feat: Add Java support

--story=1

Author: studyzy

.codei18n/mappings.json

@@ -0,0 +1,124 @@
+package java
+
+import (
+	"context"
+	"os"
+	"strings"
+
+	sitter "github.com/smacker/go-tree-sitter"
+	"github.com/smacker/go-tree-sitter/java"
+	"github.com/studyzy/codei18n/core/domain"
+)
+
+// Adapter 实现 Java 语言的 LanguageAdapter 接口
+// 使用 Tree-sitter 解析 Java 源代码并提取注释及其上下文
+type Adapter struct {
+	parser *sitter.Parser
+}
+
+// NewAdapter 创建一个新的 Java 适配器实例
+// 初始化 Tree-sitter 解析器并加载 Java 语法
+func NewAdapter() *Adapter {
+	p := sitter.NewParser()
+	p.SetLanguage(java.GetLanguage())
+	return &Adapter{parser: p}
+}
+
+// Language 返回语言标识符 ("java")
+func (a *Adapter) Language() string {
+	return "java"
+}
+
+// Parse 解析提供的 Java 源代码并提取注释
+// 返回包含注释文本、位置和上下文符号的 domain.Comment 列表
+// 如果 src 为 nil，则从文件路径读取源代码
+func (a *Adapter) Parse(file string, src []byte) ([]*domain.Comment, error) {
+	if src == nil {
+		data, err := os.ReadFile(file)
+		if err != nil {
+			return nil, err
+		}
+		src = data
+	}
+
+	// Parse the source code using Tree-sitter
+	tree, err := a.parser.ParseCtx(context.Background(), nil, src)
+	if err != nil {
+		return nil, err
+	}
+	defer tree.Close()
+
+	return a.extractComments(tree.RootNode(), src, file)
+}
+
+// extractComments 从语法树中提取注释
+func (a *Adapter) extractComments(root *sitter.Node, src []byte, file string) ([]*domain.Comment, error) {
+	// 首先提取包名
+	packageName := extractPackageName(root, src)
+
+	// 创建查询
+	q, err := sitter.NewQuery([]byte(javaCommentQuery), java.GetLanguage())
+	if err != nil {
+		return nil, err
+	}
+	defer q.Close()
+
+	qc := sitter.NewQueryCursor()
+	defer qc.Close()
+
+	qc.Exec(q, root)
+
+	var comments []*domain.Comment
+
+	for {
+		m, ok := qc.NextMatch()
+		if !ok {
+			break
+		}
+
+		for _, c := range m.Captures {
+			node := c.Node
+			text := node.Content(src)
+
+			// 规范化文本并识别注释类型
+			cType := domain.CommentTypeLine
+			normalized := text
+
+			if strings.HasPrefix(text, "//") {
+				cType = domain.CommentTypeLine
+				normalized = strings.TrimPrefix(text, "//")
+			} else if strings.HasPrefix(text, "/*") {
+				if strings.HasPrefix(text, "/**") {
+					cType = domain.CommentTypeDoc
+					normalized = strings.TrimPrefix(text, "/**")
+				} else {
+					cType = domain.CommentTypeBlock
+					normalized = strings.TrimPrefix(text, "/*")
+				}
+				normalized = strings.TrimSuffix(normalized, "*/")
+			}
+			normalized = strings.TrimSpace(normalized)
+
+			// 解析符号路径
+			symbol := resolveSymbol(node, src, packageName)
+
+			comment := &domain.Comment{
+				File:     file,
+				Language: "java",
+				Symbol:   symbol,
+				Range: domain.TextRange{
+					StartLine: int(node.StartPoint().Row) + 1,
+					StartCol:  int(node.StartPoint().Column) + 1,
+					EndLine:   int(node.EndPoint().Row) + 1,
+					EndCol:    int(node.EndPoint().Column) + 1,
+				},
+				SourceText: text,
+				Type:       cType,
+			}
+
+			comments = append(comments, comment)
+		}
+	}
+
+	return comments, nil
+}

adapters/java/adapter.go

@@ -0,0 +1,235 @@
+package java
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/studyzy/codei18n/core/domain"
+)
+
+func TestAdapter_Parse_Class(t *testing.T) {
+	src := `package com.example;
+
+// 计算器类
+public class Calculator {
+}
+`
+	adapter := NewAdapter()
+	comments, err := adapter.Parse("test.java", []byte(src))
+	assert.NoError(t, err)
+	assert.Len(t, comments, 1)
+
+	assert.Equal(t, "// 计算器类", comments[0].SourceText)
+	assert.Equal(t, "com.example.Calculator", comments[0].Symbol)
+	assert.Equal(t, domain.CommentTypeLine, comments[0].Type)
+}
+
+func TestAdapter_Parse_Method(t *testing.T) {
+	src := `package com.example;
+
+public class Calculator {
+    // 计算两个数的和
+    public int add(int a, int b) {
+        return a + b;
+    }
+}
+`
+	adapter := NewAdapter()
+	comments, err := adapter.Parse("test.java", []byte(src))
+	assert.NoError(t, err)
+	assert.Len(t, comments, 1)
+
+	assert.Equal(t, "// 计算两个数的和", comments[0].SourceText)
+	assert.Equal(t, "com.example.Calculator#add", comments[0].Symbol)
+	assert.Equal(t, domain.CommentTypeLine, comments[0].Type)
+}
+
+func TestAdapter_Parse_Field(t *testing.T) {
+	src := `package com.example;
+
+public class Config {
+    // 默认超时时间（毫秒）
+    private int timeoutMs = 1000;
+}
+`
+	adapter := NewAdapter()
+	comments, err := adapter.Parse("test.java", []byte(src))
+	assert.NoError(t, err)
+	assert.Len(t, comments, 1)
+
+	assert.Equal(t, "// 默认超时时间（毫秒）", comments[0].SourceText)
+	assert.Equal(t, "com.example.Config#timeoutMs", comments[0].Symbol)
+	assert.Equal(t, domain.CommentTypeLine, comments[0].Type)
+}
+
+func TestAdapter_Parse_Interface(t *testing.T) {
+	src := `package com.example;
+
+// 用户接口
+public interface User {
+    String getName();
+}
+`
+	adapter := NewAdapter()
+	comments, err := adapter.Parse("test.java", []byte(src))
+	assert.NoError(t, err)
+	assert.Len(t, comments, 1)
+
+	assert.Equal(t, "// 用户接口", comments[0].SourceText)
+	assert.Equal(t, "com.example.User", comments[0].Symbol)
+	assert.Equal(t, domain.CommentTypeLine, comments[0].Type)
+}
+
+func TestAdapter_Parse_Enum(t *testing.T) {
+	src := `package com.example;
+
+// 账户类型
+public enum AccountType {
+    STANDARD,
+    PREMIUM
+}
+`
+	adapter := NewAdapter()
+	comments, err := adapter.Parse("test.java", []byte(src))
+	assert.NoError(t, err)
+	assert.Len(t, comments, 1)
+
+	assert.Equal(t, "// 账户类型", comments[0].SourceText)
+	assert.Equal(t, "com.example.AccountType", comments[0].Symbol)
+	assert.Equal(t, domain.CommentTypeLine, comments[0].Type)
+}
+
+func TestAdapter_Parse_BlockComment(t *testing.T) {
+	src := `package com.example;
+
+/* 这是一个块注释 */
+public class Test {
+}
+`
+	adapter := NewAdapter()
+	comments, err := adapter.Parse("test.java", []byte(src))
+	assert.NoError(t, err)
+	assert.Len(t, comments, 1)
+
+	assert.Equal(t, "/* 这是一个块注释 */", comments[0].SourceText)
+	assert.Equal(t, "com.example.Test", comments[0].Symbol)
+	assert.Equal(t, domain.CommentTypeBlock, comments[0].Type)
+}
+
+func TestAdapter_Parse_Javadoc(t *testing.T) {
+	src := `package com.example;
+
+/**
+ * 计算器类
+ * 提供基本的数学运算
+ */
+public class Calculator {
+    /**
+     * 计算两个数的和
+     * @param a 第一个数
+     * @param b 第二个数
+     * @return 两数之和
+     */
+    public int add(int a, int b) {
+        return a + b;
+    }
+}
+`
+	adapter := NewAdapter()
+	comments, err := adapter.Parse("test.java", []byte(src))
+	assert.NoError(t, err)
+	assert.Len(t, comments, 2)
+
+	// 类的 Javadoc
+	assert.Contains(t, comments[0].SourceText, "/**")
+	assert.Equal(t, "com.example.Calculator", comments[0].Symbol)
+	assert.Equal(t, domain.CommentTypeDoc, comments[0].Type)
+
+	// 方法的 Javadoc
+	assert.Contains(t, comments[1].SourceText, "/**")
+	assert.Equal(t, "com.example.Calculator#add", comments[1].Symbol)
+	assert.Equal(t, domain.CommentTypeDoc, comments[1].Type)
+}
+
+func TestAdapter_Parse_NoPackage(t *testing.T) {
+	src := `
+// 简单类
+public class Simple {
+}
+`
+	adapter := NewAdapter()
+	comments, err := adapter.Parse("test.java", []byte(src))
+	assert.NoError(t, err)
+	assert.Len(t, comments, 1)
+
+	assert.Equal(t, "// 简单类", comments[0].SourceText)
+	assert.Equal(t, "Simple", comments[0].Symbol)
+	assert.Equal(t, domain.CommentTypeLine, comments[0].Type)
+}
+
+func TestAdapter_Parse_FileLevel(t *testing.T) {
+	src := `// Copyright (c) 2025 Example Corp
+// 本文件包含示例代码
+package com.example;
+
+public class Test {
+}
+`
+	adapter := NewAdapter()
+	comments, err := adapter.Parse("test.java", []byte(src))
+	assert.NoError(t, err)
+	// 文件顶部的注释可能不会附加到 package 声明
+	// 具体行为取决于 Tree-sitter 的解析
+	assert.Greater(t, len(comments), 0)
+}
+
+func TestAdapter_Parse_Range(t *testing.T) {
+	src := `package com.example;
+
+public class Test {
+    // 测试方法
+    public void test() {
+    }
+}
+`
+	adapter := NewAdapter()
+	comments, err := adapter.Parse("test.java", []byte(src))
+	assert.NoError(t, err)
+	assert.Len(t, comments, 1)
+
+	// 验证位置信息
+	comment := comments[0]
+	assert.Greater(t, comment.Range.StartLine, 0)
+	assert.Greater(t, comment.Range.StartCol, 0)
+	assert.Greater(t, comment.Range.EndLine, 0)
+	assert.Greater(t, comment.Range.EndCol, 0)
+	assert.GreaterOrEqual(t, comment.Range.EndLine, comment.Range.StartLine)
+	if comment.Range.EndLine == comment.Range.StartLine {
+		assert.Greater(t, comment.Range.EndCol, comment.Range.StartCol)
+	}
+}
+
+func TestAdapter_Parse_SyntaxError(t *testing.T) {
+	src := `package com.example;
+
+public class Broken {
+    // 未完成的方法定义
+    public void doSomething( {
+}
+`
+	adapter := NewAdapter()
+	comments, err := adapter.Parse("test.java", []byte(src))
+	// Tree-sitter 应该能够容错处理，至少返回注释
+	assert.NoError(t, err)
+	assert.Greater(t, len(comments), 0)
+	
+	// 验证至少能提取到注释
+	found := false
+	for _, c := range comments {
+		if c.SourceText == "// 未完成的方法定义" {
+			found = true
+			break
+		}
+	}
+	assert.True(t, found, "应该能提取到注释，即使代码有语法错误")
+}

adapters/java/adapter_test.go

@@ -0,0 +1,8 @@
+package java
+
+// javaCommentQuery 定义用于提取 Java 注释的 Tree-sitter 查询
+// 匹配行注释 (//)、块注释 (/* */) 和 Javadoc (/** */)
+const javaCommentQuery = `
+(line_comment) @comment
+(block_comment) @comment
+`