You might want to hire a consultant. Tika can deal with Word documents. Ids needs to be unique. One index might work, not sure based on your info below.
For database you need to use a Java db thin connector to SQL server. Throw the jar in the lib directory and restart. Then setup dih settings to get data from the database. There is also web crawling solutions like Nutch. I would get some data indexed and try searching it first. Lucid has a tutorial. Bill Bell Sent from mobile On Mar 2, 2011, at 4:39 AM, "Ken Foskey" <kfos...@tpg.com.au> wrote: > > I have read the solr book and the other book is on its way for me to read. I > need some help in the mean time. > > a) Using the example solr system how do I send the word document using curl > into the system. I want to have the ID as the full path of the document. I > have tried various commands but it gives me stream errors, documents are on > one server, solr is on a second server. I would like to experiment at home > indexing all the documents on the server, giving me experience outside work. > > b) I am trying to grasp a few issues with my design, I have read the books > but I am struggling to grasp the ideas yet. > > - I am using a .Net connector, which one is recommended? > > - I have an aspx webpage that connects to solr, adds some extra information > to the solr query (eg client id see later) and returns the data. Is this a > normal design? > > I am using a shopping trolley design. Orders are created and then comments > are added as the order is processed. I want to search all the order > information. (this is one of many indexing requirements, then there will be > a 'site search' facility.) > > I want to index many database tables, I am proposing to use one SOLR > instance that indexes all the data. I will use the example TEXT field and > copydata all the extra fields I have for the tables. The SOLR book gave me > short shift on the database indexing tool, is there a good tutorial on using > SOLR with SQLServer? I have a many tables, I was also thinking I might have > to join a lot, it might be easier to create an XML output then send that, > is there pros and cons. > > I am thinking that I will search client = 578 and ( text = keywords). This > way I am restricting my searches to the client signed on. Is this a good > idea? (it is secured behind my application so I am adding the extras) > > I was also thinking that I would index the order notes separately from the > order, The other alternative is to multi value a note field in the order > and keep deleting it. Which is best? One way I have to keep replacing the > whole order, the other I have to some how 'join' up the notes in the search. > (more like this feature I suppose). > > When you come out of the SOLR index, is it normal to map the ID to a URL, > for example id="product: 1234" => href="product.aspx?productId=1234" > in code using the return data from solr. > > Hope you can help me, > > Ta > Ken > > >
