Personal Video Database

Hey, thanks, joc!

Checking it out now... works mostly, though it seems to mix up the actors and directors a little. It selects the last actor listed under "Cast" as the director if no director is specifically listed. And even where a director(s) ARE given, it still selects the last (non-director) cast member as a director anyway. In both cases, the erroneous "director" isn't properly added to the "Actors" field in PVD.

But it's a definite improvement over the non-working old script. (Beats entering EVERYthing manually.)

Aimhere

Yes, it's true... the layout and HTML code of the search-results pages have completely changed, as have the actual movie pages. This has completely broken the adultdvdempire script. 

I've been kind of maintaining the adultdvdempire script for a couple of years, but I've only had to make minor tweaks as the site layout changed. I'm no programmer, and a site revision of this magnitude is simply beyond my capabilities. I just don't have the time or skills to deal with it, unfortunately.

I really, really wish I knew WHO had written the ORIGINAL adultdvdempire script. Whomever it was would be better equipped to deal with this change. Anyone know???

Barring that, any aspiring script-writers out there who want to pick up the ball and run with it?

Aimhere

Hi,

I have a minor problem in PVD 0.9.9.21 I'd like to solve. There was a time a while back when the IMDB plugin was returning URLs of the form "http://us.imdb.com/...", instead of "http://www.imdb.com/...", because that's what IMDB itself was returning for search results at the time. Eventually, IMDB went back to using "www.imdb.com", though. But now, I have a number of performers in my People database who still have the "us.imdb.com" flavor of URLs, and every time I update a record from IMDB (via the plugin) without remembering to change the "us" to "www", I wind up with two URLs in the URL field, the original "us" and the current/standard "www".

What I'd like to do is filter my People database to only show records that still have "us.imdb.com", so I can update them all and be done with it. But there doesn't seem to be any way to search by URL in the "Search" box. "URL" is not an option in the drop-down list next to the "Search" box, and choosing "All" doesn't actually include the URL field in the search (apparently "All" only includes the fields listed in the drop-down list, which, again, doesn't have "URL".)

I thought that maybe I could export the URLs to a CSV file (which I can), edit that, then re-import it... but discovered there is no CSV import for People, only for Movies. 

So my question is, IS there a simple way to do a search (and replace) on the URL field in People view?

Aimhere

See, the default set of RegExes in PVD's file scanner configuration already removes things like "CD1" or "CD01" when reading the filenames, using the pattern "(?i)\bCD\d{1,2}\b" in the "find and replace" section. I'm just trying to get it to do the same for "D1CD1" and the like. Without the additions I've made to the RegExes, the file scanner would give me titles like "Some Movie D1CD". (This is why the file scanner configuration HAS a "find and replace" section... it strips out a lot of garbage that would otherwise get included in the extracted titles, BEFORE even attempting the extraction.)

A lot of the movies in my collection were released as multi-disc collections, with each disc in multiple 700MB AVI chunks. The filenames HAVE to include strings like "D1CD1" so each file has a unique filename. I could use something like VirtualDub to combine the chunks into a single AVI, but this is an awful lot of work given the number of files involved (with me acquiring more all the time), not to mention presenting the danger of exceeding file-size limits that may exist when I want to back up these files on DVD-R. Also, the second disc for each title often contains ONLY "extras", not any part of the main movie, so I wouldn't want to combine that with the first disc (usually I only want to watch the main movie, but I still want to keep the extras around in case I ever feel the urge to watch them). In any event, I'm still going to have multiple files for any particular movie.

I'm already renaming the files as I acquire them... believe me, the original filenames as posted on Usenet are practically worthless.  (They don't include readable movie titles or anything useful.)  

Edit: I should mention, all of my movies have filenames of the general form "Title Discnumber (Year, Studio).avi", where "Discnumber" is the "CD1" or "D1A" or whatever. I suppose I could move the "Discnumber" part to the very end, maybe that would keep it from being included in the title without all of the fiddly RegEx stuff. But then I'd want to rename all of my existing AVI files (several thousand, burned onto DVD-R) to the new format for the sake of consistency, and besides, the RegExes I came up with seem to work fine...

Aimhere

Okay, I managed to figure out the strings I needed to add to the "find and replace" section in the File Scanner preferences to strip out the strings I wanted to remove:

(?i)\bD\d{1,2}CD\d{1,2}\b
(?i)\bD\d{1,2}[a-z]\b
(?i)\bD\d{1,2}\b

I just added these after the existing "(?i)\bCD\d{1,2}\b" line, and tested on a folder full of files named as we've been discussing. This seems to remove all of the "D1CD1" and "D1A" style strings from the filenames. Maybe there's a more elegant way to do the same thing with fewer RegEx's, but I'll take what works.

I've noticed another problem, though: the first line in the default set of "find and replace" expressions, "(?i).?((?<!\b).){0,5}Rip", is clearly intended to remove strings like "DVDRip" and "BDRip". But, I'm finding it ALSO removes any unrelated string that happens to end with the substring "rip". In other words, it removes words like "Trip", "Strip", "Grip", and so on. (And words like "Gripped" get mangled to "ped".) After some wrangling with it, I managed to come up with a RegEx that removes things like "DVDRip" while leaving words like "ripper", "tripping", etc. alone:

(?i)\b(CD|DVD|BD|Blu-Ray|BluRay|HD-DVD|HDDVD|VHS|Vinyl|Cassette|Tape|.{0})Rip(\s|\b)

Yeah, it's not pretty, but I figure it covers all the "rip" bases I'll ever encounter. (I had to add the fiddly bit at the end to strip out spaces that were creeping into the modified string.)

Aimhere

Hi,

I need some help with the configuration for the File Scanner, specifically, regular expressions.

I like how, when importing video files, PVD strips things like "CD1", "CD2", and so on from the filenames when generating a movie title. So, for example, "This Movie CD1 (2011, SomeStudio)" becomes "This Movie" in the title field.

Thing is, I have more and more movies that came from multiple DVD sets, where each DVD is broken into CD-sized files as well. They have filenames like "That Movie D1CD1", "That Movie D1CD2", "That Movie D2CD1", etc. ... OR, "Another Movie D1A", "Another Movie D1B", "Another Movie D2A", etc.

When I point PVD's file scanner to movies like these, I wind up with titles like "That Movie D1CD" or "Another Movie D" in the Scan Results window. Note how it retains part of the disc/CD identification tag. I have to edit the title and strip them out by hand.

Also, with the "DxCDy" naming convention, PVD doesn't group all the parts together under the same title, instead I get a separate item for each DVD ("That Movie D1CD" AND "That Movie D2CD"). So, I have to select both items and right-click to choose "Same Movie".

Now, I don't really know much about regular expressions, so I'm not sure how to go about fixing this in the preferences for the File Scanner. I'd like to retain the "D1CD1" or "D1A" tagging for the filenames, but keep it from carrying over to the titles in the File Scanner.

Any ideas?

Aimhere

Great idea. I often wish synchronized movie and people panes could be shown at the same time. If this is not to be, synchronized panes not shown at the same time would be the next best thing.

It should be noted the simple solution of running the necessary search in the other view is not going to work. A filmography search is much too slow. And then anything done in the view that requires a refresh results in the search being run again. No, the people in the movie record would have to be programmatically selected in the people list. For simplicity, it should probably select all credits, rather than be restricted to actors or applied to each type separately.

Agreed. The filmography search IS slow, but it's all I have available right now. Still, on a reasonably modern system, it only takes a few seconds.

I'd still prefer for the people list to be "boiled down" to the ones associated with the chosen movie (filtering out irrelevant ones), rather than merely selecting the records. That way, I can see at a glance exactly which people and how many I'm dealing with, rather than scrolling through my entire people list and looking for highlighted records.

Also, most of the time I only want just the actors. But having the option to show other credited people too would be a plus.

The reverse function isn't as important. It's not much trouble to switch to a particular movie using a link in a person's filmography. From there, a search can be run from the person's context menu—and those searches are reasonably fast.

Maybe so, but again, it'd be nice to be able to do it in a single step.

What would be handy is a simple search (also available on the context menu) for searching a person for any credit type (i.e., producer, director, actor..., just like there is for titles).

True.

Aimhere

Success! The number of visible movie records didn't change after this optimization pass. PVD did delete 368 movies, but they were all invisible ones.

I guess now it's just a matter of preventing future duplications.

Aimhere

I've never noticed it delete a visible movie. You could export before and after lists of visible moves and compare those in Excel to determine what was deleted. That should indicate why the deletions are happening.

Okay, did before-and-after exports, and compared them. PVD deleted eight out of 2022 visible movie records. I can't find any rhyme or reason why it would think they are duplicates, either. Looking at my "before" database, the movies all have proper titles and "original titles", all unique as far as I can tell. (A couple are parts of series, e.g. "Some Movie", "Some Movie 2", etc., but they shouldn't be considered duplicates.)

I wonder if we could get Nostra to weigh in on the methodology used to determine whether movies are duplicates?

Aimhere

Quote

The plugin is set to overwrite filmographies, and "Merge filmography" selected in Preferences.

Why do you have "Merge filmography" selected? If it wasn't, running the plugin would fix the problem by replacing everything with the person's current filmography.

What happens if I don't have "merge filmography" selected, and I have a movie in my database that ISN'T in IMDB's filmography? (IMDB isn't always complete, after all.) What happens to my movie record? Would the actor's filmography in PVD continue to list that movie, even if it wasn't found on IMDB? Or would the filmography omit it, thus disconnecting the movie from the actor? I thought that's what "Merge filmographies" was supposed to prevent?

Quote

IMDB lists a different release year than the one I already have in my database record for that movie.

Why are they different?

Because I import most of the movie info (posters/box art, studio, descriptions, etc.) from another source, AdultDVDEmpire. (In this genre, IMDB rarely has any info other than the title and cast.) And the two sources just happen to disagree on release year.

In my typical workflow, I'll add a movie, import the movie info from ADE, confirm the list of cast members, go to each cast member in turn, and update the person's filmography from IMDB. It's that last step that's giving me the duplicate movies in the filmography, when ADE and IMDB disagree on the release year.

Quote

I don't know if the "Delete movie duplicates" in "Tools/Optimize Database" will do what I want, and I'd rather not risk having it delete any "complete" movie records.

I believe the only way that could happen is if both records are visible. That's not the case in the scenario you describe. But if you're concerned about that possibility, you can still run the utility with the option to remove only invisible records.

"Delete movie duplicates" (unlike "Delete movie orphans") doesn't have an option to only remove invisible records.

Okay, as a test I made a copy of my database, then ran "Optimize database" with "Delete movie duplicates" selected. It did delete some 300+ movies. Unfortunately, it removed eight (out of 2009) visible movie records, e.g. movies that were legitimately in my collection. I don't know why this would happen, I'm certain I didn't have any actual duplicates. It's possible there may have been a couple of movies (from different studios) with identical titles, but whenever I run across those, I add the studios to the titles to avoid confusion...

Aimhere