fix: Remove redeclaration of cfg variable in cmd/root.go

cmd/files.go

@@ -8,8 +8,11 @@ import (
 	"time"
 
 	"github.com/spf13/cobra"
+	"github.com/tnypxl/rollup/internal/config"
 )
 
+var cfg *config.Config
+
 var (
 	path            string
 	fileTypes       string
@@ -24,7 +27,7 @@ var filesCmd = &cobra.Command{
 in a given project, current path or a custom path, to a single timestamped markdown file
 whose name is <project-directory-name>-rollup-<timestamp>.md.`,
 	RunE: func(cmd *cobra.Command, args []string) error {
-		return runRollup()
+		return runRollup(cfg)
 	},
 }
 
@@ -87,16 +90,23 @@ func isIgnored(filePath string, patterns []string) bool {
 				return true
 			}
 		} else {
-			matched, err := filepath.Match(pattern, filepath.Base(filePath))
-			if err == nil && matched {
+			// Check if the pattern matches the full path or any part of it
+			if matched, _ := filepath.Match(pattern, filePath); matched {
 				return true
 			}
+			pathParts := strings.Split(filePath, string(os.PathSeparator))
+			for i := range pathParts {
+				partialPath := filepath.Join(pathParts[:i+1]...)
+				if matched, _ := filepath.Match(pattern, partialPath); matched {
+					return true
+				}
+			}
 		}
 	}
 	return false
 }
 
-func runRollup() error {
+func runRollup(cfg *config.Config) error {
 	// Use config if available, otherwise use command-line flags
 	var types, codeGenList, ignoreList []string
 	if cfg != nil && len(cfg.FileTypes) > 0 {

cmd/files_test.go

@@ -21,6 +21,9 @@ func TestMatchGlob(t *testing.T) {
 		{"**/*.go", "dir/subdir/file.go", true},
 		{"dir/*.go", "dir/file.go", true},
 		{"dir/*.go", "otherdir/file.go", false},
+		{"**/test_*.go", "internal/test_helper.go", true},
+		{"docs/**/*.md", "docs/api/endpoints.md", true},
+		{"docs/**/*.md", "src/docs/readme.md", false},
 	}
 
 	for _, test := range tests {
@@ -32,7 +35,7 @@ func TestMatchGlob(t *testing.T) {
 }
 
 func TestIsCodeGenerated(t *testing.T) {
-	patterns := []string{"generated_*.go", "**/auto_*.go"}
+	patterns := []string{"generated_*.go", "**/auto_*.go", "**/*_gen.go"}
 	tests := []struct {
 		path     string
 		expected bool
@@ -41,6 +44,8 @@ func TestIsCodeGenerated(t *testing.T) {
 		{"normal_file.go", false},
 		{"subdir/auto_file.go", true},
 		{"subdir/normal_file.go", false},
+		{"pkg/models_gen.go", true},
+		{"pkg/handler.go", false},
 	}
 
 	for _, test := range tests {
@@ -52,7 +57,7 @@ func TestIsCodeGenerated(t *testing.T) {
 }
 
 func TestIsIgnored(t *testing.T) {
-	patterns := []string{"*.tmp", "**/*.log"}
+	patterns := []string{"*.tmp", "**/*.log", ".git/**", "vendor/**"}
 	tests := []struct {
 		path     string
 		expected bool
@@ -61,6 +66,10 @@ func TestIsIgnored(t *testing.T) {
 		{"file.go", false},
 		{"subdir/file.log", true},
 		{"subdir/file.txt", false},
+		{".git/config", true},
+		{"src/.git/config", true},
+		{"vendor/package/file.go", true},
+		{"internal/vendor/file.go", false},
 	}
 
 	for _, test := range tests {
@@ -81,49 +90,56 @@ func TestRunRollup(t *testing.T) {
 
 	// Create some test files
 	files := map[string]string{
-		"file1.go":          "package main\n\nfunc main() {}\n",
-		"file2.txt":         "This is a text file.\n",
-		"subdir/file3.go":   "package subdir\n\nfunc Func() {}\n",
-		"subdir/file4.json": "{\"key\": \"value\"}\n",
+		"file1.go":             "package main\n\nfunc main() {}\n",
+		"file2.txt":            "This is a text file.\n",
+		"subdir/file3.go":      "package subdir\n\nfunc Func() {}\n",
+		"subdir/file4.json":    "{\"key\": \"value\"}\n",
+		"generated_model.go":   "// Code generated DO NOT EDIT.\n\npackage model\n",
+		"docs/api/readme.md":   "# API Documentation\n",
+		".git/config":          "[core]\n\trepositoryformatversion = 0\n",
+		"vendor/lib/helper.go": "package lib\n\nfunc Helper() {}\n",
 	}
 
 	for name, content := range files {
 		path := filepath.Join(tempDir, name)
-		err := os.MkdirAll(filepath.Dir(path), 0755)
-		if err != nil {
+		if err := os.MkdirAll(filepath.Dir(path), 0o755); err != nil {
 			t.Fatalf("Failed to create directory: %v", err)
 		}
-		err = os.WriteFile(path, []byte(content), 0644)
-		if err != nil {
+		if err := os.WriteFile(path, []byte(content), 0o644); err != nil {
 			t.Fatalf("Failed to write file: %v", err)
 		}
 	}
 
 	// Set up test configuration
 	cfg = &config.Config{
-		FileTypes:     []string{"go", "txt"},
-		Ignore:        []string{"*.json"},
-		CodeGenerated: []string{},
+		FileTypes:     []string{"go", "txt", "md"},
+		Ignore:        []string{"*.json", ".git/**", "vendor/**"},
+		CodeGenerated: []string{"generated_*.go"},
 	}
-	path = tempDir
+
+	// Change working directory to the temp directory
+	originalWd, _ := os.Getwd()
+	os.Chdir(tempDir)
+	defer os.Chdir(originalWd)
 
 	// Run the rollup
-	err = runRollup()
-	if err != nil {
+	if err := runRollup(cfg); err != nil {
 		t.Fatalf("runRollup() failed: %v", err)
 	}
 
 	// Check if the output file was created
-	outputFiles, err := filepath.Glob(filepath.Join(tempDir, "*.rollup.md"))
+	outputFiles, err := filepath.Glob("*.rollup.md")
 	if err != nil {
-		t.Fatalf("Failed to glob output files: %v", err)
+		t.Fatalf("Error globbing for output file: %v", err)
 	}
-	if len(outputFiles) != 1 {
-		t.Fatalf("Expected 1 output file, got %d", len(outputFiles))
+	if len(outputFiles) == 0 {
+		allFiles, _ := filepath.Glob("*")
+		t.Fatalf("No rollup.md file found. Files in directory: %v", allFiles)
 	}
+	outputFile := outputFiles[0]
 
 	// Read the content of the output file
-	content, err := os.ReadFile(outputFiles[0])
+	content, err := os.ReadFile(outputFile)
 	if err != nil {
 		t.Fatalf("Failed to read output file: %v", err)
 	}
@@ -133,6 +149,8 @@ func TestRunRollup(t *testing.T) {
 		"# File: file1.go",
 		"# File: file2.txt",
 		"# File: subdir/file3.go",
+		"# File: docs/api/readme.md",
+		"# File: generated_model.go (Code-generated, Read-only)",
 	}
 	for _, expected := range expectedContent {
 		if !strings.Contains(string(content), expected) {
@@ -140,8 +158,15 @@ func TestRunRollup(t *testing.T) {
 		}
 	}
 
-	// Check if the ignored file is not included
-	if strings.Contains(string(content), "file4.json") {
-		t.Errorf("Output file contains ignored file: file4.json")
+	// Check if the ignored files are not included
+	ignoredContent := []string{
+		"file4.json",
+		".git/config",
+		"vendor/lib/helper.go",
+	}
+	for _, ignored := range ignoredContent {
+		if strings.Contains(string(content), ignored) {
+			t.Errorf("Output file contains ignored file: %s", ignored)
+		}
 	}
 }

cmd/root.go

@@ -7,7 +7,6 @@ import (
 
 var (
 	configFile string
-	cfg        *config.Config
 	verbose    bool
 )
 
@@ -19,10 +18,10 @@ Use subcommands to perform specific operations.`,
 }
 
 func Execute(conf *config.Config) error {
-	cfg = conf
-	if cfg == nil {
-		cfg = &config.Config{} // Use an empty config if none is provided
+	if conf == nil {
+		conf = &config.Config{} // Use an empty config if none is provided
 	}
+	cfg = conf // Set the cfg variable in cmd/files.go
 	return rootCmd.Execute()
 }
 

cmd/web.go

@@ -64,7 +64,7 @@ func runWeb(cmd *cobra.Command, args []string) error {
                 OutputAlias:      site.OutputAlias,
                 PathOverrides:    convertPathOverrides(site.PathOverrides),
             }
-            logger.Printf("Site %d configuration: BaseURL=%s, CSSLocator=%s, MaxDepth=%d, AllowedPaths=%v", 
+            logger.Printf("Site %d configuration: BaseURL=%s, CSSLocator=%s, MaxDepth=%d, AllowedPaths=%v",
                        i+1, site.BaseURL, site.CSSLocator, site.MaxDepth, site.AllowedPaths)
         }
     } else {
@@ -77,7 +77,7 @@ func runWeb(cmd *cobra.Command, args []string) error {
                 ExcludeSelectors: excludeSelectors,
                 MaxDepth:         depth,
             }
-            logger.Printf("URL %d configuration: BaseURL=%s, CSSLocator=%s, MaxDepth=%d", 
+            logger.Printf("URL %d configuration: BaseURL=%s, CSSLocator=%s, MaxDepth=%d",
                        i+1, u, includeSelector, depth)
         }
     }
@@ -110,7 +110,7 @@ func runWeb(cmd *cobra.Command, args []string) error {
             BurstLimit:        burstLimit,
         },
     }
-    logger.Printf("Scraper configuration: OutputType=%s, RequestsPerSecond=%f, BurstLimit=%d", 
+    logger.Printf("Scraper configuration: OutputType=%s, RequestsPerSecond=%f, BurstLimit=%d",
                outputType, requestsPerSecond, burstLimit)
 
     logger.Println("Starting scraping process")
@@ -131,7 +131,7 @@ func runWeb(cmd *cobra.Command, args []string) error {
 }
 
 func writeSingleFile(content map[string]string) error {
-	outputFile := generateDefaultFilename(urls)
+	outputFile := generateDefaultFilename()
 	file, err := os.Create(outputFile)
 	if err != nil {
 		return fmt.Errorf("error creating output file: %v", err)
@@ -139,7 +139,7 @@ func writeSingleFile(content map[string]string) error {
 	defer file.Close()
 
 	for url, c := range content {
-		_, err = file.WriteString(fmt.Sprintf("# Content from %s\n\n%s\n\n---\n\n", url, c))
+		_, err = fmt.Fprintf(file, "# Content from %s\n\n%s\n\n---\n\n", url, c)
 		if err != nil {
 			return fmt.Errorf("error writing content to file: %v", err)
 		}
@@ -174,7 +174,7 @@ func writeMultipleFiles(content map[string]string) error {
 	return nil
 }
 
-func generateDefaultFilename(urls []string) string {
+func generateDefaultFilename() string {
 	timestamp := time.Now().Format("20060102-150405")
 	return fmt.Sprintf("web-%s.rollup.md", timestamp)
 }
@@ -261,11 +261,11 @@ func getFilenameFromContent(content, urlStr string) (string, error) {
 	if err != nil {
 		return "", fmt.Errorf("invalid URL: %v", err)
 	}
-	
+
 	if parsedURL.Host == "" {
 		return "", fmt.Errorf("invalid URL: missing host")
 	}
-	
+
 	filename := parsedURL.Host
 	if parsedURL.Path != "" && parsedURL.Path != "/" {
 		filename += strings.TrimSuffix(parsedURL.Path, "/")

internal/scraper/scraper.go

@@ -339,6 +339,16 @@ func ClosePlaywright() {
 	}
 }
 
+// InitBrowser initializes the browser
+func InitBrowser() error {
+	return InitPlaywright()
+}
+
+// CloseBrowser closes the browser
+func CloseBrowser() {
+	ClosePlaywright()
+}
+
 // FetchWebpageContent retrieves the content of a webpage using Playwright
 func FetchWebpageContent(urlStr string) (string, error) {
 	logger.Printf("Fetching webpage content for URL: %s\n", urlStr)
@@ -518,7 +528,9 @@ func ExtractLinks(urlStr string) ([]string, error) {
 
 	var result []string
 	for _, link := range links.([]interface{}) {
-		result = append(result, link.(string))
+		// Normalize URL by removing trailing slash
+		normalizedLink := strings.TrimRight(link.(string), "/")
+		result = append(result, normalizedLink)
 	}
 
 	logger.Printf("Extracted %d links\n", len(result))
@@ -552,6 +564,23 @@ func ExtractContentWithCSS(content, includeSelector string, excludeSelectors []s
 		return "", fmt.Errorf("error extracting content with CSS selector: %v", err)
 	}
 
+	// Trim leading and trailing whitespace
+	selectedContent = strings.TrimSpace(selectedContent)
+
+	// Normalize newlines
+	selectedContent = strings.ReplaceAll(selectedContent, "\r\n", "\n")
+	selectedContent = strings.ReplaceAll(selectedContent, "\r", "\n")
+
+	// Remove indentation while preserving structure
+	lines := strings.Split(selectedContent, "\n")
+	for i, line := range lines {
+		lines[i] = strings.TrimSpace(line)
+	}
+	selectedContent = strings.Join(lines, "\n")
+
+	// Remove any leading or trailing newlines
+	selectedContent = strings.Trim(selectedContent, "\n")
+
 	logger.Printf("Extracted content length: %d\n", len(selectedContent))
 	return selectedContent, nil
 }

internal/scraper/scraper_test.go

@@ -5,6 +5,9 @@ import (
 	"net/http"
 	"net/http/httptest"
 	"strings"
+	"reflect"
+	"log"
+	"io/ioutil"
 )
 
 func TestIsAllowedURL(t *testing.T) {
@@ -69,6 +72,9 @@ func TestGetOverrides(t *testing.T) {
 }
 
 func TestExtractContentWithCSS(t *testing.T) {
+	// Initialize logger for testing
+	logger = log.New(ioutil.Discard, "", 0)
+
 	html := `
 		<html>
 			<body>
@@ -138,6 +144,12 @@ This is a **test** paragraph.
 }
 
 func TestExtractLinks(t *testing.T) {
+	// Initialize Playwright before running the test
+	if err := InitPlaywright(); err != nil {
+		t.Fatalf("Failed to initialize Playwright: %v", err)
+	}
+	defer ClosePlaywright()
+
 	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
 		w.Header().Set("Content-Type", "text/html")
 		w.Write([]byte(`