Twint Twitter OSINT ツール チートシート

概要

Twintは、Twitter APIを使用せずにTwitterプロファイルからツイートをスクレイピングできる、Pythonで書かれた高度なTwitterスクレイピングツールです。Twitterの多くの制限を回避しながら、ツイート、フォロワー、フォロー、リツイートなどを取得できます。TwintはOSINT調査、ソーシャルメディアモニタリング、研究目的に特に有用です。

⚠️ 法的注意: Twintは正当な研究、OSINT調査、または承認されたセキュリティテストにのみ使用してください。Twitterの利用規約および適用される個人情報保護法を尊重してください。

インストール

Python pipインストール

# Install via pip
pip3 install twint

# Install development version
pip3 install --user --upgrade git+https://github.com/twintproject/twint.git@origin/master#egg=twint

# Install with additional dependencies
pip3 install twint[all]

# Verify installation
twint --help

Dockerインストール

# Pull Docker image
docker pull twintproject/twint

# Run with Docker
docker run -it --rm twintproject/twint

# Build from source
git clone https://github.com/twintproject/twint.git
cd twint
docker build -t twint .

# Run with volume mount
docker run -it --rm -v $(pwd)/output:/output twint

手動インストール

# Clone repository
git clone https://github.com/twintproject/twint.git
cd twint

# Install dependencies
pip3 install -r requirements.txt

# Install package
python3 setup.py install

# Alternative: Run directly
python3 -m twint --help

仮想環境のセットアップ

# Create virtual environment
python3 -m venv twint-env
source twint-env/bin/activate

# Install Twint
pip install twint

# Verify installation
twint --version

基本的な使用方法

コマンドラインインターフェース

# Basic tweet scraping
twint -u username

# Scrape tweets with specific search term
twint -s "search term"

# Scrape tweets from specific user
twint -u elonmusk

# Limit number of tweets
twint -u username --limit 100

# Save to file
twint -u username -o tweets.csv --csv

# Search with date range
twint -s "cybersecurity" --since "2023-01-01" --until "2023-12-31"

Python API の使用

import twint

# Configure Twint
c = twint.Config()
c.Username = "username"
c.Limit = 100
c.Store_csv = True
c.Output = "tweets.csv"

# Run search
twint.run.Search(c)

高度な検索オプション

ユーザーベースの検索

# Get user's tweets
twint -u username

# Get user's followers
twint -u username --followers

# Get user's following
twint -u username --following

# Get user's favorites/likes
twint -u username --favorites

# Get user information
twint -u username --user-full

# Get verified users only
twint -s "search term" --verified

コンテンツベースの検索

# Search by keyword
twint -s "cybersecurity"

# Search with hashtag
twint -s "#infosec"

# Search with multiple keywords
twint -s "cybersecurity OR infosec"

# Search for exact phrase
twint -s '"exact phrase"'

# Search excluding terms
twint -s "cybersecurity -spam"

# Search for tweets with links
twint -s "cybersecurity" --links

# Search for tweets with media
twint -s "cybersecurity" --media

地理的および言語フィルター

# Search by location
twint -s "cybersecurity" --near "New York"

# Search with specific language
twint -s "cybersecurity" --lang en

# Search with geolocation
twint -s "cybersecurity" --geo "40.7128,-74.0060,10km"

# Search popular tweets only
twint -s "cybersecurity" --popular

# Search for tweets with minimum likes
twint -s "cybersecurity" --min-likes 10

# Search for tweets with minimum retweets
twint -s "cybersecurity" --min-retweets 5

日付と時間のフィルター

# Search with date range
twint -s "cybersecurity" --since "2023-01-01" --until "2023-12-31"

# Search tweets from specific year
twint -s "cybersecurity" --year 2023

# Search tweets from specific hour
twint -s "cybersecurity" --hour 14

# Search tweets from today
twint -s "cybersecurity" --since $(date +%Y-%m-%d)

# Search tweets from last week
twint -s "cybersecurity" --since $(date -d '7 days ago' +%Y-%m-%d)

出力形式とストレージ

ファイル出力オプション

# Save as CSV
twint -u username -o output.csv --csv

# Save as JSON
twint -u username -o output.json --json

# Save as text file
twint -u username -o output.txt

# Custom CSV format
twint -u username --csv --output tweets.csv --custom-csv "date,time,username,tweet"

# Hide output (silent mode)
twint -u username --hide-output

# Debug mode
twint -u username --debug

データベースストレージ

# Store in Elasticsearch
twint -u username --elasticsearch localhost:9200

# Store in SQLite database
twint -u username --database tweets.db

# Store with custom database table
twint -u username --database tweets.db --table-tweets custom_tweets

高度な出力設定

import twint

# Configure advanced output
c = twint.Config()
c.Username = "username"
c.Store_csv = True
c.Output = "detailed_tweets.csv"
c.Custom_csv = ["date", "time", "username", "tweet", "replies_count", "retweets_count", "likes_count", "hashtags", "urls"]
c.Hide_output = True

# Run search
twint.run.Search(c)

Python API 高度な使用

基本設定

import twint
import pandas as pd

def scrape_user_tweets(username, limit=100):
    """Scrape tweets from specific user"""
    c = twint.Config()
    c.Username = username
    c.Limit = limit
    c.Store_pandas = True
    c.Hide_output = True

    twint.run.Search(c)

    # Get pandas dataframe
    tweets_df = twint.storage.panda.Tweets_df
    return tweets_df

# Usage
tweets = scrape_user_tweets("elonmusk", 50)
print(f"Scraped \\\\{len(tweets)\\\\} tweets")

高度な検索設定

import twint
from datetime import datetime, timedelta

def advanced_search(search_term, days_back=7, min_likes=5):
    """Advanced search with multiple filters"""
    c = twint.Config()

    # Search configuration
    c.Search = search_term
    c.Lang = "en"
    c.Min_likes = min_likes
    c.Popular_tweets = True

    # Date range (last N days)
    end_date = datetime.now()
    start_date = end_date - timedelta(days=days_back)
    c.Since = start_date.strftime("%Y-%m-%d")
    c.Until = end_date.strftime("%Y-%m-%d")

    # Output configuration
    c.Store_pandas = True
    c.Hide_output = True

    # Run search
    twint.run.Search(c)

    # Process results
    if twint.storage.panda.Tweets_df is not None:
        tweets_df = twint.storage.panda.Tweets_df
        return tweets_df
    else:
        return pd.DataFrame()

# Usage
cybersec_tweets = advanced_search("cybersecurity", days_back=30, min_likes=10)
print(f"Found \\\\{len(cybersec_tweets)\\\\} popular cybersecurity tweets")

ユーザー分析機能

import twint
import pandas as pd
from collections import Counter

class TwitterOSINT:
    def __init__(self):
        self.tweets_df = None
        self.users_df = None

    def analyze_user(self, username):
        """Comprehensive user analysis"""
        # Get user tweets
        c = twint.Config()
        c.Username = username
        c.Limit = 1000
        c.Store_pandas = True
        c.Hide_output = True

        twint.run.Search(c)
        self.tweets_df = twint.storage.panda.Tweets_df

        if self.tweets_df is not None and not self.tweets_df.empty:
            analysis = \\\\{
                'username': username,
                'total_tweets': len(self.tweets_df),
                'date_range': \\\\{
                    'earliest': self.tweets_df['date'].min(),
                    'latest': self.tweets_df['date'].max()
                \\\\},
                'engagement': \\\\{
                    'avg_likes': self.tweets_df['likes_count'].mean(),
                    'avg_retweets': self.tweets_df['retweets_count'].mean(),
                    'avg_replies': self.tweets_df['replies_count'].mean()
                \\\\},
                'top_hashtags': self.get_top_hashtags(),
                'top_mentions': self.get_top_mentions(),
                'posting_patterns': self.analyze_posting_patterns()
            \\\\}
            return analysis
        else:
            return None

    def get_top_hashtags(self, top_n=10):
        """Extract top hashtags from tweets"""
        if self.tweets_df is None:
            return []

        all_hashtags = []
        for hashtags in self.tweets_df['hashtags'].dropna():
            if hashtags:
                all_hashtags.extend(hashtags)

        return Counter(all_hashtags).most_common(top_n)

    def get_top_mentions(self, top_n=10):
        """Extract top mentions from tweets"""
        if self.tweets_df is None:
            return []

        all_mentions = []
        for mentions in self.tweets_df['mentions'].dropna():
            if mentions:
                all_mentions.extend(mentions)

        return Counter(all_mentions).most_common(top_n)

    def analyze_posting_patterns(self):
        """Analyze posting time patterns"""
        if self.tweets_df is None:
            return \\\\{\\\\}

        # Convert time to hour
        self.tweets_df['hour'] = pd.to_datetime(self.tweets_df['time']).dt.hour

        patterns = \\\\{
            'hourly_distribution': self.tweets_df['hour'].value_counts().to_dict(),
            'most_active_hour': self.tweets_df['hour'].mode().iloc[0] if not self.tweets_df['hour'].empty else None,
            'daily_tweet_count': self.tweets_df.groupby('date').size().mean()
        \\\\}

        return patterns

    def search_and_analyze(self, search_term, limit=500):
        """Search for tweets and analyze patterns"""
        c = twint.Config()
        c.Search = search_term
        c.Limit = limit
        c.Store_pandas = True
        c.Hide_output = True

        twint.run.Search(c)
        self.tweets_df = twint.storage.panda.Tweets_df

        if self.tweets_df is not None and not self.tweets_df.empty:
            analysis = \\\\{
                'search_term': search_term,
                'total_tweets': len(self.tweets_df),
                'unique_users': self.tweets_df['username'].nunique(),
                'top_users': self.tweets_df['username'].value_counts().head(10).to_dict(),
                'engagement_stats': \\\\{
                    'total_likes': self.tweets_df['likes_count'].sum(),
                    'total_retweets': self.tweets_df['retweets_count'].sum(),
                    'avg_engagement': (self.tweets_df['likes_count'] + self.tweets_df['retweets_count']).mean()
                \\\\},
                'top_hashtags': self.get_top_hashtags(),
                'sentiment_indicators': self.basic_sentiment_analysis()
            \\\\}
            return analysis
        else:
            return None

    def basic_sentiment_analysis(self):
        """Basic sentiment analysis using keyword matching"""
        if self.tweets_df is None:
            return \\\\{\\\\}

        positive_words = ['good', 'great', 'excellent', 'amazing', 'love', 'best', 'awesome']
        negative_words = ['bad', 'terrible', 'awful', 'hate', 'worst', 'horrible', 'disgusting']

        positive_count = 0
        negative_count = 0

        for tweet in self.tweets_df['tweet'].str.lower():
            if any(word in tweet for word in positive_words):
                positive_count += 1
            if any(word in tweet for word in negative_words):
                negative_count += 1

        total_tweets = len(self.tweets_df)
        return \\\\{
            'positive_tweets': positive_count,
            'negative_tweets': negative_count,
            'neutral_tweets': total_tweets - positive_count - negative_count,
            'positive_ratio': positive_count / total_tweets if total_tweets > 0 else 0,
            'negative_ratio': negative_count / total_tweets if total_tweets > 0 else 0
        \\\\}

# Usage example
osint = TwitterOSINT()

# Analyze specific user
user_analysis = osint.analyze_user("elonmusk")
if user_analysis:
    print(f"User Analysis for \\\\{user_analysis['username']\\\\}:")
    print(f"Total tweets: \\\\{user_analysis['total_tweets']\\\\}")
    print(f"Average likes: \\\\{user_analysis['engagement']['avg_likes']:.2f\\\\}")
    print(f"Top hashtags: \\\\{user_analysis['top_hashtags'][:5]\\\\}")

# Search and analyze topic
topic_analysis = osint.search_and_analyze("cybersecurity", limit=200)
if topic_analysis:
    print(f"\nTopic Analysis for '\\\\{topic_analysis['search_term']\\\\}':")
    print(f"Total tweets: \\\\{topic_analysis['total_tweets']\\\\}")
    print(f"Unique users: \\\\{topic_analysis['unique_users']\\\\}")
    print(f"Average engagement: \\\\{topic_analysis['engagement_stats']['avg_engagement']:.2f\\\\}")

OSINT調査ワークフロー

ターゲットユーザー調査

#!/usr/bin/env python3
# twitter-user-investigation.py

import twint
import pandas as pd
import json
from datetime import datetime, timedelta
import matplotlib.pyplot as plt
import seaborn as sns

class TwitterUserInvestigation:
    def __init__(self, username):
        self.username = username
        self.tweets_df = None
        self.followers_df = None
        self.following_df = None
        self.results = \\\\{\\\\}

    def collect_user_data(self):
        """Collect comprehensive user data"""
        print(f"Investigating Twitter user: \\\\{self.username\\\\}")

        # Collect tweets
        self.collect_tweets()

        # Collect followers (limited)
        self.collect_followers()

        # Collect following (limited)
        self.collect_following()

        # Analyze collected data
        self.analyze_data()

    def collect_tweets(self, limit=1000):
        """Collect user tweets"""
        print("Collecting tweets...")

        c = twint.Config()
        c.Username = self.username
        c.Limit = limit
        c.Store_pandas = True
        c.Hide_output = True

        try:
            twint.run.Search(c)
            self.tweets_df = twint.storage.panda.Tweets_df
            print(f"Collected \\\\{len(self.tweets_df)\\\\} tweets")
        except Exception as e:
            print(f"Error collecting tweets: \\\\{e\\\\}")

    def collect_followers(self, limit=100):
        """Collect user followers"""
        print("Collecting followers...")

        c = twint.Config()
        c.Username = self.username
        c.Limit = limit
        c.Store_pandas = True
        c.Hide_output = True

        try:
            twint.run.Followers(c)
            self.followers_df = twint.storage.panda.Follow_df
            print(f"Collected \\\\{len(self.followers_df)\\\\} followers")
        except Exception as e:
            print(f"Error collecting followers: \\\\{e\\\\}")

    def collect_following(self, limit=100):
        """Collect users being followed"""
        print("Collecting following...")

        c = twint.Config()
        c.Username = self.username
        c.Limit = limit
        c.Store_pandas = True
        c.Hide_output = True

        try:
            twint.run.Following(c)
            self.following_df = twint.storage.panda.Follow_df
            print(f"Collected \\\\{len(self.following_df)\\\\} following")
        except Exception as e:
            print(f"Error collecting following: \\\\{e\\\\}")

    def analyze_data(self):
        """Analyze collected data"""
        if self.tweets_df is not None and not self.tweets_df.empty:
            self.results = \\\\{
                'basic_stats': self.get_basic_stats(),
                'temporal_analysis': self.analyze_temporal_patterns(),
                'content_analysis': self.analyze_content(),
                'network_analysis': self.analyze_network(),
                'behavioral_patterns': self.analyze_behavior()
            \\\\}

    def get_basic_stats(self):
        """Get basic statistics"""
        return \\\\{
            'total_tweets': len(self.tweets_df),
            'date_range': \\\\{
                'first_tweet': self.tweets_df['date'].min(),
                'last_tweet': self.tweets_df['date'].max()
            \\\\},
            'engagement': \\\\{
                'total_likes': self.tweets_df['likes_count'].sum(),
                'total_retweets': self.tweets_df['retweets_count'].sum(),
                'total_replies': self.tweets_df['replies_count'].sum(),
                'avg_likes': self.tweets_df['likes_count'].mean(),
                'avg_retweets': self.tweets_df['retweets_count'].mean()
            \\\\}
        \\\\}

    def analyze_temporal_patterns(self):
        """Analyze posting time patterns"""
        # Convert datetime
        self.tweets_df['datetime'] = pd.to_datetime(self.tweets_df['date'] + ' ' + self.tweets_df['time'])
        self.tweets_df['hour'] = self.tweets_df['datetime'].dt.hour
        self.tweets_df['day_of_week'] = self.tweets_df['datetime'].dt.day_name()

        return \\\\{
            'hourly_pattern': self.tweets_df['hour'].value_counts().to_dict(),
            'daily_pattern': self.tweets_df['day_of_week'].value_counts().to_dict(),
            'most_active_hour': self.tweets_df['hour'].mode().iloc[0],
            'most_active_day': self.tweets_df['day_of_week'].mode().iloc[0],
            'posting_frequency': len(self.tweets_df) / max(1, (self.tweets_df['datetime'].max() - self.tweets_df['datetime'].min()).days)
        \\\\}

    def analyze_content(self):
        """Analyze tweet content"""
        # Extract hashtags and mentions
        all_hashtags = []
        all_mentions = []
        all_urls = []

        for _, row in self.tweets_df.iterrows():
            if row['hashtags']:
                all_hashtags.extend(row['hashtags'])
            if row['mentions']:
                all_mentions.extend(row['mentions'])
            if row['urls']:
                all_urls.extend(row['urls'])

        return \\\\{
            'top_hashtags': pd.Series(all_hashtags).value_counts().head(10).to_dict(),
            'top_mentions': pd.Series(all_mentions).value_counts().head(10).to_dict(),
            'url_domains': self.extract_domains(all_urls),
            'tweet_length_stats': \\\\{
                'avg_length': self.tweets_df['tweet'].str.len().mean(),
                'max_length': self.tweets_df['tweet'].str.len().max(),
                'min_length': self.tweets_df['tweet'].str.len().min()
            \\\\}
        \\\\}

    def extract_domains(self, urls):
        """Extract domains from URLs"""
        from urllib.parse import urlparse

        domains = []
        for url in urls:
            try:
                domain = urlparse(url).netloc
                if domain:
                    domains.append(domain)
            except:
                continue

        return pd.Series(domains).value_counts().head(10).to_dict()

    def analyze_network(self):
        """Analyze network connections"""
        network_data = \\\\{\\\\}

        if self.followers_df is not None:
            network_data['followers_count'] = len(self.followers_df)

        if self.following_df is not None:
            network_data['following_count'] = len(self.following_df)

        # Analyze interaction patterns
        if self.tweets_df is not None:
            reply_users = []
            for mentions in self.tweets_df['mentions'].dropna():
                if mentions:
                    reply_users.extend(mentions)

            network_data['frequent_interactions'] = pd.Series(reply_users).value_counts().head(10).to_dict()

        return network_data

    def analyze_behavior(self):
        """Analyze behavioral patterns"""
        if self.tweets_df is None:
            return \\\\{\\\\}

        # Retweet vs original content ratio
        retweet_count = self.tweets_df['tweet'].str.startswith('RT @').sum()
        original_count = len(self.tweets_df) - retweet_count

        # Reply patterns
        reply_count = self.tweets_df['tweet'].str.startswith('@').sum()

        return \\\\{
            'content_type_distribution': \\\\{
                'original_tweets': original_count,
                'retweets': retweet_count,
                'replies': reply_count
            \\\\},
            'retweet_ratio': retweet_count / len(self.tweets_df),
            'engagement_patterns': \\\\{
                'high_engagement_threshold': self.tweets_df['likes_count'].quantile(0.9),
                'viral_tweets': len(self.tweets_df[self.tweets_df['likes_count'] > self.tweets_df['likes_count'].quantile(0.95)])
            \\\\}
        \\\\}

    def generate_report(self):
        """Generate investigation report"""
        report = \\\\{
            'investigation_target': self.username,
            'investigation_date': datetime.now().isoformat(),
            'data_summary': \\\\{
                'tweets_collected': len(self.tweets_df) if self.tweets_df is not None else 0,
                'followers_collected': len(self.followers_df) if self.followers_df is not None else 0,
                'following_collected': len(self.following_df) if self.following_df is not None else 0
            \\\\},
            'analysis_results': self.results
        \\\\}

        # Save to JSON
        with open(f'twitter_investigation_\\\\{self.username\\\\}_\\\\{datetime.now().strftime("%Y%m%d")\\\\}.json', 'w') as f:
            json.dump(report, f, indent=2, default=str)

        # Generate HTML report
        self.generate_html_report(report)

        return report

    def generate_html_report(self, report):
        """Generate HTML investigation report"""
        html_content = f"""
<!DOCTYPE html>
<html>
<head>
    <title>Twitter Investigation Report - \\\\{self.username\\\\}</title>
    <style>
        body \\\\{\\\\{ font-family: Arial, sans-serif; margin: 20px; \\\\}\\\\}
        .section \\\\{\\\\{ margin: 20px 0; border: 1px solid #ccc; padding: 15px; \\\\}\\\\}
        .section h2 \\\\{\\\\{ color: #333; margin-top: 0; \\\\}\\\\}
        table \\\\{\\\\{ border-collapse: collapse; width: 100%; \\\\}\\\\}
        th, td \\\\{\\\\{ border: 1px solid #ddd; padding: 8px; text-align: left; \\\\}\\\\}
        th \\\\{\\\\{ background-color: #f2f2f2; \\\\}\\\\}
        .metric \\\\{\\\\{ display: inline-block; margin: 10px; padding: 10px; background: #f9f9f9; border-radius: 5px; \\\\}\\\\}
    </style>
</head>
<body>
    <h1>Twitter OSINT Investigation Report</h1>
    <div class="section">
        <h2>Investigation Summary</h2>
        <div class="metric"><strong>Target:</strong> @\\\\{self.username\\\\}</div>
        <div class="metric"><strong>Date:</strong> \\\\{report['investigation_date']\\\\}</div>
        <div class="metric"><strong>Tweets Analyzed:</strong> \\\\{report['data_summary']['tweets_collected']\\\\}</div>
    </div>
"""

        if 'basic_stats' in self.results:
            stats = self.results['basic_stats']
            html_content += f"""
    <div class="section">
        <h2>Basic Statistics</h2>
        <div class="metric"><strong>Total Tweets:</strong> \\\\{stats['total_tweets']\\\\}</div>
        <div class="metric"><strong>Total Likes:</strong> \\\\{stats['engagement']['total_likes']\\\\}</div>
        <div class="metric"><strong>Total Retweets:</strong> \\\\{stats['engagement']['total_retweets']\\\\}</div>
        <div class="metric"><strong>Average Likes:</strong> \\\\{stats['engagement']['avg_likes']:.2f\\\\}</div>
    </div>
"""

        if 'content_analysis' in self.results:
            content = self.results['content_analysis']
            html_content += """
    <div class="section">
        <h2>Content Analysis</h2>
        <h3>Top Hashtags</h3>
        <table>
            <tr><th>Hashtag</th><th>Count</th></tr>
"""
            for hashtag, count in list(content['top_hashtags'].items())[:10]:
                html_content += f"<tr><td>#\\\\{hashtag\\\\}</td><td>\\\\{count\\\\}</td></tr>"

            html_content += """
        </table>
        <h3>Top Mentions</h3>
        <table>
            <tr><th>User</th><th>Count</th></tr>
"""
            for user, count in list(content['top_mentions'].items())[:10]:
                html_content += f"<tr><td>@\\\\{user\\\\}</td><td>\\\\{count\\\\}</td></tr>"

            html_content += "</table></div>"

        html_content += """
</body>
</html>
"""

        with open(f'twitter_investigation_\\\\{self.username\\\\}_\\\\{datetime.now().strftime("%Y%m%d")\\\\}.html', 'w') as f:
            f.write(html_content)

def main():
    import sys

    if len(sys.argv) != 2:
        print("Usage: python3 twitter-user-investigation.py <username>")
        sys.exit(1)

    username = sys.argv[1].replace('@', '')  # Remove @ if present

    investigation = TwitterUserInvestigation(username)
    investigation.collect_user_data()
    report = investigation.generate_report()

    print(f"\nInvestigation completed for @\\\\{username\\\\}")
    print(f"Report saved as: twitter_investigation_\\\\{username\\\\}_\\\\{datetime.now().strftime('%Y%m%d')\\\\}.json")
    print(f"HTML report saved as: twitter_investigation_\\\\{username\\\\}_\\\\{datetime.now().strftime('%Y%m%d')\\\\}.html")

if __name__ == "__main__":
    main()

ハッシュタグとトレンド分析

#!/usr/bin/env python3
# twitter-hashtag-analysis.py

import twint
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
from datetime import datetime, timedelta
from collections import Counter
import networkx as nx

class HashtagAnalysis:
    def __init__(self):
        self.tweets_df = None
        self.hashtag_network = None

    def analyze_hashtag(self, hashtag, days_back=7, limit=1000):
        """Analyze specific hashtag usage"""
        print(f"Analyzing hashtag: #\\\\{hashtag\\\\}")

        # Configure search
        c = twint.Config()
        c.Search = f"#\\\\{hashtag\\\\}"
        c.Limit = limit
        c.Store_pandas = True
        c.Hide_output = True

        # Set date range
        end_date = datetime.now()
        start_date = end_date - timedelta(days=days_back)
        c.Since = start_date.strftime("%Y-%m-%d")
        c.Until = end_date.strftime("%Y-%m-%d")

        # Run search
        twint.run.Search(c)
        self.tweets_df = twint.storage.panda.Tweets_df

        if self.tweets_df is not None and not self.tweets_df.empty:
            analysis = \\\\{
                'hashtag': hashtag,
                'total_tweets': len(self.tweets_df),
                'unique_users': self.tweets_df['username'].nunique(),
                'date_range': f"\\\\{start_date.strftime('%Y-%m-%d')\\\\} to \\\\{end_date.strftime('%Y-%m-%d')\\\\}",
                'engagement_stats': self.calculate_engagement_stats(),
                'top_users': self.get_top_users(),
                'related_hashtags': self.get_related_hashtags(),
                'temporal_patterns': self.analyze_temporal_patterns(),
                'influence_metrics': self.calculate_influence_metrics()
            \\\\}

            return analysis
        else:
            print(f"No tweets found for #\\\\{hashtag\\\\}")
            return None

    def calculate_engagement_stats(self):
        """Calculate engagement statistics"""
        return \\\\{
            'total_likes': self.tweets_df['likes_count'].sum(),
            'total_retweets': self.tweets_df['retweets_count'].sum(),
            'total_replies': self.tweets_df['replies_count'].sum(),
            'avg_likes': self.tweets_df['likes_count'].mean(),
            'avg_retweets': self.tweets_df['retweets_count'].mean(),
            'avg_replies': self.tweets_df['replies_count'].mean(),
            'engagement_rate': (self.tweets_df['likes_count'] + self.tweets_df['retweets_count'] + self.tweets_df['replies_count']).mean()
        \\\\}

    def get_top_users(self, top_n=10):
        """Get top users by tweet count and engagement"""
        user_stats = self.tweets_df.groupby('username').agg(\\\\{
            'tweet': 'count',
            'likes_count': 'sum',
            'retweets_count': 'sum',
            'replies_count': 'sum'
        \\\\}).reset_index()

        user_stats['total_engagement'] = user_stats['likes_count'] + user_stats['retweets_count'] + user_stats['replies_count']

        return \\\\{
            'by_tweet_count': user_stats.nlargest(top_n, 'tweet')[['username', 'tweet']].to_dict('records'),
            'by_engagement': user_stats.nlargest(top_n, 'total_engagement')[['username', 'total_engagement']].to_dict('records')
        \\\\}

    def get_related_hashtags(self, top_n=20):
        """Get hashtags that appear with the target hashtag"""
        all_hashtags = []

        for hashtags in self.tweets_df['hashtags'].dropna():
            if hashtags:
                all_hashtags.extend(hashtags)

        hashtag_counts = Counter(all_hashtags)
        return hashtag_counts.most_common(top_n)

    def analyze_temporal_patterns(self):
        """Analyze temporal posting patterns"""
        self.tweets_df['datetime'] = pd.to_datetime(self.tweets_df['date'] + ' ' + self.tweets_df['time'])
        self.tweets_df['hour'] = self.tweets_df['datetime'].dt.hour
        self.tweets_df['day'] = self.tweets_df['datetime'].dt.date

        return \\\\{
            'hourly_distribution': self.tweets_df['hour'].value_counts().sort_index().to_dict(),
            'daily_volume': self.tweets_df['day'].value_counts().sort_index().to_dict(),
            'peak_hour': self.tweets_df['hour'].mode().iloc[0],
            'peak_day': self.tweets_df['day'].value_counts().index[0].strftime('%Y-%m-%d')
        \\\\}

    def calculate_influence_metrics(self):
        """Calculate influence and reach metrics"""
        # Identify influential tweets (top 10% by engagement)
        engagement_threshold = self.tweets_df['likes_count'].quantile(0.9)
        influential_tweets = self.tweets_df[self.tweets_df['likes_count'] >= engagement_threshold]

        return \\\\{
            'influential_tweets_count': len(influential_tweets),
            'influential_users': influential_tweets['username'].unique().tolist(),
            'viral_threshold': engagement_threshold,
            'reach_estimate': self.tweets_df['retweets_count'].sum() * 100  # Rough estimate
        \\\\}

    def create_hashtag_network(self, min_cooccurrence=2):
        """Create network of co-occurring hashtags"""
        hashtag_pairs = []

        for hashtags in self.tweets_df['hashtags'].dropna():
            if hashtags and len(hashtags) > 1:
                # Create pairs of hashtags that appear together
                for i in range(len(hashtags)):
                    for j in range(i + 1, len(hashtags)):
                        pair = tuple(sorted([hashtags[i], hashtags[j]]))
                        hashtag_pairs.append(pair)

        # Count co-occurrences
        pair_counts = Counter(hashtag_pairs)

        # Create network graph
        G = nx.Graph()

        for (hashtag1, hashtag2), count in pair_counts.items():
            if count >= min_cooccurrence:
                G.add_edge(hashtag1, hashtag2, weight=count)

        self.hashtag_network = G
        return G

    def visualize_hashtag_network(self, output_file="hashtag_network.png"):
        """Visualize hashtag co-occurrence network"""
        if self.hashtag_network is None:
            self.create_hashtag_network()

        plt.figure(figsize=(12, 8))

        # Calculate node sizes based on degree
        node_sizes = [self.hashtag_network.degree(node) * 100 for node in self.hashtag_network.nodes()]

        # Draw network
        pos = nx.spring_layout(self.hashtag_network, k=1, iterations=50)
        nx.draw(self.hashtag_network, pos,
                node_size=node_sizes,
                node_color='lightblue',
                font_size=8,
                font_weight='bold',
                with_labels=True,
                edge_color='gray',
                alpha=0.7)

        plt.title("Hashtag Co-occurrence Network")
        plt.axis('off')
        plt.tight_layout()
        plt.savefig(output_file, dpi=300, bbox_inches='tight')
        plt.close()

        print(f"Network visualization saved as: \\\\{output_file\\\\}")

def main():
    import sys

    if len(sys.argv) < 2:
        print("Usage: python3 twitter-hashtag-analysis.py <hashtag> [days_back] [limit]")
        sys.exit(1)

    hashtag = sys.argv[1].replace('#', '')  # Remove # if present
    days_back = int(sys.argv[2]) if len(sys.argv) > 2 else 7
    limit = int(sys.argv[3]) if len(sys.argv) > 3 else 1000

    analyzer = HashtagAnalysis()
    analysis = analyzer.analyze_hashtag(hashtag, days_back, limit)

    if analysis:
        print(f"\nHashtag Analysis Results for #\\\\{hashtag\\\\}")
        print("=" * 50)
        print(f"Total tweets: \\\\{analysis['total_tweets']\\\\}")
        print(f"Unique users: \\\\{analysis['unique_users']\\\\}")
        print(f"Average engagement: \\\\{analysis['engagement_stats']['engagement_rate']:.2f\\\\}")
        print(f"Peak hour: \\\\{analysis['temporal_patterns']['peak_hour']\\\\}:00")

        # Create network visualization
        analyzer.visualize_hashtag_network(f"hashtag_network_\\\\{hashtag\\\\}.png")

        # Save detailed results
        import json
        with open(f"hashtag_analysis_\\\\{hashtag\\\\}_\\\\{datetime.now().strftime('%Y%m%d')\\\\}.json", 'w') as f:
            json.dump(analysis, f, indent=2, default=str)

        print(f"\nDetailed analysis saved as: hashtag_analysis_\\\\{hashtag\\\\}_\\\\{datetime.now().strftime('%Y%m%d')\\\\}.json")

if __name__ == "__main__":
    main()

ベストプラクティスとOPSEC

オペレーショナルセキュリティ

Would you like me to fill in the remaining placeholders with translations or keep them as is?```bash #!/bin/bash

twint-opsec-setup.sh

echo “Twint OPSEC Configuration” echo ”========================“

Use VPN or proxy

echo “1. Network Security:” echo ” □ Configure VPN connection” echo ” □ Use SOCKS proxy if needed” echo ” □ Rotate IP addresses periodically”

Rate limiting

echo -e “\n2. Rate Limiting:” echo ” □ Add delays between requests” echo ” □ Limit concurrent searches” echo ” □ Monitor for rate limiting”

Data security

echo -e “\n3. Data Security:” echo ” □ Encrypt stored data” echo ” □ Use secure file permissions” echo ” □ Regular data cleanup”

Legal compliance

echo -e “\n4. Legal Compliance:” echo ” □ Verify investigation scope” echo ” □ Document methodology” echo ” □ Respect privacy laws”

```python
import twint
import time
import random

def safe_twint_search(config, delay_range=(1, 3)):
    """Run Twint search with random delays"""
    try:
        # Add random delay
        delay = random.uniform(delay_range[0], delay_range[1])
        time.sleep(delay)

        # Run search
        twint.run.Search(config)
        return True

    except Exception as e:
        print(f"Search failed: \\\\{e\\\\}")
        # Longer delay on failure
        time.sleep(random.uniform(5, 10))
        return False

def batch_user_analysis(usernames, delay_range=(2, 5)):
    """Analyze multiple users with delays"""
    results = \\\\{\\\\}

    for username in usernames:
        print(f"Analyzing @\\\\{username\\\\}")

        c = twint.Config()
        c.Username = username
        c.Limit = 100
        c.Store_pandas = True
        c.Hide_output = True

        if safe_twint_search(c, delay_range):
            if twint.storage.panda.Tweets_df is not None:
                results[username] = len(twint.storage.panda.Tweets_df)
            else:
                results[username] = 0
        else:
            results[username] = "Failed"

        # Clear storage for next user
        twint.storage.panda.Tweets_df = None

    return results
```## トラブルシューティング

### 一般的な問題と解決策
```bash
# Issue: No tweets returned
# Solution: Check if user exists and has public tweets
twint -u username --debug

# Issue: Rate limiting
# Solution: Add delays and reduce request frequency
twint -u username --limit 50

# Issue: SSL/TLS errors
# Solution: Update certificates or disable SSL verification
pip install --upgrade certifi

# Issue: Pandas storage not working
# Solution: Clear storage and reinitialize
python3 -c "import twint; twint.storage.panda.Tweets_df = None"
```### デバッグとログ記録
```python
import twint
import logging

# Enable debug logging
logging.basicConfig(level=logging.DEBUG)

# Configure with debug mode
c = twint.Config()
c.Username = "username"
c.Debug = True
c.Verbose = True

# Run with error handling
try:
    twint.run.Search(c)
except Exception as e:
    print(f"Error: \\\\{e\\\\}")
    import traceback
    traceback.print_exc()
```## リソース

- [Twint GitHub リポジトリ](https://github.com/twintproject/twint)
- [Twint ドキュメンテーション](https://github.com/twintproject/twint/wiki)
- [Twitter OSINT テクニック](https://osintframework.com/)
- [ソーシャルメディアインテリジェンスガイド](https://www.bellingcat.com/resources/how-tos/2019/06/21/using-twitter-for-osint-investigations/)
- [Pandasを使用したPythonデータ分析](https://pandas.pydata.org/docs/)

---

*このチートシートは、Twitter OSINTの調査にTwintを使用するための包括的なガイダンスを提供します。ソーシャルメディアインテリジェンス収集活動を行う前に、常に適切な承認と法的コンプライアンスを確保してください。*