Tagger/exclude.py

110 lines
1.3 KiB
Python
Raw Permalink Normal View History

2022-05-05 17:30:56 +00:00
EXCLUDED_WORDS = {
2022-05-28 13:39:16 +00:00
# Deutsche wörter
2022-05-28 13:41:22 +00:00
"aber",
"alle",
"als",
"am",
"an",
2022-05-07 17:50:44 +00:00
"auch",
2022-05-28 13:41:22 +00:00
"auf",
"aus",
2022-05-07 17:50:44 +00:00
"bei",
2022-05-28 13:41:22 +00:00
"bis",
"dabei",
"dafür",
"damit",
"das",
"dass",
"davon",
"dem",
"den",
"der",
"des",
"die",
"diese",
"dir",
"dort",
"du",
"ein",
2022-05-07 17:50:44 +00:00
"ein",
"eine",
2022-05-28 13:41:22 +00:00
"einem",
2022-05-07 17:50:44 +00:00
"einen",
"einer",
"eines",
2022-05-28 13:41:22 +00:00
"einige",
2022-05-07 17:50:44 +00:00
"er",
"es",
2022-05-28 13:41:22 +00:00
"etwas",
"euch",
"für",
"gibt",
"haben",
"hat",
2022-05-28 13:39:16 +00:00
"hier",
2022-05-28 13:41:22 +00:00
"ich",
"ihr",
"im",
"immer",
"in",
"ins",
"ist",
"jetzt",
"kann",
"man",
"mehr",
2022-05-28 13:39:16 +00:00
"mein",
2022-05-28 13:41:22 +00:00
"meine",
2022-05-28 13:39:16 +00:00
"mich",
2022-05-28 13:41:22 +00:00
"mir",
"mit",
"nach",
"nicht",
"noch",
"nur",
"oder",
"ohne",
"schade",
2022-05-28 13:39:16 +00:00
"sehr",
"sein",
"seit",
2022-05-28 13:41:22 +00:00
"sich",
"sie",
"sind",
"um",
"und",
"viele",
"von",
2022-05-28 13:39:16 +00:00
"vor",
2022-05-28 13:41:22 +00:00
"war",
"wenn",
"werden",
"wie",
"wir",
"wird",
"zu",
"zum",
"zur",
"über",
2022-05-28 13:39:16 +00:00
# Englische wörter
"and",
2022-05-28 13:47:13 +00:00
"about",
2022-05-28 13:39:16 +00:00
"in",
2022-05-28 13:41:22 +00:00
"more",
"much",
"of",
"or",
"this",
"your",
2022-05-28 13:39:16 +00:00
# URL Bestandteile
"https",
2022-05-28 13:41:22 +00:00
"http",
2022-05-28 13:39:16 +00:00
"www",
"com",
"de",
"org",
"net",
"it",
"ch",
2022-05-05 17:30:56 +00:00
}