import os
import sys
src_directory = os.path.abspath(os.path.join(
    os.path.dirname(__file__), "../..", "src"))
sys.path.append(src_directory)
from services import data_processing
import pandas as pd
from collections import Counter
from sqlalchemy.orm import Session


def get_trending_videos_count():
   df = data_processing.get_updated_df()
   trending_counts= df.groupby(["trending_date"])['trending_date'].count()
   return trending_counts

def get_most_popular_categories():
    df = data_processing.get_updated_df()
    category_counts = df.groupby(["category_name"])["category_name"].count()
    return category_counts


def get_views_vs_likes():
   df = data_processing.get_updated_df()
   scatter_data = df[["views", "likes"]].dropna()
   return scatter_data


def get_like_ratio_distribution():
    df = data_processing.get_updated_df()
    data= df[["video_id","views","likes"]].copy()
    data = data[data["views"] > 0]
    data["like_ratio"] = data["likes"] / data["views"]
    return data

def get_top_liked_videos(top_n=10):
    df = data_processing.get_updated_df()
    top_videos = df[["title", "likes"]].dropna().sort_values(by="likes", ascending=False).head(10)
    return top_videos

def get_trending_channels():
    df = data_processing.get_updated_df()
    data = df[["channelTitle","publishedAt"]].copy()
    return data

def calculate_channel_growth():
    data = data_processing.get_updated_df()
    data["publishedAt"] = pd.to_datetime(data["publishedAt"], errors="coerce")
    data.dropna(subset=["publishedAt"], inplace=True)
    data["published_month"] = data["publishedAt"].dt.to_period("M").astype(str)
    grouped_data = data.groupby(["published_month", "channelTitle"]).size().reset_index(name="video_count")
    return grouped_data


def process_tags():
        data = data_processing.get_updated_df()
        tags = data["tags"].dropna().str.lower().str.split("|")
        all_tags = [tag.strip() for sublist in tags for tag in sublist if tag.strip()]
        tag_counts = Counter(all_tags)        
        tag_data = [{"tag": tag, "count": count} for tag, count in tag_counts.items()]
        return {"tags": tag_data}


def analyze_trending_duration():
    data = data_processing.get_updated_df()
    trending_days = data.groupby("video_id")["trending_date"].count().reset_index()
    trending_days.columns = ["video_id", "days_trending"]
    views_growth = data.groupby("trending_date")["views"].mean().reset_index()
    
    return {
        "lifespan": trending_days.to_dict(orient="records"),
        "views_growth": views_growth.to_dict(orient="records")
    }


def analyze_upload_patterns(mode: str): 
    data = data_processing.get_updated_df()   
    data["publishedAt"] = pd.to_datetime(data["publishedAt"], errors="coerce")
    if mode == "hour":
        data["upload_hour"] = data["publishedAt"].dt.hour
        hourly_counts = data["upload_hour"].value_counts().sort_index().reset_index()
        hourly_counts.columns = ["hour", "count"]
        return hourly_counts.to_dict(orient="records")
    
    elif mode == "day":
        data["upload_day"] = data["publishedAt"].dt.day_name()
        daily_counts = data["upload_day"].value_counts().reindex(
            ["Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday", "Sunday"]
        ).reset_index()
        daily_counts.columns = ["day", "count"]
        return daily_counts.to_dict(orient="records")

    return {"error": "Invalid mode"}


def category_like_view_ratio():
    data = data_processing.get_updated_df()
    data["like_view_ratio"] = data["likes"] / data["views"]
    category_data = data.groupby("category_name")["like_view_ratio"].mean().reset_index()

    return category_data.to_dict(orient="records")

def category_comment_engagement():
    data = data_processing.get_updated_df()
    return data[["category_name", "comment_count", "views", "likes"]].dropna().to_dict(orient="records")



    


if __name__ == "__main__":
    pass