Enhance clipboard parsing to correctly extract file paths from XML tags, including unquoted paths and diff content embedded within XML structures

Author: robertpiosik

packages/vscode/package.json

@@ -2,7 +2,7 @@
   "name": "gemini-coder",
   "displayName": "Code Web Chat",
   "description": "Superfast AI coding (CWC)",
-  "version": "1.623.0",
+  "version": "1.624.0",
   "scripts": {
     "build": "npx vsce package --no-dependencies",
     "vscode:prepublish": "rimraf out && npm run compile",
@@ -578,9 +578,9 @@
           "scope": "resource",
           "description": "Style of generated code blocks in \"Edit context\" mode.",
           "default": {
-            "whole": "Whenever showing a new or updated file: provide brief explanations, then print full path in a heading, and a markdown code block with its full contents—as I have a disability which means I can't type and need to be able to just copy and paste.",
-            "truncated": "Whenever showing a new or updated file: provide brief explanations, then print full path in a heading, and a markdown code block with its truncated contents—as a space saving measure (unchanged parts should be hidden behind ellipsis comments, e.g. \"// ...\").",
-            "diff": "Whenever showing a new or updated file: provide brief explanations, then print unified diff within markdown code block."
+            "whole": "Whenever showing a new or updated file, provide brief explanation, then print path in a markdown heading (e.g. ### File: `src/index.ts`), followed by a markdown code block with full contents—as I have a disability which means I can't type and need to be able to just copy and paste.",
+            "truncated": "Whenever showing a new or updated file, provide brief explanation, then print path in a markdown heading (e.g. ### File: `src/index.ts`), followed by a markdown code block with truncated contents—as a space saving measure (unchanged parts should be hidden behind ellipsis comments, e.g. \"// ...\").",
+            "diff": "Whenever showing a new or updated file, provide brief explanation and print the Unified Diff within a markdown code block. Example: \n[brief explanation]\n```diff\n--- a/file.py\n+++ b/file.py\n@@ -2,1 +2,1 @@\n-old_value = 10\n+new_value = 20\n```"
           },
           "properties": {
             "whole": {

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

@@ -618,6 +618,26 @@ describe('clipboard-parser', () => {
       })
     })
 
+    it('parses file path from a code block preceding a content code block without blank line', () => {
+      const test_case = 'path-above-code-block-enclosed-in-code-block-2'
+      const text = load_test_case_file(test_case, `${test_case}.txt`)
+      const result = parse_multiple_files({
+        response: text,
+        is_single_root_folder_workspace: true
+      })
+
+      expect(result).toHaveLength(2)
+      expect(result[0]).toMatchObject({
+        type: 'text',
+        content: load_test_case_file(test_case, '1-text.txt')
+      })
+      expect(result[1]).toMatchObject({
+        type: 'file',
+        file_path: 'src/hello-world.ts',
+        content: load_test_case_file(test_case, '2-file.txt')
+      })
+    })
+
     it('extracts file path from backticks when there are no intermediate lines before code block', () => {
       const test_case = 'path-in-backticks-no-intermediate-empty-lines'
       const text = load_test_case_file(test_case, `${test_case}.txt`)
@@ -1527,5 +1547,53 @@ describe('clipboard-parser', () => {
         content: load_test_case_file(test_case, '1-file.txt')
       })
     })
+
+    it('parses diff within XML tags', () => {
+      const test_case = 'diff-within-xml-tags'
+      const text = load_test_case_file(test_case, `${test_case}.txt`)
+      const result = parse_response({
+        response: text,
+        is_single_root_folder_workspace: true
+      })
+
+      expect(result).toHaveLength(1)
+      expect(result[0]).toMatchObject({
+        type: 'diff',
+        file_path: 'src/index.ts',
+        content: load_test_case_file(test_case, '1-file.txt')
+      })
+    })
+
+    it('parses diff within XML tags with CDATA', () => {
+      const test_case = 'diff-within-xml-tags-with-cdata'
+      const text = load_test_case_file(test_case, `${test_case}.txt`)
+      const result = parse_response({
+        response: text,
+        is_single_root_folder_workspace: true
+      })
+
+      expect(result).toHaveLength(1)
+      expect(result[0]).toMatchObject({
+        type: 'diff',
+        file_path: 'src/index.ts',
+        content: load_test_case_file(test_case, '1-file.txt')
+      })
+    })
+
+    it('parses diff within XML tags with unquoted path', () => {
+      const test_case = 'diff-within-xml-tags-unquoted-path'
+      const text = load_test_case_file(test_case, `${test_case}.txt`)
+      const result = parse_response({
+        response: text,
+        is_single_root_folder_workspace: true
+      })
+
+      expect(result).toHaveLength(1)
+      expect(result[0]).toMatchObject({
+        type: 'diff',
+        file_path: 'src/index.ts',
+        content: load_test_case_file(test_case, '1-file.txt')
+      })
+    })
   })
 })

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

@@ -336,7 +336,7 @@ const convert_code_block_to_new_file_diff = (params: {
   // Regex to find file path. It can be commented or not.
   // It handles paths with slashes, backslashes, dots, alphanumerics, underscores, and hyphens.
   const path_regex = /(?:(?:\/\/|#|--|<!--)\s*)?([\w./\\-]+)/
-  const xml_path_regex = /<(\w+)\s+path=["']([^"']+)["']/
+  const xml_path_regex = /<([\w-]+)\s+path=(?:["']([^"']+)["']|([^>\s]+))/
 
   let file_path: string | undefined = params.file_path_hint
   let path_line_index = -1
@@ -346,9 +346,9 @@ const convert_code_block_to_new_file_diff = (params: {
   const first_line = params.lines.length > 0 ? params.lines[0].trim() : ''
   const xml_match = first_line.match(xml_path_regex)
 
-  if (xml_match && xml_match[1] && xml_match[2]) {
+  if (xml_match && xml_match[1] && (xml_match[2] || xml_match[3])) {
     const xml_tag_name = xml_match[1]
-    const potential_path = normalize_path(xml_match[2])
+    const potential_path = normalize_path(xml_match[2] || xml_match[3])
     if (potential_path.endsWith('/')) {
       // This is a directory, not a file.
       return null
@@ -446,8 +446,11 @@ const convert_code_block_to_new_file_diff = (params: {
       raw_file_path: file_path,
       is_single_root_folder_workspace: params.is_single_root
     })
+
+    const is_new_file = first_real_content_line.trim().startsWith('@@ -0,0')
+
     const patch_content = [
-      `--- a/${relative_path}`,
+      is_new_file ? '--- /dev/null' : `--- a/${relative_path}`,
       `+++ b/${relative_path}`,
       ...content_lines
     ].join('\n')
@@ -557,7 +560,7 @@ const extract_all_code_block_patches = (params: {
           code_blocks[i - 1].start + 1,
           code_blocks[i - 1].end
         )
-        const xml_path_regex = /<file\s+path=["']([^"']+)["']([^>]*)>/
+        const xml_path_regex = /<[\w-]+\s+path=["']([^"']+)["']([^>]*)>/
         const non_empty_lines = prev_block_lines.filter((l) => l.trim() != '')
         if (non_empty_lines.length === 1) {
           const match = non_empty_lines[0].match(xml_path_regex)
@@ -655,7 +658,7 @@ const extract_all_code_block_patches = (params: {
       if (i < code_blocks.length - 1) {
         const next_block = code_blocks[i + 1]
         if (next_block.type === 'diff' || next_block.type === 'patch') {
-          const xml_path_regex = /<file\s+path=["']([^"']+)["']/
+          const xml_path_regex = /<[\w-]+\s+path=["']([^"']+)["']/
           const non_empty_lines = block_lines.filter((l) => l.trim() !== '')
           if (non_empty_lines.length === 1) {
             const match = non_empty_lines[0].match(xml_path_regex)
@@ -826,6 +829,28 @@ export const extract_diffs = (params: {
       is_single_root: params.is_single_root
     })
   } else {
+    const xml_file_tag_start_regex = /^\s*<([\w-]+)\s+path=(?:["']([^"']+)["']|([^>\s]+))/
+    const start_match =
+      lines.length > 0 ? lines[0].trim().match(xml_file_tag_start_regex) : null
+
+    if (start_match) {
+      const tag_name = start_match[1]
+      const xml_file_tag_end_regex = new RegExp(`^\\s*<\\/${tag_name}>`)
+      if (
+        lines.length >= 2 &&
+        xml_file_tag_end_regex.test(lines[lines.length - 1].trim()) &&
+        lines.some((l) => l.trim().startsWith('@@'))
+      ) {
+        const result = convert_code_block_to_new_file_diff({
+          lines: lines,
+          is_single_root: params.is_single_root
+        })
+        if (result) {
+          return [result]
+        }
+      }
+    }
+
     return parse_multiple_raw_patches({
       all_lines: lines,
       is_single_root: params.is_single_root

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

@@ -115,6 +115,7 @@ export const parse_multiple_files = (params: {
   let is_markdown_container_block = false
   let current_language = ''
   let current_xml_tag: string | null = null
+  let current_file_name_was_comment = false
 
   const lines = params.response.split('\n')
 
@@ -199,6 +200,7 @@ export const parse_multiple_files = (params: {
           if (workspace_name) {
             current_workspace_name = workspace_name
           }
+          current_file_name_was_comment = false
         }
         last_seen_file_path_comment = null
         current_content = ''
@@ -417,6 +419,25 @@ export const parse_multiple_files = (params: {
       const trimmed_line = line.trim()
 
       if (trimmed_line.startsWith('```') && trimmed_line !== '```') {
+        if (
+          backtick_nesting_level === 1 &&
+          current_file_name &&
+          current_content.trim() === '' &&
+          !current_file_name_was_comment &&
+          !xml_file_mode
+        ) {
+          state = 'TEXT'
+          last_seen_file_path_comment = current_file_name
+          current_file_name = ''
+          current_content = ''
+          is_first_content_line = false
+          current_workspace_name = undefined
+          xml_file_mode = false
+          in_cdata = false
+          i--
+          continue
+        }
+
         if (
           backtick_nesting_level == 1 &&
           (current_language == 'markdown' || current_language == 'md')
@@ -474,7 +495,8 @@ export const parse_multiple_files = (params: {
           backtick_nesting_level === 1 &&
           current_file_name &&
           i + 1 < lines.length &&
-          lines[i + 1].trim() !== ''
+          lines[i + 1].trim() !== '' &&
+          !lines[i + 1].trim().startsWith('```')
         ) {
           should_close = false
         }
@@ -557,6 +579,7 @@ export const parse_multiple_files = (params: {
             xml_file_mode = true
             current_xml_tag = xml_info.tagName
             is_first_content_line = false
+            current_file_name_was_comment = true
             continue
           }
         }
@@ -594,6 +617,7 @@ export const parse_multiple_files = (params: {
             }
 
             is_first_content_line = false
+            current_file_name_was_comment = true
             continue
           }
 
@@ -650,6 +674,7 @@ export const parse_multiple_files = (params: {
                   params.is_single_root_folder_workspace
               })
             current_file_name = relative_path
+            current_file_name_was_comment = is_comment
             if (
               backtick_nesting_level > 1 &&
               (current_language == 'markdown' || current_language == 'md')

packages/vscode/src/commands/apply-chat-response-command/utils/clipboard-parser/test-cases/diff-within-xml-tags-unquoted-path/1-file.txt

@@ -0,0 +1,4 @@
+--- /dev/null
++++ b/src/index.ts
+@@ -0,0 +1,1 @@
++console.log('hello')

packages/vscode/src/commands/apply-chat-response-command/utils/clipboard-parser/test-cases/diff-within-xml-tags-unquoted-path/diff-within-xml-tags-unquoted-path.txt

@@ -0,0 +1,4 @@
+<file path=src/index.ts>
+@@ -0,0 +1,1 @@
++console.log('hello')
+</file>

packages/vscode/src/commands/apply-chat-response-command/utils/clipboard-parser/test-cases/diff-within-xml-tags-with-cdata/1-file.txt

@@ -0,0 +1,4 @@
+--- /dev/null
++++ b/src/index.ts
+@@ -0,0 +1,1 @@
++console.log('hello')

packages/vscode/src/commands/apply-chat-response-command/utils/clipboard-parser/test-cases/diff-within-xml-tags-with-cdata/diff-within-xml-tags-with-cdata.txt

@@ -0,0 +1,6 @@
+<file path="src/index.ts">
+<![CDATA[
+@@ -0,0 +1,1 @@
++console.log('hello')
+]]>
+</file>

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

@@ -0,0 +1,4 @@
+--- /dev/null
++++ b/src/index.ts
+@@ -0,0 +1,1 @@
++console.log('hello')

packages/vscode/src/commands/apply-chat-response-command/utils/clipboard-parser/test-cases/diff-within-xml-tags/diff-within-xml-tags.txt

@@ -0,0 +1,4 @@
+<file path="src/index.ts">
+@@ -0,0 +1,1 @@
++console.log('hello')
+</file>