Create Spacy python NLP model to detect interests

I need you to crater spacy model (NLP framework in python) to detect any interests in Twitter like posts without any classification or post calculations. I mean, I need script that only extracts "hot phrases".
1. Detect hobbies, sport, entertainment:
I love soccer! - interests: soccer.
Buy the way rope jumping is awesome! - interests: rope jumping.
Yesterday I was on basketball game - interests: basketball.
2. Detect films, music, art or any other content:
I have seen movie about ants - interests: movie, ants, movie about ants.
Only imagine dragons, only rock! - interests: rock, imagine dragons.
I want to declare my Netflix status, which is Peaky blinders fan - interests: Netflix, Peaky blinders
3. Any words or phrases that is not common:
I can tell you what bugaga is - interests: bugaga
My heart is broken he is hjagaf - interest: heart is broken, hjagaf
4. Hashtag phrases:
Stop racism! - interests: racism, stop racism
Pizza is super! - interests: pizza
I am Superman! - interests: Superman
5. Names, companies, geographic places:
New York is the best - interests: new York
Johnny Depp is so cute - interests: Johnny Depp
I want to buy Google stocks - interests: Google, stocks
6. Occupation, education or major:
I am engineer - interests: engineer
I have PhD in biology - biology, PhD, PhD in biology
We had meeting today in the office - interests: meeting, office
7. Any, literally any interest, characteristic or key word:
I believe that the superior creature is watching how I am having sex with beautiful models because I am the richest plumber in the California - interests: superior creature, sex, beautiful models, richest plumber, California.