IMDB has stopped updating the files that this script processes. You can still download a frozen set as of late late 2017 at ftp://ftp.fu-berlin.de/pub/misc/movies/database/frozendata/ .

IMDB -> JSON

IMDB provides an up-to-date dump of their database which is a very interesting source of data. However the format it is provided in is outright rubbish. Each file has a different free text format and polluted by documentation. imdb2json seeks to remedy this by parsing the files and outputting fairly sane JSON.

imdb2json requires python 3.5 and I don't have any plans to make this work with older versions of python. Also, this is a work in progress and not every file and/or every piece of data is currently parsed. Feel free to send patches/pull requests.

Usage

You need to download the .list.gz files from the mirrors provided on IMDB website onto your machine first. Say you pick the Germany mirror, below is an easy way to get all the files:

BASE='ftp://ftp.fu-berlin.de/pub/misc/movies/database/frozendata/'
curl -sl "$BASE" | grep '\.list\.gz$' |
  parallel -j2 curl -s -o {} "$BASE"{}

imdb2json can read both gzipped and plain text files. It detects the type of file based on its content so file names don't matter. imdb2json categorizes the files into two kinds: names and titles. Names refers to people — e.g. actors, directors, stunt people. Titles refers to movies, TV series, episodes, etc. When running imdb2json, you pass the kind of file and it will only process the files that match that kind. Here's an example:

python imdb2json.py convert title ratings.list.gz actors.list.gz

actors.list.gz is ignored because it's not the title kind. The above will output one JSON object per line, containing all the info about a single title. If more files are given, all the info about the same title across all files are merged into a single JSON. To get all the info about all titles, run:

python imdb2json.py convert title *.list.gz

The above will take some time to run. Even with one file, imdb2json still needs to sort the file first and so there will be a delay till the first line is output.

If you have a multicore machine, you should be able to get a faster processing by running multiple copies of imdb2json:

python imdb2json.py list title |
  parallel python imdb2json.py convert title {}.list.gz '>' {}.json

list command outputs the supported IMDB files one per line and GNU parallel runs as many instances of imdb2json as there are CPU cores. After the above is complete, you can merge all the JSON files into one:

python imdb2json.py merge *.json > all.json

Fun stuff

For starters, you can pretty print the output with something like the awesome jq:

python imdb2json.py convert title ratings.list.gz | jq .

Now let's do something semi-useful. Let's find out the top 20 shows/movies with 100K+ votes:

python imdb2json.py convert title ratings.list.gz | jq -r '
  [.] | map(select(.rating.votes > 100000)) |
  .[] | [.id, .rating.rank] | @tsv
' | sort -t$'\t' -k2 -rn | head -20

Movie by Soundtrack

Another way to find movies to watch is to look up titles with soundtracks from your favorite artists. Here's an example of finding titles with >1000 ratings with soundtracks performed by Depeche Mode:

grep -i 'depeche mode' all.json |
  jq -r 'select(.rating.votes > 1000) | [.id, .rating.rank] | @tsv' |
  sort -t$'\t' -k2 -rn