Integrate workspace file filtering into path extraction and update relevant file parsing logic for code blocks

- we don't need to support relative paths, as workspace files are always absolute

Author: robertpiosik

apps/editor/src/commands/apply-response-command/utils/response-parser/parsers/relevant-files-parser/relevant-files-parser.ts

@@ -9,15 +9,17 @@ export const parse_relevant_files = (params: {
   response: string
   workspace_files: string[]
 }): RelevantFilesItem | null => {
-  if (params.response.includes('```')) {
+  if (
+    params.response.includes('```') &&
+    !params.response.trimStart().startsWith('**Relevant files:**')
+  ) {
     return null
   }
 
-  const found_paths = extract_paths_from_text(params.response)
-
-  const valid_paths = found_paths
-    .filter((path) => params.workspace_files.includes(path))
-    .sort((a, b) => params.response.indexOf(a) - params.response.indexOf(b))
+  const valid_paths = extract_paths_from_text(
+    params.response,
+    params.workspace_files
+  )
 
   if (valid_paths.length == 0) {
     return null

apps/editor/src/commands/apply-response-command/utils/response-parser/parsers/relevant-files-parser/test/cases/code-blocks-valid/relevant-files.txt

@@ -0,0 +1,7 @@
+**Relevant files:**
+
+`src/hello.ts`
+
+```typescript
+const x = 1;
+```

apps/editor/src/commands/apply-response-command/utils/response-parser/parsers/relevant-files-parser/test/relevant-files-parser.spec.ts

@@ -20,4 +20,16 @@ describe('relevant files parser', () => {
 
     expect(result).toBeNull()
   })
+
+  it('accepts when including code blocks if it starts with **Relevant files:**', () => {
+    const test_case = 'code-blocks-valid'
+    const text = load_test_case_file(test_case, 'relevant-files.txt')
+    const result = parse_relevant_files({
+      response: text,
+      workspace_files: ['src/hello.ts', 'src/main.ts']
+    })
+
+    expect(result).not.toBeNull()
+    expect(result?.file_paths).toEqual(['src/hello.ts'])
+  })
 })

apps/editor/src/commands/select-clipboard-paths-command.ts

@@ -28,11 +28,7 @@ export const select_clipboard_paths_command = (
           workspace_provider
         })
 
-        const found_paths = extract_paths_from_text(text)
-
-        const valid_paths = found_paths
-          .filter((p) => workspace_files.includes(p))
-          .sort((a, b) => text.indexOf(a) - text.indexOf(b))
+        const valid_paths = extract_paths_from_text(text, workspace_files)
 
         if (valid_paths.length === 0) {
           vscode.window.showInformationMessage(

apps/editor/src/utils/extract-paths-from-text/extract-paths-from-text.spec.ts

@@ -9,7 +9,8 @@ describe('extract_paths_from_text', () => {
 
 Lorem ipsum.`
 
-    const result = extract_paths_from_text(text)
+    const workspace_files = ['src/hello.ts', 'src/welcome.ts']
+    const result = extract_paths_from_text(text, workspace_files)
 
     expect(result).toContain('src/hello.ts')
     expect(result).toContain('src/welcome.ts')
@@ -20,29 +21,38 @@ Lorem ipsum.`
 
 Lorem ipsum.`
 
-    const result = extract_paths_from_text(text)
+    const workspace_files = ['src/hello.ts', 'src/welcome.ts', 'src/main.ts']
+    const result = extract_paths_from_text(text, workspace_files)
 
     expect(result).toContain('src/hello.ts')
     expect(result).toContain('src/welcome.ts')
     expect(result).toContain('src/main.ts')
   })
 
-  it('handles paths with ./ prefix', () => {
-    const text = 'Please check ./src/main.ts and ./tests/main.spec.ts.'
-    const result = extract_paths_from_text(text)
-
-    expect(result).toContain('./src/main.ts')
-    expect(result).toContain('src/main.ts')
-    expect(result).toContain('./tests/main.spec.ts')
-    expect(result).toContain('tests/main.spec.ts')
-  })
-
   it('strips trailing punctuation', () => {
     const text = 'Look at src/utils.ts, src/types.ts! Also see src/index.ts.'
-    const result = extract_paths_from_text(text)
+    const workspace_files = ['src/utils.ts', 'src/types.ts', 'src/index.ts']
+    const result = extract_paths_from_text(text, workspace_files)
 
     expect(result).toContain('src/utils.ts')
     expect(result).toContain('src/types.ts')
     expect(result).toContain('src/index.ts')
   })
+
+  it('extracts files in the root directory', () => {
+    const text =
+      'Check out package.json, README.md, and .gitignore for project configurations.'
+    const workspace_files = [
+      'package.json',
+      'README.md',
+      '.gitignore',
+      'src/main.ts'
+    ]
+    const result = extract_paths_from_text(text, workspace_files)
+
+    expect(result).toContain('package.json')
+    expect(result).toContain('README.md')
+    expect(result).toContain('.gitignore')
+    expect(result).not.toContain('src/main.ts')
+  })
 })

apps/editor/src/utils/extract-paths-from-text/extract-paths-from-text.ts

@@ -1,5 +1,9 @@
-export const extract_paths_from_text = (text: string): string[] => {
+export const extract_paths_from_text = (
+  text: string,
+  workspace_files: string[]
+): string[] => {
   const found_paths = new Set<string>()
+  const workspace_files_set = new Set(workspace_files)
 
   const inline_matches = text.match(/`([^`]+)`/g)
   if (inline_matches) {
@@ -27,12 +31,10 @@ export const extract_paths_from_text = (text: string): string[] => {
     if (word) {
       const cleaned = word.trim().replace(/[.?!]+$/, '')
       found_paths.add(cleaned)
-
-      if (cleaned.startsWith('./')) {
-        found_paths.add(cleaned.substring(2))
-      }
     }
   }
 
   return Array.from(found_paths)
+    .filter((p) => workspace_files_set.has(p))
+    .sort((a, b) => text.indexOf(a) - text.indexOf(b))
 }