ansaurus

Question

LINQ 4 XML - What is the proper way to query deep in the tree structure?

Answer 1

+1 A:

This code you have here:

XmlSerializer s = new XmlSerializer(typeof(EmailRuleList)); 
TextReader r = new StreamReader(path); 
_emailRuleList = (EmailRuleList)s.Deserialize(r); 

TargetPST[] PSTList = _emailRuleList.Items; 
foreach (TargetPST targetPST in PSTList) 
{ 
    olRoot = GetRootFolder(targetPST.name); 
    if (olRoot != null) 
    { 
        Parse[] ParseList = targetPST.Items; 
        foreach (Parse parseRules in ParseList) 
        { 
            EmailRule[] EmailRuleList = parseRules.Items; 
            foreach (EmailRule targetFolders in EmailRuleList) 
            { 
            } 
        } 
    } 
}

Is effectively just the following in LINQ:

var query = from targetPST in _emailRuleList.Items
            let olRoot = GetRootFolder(targetPST.name)
            where olRoot != null
            from parseList in targetPST.Items
            from emailRule in parseList.Items
            select [whatever you want to select];

From here, just include the appropriate where clauses. At every "level" you have access to the objects previously specified in a from clause, so if you wanted to do one of your queries (your first one), it'd be something like this:

where emailRule.address == ToAddress && 
    parseList.toAddress == true &&
    parseList.useJustDomain == false
select new {
    EmailRule = emailRule,
    Parse = parseList,
    TargetPST = targetPST
}

To do this in methods, I think you're doing more harm than good (esp in readability), but here it is. The important thing to note on why this is so complicated is that a nested set of FROM clauses translates into a SelectMany, and because ultimately you need the EmailRule with its associated parent objects, you have to do all your selecting inside a set of nested lambdas in order to have references to the parent objects (since the child objects themselves do not have backreferences).

_emailRuleList
    .Where( targetPst => GetRootFolder( targetPst.Name ) != null )
    .SelectMany( targetPst => {
        return targetPst.Items.SelectMany( parse => {
            return parse.Items.Select( rule => {
                return new {
                    TargetPST = targetPst,
                    Parse = parse,
                    EmailRule = rule
                };
            } );
        } );
    } )
    .Where( x => x.EmailRule.address == ToAddress &&
                 x.Parse.toAddress == true &&
                 x.Parse.useJustDomain == false );

Adam Sills 2010-03-25 20:45:48

anyway to do something like: var pRules = _emailRuleList.Items.Where(a => a.Items.Where(b => b.Items.Where(c => c.address == toAddress))).ToList();

Keith Barrows 2010-03-25 23:16:04

Well, aside from that being much harder to read, yes, you can sort of do that. But you can't nest the WHERE methods like that. The problem is you want the EmailRule with its Parse and TargetPST. What you've got here is the TargetPSTs where all the various conditions are true.

Adam Sills 2010-03-26 13:18:58

If you want to structure your query using a set of methods, I think you're going to make a harder time for yourself just to get the backreferences working correctly (to do so, the entire thing must be a set of nested lambdas, as you are doing) whereas if you use the LINQ query it handles that for you.

Adam Sills 2010-03-26 13:22:37

Yea. Found that out. Also fought with returning anonymous types from a method. lol. Since the variables change but the linq statement does not, I moved it to a method. I need to return 2 stings and a bool out of the entire mess so created a class and dropped the var in favor of a defined class. Works like a champ!

Keith Barrows 2010-03-26 19:07:12

ansaurus

tags:

views:

answers:

LINQ 4 XML - What is the proper way to query deep in the tree structure?

related questions