I have three tables that I have to present in one Android ListView. To get the data I use the SQL UNION operator to "merge" all three tables together, so that in my ViewBinder I can make each timeline item look distinct.
These items need to be sorted in chronological order. These three tables do not have a common base class.
Here is the SQL that I have in mind:
SELECT * FROM (
SELECT id, startTime as time, username, comment, "CustomerInteraction" FROM CustomerInteraction
UNION
SELECT id, date as time, "" as username, "" as comment, "Sale" FROM Sale
UNION
SELECT id, claimDate as time, username, comment, "TravelClaim" FROM TravelClaim)
ORDER BY time DESC LIMIT 100
How can I express the above query in ORMLite?
I know I can use Dao.executeRaw, but I don't want to populate my entire list in one go. I would much rather use the trick to get the underlying cursor from ORMLite, and then just pass that to my Adapter. (Lazy loading, makes initial display of long lists much faster.)
Is there a way I can do something like Dao.getPreparedQueryFromRaw(String statement) ? Or better yet QueryBuilder.union(QueryBuilder qb)?
You can get a Cursor by calling rawQuery on the SQLiteDatabase. I do something like this:
final SQLiteDatabase db = getHelper().getReadableDatabase();
Cursor cursor = db.rawQuery(MY_SQL_QUERY, null);
You don't need to do anything much more than that.
Is there a way I can do something like Dao.getPreparedQueryFromRaw(String statement) ? Or better yet QueryBuilder.union(QueryBuilder qb)?
The best way to do this with ORMLite is with one of the the queryRaw(...) methods. They return a GenericRawResults class which you can iterate across. The iterator gives you a number of different methods to help with moving around the list.
The problem is that the generic results are not of a certain type so I'm not sure if you can map it into the Android ListView. You can provide a RawRowMapper to queryRaw(...). You can get the mapper for a particular type by using the dao.getRawRowMapper() method.
Hope something here is helpful.
Related
i'm trying to find a good way to sort the search results according to relevance after performing a search with a SearchView in Android. For me relevance means the number of matches in two SQLite text columns
I'm using a CursorLoader and there the sort order can be given to the constructor at the end
CursorLoader tLoader = new CursorLoader(
getActivity(), ContentProviderM.ARTICLE_CONTENT_URI,
tProj, tSel, tSelArgs, SORT_ORDER);
(or set using the setSortOrder (String sortOrder) method)
But i need more flexibility than this because i'm looking to sort on the number of matches rather than just on one or two columns
The only solution i can see myself is to add another column in my SQLite table, do some processing, and supply that column as the sort column to the CursorLoader
Now for my question: What is the best way to supply the sort order information to the CursorLoader using SQLite syntax, avoiding having to add a new column? (And what could this SQLite code look like?) Also, i'd like to ask more in general: Is there a different solution to this problem that i've missed?
Grateful for any help! And with kind regards,
Tord
Depending on the content provider, if it just pass to the orderBy field, you can do anything.
SQLiteDatabase query
orderBy How to order the rows, formatted as an SQL ORDER BY clause
(excluding the ORDER BY itself). Passing null will use the default
sort order, which may be unordered.
you can do whatever you want, this is just the line after ORDER BY
P.S. It is totally depending on the Content Provider, it it choose to ignore the parameter, you can do nothing.
i found a "workaround" for this problem.
After investigating different ways to write sqlite code i ended up just adding a new table column just for sorting. This column simply stores an integer and is updated every time that the user performs a search, right before the CursorLoader is created
Advantages:
We can now do all of the relevance calculations in Java code
Drawbacks:
Relevance calculation is done as the search is done so if we have a large number of items it may take some time to process everything
I am making an app which is pretty much like this:
Register a customer;
Add different types of math calculations to this customer;
Now this customer has a list full of specific calculations.
Everything is working pretty good so far. But the thing is, how could I make a sqlite function in my db class that would list ONLY the calcs of a specific customer? e.g. when I select one specific customer, I don't want to see the calcs of other customers, only of this specific one.
I have 2 tables:
Customer (_id, name, email)
Calculations (_id, customer_id, date, value1, value2, result)
is there anyway I could like add a new column to Customer which would be like "calcs_made" and then it would call ALL "Calculations" table's columns?
I thought of searching the "customer_id" that the user is looking for and showing only the lines with that matching information, but it doesn't sound like a good practice.
The "Calculations" table is the first and only one so far, but there will be many others with different types of calcs as I update the app, therefore it would have to be really flexible.
Try using a sql selection to get what you need. It may look like this
select * from Calculations where customer_id=id;
The id is the specific customer id you choose.
This is the query that I use to create a table
create table site_table(
_id integer primary key autoincrement,
name_site text,
url text,
login text,
pass text
);
I called Cursor.getColumnNames() and noticed that columns order are id, login, pass, name, url.
So, if I want a value I have to get it by the index Cursor.getString(index). Until I debugged I was messing up calling the wrong index, but now I wonder, why SQLite saves that way? Why it does not follow that way I created id, name_site, url, login and pass?
Thanks
So, if I want a value I have to get it by the index
Cursor.getString(index)
So for example for this reason you should always use
c.getString(c.getColumnIndex("ColName")); // or better getColumnIndex(CONSTANT)
This method saves all of us and ensure that you never get wrong results. Generally this method is recommended and also storing COLUMN_NAMES as CONSTANTS in separated class is very, very useful and efficient practise.
Note: Order depends on projection i.e. select name, lastname from table
That data is ordered by the order your requested it in your query, not the order you created the table with. So you probably changed the order in your query that generated said cursor.
Columns order in your cursor depends on projection. To be sure you use correct column index use c.getString(c.getColumnIndexOrThrow("COLUMN_NAME")) where c is your cursor.
I just made the experience first hand:
The indices of the columns of the cursor as a result of a
SELECT * FROM mytable WHERE ...
query have sometimes (not always) a different order that what SQLITE Database Browser shows as column order in the Database Structure tab. So referencing the columns via getColumnIndex seems to be the only safe way.
Here's an interesting question that I'm shocked hasn't been asked more often on the internet. Android's CursorAdapters are extremely useful once you get a ContentProvider up and running and learn how to use it, but they are limited due to their requirement on having the _id field as part of their query (an error is thrown without it). Here's why:
My specific problem is that I have two spinners: One spinner should contain unique "category" names from the database, and the other should populate with all the database entries from the selected "category" (category being the column name, here). This seems like a pretty simple setup that many programs might use, no? Trying to implement that first spinner is where I've run into problems.
Here's the query that I would like for that first spinner:
SELECT DISTINCT category FROM table;
Making this query throws an error on CursorAdapter because the _id column is required as part of the query. Adding the _id column to the projection naturally returns every row of the table, since you're now asking for distinct id's as well, and every id is distinct (by definition). Obviously I would rather see only one entry per category name.
I've already implemented a work around, which is to simply make the query above and then copy the results into an ArrayAdapter. My reason for this post is to see if there was a more elegant solution to this odd little problem and start a discussion on what I could be doing better. Alternate implementation suggestions, such as using different kinds of controls or adapters, are very welcome.
Here's the query I ended up with:
SELECT _id, category FROM table_name GROUP BY category;
I used the rawQuery() function on an SQLiteDatabase object to carry this out. The "GROUP BY" piece was the key towards getting the right results, so thanks to user Sagar for pointing me in the right direction.
Do consider user Al Sutton's answer to this question as well, as it may be a more elegant solution to this problem.
Thanks everyone!
I'd suggest having a separate table with just _id & category in it which contains one row per unique category. Your data rows can then replace their category field with the _id from the category table.
This has the added advantage you can change the category in the categories table and it will show up in all entries in that category.
SELECT DISTINCT category,_id FROM table GROUP BY category;
I think this should give you what you are looking for. The results from this will be the category, and the first _id for that category. You can ignore the second column (_id).
You can specify an _id field alias in your select statement that is just a constant value, for example:
SELECT DISTINCT 0 _id, category FROM table;
Better yet, I solved this problem by using:
SELECT DISTINCT category AS _id FROM table
Now, you have a column with the name _id which has what you want in it
I get records from the system by quering a ContentResolver. I maintain the order of the items in the database. So I want to display the items in the order taken from my database.
How do I merge these two informations?
EDIT 1
I am looking after an alternative way now. As what I ideally want is:
get order of contacts by a custom order held in my database (this involves joining CR with my DB cursor, and doing an order by, later seams it's not possible with CursorJoiner)
but there is more, if the join is not unique I want to sort by contact's name as last measure
which is impossible using Cursor and Joiners, because of the missing feature of order bys, also I need to return a Cursor, as I will be using the in an ExpandableList
Also this translated to TSQL it would look like
select * from contactsdata
left join category on contactsdata.catid=category.id
order by category.pos asc, contact.display_name asc
So I am looking now after an alternative.
I have in mind to load in a temporary DB table all data from CR, then do the query on the temporary table where I can join tables and do order bys? How does this sound to you?
Take a look at CursorJoiner.
If that doesn't work, you can roll your own equivalent with a fair amount of pain, whiskey, or both.