Skip to content

Latest commit

 

History

History
78 lines (50 loc) · 3.5 KB

02-IMDB-ID-Detection.md

File metadata and controls

78 lines (50 loc) · 3.5 KB
Raw

IMDB ID detection

In order to retrieve metadata from IMDB, Media Hoarder needs to find out the IMDB ID for each movie.

IMDB ID as part of the file- or directory name

If the IMDB ID is part of the file- or directory name, Media Hoarder will immediately use it. The ID must be of the format "tt0000000".

Examples:

The Godfather [tt0068646].mkv

/movies/Avengers Endgame (2019)(tt4154796)

IMDB ID is referenced in an .nfo file

Media Hoarder parses .nfo files and checks if an IMDB ID is available. For this, the movie must be in the same directory as the .nfo file. The .nfo filename is irrelevant.

IMDB ID is neither part of the file- nor the directory name

If the IMDB ID is neither part of the file- nor the directory name and cannot be found within an .nfo file, Media Hoarder tries to detect the correct IMDB entry by performing an IMDB search with parts of the filename.

It is greatly beneficial if the file name also provides the movie's release year.

Testing the IMDB ID Detection

We generated a controlled test set of 7576 movies, for details see https://github.com/theMK2k/media-hoarder-testset-generator. The testset and the benchmark results can be viewed at ../src/tests/testdata.

We ran tests on the following sets:

  • 01 English movie titles including year
  • 02 English movie titles without year
  • 11 German movie titles (if applicable, else the original title) including year
  • 12 German movie titles without year

Result of the tests:

  • 01 (English w/ year)
    • 7568 of 7576 movies correctly detected (99.9%)
    • 8 of 7576 movies incorrectly detected (0.1%)
  • 01 (English w/o year)
    • 7497 of 7576 movies correctly detected (99%)
    • 79 of 7576 movies incorrectly detected (1%)

Media Hoarder's Search Algorithm is as follows:

  1. Take the file- or directory name
  2. Create the filtered name by removing unnecessary characters (. , _ -) and replace them with space
  3. Extract possible mentions of a year from the filtered name, i.e. substrings containing extactly 4 digits
  4. Furthermore remove anything enclosed in brackets from filtered name, we assume attributes like (HD), [BluRay] etc. within brackets
  5. Feed the filtered name into a search query to IMDB
  • the search returns no result: remove the last word from the filtered name and run 5. again
  • the search returns exactly one result: use this result's IMDB ID
  • the search returns more than one result:
    • if possible exclude entries from the result that don't match the years extracted from the file- or directory name (see 3.)
    • if possible match the result's entries runtime to the actual runtime of the movie, sort the result accordingly
    • take the best match (by year and/or runtime) or just take the first element in the result

IMDB ID Detection for Series and Episodes

Series

For series, Media Hoarder will try to detect the IMDB ID for the series itself.

If the series' directory name contains the IMDB ID, it will be used directly.

Otherwise, Media Hoarder will try to find the correct IMDB data by searching for the series' name and year.

Episodes

For episodes, if the file- or directoryname contains the IMDB ID, it will be directly used. This also means, you can provide any IMDB imaginable (even those of stand-alone movies) and Media Hoarder will use it.

Otherwise, it is crucial that the series is correctly linked to the IMDB ID. With that, each episode will be matched to its IMDB entry using the season and episode numbers.