fix(parser): handle indentation in code blocks #259

- Update regex to account for leading whitespace in code blocks.
- Add support for indentation in code block parsing, ensuring proper handling of nested or indented code.
- Add test case for code blocks within lists with varying indentation.

Author: phodal

core/src/main/kotlin/cc/unitmesh/devti/observer/TestAgentObserver.kt

@@ -1,7 +1,5 @@
 package cc.unitmesh.devti.observer
 
-import cc.unitmesh.devti.gui.chat.message.ChatActionType
-import cc.unitmesh.devti.gui.sendToChatWindow
 import cc.unitmesh.devti.provider.observer.AgentObserver
 import cc.unitmesh.devti.util.relativePath
 import com.intellij.execution.testframework.sm.runner.SMTRunnerEventsAdapter
@@ -27,8 +25,8 @@ class TestAgentObserver : AgentObserver, Disposable {
     }
 
     private fun sendResult(test: SMTestProxy, project: Project, searchScope: GlobalSearchScope) {
-        val sourceCode = test.getLocation(project, searchScope)
         runInEdt {
+            val sourceCode = test.getLocation(project, searchScope)
             val psiElement = sourceCode?.psiElement
             val language = psiElement?.language?.displayName ?: ""
             val filepath = psiElement?.containingFile?.virtualFile?.relativePath(project) ?: ""

core/src/main/kotlin/cc/unitmesh/devti/util/parser/CodeFence.kt

@@ -16,7 +16,7 @@ class CodeFence(
         val devinEndRegex = Regex("</devin>")
 
         fun parse(content: String): CodeFence {
-            val languageRegex = Regex("```([\\w#+ ]*)")
+            val languageRegex = Regex("\\s*```([\\w#+ ]*)")
             val lines = content.lines()
 
             val startMatch = devinStartRegex.find(content)
@@ -38,21 +38,28 @@ class CodeFence(
             var codeStarted = false
             var codeClosed = false
             var languageId: String? = null
+            var codeIndentation = ""
             val codeBuilder = StringBuilder()
 
             for (line in lines) {
                 if (!codeStarted) {
-                    val matchResult: MatchResult? = languageRegex.find(line.trimStart())
+                    val trimmedLine = line.trimStart()
+                    val matchResult: MatchResult? = languageRegex.find(trimmedLine)
                     if (matchResult != null) {
-                        val substring = matchResult.groups[1]?.value
+                        // Store the indentation to match it when looking for the closing fence
+                        codeIndentation = line.substring(0, line.length - trimmedLine.length)
+                        val substring = matchResult.groups[1]?.value?.trim()
                         languageId = substring
                         codeStarted = true
                     }
-                } else if (line.startsWith("```")) {
-                    codeClosed = true
-                    break
                 } else {
-                    codeBuilder.append(line).append("\n")
+                    val trimmedLine = line.trimStart()
+                    if (trimmedLine == "```") {
+                        codeClosed = true
+                        break
+                    } else {
+                        codeBuilder.append(line).append("\n")
+                    }
                 }
             }
 
@@ -112,7 +119,7 @@ class CodeFence(
         }
 
         val devinRegexBlock = Regex("(?<=^|\\n)```devin\\n([\\s\\S]*?)\\n```\\n")
-        val normalCodeBlock = Regex("```([\\w#+ ]*)\\n")
+        val normalCodeBlock = Regex("\\s*```([\\w#+ ]*)\\n")
 
         fun preProcessDevinBlock(content: String): String {
             var currentContent = content
@@ -135,17 +142,21 @@ class CodeFence(
         }
 
         private fun parseMarkdownContent(content: String, codeFences: MutableList<CodeFence>) {
-            val languageRegex = Regex("```([\\w#+ ]*)")
+            val languageRegex = Regex("\\s*```([\\w#+ ]*)")
             val lines = content.lines()
 
             var codeStarted = false
             var languageId: String? = null
             val codeBuilder = StringBuilder()
             val textBuilder = StringBuilder()
+            var codeIndentation = ""
 
-            for (line in lines) {
+            for (i in lines.indices) {
+                val line = lines[i]
                 if (!codeStarted) {
-                    val matchResult = languageRegex.find(line.trimStart())
+                    // Check for code block start with any indentation
+                    val trimmedLine = line.trimStart()
+                    val matchResult = languageRegex.find(trimmedLine)
                     if (matchResult != null) {
                         if (textBuilder.isNotEmpty()) {
                             val textBlock = CodeFence(
@@ -156,13 +167,20 @@ class CodeFence(
                             textBuilder.clear()
                         }
 
-                        languageId = matchResult.groups[1]?.value
+                        // Store the indentation to match it when looking for the closing fence
+                        codeIndentation = line.substring(0, line.length - trimmedLine.length)
+                        languageId = matchResult.groups[1]?.value?.trim()
                         codeStarted = true
                     } else {
                         textBuilder.append(line).append("\n")
                     }
                 } else {
-                    if (line.startsWith("```")) {
+                    // Check if this line contains the closing fence with the same or similar indentation
+                    val trimmedLine = line.trimStart()
+                    
+                    // Allow for some flexibility in indentation for the closing fence
+                    // This helps with numbered lists where indentation might vary slightly
+                    if (trimmedLine == "```") {
                         val codeContent = codeBuilder.trim().toString()
                         val codeFence = CodeFence(
                             findLanguage(languageId ?: "markdown"),
@@ -176,6 +194,7 @@ class CodeFence(
                         codeBuilder.clear()
                         codeStarted = false
                         languageId = null
+                        codeIndentation = ""
                     } else {
                         codeBuilder.append(line).append("\n")
                     }

core/src/main/resources/icons/send.svg

@@ -333,4 +333,47 @@ npm run dev
         )
         assertTrue(codeFences[0].isComplete)
     }
+
+    fun testShouldHandleCodeUnderListWithErrorIndent() {
+        val content = """
+1. **通过代码设置属性**：
+   你可以在你的插件代码或项目中的某个地方直接使用这段代码来设置属性。例如：
+
+   ```java
+   PropertiesComponent.getInstance().setValue("matterhorn.junie.untestedIdeAccepted", "true");
+   ```
+
+   这会将 `matterhorn.junie.untestedIdeAccepted` 的值设置为 `"true"`。
+
+2. **通过 IDEA 的配置文件手动设置**：
+   如果你不想通过代码来设置这个属性，你可以手动编辑 IDEA 的配置文件。IDEA 的全局属性通常存储在 `idea.properties` 文件中，或者在某些情况下存储在 `options` 目录下的 XML 文件中。
+```
+        """.trimMargin()
+
+        val codeFences = CodeFence.parseAll(content)
+        assertEquals(codeFences.size, 3)
+        assertEquals(
+            codeFences[0].text, """
+            |1. **通过代码设置属性**：
+            |   你可以在你的插件代码或项目中的某个地方直接使用这段代码来设置属性。例如：
+        """.trimMargin()
+        )
+        assertTrue(codeFences[0].isComplete)
+
+        assertEquals(
+            codeFences[1].text, """
+            |PropertiesComponent.getInstance().setValue("matterhorn.junie.untestedIdeAccepted", "true");
+        """.trimMargin()
+        )
+        assertEquals(codeFences[1].originLanguage, "java")
+
+        assertEquals(
+            codeFences[2].text, """
+            |这会将 `matterhorn.junie.untestedIdeAccepted` 的值设置为 `"true"`。
+            |
+            |2. **通过 IDEA 的配置文件手动设置**：
+            |   如果你不想通过代码来设置这个属性，你可以手动编辑 IDEA 的配置文件。IDEA 的全局属性通常存储在 `idea.properties` 文件中，或者在某些情况下存储在 `options` 目录下的 XML 文件中。
+        """.trimMargin()
+        )
+    }
 }