Quoting GSyren:

Quote:
Quoting mediadogg:

Quote:
I was hoping that you would offer to sanitize on the way in, then never have to worry about it for subsequent queries.

That requires that the output is run through CltInfo before being used for other purposes. But maybe that's a limitation that we can live with.

I think for now, CLTInfo should be viewed as the "go to" successor to CLTPlus. Anybody else who want to use the raw XML will undoubtedly also have the skills to do the search.

Maybe I can also add a checkbox (here we go again). User can decide whether to suffer my search or leave it to the experts.

Had an idea that is bearing fruit. Since I must make at least one database call per scraped UPC, I am using as much as possible from that one call, trying to avoid any time spent in parsing the xml or making additional database calls.

(1) I get a UPC from the CLT scrape, but there is a small % chance it is not valid (explained in detail earlier)
(2) I make the database call to retrieve the profiles associated with the UPC
(3) In lieu of scanning the XML, I do the following quick comparisons with data already collected

- if there is only one profile, by definition its a match
- otherwise try a simple match on ( (title or title:edition) and production year ) which indicates a match
- if not either of those, then whether credits exist can only be determined by XML scan or database calls

I am finding only a very small percentage of profiles that need scanning, so maybe it is better if I just go ahead and do it, by default. but allow the user to disable the validation step and just deliver all the profiles found by the database calls, to the XML, which can result is a small percentage of profiles that actually have no credits for the variants in question.

I've been playing around a bit with XPath to see how easy it would be to create the common name statistics from the Profiler XML format. It turns out it works quite well. I realized that I could quite easily use it for either format, but it would be a whole lot easier if the CLTBoss output had Actors/Actor and Credits/Credit instead of CLTCredits/CLTCredit.

I don't suppose you would consider changing that, this late in the game?

Oh and just to be clear, Variants/Variant goes to Actors/Actor in both forms of XML output?

Sample that I will delete to reduce clutter, after you take a look.


Content deleted.

Quoting Kathy:

Quote:
.

Hey Kathy, glad to see you monitoring this thread. Any thoughts?

Quoting GSyren:

Quote:
OK, a couple of thought.

I would prefer if the variants went into a node with a name that isn't also used later, so perhaps Arguments/Argument, or something similar. And there seems no point in having FirstName, MiddleName and LastName there if they are not used as such (unless you have something else in mind for them). If you actually have them properly parsed, that would be really helpful.

And what I had in mind instead of CLTCredits/CLTCredit was Actors/Actor for cast credits and Credit/Credits for crew credits, just like in the Profiler output. Is that doable?

Oh of course it is doable, actually I had forgotten they were different. I already made the change, but it was the same for both. Easy to fix.

I agree about the variants, but I thought I was doing what you wanted. I'll leave it as variants then. If it needs to be different, then it already is (was - now I'll change it back). The names are properly parsed, and I have a Class coded with methods that I can tweak if anybody alerts me to a problem.

Edit: I should say that I attempt to properly parse them. If it doesn't happen, it is a bug. For example, if the user types in "Queen Elizabeth" into the firstname field, the XML will have fn="Queen", mn="" and ln="Elizabeth".

Have I tested all the ways a user could break my code? Probably not. My focus is on getting the CLT results correct at the moment.

Quoting GSyren:

Quote:
Just to be clear what I meant:

Quote:
<Actor FirstName="melissa locsin" MiddleName="" LastName="" BirthYear="0" HashName="locsin_melissa_">melissa locsin</Actor>

In this example you put the whole name in FirstName. I would have expected:

<Actor FirstName="melissa" MiddleName="" LastName="locsin" BirthYear="0" HashName="locsin_melissa_">melissa locsin</Actor>

(And Variant instead of Actor, of course)

That was not intended. I thought I had that covered, as my prior post indicates. Consider it a bug. My code has been through so many iterations, there was probably a time that it worked. I think my error is that for the XML, I am pulling the fields from the grid. Instead, all I need to do is populate an instance of my Actor class, and it will "automatically" fix the parsing.

Quoting Kathy:

Quote:
Quoting mediadogg:

Quote:
Quoting Kathy:

Quote:
.

Hey Kathy, glad to see you monitoring this thread. Any thoughts?

None. I have no idea what any of this is about. 

I just posted the period because the data took up so much of my screen.

Yeah, I don't usually do that. I will be deleting it soon.