Support parsing file content enclosed in non-standard XML tags e.g., addfile path="..." within chat responses

Author: robertpiosik

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

@@ -80,6 +80,21 @@ describe('clipboard-parser', () => {
       })
     })
 
+    it('parses file when using file-xml format with a non-standard tag', () => {
+      const test_case = 'file-xml-non-standard-tag'
+      const text = load_test_case_file(test_case, `${test_case}.txt`)
+      const result = parse_multiple_files({
+        response: text,
+        is_single_root_folder_workspace: true
+      })
+
+      expect(result).toHaveLength(1)
+      expect(result[0]).toMatchObject({
+        file_path: 'src/index.ts',
+        content: load_test_case_file(test_case, '1-file.txt')
+      })
+    })
+
     it('parses file when CDATA is present inside a markdown code block with a comment filename', () => {
       const test_case = 'cdata-inside-code-block'
       const text = load_test_case_file(test_case, `${test_case}.txt`)

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

@@ -167,7 +167,7 @@ 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 = /<file\s+path=["']([^"']+)["']/
+  const xml_path_regex = /<(\w+)\s+path=["']([^"']+)["']/
 
   let file_path: string | undefined = params.file_path_hint
   let path_line_index = -1
@@ -177,20 +177,21 @@ 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]) {
-    const potential_path = xml_match[1].replace(/\\/g, '/')
+  if (xml_match && xml_match[1] && xml_match[2]) {
+    const xml_tag_name = xml_match[1]
+    const potential_path = xml_match[2].replace(/\\/g, '/')
     if (potential_path.endsWith('/')) {
       // This is a directory, not a file.
       return null
     }
     file_path = potential_path
 
     const file_tag_start = params.lines.findIndex((l) =>
-      l.trim().startsWith('<file')
+      l.trim().startsWith(`<${xml_tag_name}`)
     )
     let file_tag_end = -1
     for (let i = params.lines.length - 1; i >= 0; i--) {
-      if (params.lines[i].trim().startsWith('</file>')) {
+      if (params.lines[i].trim().startsWith(`</${xml_tag_name}>`)) {
         file_tag_end = i
         break
       }

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

@@ -7,9 +7,14 @@ import {
 } from '../clipboard-parser'
 import { parse_file_content_only } from './file-content-only-parser'
 
-const extract_file_path_from_xml = (line: string): string | null => {
-  const match = line.match(/<file\s+path=["']([^"']+)["']/)
-  return match ? match[1] : null
+const extract_file_path_from_xml = (
+  line: string
+): { tagName: string; path: string } | null => {
+  const match = line.match(/<(\w+)\s+path=["']([^"']+)["']/)
+  if (match && match[1] && match[2]) {
+    return { tagName: match[1], path: match[2] }
+  }
+  return null
 }
 
 export const parse_multiple_files = (params: {
@@ -37,6 +42,7 @@ export const parse_multiple_files = (params: {
   let last_seen_file_path_comment: string | null = null
   let is_markdown_container_block = false
   let current_language = ''
+  let current_xml_tag: string | null = null
 
   const lines = params.response.split('\n')
 
@@ -136,18 +142,19 @@ export const parse_multiple_files = (params: {
         in_cdata = false
         is_markdown_container_block = false
         continue
-      } else if (line.trim().startsWith('<file')) {
-        const extracted_filename = extract_file_path_from_xml(line)
-        if (extracted_filename) {
+      } else if (line.trim().startsWith('<')) {
+        const xml_info = extract_file_path_from_xml(line)
+        if (xml_info) {
           if (current_text_block.trim()) {
             results.push({ type: 'text', content: current_text_block.trim() })
           }
           current_text_block = ''
 
           state = 'CONTENT'
+          current_xml_tag = xml_info.tagName
           top_level_xml_file_mode = true
           const { workspace_name, relative_path } = extract_workspace_and_path({
-            raw_file_path: extracted_filename,
+            raw_file_path: xml_info.path,
             is_single_root_folder_workspace:
               params.is_single_root_folder_workspace
           })
@@ -228,7 +235,7 @@ export const parse_multiple_files = (params: {
       }
     } else if (state == 'CONTENT') {
       if (top_level_xml_file_mode) {
-        if (line.trim().startsWith('</file>')) {
+        if (line.trim().startsWith(`</${current_xml_tag}>`)) {
           let final_content = current_content.trim()
           const content_lines = final_content.split('\n')
           if (
@@ -265,6 +272,7 @@ export const parse_multiple_files = (params: {
           current_workspace_name = undefined
           top_level_xml_file_mode = false
           in_cdata = false
+          current_xml_tag = null
         } else if (line.trim().startsWith('<![CDATA[')) {
           in_cdata = true
         } else if (in_cdata && line.trim().includes(']]>')) {
@@ -386,12 +394,12 @@ export const parse_multiple_files = (params: {
           current_text_block = line.substring(last_backticks_index + 3)
         }
       } else {
-        if (is_first_content_line && line.trim().startsWith('<file')) {
-          const extracted_filename = extract_file_path_from_xml(line)
-          if (extracted_filename) {
+        if (is_first_content_line && line.trim().startsWith('<')) {
+          const xml_info = extract_file_path_from_xml(line)
+          if (xml_info) {
             const { workspace_name, relative_path } =
               extract_workspace_and_path({
-                raw_file_path: extracted_filename,
+                raw_file_path: xml_info.path,
                 is_single_root_folder_workspace:
                   params.is_single_root_folder_workspace
               })
@@ -400,6 +408,7 @@ export const parse_multiple_files = (params: {
               current_workspace_name = workspace_name
             }
             xml_file_mode = true
+            current_xml_tag = xml_info.tagName
             is_first_content_line = false
             continue
           }
@@ -531,7 +540,11 @@ export const parse_multiple_files = (params: {
           continue
         }
 
-        if (xml_file_mode && !in_cdata && line.trim() == '</file>') {
+        if (
+          xml_file_mode &&
+          !in_cdata &&
+          line.trim() == `</${current_xml_tag}>`
+        ) {
           continue
         }
 

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

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

packages/vscode/src/commands/apply-chat-response-command/utils/clipboard-parser/test-cases/file-xml-non-standard-tag/file-xml-non-standard-tag.txt

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