feat: Groovy 脚本解析引擎，脚本与配置一同管理，提交时支持上传 script 字段

par-api/src/main/java/com/par/api/controller/AdapterController.java

@@ -25,6 +25,7 @@ import java.util.stream.Collectors;
 public class AdapterController {
 
     private final AdapterEngine adapterEngine;
+    private final GroovyScriptEngine groovyEngine;
     private final SiteConfigMapper siteConfigMapper;
     private final AccessResultMapper accessResultMapper;
     private final CachedStorageService storageService;
@@ -137,14 +138,24 @@ public class AdapterController {
             item.put("invalidCount", allResults.stream().filter(r -> url.equals(r.getUrl()) && Boolean.FALSE.equals(r.getIsValid())).count());
             item.put("uncheckedCount", allResults.stream().filter(r -> url.equals(r.getUrl()) && r.getIsValid() == null).count());
 
-            // 解析内容
+            // 解析内容：优先 Groovy 脚本，否则用 Java 引擎
             try {
                 String content = storageService.read(ar.getStoragePath());
                 if (content != null) {
-                    Map<String, Object> parsed = adapterEngine.extract(adapterConfig, content, ar.getContentType(), url);
+                    Map<String, Object> parsed;
+                    if (siteConfig.getScriptStoragePath() != null && !siteConfig.getScriptStoragePath().isBlank()) {
+                        String script = storageService.read(siteConfig.getScriptStoragePath());
+                        if (script != null && !script.isBlank()) {
+                            parsed = groovyEngine.execute(script, content, url);
+                        } else {
+                            parsed = adapterEngine.extract(adapterConfig, content, ar.getContentType(), url);
+                        }
+                    } else {
+                        parsed = adapterEngine.extract(adapterConfig, content, ar.getContentType(), url);
+                    }
                     item.put("parsed", parsed);
                 } else {
-                    item.put("parsed", Map.of("matched", false, "message", "内容为空，可能已从存储中删除"));
+                    item.put("parsed", Map.of("matched", false, "message", "内容为空"));
                 }
             } catch (Exception e) {
                 item.put("parsed", Map.of("matched", false, "message", "解析异常: " + e.getMessage()));

par-api/src/main/java/com/par/api/controller/AdminController.java

@@ -100,6 +100,16 @@ public class AdminController {
             config.setStoragePath(approvedPath);
         }
 
+        // Groovy 脚本也移到 approved
+        if (config.getScriptStoragePath() != null && !config.getScriptStoragePath().isBlank()) {
+            String approvedScriptPath = config.getScriptStoragePath().replace("/pending/", "/approved/");
+            String script = storageService.read(config.getScriptStoragePath());
+            if (script != null) {
+                storageService.store(approvedScriptPath, script);
+                config.setScriptStoragePath(approvedScriptPath);
+            }
+        }
+
         config.setStatus(ConfigStatus.APPROVED);
         config.setReviewerId(reviewerId);
         config.setReviewedAt(LocalDateTime.now());

par-api/src/main/java/com/par/api/controller/ConfigController.java

@@ -129,6 +129,7 @@ public class ConfigController {
         String version = (String) payload.get("version");
         String configJson = (String) payload.get("config");
         String schemaVersion = (String) payload.get("schemaVersion");
+        String groovyScript = (String) payload.get("script");
 
         if (version == null || configJson == null) {
             return ApiResponse.error(400, "version and config are required");
@@ -136,7 +137,7 @@ public class ConfigController {
 
         Long submitterId = (Long) request.getAttribute("accountId");
 
-        SiteConfig config = configService.submitConfig(siteId, version, configJson, schemaVersion, submitterId);
+        SiteConfig config = configService.submitConfig(siteId, version, configJson, schemaVersion, groovyScript, submitterId);
         return ApiResponse.success(configService.toDTO(config));
     }
 }

par-core/pom.xml

@@ -102,5 +102,11 @@
             <groupId>org.jsoup</groupId>
             <artifactId>jsoup</artifactId>
         </dependency>
+
+        <!-- Groovy 脚本引擎 -->
+        <dependency>
+            <groupId>org.apache.groovy</groupId>
+            <artifactId>groovy</artifactId>
+        </dependency>
     </dependencies>
 </project>

par-core/src/main/java/com/par/core/adapter/GroovyScriptEngine.java

@@ -0,0 +1,117 @@
+package com.par.core.adapter;
+
+import com.fasterxml.jackson.databind.ObjectMapper;
+import groovy.lang.Binding;
+import groovy.lang.GroovyShell;
+import lombok.extern.slf4j.Slf4j;
+import org.jsoup.Jsoup;
+import org.springframework.stereotype.Service;
+
+import java.util.*;
+
+/**
+ * Groovy 脚本解析引擎。
+ * 脚本可访问以下内置变量：
+ *   content  - String，页面内容（HTML 或 JSON）
+ *   url      - String，访问的 URL
+ *   jsoup    - 可直接调用 Jsoup.parse() 等
+ *   json     - ObjectMapper，可直接调用 readTree() 等
+ *   utils    - 工具方法集合
+ * 脚本应返回 Map<String, Object> 或 List<Map<String, Object>>
+ */
+@Slf4j
+@Service
+public class GroovyScriptEngine {
+
+    private static final ObjectMapper objectMapper = new ObjectMapper();
+
+    @SuppressWarnings("unchecked")
+    public Map<String, Object> execute(String groovyScript, String content, String url) {
+        Map<String, Object> result = new LinkedHashMap<>();
+        result.put("engine", "groovy");
+        try {
+            Binding binding = new Binding();
+            binding.setVariable("content", content);
+            binding.setVariable("url", url);
+            binding.setVariable("jsoup", new JsoupHelper());
+            binding.setVariable("json", objectMapper);
+            binding.setVariable("utils", new ScriptUtils());
+
+            GroovyShell shell = new GroovyShell(binding);
+            Object scriptResult = shell.evaluate(groovyScript);
+
+            if (scriptResult instanceof Map) {
+                result.put("matched", true);
+                Map<String, Object> m = (Map<String, Object>) scriptResult;
+                result.put("pageName", m.getOrDefault("pageName", "Groovy脚本"));
+                result.put("fieldCount", m.containsKey("fields") ? ((Map<?, ?>) m.get("fields")).size() : 0);
+                // 有 fields map → 展示为一行
+                if (m.containsKey("fields")) {
+                    result.put("extracted", List.of(m.get("fields")));
+                    result.put("rowCount", 1);
+                } else if (m.containsKey("extracted")) {
+                    result.put("extracted", m.get("extracted"));
+                    result.put("rowCount", m.getOrDefault("rowCount", ((List<?>) m.get("extracted")).size()));
+                } else {
+                    result.put("extracted", List.of(m));
+                    result.put("rowCount", 1);
+                }
+                result.put("message", m.getOrDefault("message", "解析成功"));
+            } else if (scriptResult instanceof List) {
+                result.put("matched", true);
+                result.put("pageName", "Groovy脚本");
+                result.put("extracted", (List<?>) scriptResult);
+                result.put("rowCount", ((List<?>) scriptResult).size());
+                result.put("fieldCount", !((List<?>) scriptResult).isEmpty() && ((List<?>) scriptResult).get(0) instanceof Map
+                        ? ((Map<?, ?>) ((List<?>) scriptResult).get(0)).size() : 0);
+                result.put("message", "共 " + ((List<?>) scriptResult).size() + " 行");
+            } else if (scriptResult != null) {
+                result.put("matched", true);
+                result.put("pageName", "Groovy脚本");
+                result.put("extracted", List.of(Map.of("result", scriptResult)));
+                result.put("rowCount", 1);
+                result.put("message", String.valueOf(scriptResult));
+            } else {
+                result.put("matched", true);
+                result.put("message", "脚本返回 null");
+                result.put("extracted", List.of());
+                result.put("rowCount", 0);
+            }
+        } catch (Exception e) {
+            result.put("matched", false);
+            result.put("message", "Groovy 脚本执行异常: " + e.getMessage());
+            log.error("Groovy script error for url={}", url, e);
+        }
+        return result;
+    }
+
+    /** Jsoup 辅助类 */
+    public static class JsoupHelper {
+        public Object parse(String html) { return Jsoup.parse(html); }
+        public Object parse(String html, String baseUri) { return Jsoup.parse(html, baseUri); }
+    }
+
+    /** 脚本工具方法 */
+    public static class ScriptUtils {
+        public String text(Object node) {
+            try { return (String) node.getClass().getMethod("text").invoke(node); }
+            catch (Exception e) { return null; }
+        }
+        public String attr(Object node, String attr) {
+            try { return (String) node.getClass().getMethod("attr", String.class).invoke(node, attr); }
+            catch (Exception e) { return null; }
+        }
+        public Object select(Object node, String cssQuery) {
+            try { return node.getClass().getMethod("select", String.class).invoke(node, cssQuery); }
+            catch (Exception e) { return null; }
+        }
+        public Object selectFirst(Object node, String cssQuery) {
+            try {
+                Object elements = select(node, cssQuery);
+                if (elements instanceof List && !((List<?>) elements).isEmpty())
+                    return ((List<?>) elements).get(0);
+                return null;
+            } catch (Exception e) { return null; }
+        }
+    }
+}

par-core/src/main/java/com/par/core/entity/SiteConfig.java

@@ -28,6 +28,9 @@ public class SiteConfig {
     @TableField("storage_path")
     private String storagePath;
 
+    @TableField("script_storage_path")
+    private String scriptStoragePath;
+
     @TableField("file_size")
     private Long fileSize;
 

par-core/src/main/java/com/par/core/service/ConfigService.java

@@ -31,10 +31,10 @@ public interface ConfigService {
     SiteConfig getConfig(String siteId, String version);
 
     /**
-     * 提交新配置（需审核）
+     * 提交新配置（需审核），可选 Groovy 脚本
      */
     SiteConfig submitConfig(String siteId, String version, String configJson,
-                            String schemaVersion, Long submitterId);
+                            String schemaVersion, String groovyScript, Long submitterId);
 
     /**
      * 转换为 DTO

par-core/src/main/java/com/par/core/service/impl/ConfigServiceImpl.java

@@ -55,7 +55,7 @@ public class ConfigServiceImpl implements ConfigService {
 
     @Override
     public SiteConfig submitConfig(String siteId, String version, String configJson,
-                                   String schemaVersion, Long submitterId) {
+                                   String schemaVersion, String groovyScript, Long submitterId) {
         String configHash = HmacUtil.sha256(configJson);
         String storagePath = String.format("configs/%s/pending/%s.json", siteId, configHash.substring(0, 8));
 
@@ -70,7 +70,7 @@ public class ConfigServiceImpl implements ConfigService {
             return sameHash;
         }
 
-        // 存储到抽象存储层
+        // 存储配置
         try {
             storageService.store(storagePath, configJson);
         } catch (Exception e) {
@@ -78,11 +78,19 @@ public class ConfigServiceImpl implements ConfigService {
             throw new RuntimeException("存储配置文件失败: " + e.getMessage(), e);
         }
 
+        // 存储 Groovy 脚本（可选）
+        String scriptPath = null;
+        if (groovyScript != null && !groovyScript.isBlank()) {
+            scriptPath = String.format("configs/%s/pending/%s.groovy", siteId, configHash.substring(0, 8));
+            storageService.store(scriptPath, groovyScript);
+        }
+
         SiteConfig config = new SiteConfig();
         config.setSiteId(siteId);
-        config.setVersion(version);  // 客户端传入的版本，仅供参考；批准时会被自动分配的新版本号覆盖
+        config.setVersion(version);
         config.setConfigHash(configHash);
         config.setStoragePath(storagePath);
+        config.setScriptStoragePath(scriptPath);
         config.setFileSize((long) configJson.getBytes(StandardCharsets.UTF_8).length);
         config.setSchemaVersion(schemaVersion != null ? schemaVersion : "1.0.0");
         config.setStatus(ConfigStatus.PENDING);
@@ -90,7 +98,7 @@ public class ConfigServiceImpl implements ConfigService {
         config.setIsLatest(false);
 
         siteConfigMapper.insert(config);
-        log.info("Config submitted: siteId={}, version={}, id={}, hash={}", siteId, version, config.getId(), configHash);
+        log.info("Config submitted: siteId={}, version={}, id={}, hash={}, hasScript={}", siteId, version, config.getId(), configHash, scriptPath != null);
         return config;
     }
 

par-core/src/main/resources/db/migration/V14__add_script_to_site_configs.sql

@@ -0,0 +1,2 @@
+-- PAR Flyway V14: site_configs 增加 script_storage_path 字段存储 Groovy 脚本路径
+ALTER TABLE site_configs ADD COLUMN script_storage_path VARCHAR(500) COMMENT 'Groovy 脚本存储路径（七牛云/本地）';

pom.xml

@@ -124,6 +124,13 @@
                 <artifactId>jsoup</artifactId>
                 <version>1.18.1</version>
             </dependency>
+
+            <!-- Groovy 脚本引擎 -->
+            <dependency>
+                <groupId>org.apache.groovy</groupId>
+                <artifactId>groovy</artifactId>
+                <version>4.0.22</version>
+            </dependency>
         </dependencies>
     </dependencyManagement>