Upload data1/reporting/visualization.py with huggingface_hub

data1/reporting/visualization.py

@@ -0,0 +1,595 @@
+"""
+Visualization module: Generate publication-ready figures (PNG/SVG)
+ACL conference style, 1920x1080 or A4 landscape size
+"""
+import matplotlib
+matplotlib.use('Agg')  # 非交互式后端
+import matplotlib.pyplot as plt
+import matplotlib.font_manager as fm
+import seaborn as sns
+import pandas as pd
+import numpy as np
+from pathlib import Path
+import json
+import statistics
+from collections import Counter
+
+# Font fallback mechanism
+# Try Arial, fallback to DejaVu Sans (common on Linux) or sans-serif
+font_families_to_try = ['Arial', 'DejaVu Sans', 'Liberation Sans', 'sans-serif']
+available_fonts = [f.name for f in fm.fontManager.ttflist]
+font_found = None
+
+for font_family in font_families_to_try:
+    # Check if font exists (case-insensitive)
+    font_lower = font_family.lower()
+    if any(f.lower() == font_lower for f in available_fonts):
+        font_found = font_family
+        break
+
+if font_found is None:
+    # If no font found, use default sans-serif
+    font_found = 'sans-serif'
+
+# Nature journal style: professional, high-contrast, color-rich
+# Large fonts for axis labels and tick values, smaller for titles/legends to avoid overlap
+# Increased font sizes for PPT presentation
+plt.rcParams['font.family'] = font_found
+plt.rcParams['font.size'] = 24
+plt.rcParams['axes.labelsize'] = 42  # Large axis labels (increased from 32)
+plt.rcParams['axes.titlesize'] = 30  # Titles (increased from 20)
+plt.rcParams['xtick.labelsize'] = 36  # Large tick values (increased from 28)
+plt.rcParams['ytick.labelsize'] = 36  # Large tick values (increased from 28)
+plt.rcParams['legend.fontsize'] = 20  # Legend (increased from 16)
+plt.rcParams['figure.titlesize'] = 32  # Figure titles (increased from 24)
+plt.rcParams['axes.linewidth'] = 1.5
+plt.rcParams['axes.spines.top'] = False
+plt.rcParams['axes.spines.right'] = False
+plt.rcParams['axes.grid'] = True
+plt.rcParams['grid.alpha'] = 0.3
+plt.rcParams['grid.linewidth'] = 0.5
+plt.rcParams['axes.unicode_minus'] = False
+
+# Nature color scheme (high contrast, professional)
+NATURE_COLORS = {
+    'primary': '#2E5090',      # Nature blue
+    'secondary': '#1A5490',
+    'accent': '#4A90E2',
+    'success': '#2E7D32',
+    'warning': '#F57C00',
+    'error': '#C62828',
+    'neutral': '#424242',
+    'light': '#E3F2FD'
+}
+
+# Nature style palette
+nature_palette = ['#2E5090', '#4A90E2', '#1A5490', '#6BA3D8', '#94C4E8']
+
+
+class Visualization:
+    def __init__(self, output_dir, figsize=(19.2, 10.8), dpi=150):
+        self.output_dir = Path(output_dir)
+        self.output_dir.mkdir(parents=True, exist_ok=True)
+        self.figsize = figsize
+        self.dpi = dpi  # 提高DPI以获得更清晰的图片
+        self.fig_counter = 1
+        
+    def apply_nature_style(self, ax):
+        """Apply Nature journal style to axes"""
+        ax.spines['top'].set_visible(False)
+        ax.spines['right'].set_visible(False)
+        ax.spines['left'].set_linewidth(1.5)
+        ax.spines['bottom'].set_linewidth(1.5)
+        ax.grid(True, alpha=0.3, linestyle='--', linewidth=0.5)
+        ax.tick_params(width=1.5, length=5)
+    
+    def save_fig(self, fig, name):
+        """保存图片"""
+        fig_path_png = self.output_dir / f"fig_{self.fig_counter:02d}_{name}.png"
+        fig_path_svg = self.output_dir / f"fig_{self.fig_counter:02d}_{name}.svg"
+        fig.savefig(fig_path_png, dpi=self.dpi, bbox_inches='tight', facecolor='white')
+        fig.savefig(fig_path_svg, bbox_inches='tight', facecolor='white')
+        plt.close(fig)
+        self.fig_counter += 1
+        print(f"Saved: {fig_path_png}")
+    
+    def plot_funnel(self, stage_a_dir, stage_b_dir, repo_meta_dir, top_n=None):
+        """绘制漏斗图：搜索->过滤->深度分析"""
+        # 读取数据
+        stage_a_path = Path(stage_a_dir) / 'summary_overall.json'
+        stage_b_path = Path(stage_b_dir) / 'filter_summary.json'
+        repo_meta_path = Path(repo_meta_dir) / 'repo_meta_summary.json'
+        
+        with open(stage_a_path, 'r') as f:
+            stage_a = json.load(f)
+        
+        with open(stage_b_path, 'r') as f:
+            stage_b = json.load(f)
+        
+        # 尝试读取repo_meta数据获取实际数量
+        deep_analysis_count = top_n if top_n else 0
+        try:
+            with open(repo_meta_path, 'r') as f:
+                repo_meta = json.load(f)
+                deep_analysis_count = repo_meta.get('total_repos', deep_analysis_count)
+        except:
+            pass
+        
+        # 动态生成标签
+        if top_n:
+            deep_analysis_label = f'Deep Analysis\n({top_n:,} repos)'
+        else:
+            deep_analysis_label = f'Deep Analysis\n({deep_analysis_count:,} repos)'
+        
+        stages = ['Search Stage\n(~1.3M repos)', 'Filtered\n(~30K repos)', deep_analysis_label]
+        values = [
+            stage_a['total_records'],
+            stage_b['total'],
+            deep_analysis_count
+        ]
+        
+        fig, ax = plt.subplots(figsize=self.figsize)
+        self.apply_nature_style(ax)
+        
+        # Draw funnel - Use Nature blue gradient
+        y_pos = np.arange(len(stages))
+        colors = [NATURE_COLORS['primary'], NATURE_COLORS['accent'], NATURE_COLORS['secondary']]
+        
+        bars = ax.barh(y_pos, values, color=colors, alpha=0.85, edgecolor='white', linewidth=2)
+        ax.set_yticks(y_pos)
+        ax.set_yticklabels(stages, fontsize=36)
+        ax.set_xlabel('Number of Repositories', fontsize=42, fontweight='bold')
+        ax.set_title('Data Pipeline: Search → Filter → Deep Analysis', fontsize=30, fontweight='bold', pad=20)
+        
+        # Add value labels
+        for i, (bar, val) in enumerate(zip(bars, values)):
+            ax.text(val * 0.5, i, f'{val:,}', ha='center', va='center', 
+                   fontsize=24, fontweight='bold', color='white')
+        
+        ax.invert_yaxis()
+        plt.tight_layout()
+        
+        self.save_fig(fig, 'funnel')
+    
+    def plot_top_keywords(self, csv_path, top_n=20):
+        """绘制Top keywords条形图"""
+        df = pd.read_csv(csv_path)
+        df_top = df.head(top_n)
+        
+        fig, ax = plt.subplots(figsize=self.figsize)
+        self.apply_nature_style(ax)
+        
+        # Use Nature blue
+        bars = ax.barh(range(len(df_top)), df_top['count'], 
+                      color=NATURE_COLORS['primary'], alpha=0.85, 
+                      edgecolor='white', linewidth=1.5)
+        ax.set_yticks(range(len(df_top)))
+        ax.set_yticklabels(df_top['keyword'], fontsize=36, rotation=15, ha='right')
+        ax.set_xlabel('Number of Repositories', fontsize=42, fontweight='bold')
+        ax.set_title(f'Top {top_n} Keywords (Search Stage)', fontsize=30, fontweight='bold', pad=20)
+        
+        # Add value labels
+        for i, (idx, row) in enumerate(df_top.iterrows()):
+            ax.text(row['count'] * 0.5, i, f"{int(row['count']):,}", 
+                   ha='center', va='center', fontsize=24, fontweight='bold', color='white')
+        
+        ax.invert_yaxis()
+        plt.tight_layout()
+        
+        self.save_fig(fig, 'top_keywords')
+    
+    def plot_language_distribution(self, csv_path, top_n=15):
+        """绘制语言分布"""
+        df = pd.read_csv(csv_path)
+        df = df[df['language'] != '<empty>'].head(top_n)
+        
+        fig, ax = plt.subplots(figsize=self.figsize)
+        self.apply_nature_style(ax)
+        
+        bars = ax.bar(range(len(df)), df['count'], 
+                     color=NATURE_COLORS['primary'], alpha=0.85,
+                     edgecolor='white', linewidth=1.5)
+        ax.set_xticks(range(len(df)))
+        ax.set_xticklabels(df['language'], rotation=45, ha='right', fontsize=36)
+        ax.set_ylabel('Number of Repositories', fontsize=42, fontweight='bold')
+        ax.set_title(f'Top {top_n} Programming Languages (Search Stage)', 
+                    fontsize=30, fontweight='bold', pad=20)
+        
+        # 添加数值标签（旋转45度避免重叠）
+        for i, count in enumerate(df['count']):
+            ax.text(i, count, f"{int(count):,}", ha='center', va='bottom', 
+                   fontsize=24, fontweight='bold', rotation=45)
+        
+        plt.tight_layout()
+        
+        self.save_fig(fig, 'language_distribution')
+    
+    def plot_stars_distribution(self, csv_path):
+        """绘制stars分布（对数坐标）"""
+        df = pd.read_csv(csv_path)
+        
+        fig, ax = plt.subplots(figsize=self.figsize)
+        self.apply_nature_style(ax)
+        
+        # Use log bins
+        log_bins = np.logspace(0, np.log10(df['stars'].max() + 1), 50)
+        ax.hist(df['stars'], bins=log_bins, 
+               color=NATURE_COLORS['primary'], alpha=0.75, 
+               edgecolor='white', linewidth=1)
+        ax.set_xscale('log')
+        ax.set_xlabel('Stars (log scale)', fontsize=42, fontweight='bold')
+        ax.set_ylabel('Number of Repositories', fontsize=42, fontweight='bold')
+        ax.set_title('Distribution of Repository Stars (Log Scale)', 
+                    fontsize=30, fontweight='bold', pad=20)
+        
+        plt.tight_layout()
+        
+        self.save_fig(fig, 'stars_distribution')
+    
+    def plot_filter_results(self, csv_path):
+        """绘制过滤结果（按keyword的YES/NO）"""
+        df = pd.read_csv(csv_path)
+        df = df.head(15)  # Top 15 keywords
+        
+        fig, ax = plt.subplots(figsize=self.figsize)
+        self.apply_nature_style(ax)
+        
+        x = np.arange(len(df))
+        width = 0.4
+        
+        bars1 = ax.bar(x - width/2, df['yes'], width, label='Relevant (YES)', 
+                      color=NATURE_COLORS['success'], alpha=0.85, edgecolor='white', linewidth=1.5)
+        bars2 = ax.bar(x + width/2, df['no'], width, label='Irrelevant (NO)', 
+                      color=NATURE_COLORS['error'], alpha=0.85, edgecolor='white', linewidth=1.5)
+        
+        ax.set_xlabel('Keyword', fontsize=42, fontweight='bold')
+        ax.set_ylabel('Number of Repositories', fontsize=42, fontweight='bold')
+        ax.set_title('Filter Results: YES/NO Distribution by Keyword', 
+                    fontsize=30, fontweight='bold', pad=20)
+        ax.set_xticks(x)
+        ax.set_xticklabels(df['keyword'], rotation=45, ha='right', fontsize=36)
+        ax.legend(fontsize=20, frameon=True, fancybox=True, shadow=True)
+        
+        plt.tight_layout()
+        
+        self.save_fig(fig, 'filter_results_by_keyword')
+    
+    def plot_reason_length_comparison(self, csv_path):
+        """绘制reason长度对比（YES vs NO）"""
+        df = pd.read_csv(csv_path)
+        
+        fig, ax = plt.subplots(figsize=self.figsize)
+        self.apply_nature_style(ax)
+        
+        yes_lengths = df[df['label'] == 'YES']['length']
+        no_lengths = df[df['label'] == 'NO']['length']
+        
+        bp = ax.boxplot([yes_lengths, no_lengths], labels=['YES', 'NO'], 
+                       patch_artist=True,
+                       widths=0.6,
+                       boxprops=dict(facecolor=NATURE_COLORS['primary'], alpha=0.7, linewidth=2),
+                       medianprops=dict(color='white', linewidth=3),
+                       whiskerprops=dict(linewidth=2),
+                       capprops=dict(linewidth=2),
+                       flierprops=dict(marker='o', markersize=8, alpha=0.5))
+        
+        ax.set_ylabel('Reason Length (Characters)', fontsize=42, fontweight='bold')
+        ax.set_title('Comparison of Filter Reason Length: YES vs NO', 
+                    fontsize=30, fontweight='bold', pad=20)
+        ax.set_xticklabels(['Relevant (YES)', 'Irrelevant (NO)'], fontsize=36)
+        
+        plt.tight_layout()
+        
+        self.save_fig(fig, 'reason_length_comparison')
+    
+    def plot_extension_distribution(self, csv_path, top_n=20, repo_top_n=None):
+        """绘制文件扩展名分布"""
+        df = pd.read_csv(csv_path)
+        df = df.head(top_n)
+        
+        fig, ax = plt.subplots(figsize=self.figsize)
+        self.apply_nature_style(ax)
+        
+        bars = ax.barh(range(len(df)), df['count'], 
+                      color=NATURE_COLORS['accent'], alpha=0.85,
+                      edgecolor='white', linewidth=1.5)
+        ax.set_yticks(range(len(df)))
+        ax.set_yticklabels(df['extension'], fontsize=36, rotation=15, ha='right')
+        ax.set_xlabel('Number of Files', fontsize=42, fontweight='bold')
+        
+        # 动态生成标题
+        if repo_top_n:
+            repo_label = f'Top {repo_top_n:,} Repositories'
+        else:
+            repo_label = 'All Repositories'
+        ax.set_title(f'Top {top_n} File Extension Distribution ({repo_label})', 
+                    fontsize=30, fontweight='bold', pad=20)
+        
+        ax.invert_yaxis()
+        plt.tight_layout()
+        
+        self.save_fig(fig, 'extension_distribution')
+    
+    def plot_repo_file_count_distribution(self, csv_path, repo_top_n=None):
+        """绘制仓库文件数分布"""
+        df = pd.read_csv(csv_path)
+        
+        fig, ax = plt.subplots(figsize=self.figsize)
+        self.apply_nature_style(ax)
+        
+        # Use log bins (file counts may span large ranges)
+        log_bins = np.logspace(0, np.log10(df['total_files'].max() + 1), 50)
+        ax.hist(df['total_files'], bins=log_bins, 
+               color=NATURE_COLORS['primary'], alpha=0.75, 
+               edgecolor='white', linewidth=1)
+        ax.set_xscale('log')
+        ax.set_xlabel('Number of Files (log scale)', fontsize=42, fontweight='bold')
+        ax.set_ylabel('Number of Repositories', fontsize=42, fontweight='bold')
+        
+        # 动态生成标题
+        if repo_top_n:
+            repo_label = f'Top {repo_top_n:,} Repositories'
+        else:
+            repo_label = 'All Repositories'
+        ax.set_title(f'Distribution of Repository File Counts ({repo_label})', 
+                    fontsize=30, fontweight='bold', pad=20)
+        
+        plt.tight_layout()
+        
+        self.save_fig(fig, 'repo_file_count_distribution')
+    
+    def plot_stars_vs_code_size(self, repos_searched_csv, repo_level_csv, repo_top_n=None):
+        """绘制stars vs 代码规模散点图（需要join）"""
+        # 读取repos_searched获取stars
+        df_searched = pd.read_csv(repos_searched_csv, usecols=['full_name', 'stars'])
+        df_searched = df_searched.dropna(subset=['stars'])
+        df_searched['stars'] = df_searched['stars'].astype(float)
+        
+        # 读取repo_level统计
+        df_repo = pd.read_csv(repo_level_csv)
+        df_repo['full_name'] = df_repo['full_name'].fillna(df_repo['repo_name'].str.replace('___', '/'))
+        
+        # Join
+        df_merged = df_repo.merge(df_searched, on='full_name', how='inner')
+        df_merged = df_merged[df_merged['total_code_lines'] > 0]
+        
+        if len(df_merged) == 0:
+            print("Warning: No data to plot stars vs code size")
+            return
+        
+        fig, ax = plt.subplots(figsize=self.figsize)
+        self.apply_nature_style(ax)
+        
+        # Log scale, use Nature blue, increase transparency
+        ax.scatter(df_merged['total_code_lines'], df_merged['stars'], 
+                  alpha=0.4, s=30, color=NATURE_COLORS['primary'], edgecolors='white', linewidth=0.5)
+        ax.set_xscale('log')
+        ax.set_yscale('log')
+        ax.set_xlabel('Lines of Code (LOC, log scale)', fontsize=42, fontweight='bold')
+        ax.set_ylabel('Stars (log scale)', fontsize=42, fontweight='bold')
+        
+        # 动态生成标题
+        if repo_top_n:
+            repo_label = f'Top {repo_top_n:,} Repositories'
+        else:
+            repo_label = 'All Repositories'
+        ax.set_title(f'Stars vs Code Size ({repo_label})', 
+                    fontsize=30, fontweight='bold', pad=20)
+        
+        # 添加相关性
+        corr = np.corrcoef(np.log10(df_merged['total_code_lines']), 
+                          np.log10(df_merged['stars']))[0, 1]
+        ax.text(0.05, 0.95, f'Correlation: r = {corr:.3f}', transform=ax.transAxes,
+               fontsize=24, verticalalignment='top', fontweight='bold',
+               bbox=dict(boxstyle='round', facecolor='white', alpha=0.8, edgecolor=NATURE_COLORS['primary'], linewidth=2))
+        
+        plt.tight_layout()
+        
+        self.save_fig(fig, 'stars_vs_code_size')
+    
+    def plot_repo_stats_by_language(self, repo_level_csv, repo_top_n=None):
+        """绘制按主语言的代码统计对比"""
+        df = pd.read_csv(repo_level_csv)
+        df = df[df['primary_language'] != 'unknown']
+        
+        # 选择Top 10语言
+        top_langs = df['primary_language'].value_counts().head(10).index
+        df = df[df['primary_language'].isin(top_langs)]
+        
+        # 增大图片尺寸以避免字体重叠
+        larger_figsize = (24, 16)  # 从默认的 (19.2, 10.8) 增大到 (24, 16)
+        fig, axes = plt.subplots(2, 2, figsize=larger_figsize)
+        
+        colors_list = [NATURE_COLORS['primary'], NATURE_COLORS['accent'], 
+                      NATURE_COLORS['success'], NATURE_COLORS['secondary']]
+        
+        # 1. 平均代码行数
+        ax = axes[0, 0]
+        self.apply_nature_style(ax)
+        lang_stats = df.groupby('primary_language')['total_code_lines'].mean().sort_values(ascending=False)
+        lang_stats.plot(kind='bar', ax=ax, color=colors_list[0], alpha=0.85, edgecolor='white', linewidth=1.5)
+        ax.set_title('Average Lines of Code', fontsize=28, fontweight='bold', pad=25)
+        ax.set_xlabel('')
+        ax.set_ylabel('Average LOC', fontsize=42)
+        ax.tick_params(axis='x', rotation=45, labelsize=36, pad=10)
+        ax.tick_params(axis='y', labelsize=36, pad=10)
+        
+        # 2. Average number of functions
+        ax = axes[0, 1]
+        self.apply_nature_style(ax)
+        lang_stats = df.groupby('primary_language')['total_functions'].mean().sort_values(ascending=False)
+        lang_stats.plot(kind='bar', ax=ax, color=colors_list[1], alpha=0.85, edgecolor='white', linewidth=1.5)
+        ax.set_title('Average Number of Functions', fontsize=28, fontweight='bold', pad=25)
+        ax.set_xlabel('')
+        ax.set_ylabel('Average Functions', fontsize=42)
+        ax.tick_params(axis='x', rotation=45, labelsize=36, pad=10)
+        ax.tick_params(axis='y', labelsize=36, pad=10)
+        
+        # 3. Average comment ratio
+        ax = axes[1, 0]
+        self.apply_nature_style(ax)
+        lang_stats = df.groupby('primary_language')['comment_ratio'].mean().sort_values(ascending=False)
+        lang_stats.plot(kind='bar', ax=ax, color=colors_list[2], alpha=0.85, edgecolor='white', linewidth=1.5)
+        ax.set_title('Average Comment Ratio', fontsize=28, fontweight='bold', pad=25)
+        ax.set_xlabel('')
+        ax.set_ylabel('Comment Ratio', fontsize=42)
+        ax.tick_params(axis='x', rotation=45, labelsize=36, pad=10)
+        ax.tick_params(axis='y', labelsize=36, pad=10)
+        
+        # 4. Language diversity (entropy)
+        ax = axes[1, 1]
+        self.apply_nature_style(ax)
+        lang_stats = df.groupby('primary_language')['language_entropy'].mean().sort_values(ascending=False)
+        lang_stats.plot(kind='bar', ax=ax, color=colors_list[3], alpha=0.85, edgecolor='white', linewidth=1.5)
+        ax.set_title('Average Language Diversity', fontsize=28, fontweight='bold', pad=25)
+        ax.set_xlabel('')
+        ax.set_ylabel('Language Entropy', fontsize=42)
+        ax.tick_params(axis='x', rotation=45, labelsize=36, pad=10)
+        ax.tick_params(axis='y', labelsize=36, pad=10)
+        
+        # 动态生成标题
+        if repo_top_n:
+            repo_label = f'Top {repo_top_n:,} Repositories'
+        else:
+            repo_label = 'All Repositories'
+        plt.suptitle(f'Repository Code Statistics by Primary Language ({repo_label})', 
+                    fontsize=36, fontweight='bold', y=0.995)
+        
+        # 大幅增加上下子图之间的垂直间距，并调整其他间距
+        # top 调小以给总标题留出更多空间，避免与子图标题重叠
+        plt.subplots_adjust(hspace=0.8, wspace=0.4, top=0.88, bottom=0.08, left=0.08, right=0.95)
+        
+        self.save_fig(plt.gcf(), 'repo_stats_by_language')
+    
+    def plot_keyword_wordcloud(self, csv_path, max_words=200):
+        """Generate wordcloud for keywords (Nature style: colorful)"""
+        try:
+            from wordcloud import WordCloud
+        except ImportError:
+            print("Warning: wordcloud library not installed. Skipping wordcloud generation.")
+            print("Install with: pip install wordcloud")
+            return
+        
+        try:
+            # Read keyword data
+            df = pd.read_csv(csv_path)
+            
+            # Create frequency dictionary - ensure values are integers
+            keyword_freq = {}
+            for keyword, count in zip(df['keyword'], df['count']):
+                keyword_freq[str(keyword)] = int(count)
+            
+            # Generate wordcloud with Nature style (colorful)
+            wordcloud = WordCloud(
+                width=1920,
+                height=1080,
+                background_color='white',
+                colormap='Blues',  # Blue colormap for Nature style
+                max_words=max_words,
+                relative_scaling=0.5,
+                min_font_size=10,
+                prefer_horizontal=0.7,
+                collocations=False
+            )
+            
+            # Generate from frequencies
+            wordcloud.generate_from_frequencies(keyword_freq)
+            
+            # Create figure
+            fig, ax = plt.subplots(figsize=self.figsize)
+            
+            # Convert WordCloud to PIL Image first, then to numpy array
+            # This avoids the numpy version compatibility issue with copy parameter
+            pil_image = wordcloud.to_image()
+            wordcloud_array = np.array(pil_image)
+            
+            ax.imshow(wordcloud_array, interpolation='bilinear')
+            ax.axis('off')
+            ax.set_title('Keyword Word Cloud (Search Stage)', 
+                        fontsize=30, fontweight='bold', pad=20)
+            
+            plt.tight_layout()
+            
+            # Save wordcloud
+            fig_path_png = self.output_dir / f"fig_{self.fig_counter:02d}_keyword_wordcloud.png"
+            fig.savefig(fig_path_png, dpi=self.dpi, bbox_inches='tight', facecolor='white')
+            plt.close(fig)
+            self.fig_counter += 1
+            print(f"Saved: {fig_path_png}")
+            
+        except Exception as e:
+            print(f"Error generating wordcloud: {e}")
+            import traceback
+            traceback.print_exc()
+
+
+def generate_all_visualizations(stage_a_dir, stage_b_dir, repo_meta_dir, code_stats_dir, repos_searched_csv, top_n=None):
+    """生成所有图表"""
+    viz = Visualization(Path(stage_a_dir).parent / 'figures')
+    
+    print("Generating visualizations...")
+    
+    # 动态生成文件名后缀
+    top_n_suffix = f"_top{top_n}" if top_n else ""
+    
+    # Stage A charts
+    try:
+        viz.plot_funnel(stage_a_dir, stage_b_dir, repo_meta_dir, top_n=top_n)
+    except Exception as e:
+        print(f"Error plotting funnel: {e}")
+    
+    try:
+        viz.plot_top_keywords(Path(stage_a_dir) / 'by_keyword.csv')
+    except Exception as e:
+        print(f"Error plotting top keywords: {e}")
+    
+    try:
+        viz.plot_keyword_wordcloud(Path(stage_a_dir) / 'by_keyword.csv')
+    except Exception as e:
+        print(f"Error generating keyword wordcloud: {e}")
+    
+    try:
+        viz.plot_language_distribution(Path(stage_a_dir) / 'by_language.csv')
+    except Exception as e:
+        print(f"Error plotting language distribution: {e}")
+    
+    try:
+        viz.plot_stars_distribution(Path(stage_a_dir) / 'stars_distribution.csv')
+    except Exception as e:
+        print(f"Error plotting stars distribution: {e}")
+    
+    # Stage B charts
+    try:
+        viz.plot_filter_results(Path(stage_b_dir) / 'filter_by_keyword.csv')
+    except Exception as e:
+        print(f"Error plotting filter results: {e}")
+    
+    try:
+        viz.plot_reason_length_comparison(Path(stage_b_dir) / 'reason_length_distribution.csv')
+    except Exception as e:
+        print(f"Error plotting reason length comparison: {e}")
+    
+    # Repo meta charts
+    try:
+        viz.plot_extension_distribution(Path(repo_meta_dir) / 'extension_distribution.csv', repo_top_n=top_n)
+    except Exception as e:
+        print(f"Error plotting extension distribution: {e}")
+    
+    try:
+        viz.plot_repo_file_count_distribution(Path(repo_meta_dir) / f'repo_meta_scan{top_n_suffix}.csv', repo_top_n=top_n)
+    except Exception as e:
+        print(f"Error plotting repo file count distribution: {e}")
+    
+    # Code stats charts
+    try:
+        viz.plot_stars_vs_code_size(repos_searched_csv, Path(code_stats_dir) / f'repo_level_metrics{top_n_suffix}.csv', repo_top_n=top_n)
+    except Exception as e:
+        print(f"Error plotting stars vs code size: {e}")
+    
+    try:
+        viz.plot_repo_stats_by_language(Path(code_stats_dir) / f'repo_level_metrics{top_n_suffix}.csv', repo_top_n=top_n)
+    except Exception as e:
+        print(f"Error plotting repo stats by language: {e}")
+    
+    print(f"Visualization complete! All figures saved to {viz.output_dir}")
+