Upload llm_services.py

app/llm_services.py

@@ -24,17 +24,18 @@ Focus on:
 
 ### Response Format (in markdown):
 
-1. Start with a brief 2 sentences **opening paragraph** that contextualizes the theme and the overall viewing experience the user is seeking.
+1. Start with a concise 2 sentences **opening paragraph** that contextualizes the theme and the overall viewing experience the user is seeking. At the end of this paragraph, insert the token: <!-- END_INTRO -->.
 
 2. Then, for each recommendation, use the following format (repeat for each title). At the end of each movie recommendation block, insert the token: <!-- END_MOVIE -->:
 
 ```
 ### <Number>. <Movie Title>
-![<Movie Title>](<Poster URL>)
-- **Genres:** Genre1, Genre2, ...
-- **IMDB Rating:** X.X
-- **Rotten Tomatoes Rating:** XX%
-**Why You Might Enjoy It:** <Short paragraph explaining the appeal based on character, themes, tone, and relevance to the user's intent.>
+- POSTER_PATH: /abc123.jpg
+- GENRES: Genre1, Genre2, ...
+- IMDB_RATING: X.X
+- ROTTEN_TOMATOES_RATING: XX%
+- TRAILER_KEY: abc123
+- WHY_YOU_MIGHT_ENJOY_IT: <Short paragraph explaining the appeal based on character, themes, tone, and relevance to the user's intent.>
 <!-- END_MOVIE -->
 ```