point, Excel or PDF and clicks Submit
4. A PlaceBot runs on a scheduled basis based on Unprocessed documents as a collection.
   The final straw here was to not call UpdateProcessedDoc(document) to ensure that the entire collection of documents is processed every time the agent runs, this could obviously become a performance hit, but was a trade for ensuring that updates to a document were processed in a timely fashion,
5. The PlaceBot collects meta data about the document, first looking to the existence of the c_AvailableInPortalSearch field. If the field exists, the next field evaluated is the action. The action complies with the GSA (Google Search Appliance) scheme for <record action=”%” /> If this is a new or update then the action is set to “add.” “Add” in the Google world means if it already exists, update it, if it does not exist, add it. All of this meta data is processed and posted to a Domino Database.
6. The GSA Domino Database has a series of agents that run continuously. The first agent processes documents. Each document is examined for meta data and a XML snippet is created accordingly based on GSA Feed DTD.
7. The final agent processes the record XML elements and assembles them into the XML feed that the GSA can “read”
8. GSA periodically reads the XML feed and crawls each record URL, attaching the specified meta data to the indexed entry.
   Now the fun part begins…
   A user accesses the internal Portal (6.0 with plans to upgrade to 6.1) and navigates to the “Policy Search” portlet
   The user enters the search criteria for “vacation policy” and clicks “Submit”
   The GSA looks up in it’s index and does magical Google formula processing and says “OK, here are 100 indexed files that meet the search criteria ;vacation policy’”
   Next is the really interesting part… The GSA has the LTPA token and can access the users context.
   GSA does a brute force query against the items in it’s index that match ‘vacation policy’ and looks for either HTTP 200 OK or HTTP 401 Access Denied. If the response is HTTP 401 Access Denied, GSA drops that specific entry our of the result set. This process continues until 10 results are matched that the user has access to. At this point, the results are returned to the UI. In the end, the list of search results the user is presented with has already been verified based on the users access to the documents.

The most interesting and time consuming code piece which turned out to be the most simplistic:

Extended thanks to many for answering questions that I was not sure how to ask in this POC.  The customer has marked the POC as fulfilled in both the business requirements and the infrastructure integration thus providing another reference account.