@@ -31,30 +31,27 @@ def all_claims
3131 [ humans_claim ] + movies
3232end
3333
34+ def output_dir = ENV . fetch ( "DIR" )
35+
3436desc "Update the movie database"
3537namespace :movies do
3638 desc "(1) Download the wiki data dump"
3739 task download : [ :environment ] do
3840 with_log do
39- output_dir = ENV . fetch ( "DIR" )
40-
4141 `curl -O https://dumps.wikimedia.org/wikidatawiki/entities/latest-all.json.bz2 --output-dir #{ output_dir } `
4242 end
4343 end
4444
4545 desc "(2) Decompress the downloaded wiki data dump"
4646 task decompress : [ :environment ] do
4747 with_log do
48- output_dir = ENV . fetch ( "DIR" )
49-
5048 `(cd #{ output_dir } && lbzcat latest-all.json.bz2 | rg '(#{ all_claims . map { "\" #{ _1 } \" " } . join ( "|" ) } )' > latest-all-reduced.json)`
5149 end
5250 end
5351
5452 desc "(3) Generate the movies file"
5553 task generate_movies : [ :environment ] do
5654 with_log do
57- output_dir = ENV . fetch ( "DIR" )
5855 input_file = File . join ( output_dir , "latest-all-reduced.json" )
5956 output_file = File . join ( output_dir , "movies.json" )
6057
@@ -65,7 +62,6 @@ namespace :movies do
6562 desc "(4) Generate the humans file"
6663 task generate_humans : [ :environment ] do
6764 with_log do
68- output_dir = ENV . fetch ( "DIR" )
6965 input_file = File . join ( output_dir , "latest-all-reduced.json" )
7066 output_file = File . join ( output_dir , "humans.json" )
7167
@@ -76,7 +72,6 @@ namespace :movies do
7672 desc "(5) Generate the minimized humans file"
7773 task generate_humans_minimized : [ :environment ] do
7874 with_log do
79- output_dir = ENV . fetch ( "DIR" )
8075 input_file = File . join ( output_dir , "humans.json" )
8176 output_file = File . join ( output_dir , "humans-min.json" )
8277
@@ -87,8 +82,6 @@ namespace :movies do
8782 desc "(6) Import movies from a wikidata json dump"
8883 task import : [ :environment ] do
8984 with_log do
90- output_dir = ENV . fetch ( "DIR" )
91-
9285 series_classes = File . readlines ( Rails . root . join ( "config/wikidata_series_classes" ) ) . map ( &:strip )
9386
9487 puts "Reading humans..."
0 commit comments