fix: Improve error handling for permission errors and special tokens (Issue #868)

biome.json

@@ -1,5 +1,5 @@
 {
-  "$schema": "https://biomejs.dev/schemas/2.2.6/schema.json",
+  "$schema": "https://biomejs.dev/schemas/latest/schema.json",
   "files": {
     "includes": [
       "bin/**",

src/cli/actions/remoteAction.ts

@@ -206,6 +206,21 @@ export const copyOutputToCurrentDirectory = async (
 
     await fs.copyFile(sourcePath, targetPath);
   } catch (error) {
+    const nodeError = error as NodeJS.ErrnoException;
+
+    // Provide helpful message for permission errors
+    if (nodeError.code === 'EPERM' || nodeError.code === 'EACCES') {
+      throw new RepomixError(
+        `Failed to copy output file to ${targetPath}: Permission denied.
+
+The current directory may be protected or require elevated permissions.
+Please try one of the following:
+  • Run from a different directory (e.g., your home directory or Documents folder)
+  • Use the --output flag to specify a writable location: --output ~/repomix-output.xml
+  • Use --stdout to print output directly to the console`,
+      );
+    }
+
     throw new RepomixError(`Failed to copy output file: ${(error as Error).message}`);
   }
 };

src/core/metrics/TokenCounter.ts

@@ -18,7 +18,12 @@ export class TokenCounter {
 
   public countTokens(content: string, filePath?: string): number {
     try {
-      return this.encoding.encode(content).length;
+      // Disable special token validation to handle files that may contain
+      // special token sequences (e.g., tokenizer configs with <|endoftext|>).
+      // This treats special tokens as ordinary text rather than control tokens,
+      // which is appropriate for general code/text analysis where we're not
+      // actually sending the content to an LLM API.
+      return this.encoding.encode(content, [], []).length;
     } catch (error) {
       let message = '';
       if (error instanceof Error) {

tests/cli/actions/remoteAction.test.ts

@@ -256,5 +256,33 @@ describe('remoteAction functions', () => {
         'Failed to copy output file',
       );
     });
+
+    test('should throw helpful error message for EPERM permission errors', async () => {
+      const sourceDir = '/tmp/repomix-123';
+      const targetDir = 'C:\\Windows\\System32';
+      const fileName = 'output.xml';
+
+      const epermError = new Error('operation not permitted') as NodeJS.ErrnoException;
+      epermError.code = 'EPERM';
+      vi.mocked(fs.copyFile).mockRejectedValue(epermError);
+
+      await expect(copyOutputToCurrentDirectory(sourceDir, targetDir, fileName)).rejects.toThrow(
+        /Permission denied.*protected.*--output.*--stdout/s,
+      );
+    });
+
+    test('should throw helpful error message for EACCES permission errors', async () => {
+      const sourceDir = '/tmp/repomix-123';
+      const targetDir = '/protected/dir';
+      const fileName = 'output.xml';
+
+      const eaccesError = new Error('permission denied') as NodeJS.ErrnoException;
+      eaccesError.code = 'EACCES';
+      vi.mocked(fs.copyFile).mockRejectedValue(eaccesError);
+
+      await expect(copyOutputToCurrentDirectory(sourceDir, targetDir, fileName)).rejects.toThrow(
+        /Permission denied.*protected.*--output.*--stdout/s,
+      );
+    });
   });
 });

tests/core/metrics/TokenCounter.test.ts

@@ -47,7 +47,7 @@ describe('TokenCounter', () => {
     const count = tokenCounter.countTokens(text);
 
     expect(count).toBe(3); // Length of mockTokens
-    expect(mockEncoder.encode).toHaveBeenCalledWith(text);
+    expect(mockEncoder.encode).toHaveBeenCalledWith(text, [], []);
   });
 
   test('should handle empty string', () => {
@@ -56,7 +56,7 @@ describe('TokenCounter', () => {
     const count = tokenCounter.countTokens('');
 
     expect(count).toBe(0);
-    expect(mockEncoder.encode).toHaveBeenCalledWith('');
+    expect(mockEncoder.encode).toHaveBeenCalledWith('', [], []);
   });
 
   test('should handle multi-line text', () => {
@@ -67,7 +67,7 @@ describe('TokenCounter', () => {
     const count = tokenCounter.countTokens(text);
 
     expect(count).toBe(6);
-    expect(mockEncoder.encode).toHaveBeenCalledWith(text);
+    expect(mockEncoder.encode).toHaveBeenCalledWith(text, [], []);
   });
 
   test('should handle special characters', () => {
@@ -78,7 +78,7 @@ describe('TokenCounter', () => {
     const count = tokenCounter.countTokens(text);
 
     expect(count).toBe(3);
-    expect(mockEncoder.encode).toHaveBeenCalledWith(text);
+    expect(mockEncoder.encode).toHaveBeenCalledWith(text, [], []);
   });
 
   test('should handle unicode characters', () => {
@@ -89,7 +89,7 @@ describe('TokenCounter', () => {
     const count = tokenCounter.countTokens(text);
 
     expect(count).toBe(4);
-    expect(mockEncoder.encode).toHaveBeenCalledWith(text);
+    expect(mockEncoder.encode).toHaveBeenCalledWith(text, [], []);
   });
 
   test('should handle code snippets', () => {
@@ -104,7 +104,7 @@ describe('TokenCounter', () => {
     const count = tokenCounter.countTokens(text);
 
     expect(count).toBe(10);
-    expect(mockEncoder.encode).toHaveBeenCalledWith(text);
+    expect(mockEncoder.encode).toHaveBeenCalledWith(text, [], []);
   });
 
   test('should handle markdown text', () => {
@@ -122,7 +122,7 @@ describe('TokenCounter', () => {
     const count = tokenCounter.countTokens(text);
 
     expect(count).toBe(15);
-    expect(mockEncoder.encode).toHaveBeenCalledWith(text);
+    expect(mockEncoder.encode).toHaveBeenCalledWith(text, [], []);
   });
 
   test('should handle very long text', () => {
@@ -133,7 +133,7 @@ describe('TokenCounter', () => {
     const count = tokenCounter.countTokens(text);
 
     expect(count).toBe(100);
-    expect(mockEncoder.encode).toHaveBeenCalledWith(text);
+    expect(mockEncoder.encode).toHaveBeenCalledWith(text, [], []);
   });
 
   test('should properly handle encoding errors without file path', () => {