Hi Igor,
First of all, I would like to express my appreciation for your product. as a programmer, I appreciate your success in finding the place where software is mostly needed. I also know that interfacing into Microsoft's products is not an easy task. You have done a remarkable job with your application. I really hope you will be able to go commercial some day. If so - I will probably be one of the first ones to buy a license.
As for my question - I am using Vista, with HOT cable provider. Downloading the EPG from the sites you recommended, I found out that sometimes the episode name or number appear in the title of the show. For example, you might have a show named "Friends - episode no. 66" (In hebrew, of course). This gives me hard times in recording the whole series, becasue the system can not recognize that episode 66 and 67 belong to the same series.
I have tried defining the series by providing Vista with a specific keyword contained in the title (something like: record any show that has the word "Friends" in it), but unfortunetly, I can not enter words in Hebrew in Vista's interface (This is probably a bug of the Hebrew UI of Vista).
Is there any way to workaround this issue?
I try to parse episode
I try to parse episode numbers and believe me it's a pain in ass. There is no standard way HOT/YES prepare episode numbers, it can be "Title - episode NNN", it can be "Title - NNN", it can be "Title NNN." and few other formats. I parse as much as I can but still miss episode numbers in some of the series.
I'm thinking about a better algorithm now and hope to introduce it in a upcoming release.
Only one pattern is missing
Seems like you are doing a good job in parsing it. You are missing one pattern(from what I have seen). It goes like that: [any_charater]*[whitespace]*[number].[any_character]*
e.g: Firends 37.Bla Bla
Notice the dot after the number of the episode. It will help you identify the pattern. If after the dot there is something (and not the end of the sentence), then it's probably an episode number. In this case you should remove the [number].[any_character]* sequence
By the way - did you think about publishing the sources, making your product an open source? That way people can contribute to your porduct... How about it?
Gal
regexp for "any
regexp for "any character-something-any character" is very problematic and extremely slow for a large amount of data.
There will be false matches since quite a few movies contain numbers in name (I checked and found many surprisingly).
Regarding the open source: Of course I thought about it. There is a problem I didn't find a solution yet: I don't feel ok to publish some internal things of MS despite I discovered them on the Internet as well. When and if I find a way to publish sources for everything else but them, I'd probably do that. But now it seems rather problematic than appealing.
I see the problem
I understand the problem.
Are those movies contain number, and a dot after it in the description? I think that number and a dot after it will probably not give false match, but I might be wrong (I did not check it).
I see the problem in open source issues.
As for regexp performance - what exactly are you using? C++? Perl? Any library (such as QT?)
Anyway - please let me know if and when there will be a release that fix this problem.
I really appreciate your efforts.
Gal