viper3400 / moviesearchengine Goto Github PK

OfdbDetailsPlotResolver

var searchResults = request.GetXPathValues("/html//p[@class=\"Blocksatz\"]/font/b[contains(text(),\"Eine Inhaltsangabe\")]/following-sibling::text()[last()]");

XPath has a problem with the nested tag which should emphasize a word in html page, but breaks the logic of the parser

• Port lib to netstandard 2.0
• remove ofdbwebgw
• separate nuget packages
• integrate with appveyor
• test with videodb webapi
• test with legacy homweb
• deploy to nuget.org
• document changes

The OFDB parser tries to get all links from the movie page, The one containing the word plot will lead to the complete plot page.

This looks normally like this:

<a href="plot/208113,453921,Thor"><b>[mehr]</b></a>

Eventually I've got a link which contains "plot" in but was not plot:

<a href="http://www.moviepilot.de/movies/plotzlich-papa" target="_blank">moviepilot.de - Plötzlich Papa</a>

This ends with a call to an empty URI and an UriFormatException

The result for the cover img url still returns the http protocol. Should be https.

TheMovieDbGenres.json is not copied to destination solution

In TheMovieDb API there is no logging at all. Ofdb has a bit old fashioned static logging. Logging could replaced with Microsoft.Logging and could be injected per DI.

At least, there is no error model at all. So the caller of the API doesn't know, why there is no result, just that there is no result.

SearchMovieByTitle shall be enhanced with the following fields:

• MetaEngine
• Rating
• Reference
• OriginalTitle

SearchMovieById shall be enhanced with the following fields:

• all from SearchMovieByTitle +
• Runtime
• ProductionCountry

When searching just by title, the result does not contain all necessary information. To get i.e. the production countries it will be necessary to append another call with the id.

(In a way, this is valid for both of the engines, Ofdb and TheMovieDb. When using Ofdb some informations are just available after query the edition. Edition concept is unknown in TheMovieDb. So filling the generic interface model is not easy).

Further the call for movie search by title in TheMovieDbApi returns a paged result, while just the first 20 results where presented on the first page.

Saying all that,

• we should just return the first 20 results
  • if the wanted result is not in there, one should refine the search
• It should be applicable to append a detail search for those 20 result within the title search
• This will come with the cost of 20 more calls to the API but with the value of having a mostly complete filled interface model

Search should return a barcode, at least, when effectively searching for one.

Now, the build runs just on new commits and tests are disabled. Build & test shall run periodically to ensure that APIs are not broken (maybe once a week).

There is a dependcy to libjfunxx.dll. I think this is just because of some logging functions. I shoul get rid of this.

Solution contains projects targeting .NET framework version 4.5. They shall target 4.8

Check, if we should switch the ci integration from appveyor to GitHub actions

Language is set to 'de-DE' and not configurable.
Language should be configurable.
https://developers.themoviedb.org/3/getting-started/languages

There are breaking new version of AutoMapper an HTML Agility Pack which needs to be implemented for maintenance reasons

Base url to OFDB is hardcoded in different places. This can be centralized and at least the baseUrl shall be configurable.

Because OFDB changed Url to SSL (https://ssl.ofdb.de) some functions are broken.

When using SearchMovieByEngineId it's assumed that there are two possible formats:

1. 123456
2. 123456;3426246

While the first one ist just the movie id, the second contains the edition, separated with the semicolon.

The case, that the value after the semicolon is null or unvalid is not covered. As well, the case, that the edition number doesn't exist any longer at ofdb or is not resolvable for other reasons is not covered.

In this cases the result should contain at least the informations form movie page - though the edition details will be missing.

There are entries without production year which will cause and uncaught exception. Result will be empty. As the year in the output model is a string, the engine shall return an empty string as year.

Because field plot is not initialized as "text" Entity Framework seems to cut it after 256 chars.

The url for ofdb.de changed from ssl.ofdb.de back to ofdb.de. This breaks the whole OFDB integration.

If nothing is found an generic exception is thrown. User don't really know what was the problem. This would need some more refining.

There is an exception when search for a movie, i.e. with id 461049 because it has no runtime.

https://www.themoviedb.org/documentation/api

GenreListe (german)
Abenteuer
Action
Animation
Dokumentarfilm
Drama
Familie
Fantasy
Historie
Horror
Komödie
Kriegsfilm
Krimi
Liebesfilm
Musik
Mystery
Science-Fiction
Thriller
TV-Film
Western

Year is still missing in result

I should think about creating a NuGet Package.

• Which files need to be delivered?
• Which versioning should be used?

PLEASE USE THE TMDB INTERFACE OF THIS LIB!

Due to a redesign of OFDB it's not longer working with this library!

Me for myself, I meanwhile just use TMDB.

I always knew, the OFDB support was very brittle.

I decided to retire the OFDB support.

For now, I just disabled the unit tests.
Depending on some personal projects I'm currently thinking about removing the code
with a next major version.

Thanks for understanding.