Refine clipboard parsing logic to correctly handle file paths specified in XML tags, markdown headings, or backticks immediately preceding code blocks, and remove redundant test cases

Author: robertpiosik

packages/vscode/src/commands/apply-chat-response-command/utils/clipboard-parser/clipboard-parser.spec.ts

@@ -65,6 +65,22 @@ describe('clipboard-parser', () => {
       })
     })
 
+    it('parses file when using code-block-xml format within a markdown code block', () => {
+      const test_case = 'code-block-xml'
+      const text = load_test_case_file(test_case, `${test_case}.txt`)
+      const result = parse_multiple_files({
+        response: text,
+        is_single_root_folder_workspace: true
+      })
+
+      expect(result).toHaveLength(1)
+      expect(result[0]).toMatchObject({
+        type: 'file',
+        file_path: 'src/index.ts',
+        content: load_test_case_file(test_case, '1-file.txt')
+      })
+    })
+
     it('parses file when using file-xml format with CDATA outside a markdown code block', () => {
       const test_case = 'file-xml-with-cdata'
       const text = load_test_case_file(test_case, `${test_case}.txt`)
@@ -574,22 +590,6 @@ describe('clipboard-parser', () => {
       })
     })
 
-    it('extracts file path from backticks when there are only empty intermediate lines before code block', () => {
-      const test_case = 'path-in-backticks-with-intermediate-empty-lines'
-      const text = load_test_case_file(test_case, `${test_case}.txt`)
-      const result = parse_multiple_files({
-        response: text,
-        is_single_root_folder_workspace: true
-      })
-
-      expect(result).toHaveLength(1)
-      expect(result[0]).toMatchObject({
-        type: 'file',
-        file_path: 'src/main.py',
-        content: load_test_case_file(test_case, '1-file.txt')
-      })
-    })
-
     it('does not extract file path from backticks when there is intermediate text before code block', () => {
       const test_case = 'path-in-backticks-with-intermediate-text'
       const text = load_test_case_file(test_case, `${test_case}.txt`)
@@ -1420,6 +1420,29 @@ describe('clipboard-parser', () => {
       })
     })
 
+    it('parses diff format with file path specified in plain text above and inside a markdown code block', () => {
+      const test_case = 'diff-markdown-repeated-path-above'
+      const text = load_test_case_file(
+        test_case,
+        'diff-markdown-repeated-path-above.txt'
+      )
+      const result = parse_response({
+        response: text,
+        is_single_root_folder_workspace: true
+      })
+
+      expect(result).toHaveLength(2)
+      expect(result[0]).toMatchObject({
+        type: 'text',
+        content: load_test_case_file(test_case, '1-text.txt')
+      })
+      expect(result[1]).toMatchObject({
+        type: 'diff',
+        file_path: 'src/index.ts',
+        content: load_test_case_file(test_case, '2-file.txt')
+      })
+    })
+
     it('parses diff with broken hunk header after git header', () => {
       const test_case = 'diff-git-broken-hunk-header'
       const text = load_test_case_file(test_case, `${test_case}.txt`)

packages/vscode/src/commands/apply-chat-response-command/utils/clipboard-parser/extract-diff-patches/extract-diffs.ts

@@ -507,16 +507,60 @@ const extract_all_code_block_patches = (params: {
       if (patches.length > 0) {
         const last_item = items.length > 0 ? items[items.length - 1] : undefined
         if (last_item && last_item.type == 'text') {
-          let potential_path = last_item.content.trim()
-          if (potential_path.endsWith(':')) {
-            potential_path = potential_path.slice(0, -1).trim()
+          const content_lines = last_item.content.split('\n')
+          let last_non_empty_line_index = -1
+          let last_non_empty_line = ''
+          for (let i = content_lines.length - 1; i >= 0; i--) {
+            if (content_lines[i].trim() !== '') {
+              last_non_empty_line_index = i
+              last_non_empty_line = content_lines[i].trim()
+              break
+            }
           }
-          const { relative_path } = extract_workspace_and_path({
-            raw_file_path: potential_path,
-            is_single_root_folder_workspace: params.is_single_root
-          })
-          if (relative_path == patches[0].file_path) {
-            items.pop()
+
+          if (last_non_empty_line) {
+            let extracted = extract_path_from_line_of_code(last_non_empty_line)
+
+            if (!extracted) {
+              let potential_path = last_non_empty_line
+              if (potential_path.endsWith(':')) {
+                potential_path = potential_path.slice(0, -1).trim()
+              }
+              const backtick_match = potential_path.match(/`([^`]+)`/)
+              if (backtick_match && backtick_match[1]) {
+                potential_path = backtick_match[1]
+              }
+
+              if (
+                potential_path &&
+                (potential_path.includes('/') ||
+                  potential_path.includes('\\') ||
+                  potential_path.includes('.')) &&
+                !potential_path.endsWith('.') &&
+                /^[a-zA-Z0-9_./@-]+$/.test(potential_path)
+              ) {
+                extracted = potential_path
+              }
+            }
+
+            if (extracted) {
+              const { relative_path } = extract_workspace_and_path({
+                raw_file_path: extracted,
+                is_single_root_folder_workspace: params.is_single_root
+              })
+              if (relative_path === patches[0].file_path) {
+                const new_content = content_lines
+                  .slice(0, last_non_empty_line_index)
+                  .join('\n')
+                  .trim()
+
+                if (new_content) {
+                  last_item.content = new_content
+                } else {
+                  items.pop()
+                }
+              }
+            }
           }
         }
       }

packages/vscode/src/commands/apply-chat-response-command/utils/clipboard-parser/parsers/multiple-files-parser.ts

@@ -10,7 +10,7 @@ import { parse_file_content_only } from './file-content-only-parser'
 const extract_file_path_from_xml = (
   line: string
 ): { tagName: string; path: string } | null => {
-  const match = line.match(/<(\w+)\s+path=["']([^"']+)["']/)
+  const match = line.match(/<([\w-]+)\s+path=["']([^"']+)["']/)
   if (match && match[1] && match[2]) {
     return { tagName: match[1], path: match[2] }
   }

packages/vscode/src/commands/apply-chat-response-command/utils/clipboard-parser/test-cases/code-block-xml/1-file.txt

@@ -0,0 +1 @@
+console.log("hello")

packages/vscode/src/commands/apply-chat-response-command/utils/clipboard-parser/test-cases/code-block-xml/code-block-xml.txt

@@ -0,0 +1,5 @@
+```typescript
+<code-block path="src/index.ts">
+console.log("hello")
+</code-block>
+```

packages/vscode/src/commands/apply-chat-response-command/utils/clipboard-parser/test-cases/diff-markdown-repeated-path-above/1-text.txt

@@ -0,0 +1 @@
+Lorem ipsum.

packages/vscode/src/commands/apply-chat-response-command/utils/clipboard-parser/test-cases/diff-markdown-repeated-path-above/2-file.txt

@@ -0,0 +1,6 @@
+--- a/src/index.ts
++++ b/src/index.ts
+@@ -1,3 +1,3 @@
+ console.log("hello")
+-console.log("old message")
++console.log("new message")

packages/vscode/src/commands/apply-chat-response-command/utils/clipboard-parser/test-cases/diff-markdown-repeated-path-above/diff-markdown-repeated-path-above.txt

@@ -0,0 +1,12 @@
+Lorem ipsum.
+
+# `src/index.ts`:
+
+```diff
+--- a/src/index.ts
++++ b/src/index.ts
+@@ -1,3 +1,3 @@
+ console.log("hello")
+-console.log("old message")
++console.log("new message")
+```