Add GitHub integration for PR activity heat maps

docs/SCHEMA.md

@@ -113,6 +113,7 @@ Array of all files and directories in the repository:
 | lastCommitDaysAgo | Days since the last commit to this file | No |
 | lastCommitDate | Date of the last commit to this file (ISO format) | No |
 | topLevelIdentifiers | Number of top-level identifiers (classes, functions, variables) | No |
+| githubActivity | GitHub activity data from open pull requests | No |
 | custom | Object for custom metrics | No |
 
 #### File Components

frontend/src/types/schema.ts

@@ -45,6 +45,8 @@ export interface FileMetrics {
   lastCommitDaysAgo?: number;
   lastCommitDate?: string;
   topLevelIdentifiers?: number;
+  testCoverageRatio?: number;
+  githubActivity?: Record<string, any>;
   custom?: Record<string, any>;
 }
 

frontend/src/types/visualization.ts

@@ -98,6 +98,16 @@ export const DATA_SOURCES: DataSource[] = [
     dataType: 'continuous',
     applicableTo: 'edge',
   },
+  {
+    id: 'github_activity',
+    name: 'GitHub Activity',
+    description: 'Activity level based on open PR changes',
+    color: '#f97316',
+    defaultWeight: 0,
+    category: 'git',
+    dataType: 'continuous',
+    applicableTo: 'node',
+  },
   {
     id: 'identifiers',
     name: 'Identifiers',
@@ -250,6 +260,7 @@ export const DEFAULT_CONFIG: VisualizationConfig = {
         filesystem_proximity: 0,
         code_references: 0,
         test_coverage_ratio: 0,
+        github_activity: 0,
       },
       threshold: 0,
       includeDirectories: false,
@@ -267,6 +278,7 @@ export const DEFAULT_CONFIG: VisualizationConfig = {
         filesystem_proximity: 0,
         code_references: 0,
         test_coverage_ratio: 0,
+        github_activity: 0,
       },
       threshold: 0,
       includeDirectories: false,
@@ -284,6 +296,7 @@ export const DEFAULT_CONFIG: VisualizationConfig = {
         filesystem_proximity: 30,
         code_references: 70,
         test_coverage_ratio: 0,
+        github_activity: 0,
       },
       threshold: 0,
       includeDirectories: true,
@@ -301,6 +314,7 @@ export const DEFAULT_CONFIG: VisualizationConfig = {
         filesystem_proximity: 0,
         code_references: 100,
         test_coverage_ratio: 0,
+        github_activity: 0,
       },
       threshold: 0,
     },
@@ -317,6 +331,7 @@ export const DEFAULT_CONFIG: VisualizationConfig = {
         filesystem_proximity: 0,
         code_references: 0,
         test_coverage_ratio: 100,
+        github_activity: 0,
       },
       threshold: 0,
       includeDirectories: true,
@@ -333,6 +348,7 @@ export const DEFAULT_CONFIG: VisualizationConfig = {
         semantic_similarity: 0,
         filesystem_proximity: 0,
         code_references: 100,
+        github_activity: 0,
       },
       threshold: 0,
     },

frontend/src/utils/visualizationUtils.ts

@@ -37,6 +37,7 @@ export interface ComputedNodeMetrics {
   identifiers: number;
   references: number;
   test_coverage_ratio?: number;
+  github_activity?: number;
 }
 
 export interface ComputedLinkMetrics {
@@ -71,6 +72,12 @@ export const computeNodeMetrics = (data: RepositoryData): Map<string, ComputedNo
     // For directories, use 'directory' as the file_type for categorical coloring
     const fileType = file.type === 'directory' ? 'directory' : file.extension || 'unknown';
 
+    // Calculate GitHub activity score (0-1)
+    let githubActivityScore = 0;
+    if (fileMetrics.githubActivity) {
+      githubActivityScore = fileMetrics.githubActivity.activity_score || 0;
+    }
+
     metrics.set(file.id, {
       file_type: fileType,
       file_size: file.size || 0,
@@ -79,6 +86,7 @@ export const computeNodeMetrics = (data: RepositoryData): Map<string, ComputedNo
       identifiers: fileMetrics.topLevelIdentifiers || 0,
       references: incomingReferences.get(file.id) || 0,
       test_coverage_ratio: fileMetrics.testCoverageRatio,
+      github_activity: githubActivityScore,
     });
 
     // Add metrics for components (classes, functions, methods) - only for files, not directories
@@ -92,6 +100,7 @@ export const computeNodeMetrics = (data: RepositoryData): Map<string, ComputedNo
           identifiers: fileMetrics.topLevelIdentifiers || 0,
           references: incomingReferences.get(component.id) || 0,
           test_coverage_ratio: fileMetrics.testCoverageRatio,
+          github_activity: githubActivityScore,
         });
       });
     }

pyproject.toml

@@ -36,6 +36,9 @@ semantic = [
     "openai>=1.0.0",
     "numpy>=1.20.0",
 ]
+github = [
+    "requests>=2.25.0",
+]
 
 [project.scripts]
 repo-visualizer = "src.repo_visualizer.cli:main"

src/repo_visualizer/analyzer.py

@@ -30,6 +30,7 @@
 except ImportError:
     NUMPY_AVAILABLE = False
 
+from .github_client import GitHubClient
 from .schema import (
     Component,
     File,
@@ -42,12 +43,19 @@
 class RepositoryAnalyzer:
     """Analyzes a local git repository and generates visualization data."""
 
-    def __init__(self, repo_path: str):
+    def __init__(
+        self,
+        repo_path: str,
+        enable_github: bool = False,
+        github_token: Optional[str] = None,
+    ):
         """
         Initialize the repository analyzer.
 
         Args:
             repo_path: Path to the local git repository
+            enable_github: Whether to fetch GitHub activity data
+            github_token: GitHub personal access token for API requests
         """
         self.repo_path = os.path.abspath(repo_path)
         if not os.path.isdir(self.repo_path):
@@ -63,6 +71,18 @@ def __init__(self, repo_path: str):
         self.relationships: List[Relationship] = []
         self.relationship_counts: Dict[Tuple[str, str, str], int] = {}
 
+        # GitHub integration
+        self.enable_github = enable_github
+        self.github_client = None
+        self.github_activity_data: Optional[Dict[str, Dict]] = None
+
+        if enable_github:
+            try:
+                self.github_client = GitHubClient(github_token)
+            except ImportError as e:
+                print(f"Warning: {e}")
+                self.enable_github = False
+
         # Load gitignore patterns
         self.gitignore_spec = self._load_gitignore_patterns()
 
@@ -76,6 +96,12 @@ def analyze(self) -> RepositoryData:
         Returns:
             RepositoryData: The complete repository data structure
         """
+        # Fetch GitHub activity data if enabled
+        if self.enable_github and self.github_client:
+            self.github_activity_data = self.github_client.analyze_repository_activity(
+                self.repo_path
+            )
+
         # Extract repository metadata
         self._extract_metadata()
 
@@ -714,6 +740,12 @@ def _analyze_file_content(
                 if coverage_ratio is not None:
                     metrics["testCoverageRatio"] = coverage_ratio
 
+                # Add GitHub activity data if available
+                if self.github_activity_data and rel_path in self.github_activity_data:
+                    if not metrics:
+                        metrics = {}
+                    metrics["githubActivity"] = self.github_activity_data[rel_path]
+
                 return components, metrics
         except Exception as e:
             print(f"Error analyzing file {rel_path}: {e}")
@@ -2339,15 +2371,24 @@ def default(self, obj):
         print(f"Repository data saved to {output_path}")
 
 
-def analyze_repository(repo_path: str, output_path: str) -> None:
+def analyze_repository(
+    repo_path: str,
+    output_path: str,
+    enable_github: bool = False,
+    github_token: Optional[str] = None,
+) -> None:
     """
     Analyze a repository and generate visualization data.
 
     Args:
         repo_path: Path to the local git repository
         output_path: Path to output JSON file
+        enable_github: Whether to fetch GitHub activity data
+        github_token: GitHub personal access token for API requests
     """
-    analyzer = RepositoryAnalyzer(repo_path)
+    analyzer = RepositoryAnalyzer(
+        repo_path, enable_github=enable_github, github_token=github_token
+    )
     analyzer.analyze()
     analyzer.save_to_file(output_path)
 

src/repo_visualizer/cli.py

@@ -65,6 +65,18 @@ def parse_args(args: Optional[List[str]] = None) -> argparse.Namespace:
         action="store_true",
     )
 
+    parser.add_argument(
+        "--github",
+        help="Enable GitHub integration to fetch PR activity data",
+        action="store_true",
+    )
+
+    parser.add_argument(
+        "--github-token",
+        help="GitHub personal access token (can also be set via GITHUB_TOKEN env var)",
+        type=str,
+    )
+
     return parser.parse_args(args)
 
 
@@ -104,7 +116,24 @@ def main(args: Optional[List[str]] = None) -> int:
 
         # Analyze repository
         logger.info(f"Analyzing repository at {repo_path}")
-        analyze_repository(repo_path, output_path)
+
+        # GitHub integration
+        enable_github = parsed_args.github
+        github_token = parsed_args.github_token
+
+        if enable_github:
+            logger.info("GitHub integration enabled - fetching PR activity data")
+            if not github_token and not os.getenv("GITHUB_TOKEN"):
+                logger.warning(
+                    "No GitHub token provided. API requests will be rate-limited."
+                )
+
+        analyze_repository(
+            repo_path,
+            output_path,
+            enable_github=enable_github,
+            github_token=github_token,
+        )
 
         logger.info(f"Analysis complete. Output saved to {output_path}")
         return 0