More robust filename parsing via stdin

fry69 · Apr 17, 2024 · 4d3746a · 4d3746a
1 parent 23c5a27
commit 4d3746a
Show file tree

Hide file tree

Showing 2 changed files with 89 additions and 3 deletions.
diff --git a/files-to-prompt.test.ts b/files-to-prompt.test.ts
@@ -257,6 +257,58 @@ describe('files-to-prompt.ts', () => {
     expect(filePathsFromStdin).toEqual(['file1.txt', 'file2.txt']);
   });
 
+  test('should parse file paths with one file path per line', () => {
+    const stdinData = `file1.txt\nfile2.txt\nfile3.txt`;
+    const filePathsFromStdin = parseFilePathsFromStdin(stdinData);
+    expect(filePathsFromStdin).toEqual(['file1.txt', 'file2.txt', 'file3.txt']);
+  });
+
+  test('should handle mixed input formats', () => {
+    const stdinData = `file1.txt:File 1 contents.\nfile2.txt\nfile3.txt:File 3 contents.`;
+    const filePathsFromStdin = parseFilePathsFromStdin(stdinData);
+    expect(filePathsFromStdin).toEqual(['file1.txt', 'file2.txt', 'file3.txt']);
+  });
+
+  test('should handle empty lines in stdin data', () => {
+    const stdinData = `file1.txt:File 1 contents.\n\nfile2.txt:File 2 contents.\n`;
+    const filePathsFromStdin = parseFilePathsFromStdin(stdinData);
+    expect(filePathsFromStdin).toEqual(['file1.txt', 'file2.txt']);
+  });
+
+  test('should handle binary data in stdin', () => {
+    const binaryData = Buffer.from([0x80, 0x81, 0x82, 0x83, 0x84, 0x85]);
+    const stdinData = `file1.txt:File 1 contents.\n${binaryData.toString('utf8')}\nfile2.txt:File 2 contents.`;
+    const filePathsFromStdin = parseFilePathsFromStdin(stdinData);
+    expect(filePathsFromStdin).toEqual(['file1.txt', 'file2.txt']);
+  });
+
+  test('should handle common text/code files in stdin', () => {
+    const textData = `console.log('Hello, world\!');`;
+    const stdinData = `file1.txt:File 1 contents.\n${textData}\nfile2.txt:File 2 contents.`;
+    const filePathsFromStdin = parseFilePathsFromStdin(stdinData);
+    expect(filePathsFromStdin).toEqual(['file1.txt', textData, 'file2.txt']);
+  });
+
+  test('should handle long file paths in stdin', () => {
+    const longFilePath = 'a'.repeat(1025);
+    const stdinData = `file1.txt:File 1 contents.\n${longFilePath}\nfile2.txt:File 2 contents.`;
+    const filePathsFromStdin = parseFilePathsFromStdin(stdinData);
+    expect(filePathsFromStdin).toEqual(['file1.txt', 'file2.txt']);
+  });
+
+  test('should ignore file paths with the null character', () => {
+    const invalidFilePath = 'invalid_file\0.txt';
+    const stdinData = `file1.txt:File 1 contents.\n${invalidFilePath}\nfile2.txt:File 2 contents.`;
+    const filePathsFromStdin = parseFilePathsFromStdin(stdinData);
+    expect(filePathsFromStdin).toEqual(['file1.txt', 'file2.txt']);
+  });
+
+  test('should ignore file paths with control characters', () => {
+    const stdinData = `file1.txt:File 1 contents.\nfile2.txt\x07.txt:File 2 contents.\nfile3.txt:File 3 contents.`;
+    const filePathsFromStdin = parseFilePathsFromStdin(stdinData);
+    expect(filePathsFromStdin).toEqual(['file1.txt', 'file3.txt']);
+  });
+
   test('should output version string when --version is passed', async () => {
     await main(['--version']);
     expect(stdoutOutput).toContain(`files-to-prompt.ts version`);

diff --git a/files-to-prompt.ts b/files-to-prompt.ts
@@ -249,12 +249,24 @@ let readStdin = async (): Promise<string> => {
  */
 export function parseFilePathsFromStdin(stdinData: string): string[] {
   const filePathsFromStdin: string[] = [];
-  const lines = stdinData.trim().split('\n');
   const seenFilePaths = new Set<string>();
+  const lines = stdinData.trim().split('\n');
 
   for (const line of lines) {
-    const filePath = line.split(':')[0];
-    if (!seenFilePaths.has(filePath)) {
+    const filePath = line.trim();
+    if (filePath === '') {
+      // Ignore empty line
+      continue;
+    }
+    if (filePath.includes(':')) {
+      // Handle grep/ripgrep output format
+      const parts = filePath.split(':');
+      if (isValidFilePath(parts[0]) && !seenFilePaths.has(parts[0])) {
+        seenFilePaths.add(parts[0]);
+        filePathsFromStdin.push(parts[0]);
+      }
+    } else if (isValidFilePath(filePath) && !seenFilePaths.has(filePath)) {
+      // Handle file path per line format
       seenFilePaths.add(filePath);
       filePathsFromStdin.push(filePath);
     }
@@ -263,6 +275,28 @@ export function parseFilePathsFromStdin(stdinData: string): string[] {
   return filePathsFromStdin;
 }
 
+/**
+ * Checks if a given string is a valid file path.
+ * @function isValidFilePath
+ * @param {string} filePath - The file path to check.
+ * @returns {boolean} - `true` if the file path is valid, `false` otherwise.
+ */
+function isValidFilePath(filePath: string): boolean {
+  // Check if the file path contains only valid characters
+  for (const char of filePath) {
+    if (char.charCodeAt(0) < 32 || char.charCodeAt(0) > 126) {
+      return false;
+    }
+  }
+
+  // Check if the file path is not too long
+  if (filePath.length > 1024) {
+    return false;
+  }
+
+  // If the file path passes the above checks, consider it valid
+  return true;
+}
 
 /**
  * The main entry point of the script.