Switch to wikipedia words

2022-02-28 19:30:49 +01:00 · 2022-02-28 19:30:49 +01:00 · 55b61ad858
commit 55b61ad858
parent 85180cd926
5 changed files with 5000 additions and 2692 deletions
--- a/blacklist.txt
+++ b/blacklist.txt
--- a/filter.sh
+++ b/filter.sh
@ -1,7 +1,7 @@
 #!/bin/sh
 set -e
-word_file=deu_news_2021_100K-words.txt
-cat $word_file | head -n15000 | cut -f2 > top10000de_utf8.txt
+word_file=deu_wikipedia_2021_1M-words.txt
+cat $word_file | head -n35000 | cut -f2 > top10000de_utf8.txt
 awk '{ if (length($0) == 5) print tolower($0) }' top10000de_utf8.txt > top10000de_utf8_len5.txt
 cat top10000de_utf8_len5.txt | rg "^([A-Za-z]|ä|ö|ü|Ä|Ö|Ü)+\$" | sort | uniq > top10000de_utf8_len5_filtered.txt
 comm -23 top10000de_utf8_len5_filtered.txt blacklist.txt > valid_words.txt
--- a/index.js
+++ b/index.js
--- a/valid_guesses.txt
+++ b/valid_guesses.txt
--- a/valid_words.txt
+++ b/valid_words.txt