From 45ec3d6418d49c3f4b8029b6a2af3c0f3ca9fe20 Mon Sep 17 00:00:00 2001
From: Alvie Rahman <alvierahman90@gmail.com>
Date: Thu, 27 Jan 2022 20:56:18 +0000
Subject: [PATCH] remove names and other weird words from wordlist

---
 Makefile                |  5 +++--
 allowed_types           |  8 ++++++++
 game.js                 |  1 +
 scripts/gen_wordlist.py | 15 ++++++++++++---
 4 files changed, 24 insertions(+), 5 deletions(-)
 create mode 100644 allowed_types

diff --git a/Makefile b/Makefile
index c9b8d59..7d141d3 100644
--- a/Makefile
+++ b/Makefile
@@ -1,9 +1,10 @@
 WORD_LENGTH=5
-MIN_FREQUENCY=3
+MIN_FREQUENCY=1
 SOURCE_WORDLIST=1_1_all_fullalpha.txt
+ALLOWED_TYPELIST=allowed_types
 
 wordlist.js: 
-	./scripts/gen_wordlist.py ${SOURCE_WORDLIST} ${WORD_LENGTH} ${MIN_FREQUENCY} > wordlist.js
+	./scripts/gen_wordlist.py ${SOURCE_WORDLIST} ${WORD_LENGTH} ${MIN_FREQUENCY} ${ALLOWED_TYPELIST} > wordlist.js
 
 clean:
 	rm -rf wordlist.js
diff --git a/allowed_types b/allowed_types
new file mode 100644
index 0000000..9d62075
--- /dev/null
+++ b/allowed_types
@@ -0,0 +1,8 @@
+adj
+prep
+adv
+num
+int
+verb
+pron
+conj
diff --git a/game.js b/game.js
index dd7c1ad..650b432 100644
--- a/game.js
+++ b/game.js
@@ -158,6 +158,7 @@ function setNextLetter(key) {
     }
 }
 
+
 function endGame(won) {
     gameCompleted = true
 
diff --git a/scripts/gen_wordlist.py b/scripts/gen_wordlist.py
index 3f8e23b..317970a 100755
--- a/scripts/gen_wordlist.py
+++ b/scripts/gen_wordlist.py
@@ -12,15 +12,23 @@ def get_args():
     parser.add_argument('wordlist')
     parser.add_argument('word_length', type=int)
     parser.add_argument('frequency_min', type=int)
+    parser.add_argument('allowedtypelist')
     return parser.parse_args()
 
 
 def main(args):
     """ Entry point for script """
-    with open(args.wordlist) as fp:
-        words = [ (word[1], int(word[4])) for word in [ word.lower().split('\t') for word in fp.read().strip().split('\n') ] ]
 
-    words = [ word[0] for word in words if word[1] > args.frequency_min and word[0].isalpha() and len(word[0]) == args.word_length ]
+    with open(args.allowedtypelist) as fp:
+        allowed_types = fp.read().split('\n')
+
+    types = set()
+    with open(args.wordlist) as fp:
+        words = [ (word[1], int(word[4]), word[2]) for word in [ word.lower().split('\t') for word in fp.read().strip().split('\n') ] ]
+
+    [ types.add(word[2]) for word in words ]
+
+    words = [ word[0] for word in words if word[1] >= args.frequency_min and word[0].isalpha() and len(word[0]) == args.word_length and word[2] in allowed_types ]
 
     words.sort(key=lambda word: word[1])
     # remove duplicates
@@ -28,6 +36,7 @@ def main(args):
     print(f"wordlist = {json.dumps(words)}")
     print(f"{args}", file=sys.stderr)
     print(f"{len(words)=}", file=sys.stderr)
+    print(f"{types=}", file=sys.stderr)
 
     return 0