import pandas as pd
import numpy as np
from textblob import TextBlob
import warnings
warnings.filterwarnings('ignore')

# Load a sample dataset of social media comments
np.random.seed(42)
comments_data = [
    {'post_id': 1, 'platform': 'YouTube',   'comment': 'Great video, learned a lot!'},
    {'post_id': 1, 'platform': 'YouTube',   'comment': 'Very clear explanation, thank you.'},
    {'post_id': 2, 'platform': 'Instagram', 'comment': 'Amazing content as always.'},
    {'post_id': 2, 'platform': 'Instagram', 'comment': 'Not what I expected, a bit confusing.'},
    {'post_id': 3, 'platform': 'TikTok',    'comment': 'This is so helpful!'},
    {'post_id': 3, 'platform': 'TikTok',    'comment': 'Could you make a follow-up video?'},
    {'post_id': 4, 'platform': 'YouTube',   'comment': 'Terrible content, waste of time.'},
    {'post_id': 4, 'platform': 'YouTube',   'comment': 'I disagree with this approach.'},
    {'post_id': 5, 'platform': 'Instagram', 'comment': 'Loved it, sharing with my friends.'},
    {'post_id': 5, 'platform': 'TikTok',    'comment': 'The best tutorial I have seen.'},
]
df = pd.DataFrame(comments_data)
df['timestamp'] = pd.date_range('2023-01-01', periods=len(df), freq='3h')
print(df.shape)
print(df.head(3))

for i, row in df.iterrows():
    print(f"Comment: {row['comment']}")

def get_sentiment(text):
    analysis = TextBlob(text)
    # Polarity ranges from -1 (negative) to 1 (positive)
    return analysis.polarity

df['sentiment'] = df['comment'].apply(get_sentiment)
print(df[['comment', 'sentiment']])

# Classify comments as Positive, Neutral, or Negative based on sentiment value
def classify_sentiment(s):
    if s > 0.2:
        return 'Positive'
    elif s < -0.2:
        return 'Negative'
    else:
        return 'Neutral'

df['sentiment_label'] = df['sentiment'].apply(classify_sentiment)
print(df[['comment', 'sentiment', 'sentiment_label']])

# Count the number of comments in each sentiment category
counts = df['sentiment_label'].value_counts()
print(counts)

# Plotting sentiment distribution
import matplotlib.pyplot as plt
plt.style.use('seaborn-v0_8-colorblind')
counts.plot(kind='bar', color=['green','grey','red'])
plt.title('Distribution of Comment Sentiment')
plt.xlabel('Sentiment')
plt.ylabel('Number of Comments')
plt.tight_layout()
plt.savefig('sentiment_distribution.png')
plt.show()

# Beginner Example: Find all negative comments
negatives = df[df['sentiment_label'] == 'Negative']
print(negatives[['platform', 'comment', 'sentiment']])

# Beginner Example: Show only positive comments
positives = df[df['sentiment_label'] == 'Positive']
print(positives[['platform', 'comment', 'sentiment']])

# Beginner Example: Calculate average sentiment per platform
avg_sent_by_platform = df.groupby('platform')['sentiment'].mean()
print(avg_sent_by_platform)

# Intermediate Example: Track sentiment over time
df['date'] = df['timestamp'].dt.date
daily_sentiment = df.groupby('date')['sentiment'].mean()
print(daily_sentiment)

# Intermediate Example: Identify influencers of negative sentiment
neg_comment_counts = df[df['sentiment_label']=='Negative'].groupby('platform')['comment'].count()
print(neg_comment_counts)

# Intermediate Example: Analyze sentiment by post_id
mean_sent_post = df.groupby('post_id')['sentiment'].mean()
print(mean_sent_post)

# Advanced Example: Word clouds for sentiment analysis
from wordcloud import WordCloud
neg_words = ' '.join(df[df['sentiment_label']=='Negative']['comment'])
pos_words = ' '.join(df[df['sentiment_label']=='Positive']['comment'])
wordcloud_neg = WordCloud(width=400, height=200, background_color='white').generate(neg_words)
wordcloud_pos = WordCloud(width=400, height=200, background_color='white', colormap='Greens').generate(pos_words)
plt.figure(figsize=(10,4))
plt.subplot(1,2,1); plt.imshow(wordcloud_neg, interpolation='bilinear'); plt.axis('off'); plt.title('Negative Comments')
plt.subplot(1,2,2); plt.imshow(wordcloud_pos, interpolation='bilinear'); plt.axis('off'); plt.title('Positive Comments')
plt.tight_layout()
plt.savefig('sentiment_wordclouds.png')
plt.show()

# Advanced Example: Platform vs sentiment crosstab analysis
cross = pd.crosstab(df['platform'], df['sentiment_label'])
print(cross)

# Advanced Example: Sentiment over time per platform
sent_time_platform = df.groupby(['platform', 'date'])['sentiment'].mean().unstack('platform')
sent_time_platform.plot(marker='o')
plt.title('Average Sentiment Over Time by Platform')
plt.ylabel('Avg Sentiment')
plt.tight_layout()
plt.savefig('sent_trend_platform.png')
plt.show()

# Error Example: Handle missing comments
df_missing = df.copy()
df_missing.loc[3, 'comment'] = None
df_missing['sentiment'] = df_missing['comment'].apply(
    lambda x: get_sentiment(x) if pd.notnull(x) else np.nan)
print(df_missing[['comment','sentiment']])

# Error Example: Incorrect aggregation mistake
try:
    # This is incorrect: taking the mean across text fields
    test = df[['comment', 'sentiment']].mean()
except Exception as e:
    print('Error:', e)

# Error Example: Confusing sentiment with comment count
grouped = df.groupby('platform').agg({'sentiment':'mean', 'comment':'count'})
print(grouped)

# End-to-end Example: Recommend content action
max_post = mean_sent_post.idxmax()
min_post = mean_sent_post.idxmin()
print(f"Post with highest avg sentiment: {max_post} ({mean_sent_post[max_post]:.2f})")
print(f"Post with lowest avg sentiment: {min_post} ({mean_sent_post[min_post]:.2f})")
if mean_sent_post[min_post] < 0:
    print(f"Review post {min_post}. Consider responding or adjusting future content.")
else:
    print(f"All posts have positive or neutral feedback. Keep up the good work!")