添加neo4j服务，修改rag服务调用neo4j

llmengine/db_neo4j.py

@@ -0,0 +1,560 @@
+from typing import Dict, List, Any, Set, Tuple
+from appPublic.jsonConfig import getConfig
+from appPublic.log import debug, error, info
+from py2neo import Graph, Node, Relationship
+import re
+from llmengine.base_db import BaseDBConnection, connection_register
+from threading import Lock
+from scipy.spatial.distance import cosine
+import numpy as np
+import aiohttp
+import asyncio
+import time
+
+# 嵌入缓存
+EMBED_CACHE = {}
+
+class Neo4jConnection(BaseDBConnection):
+    _instance = None
+    _lock = Lock()
+
+    def __new__(cls):
+        with cls._lock:
+            if cls._instance is None:
+                cls._instance = super(Neo4jConnection, cls).__new__(cls)
+                cls._instance._initialized = False
+            return cls._instance
+
+    def __init__(self):
+        if self._initialized:
+            return
+        try:
+            config = getConfig()
+            self.neo4j_uri = config['neo4j']['uri']
+            self.neo4j_user = config['neo4j']['user']
+            self.neo4j_password = config['neo4j']['password']
+            self.g = Graph(self.neo4j_uri, auth=(self.neo4j_user, self.neo4j_password))
+            info("Neo4jConnection initialized")
+            self._initialized = True
+        except KeyError as e:
+            error(f"配置文件缺少必要字段: {str(e)}")
+            self._initialized = False
+            raise RuntimeError(f"配置文件缺少必要字段: {str(e)}")
+        except Exception as e:
+            error(f"Neo4jConnection 初始化失败: {str(e)}")
+            self._initialized = False
+            raise RuntimeError(f"Neo4jConnection 初始化失败: {str(e)}")
+
+    def _normalize_label(self, entity_type: str) -> str:
+        """规范化实体类型为 Neo4j 标签"""
+        if not entity_type or not entity_type.strip():
+            return 'Entity'
+        entity_type = re.sub(r'[^\w\s]', '', entity_type.strip())
+        words = entity_type.split()
+        label = '_'.join(word.capitalize() for word in words if word)
+        return label or 'Entity'
+
+    def _clean_relation(self, relation: str) -> Tuple[str, str]:
+        """清洗关系，返回 (rel_type, rel_name)，确保 rel_type 合法"""
+        relation = relation.strip()
+        if not relation:
+            return 'RELATED_TO', '相关'
+        cleaned_relation = re.sub(r'[^\w\s]', '', relation).strip()
+        if not cleaned_relation:
+            return 'RELATED_TO', '相关'
+        if 'instance of' in relation.lower():
+            return 'INSTANCE_OF', '实例'
+        elif 'subclass of' in relation.lower():
+            return 'SUBCLASS_OF', '子类'
+        elif 'part of' in relation.lower():
+            return 'PART_OF', '部分'
+        rel_type = re.sub(r'\s+', '_', cleaned_relation).upper()
+        if rel_type and rel_type[0].isdigit():
+            rel_type = f'REL_{rel_type}'
+        if not re.match(r'^[A-Za-z][A-Za-z0-9_]*$', rel_type):
+            debug(f"非法关系类型 '{rel_type}'，替换为 'RELATED_TO'")
+            return 'RELATED_TO', relation
+        return rel_type, relation
+
+    def _read_nodes(self, triples: List[Dict]) -> Tuple[Dict[str, Set], Dict[str, List], List[Dict]]:
+        """从三元组列表中读取节点和关系"""
+        nodes_by_label = {}
+        relations_by_type = {}
+        valid_triples = []
+        try:
+            for triple in triples:
+                if not all(key in triple for key in ['head', 'head_type', 'type', 'tail', 'tail_type']):
+                    debug(f"无效三元组: {triple}")
+                    continue
+                head, relation, tail, head_type, tail_type = (
+                    triple['head'], triple['type'], triple['tail'], triple['head_type'], triple['tail_type']
+                )
+                head_label = self._normalize_label(head_type)
+                tail_label = self._normalize_label(tail_type)
+                debug(f"实体类型: {head_type} -> {head_label}, {tail_type} -> {tail_label}")
+                if head_label not in nodes_by_label:
+                    nodes_by_label[head_label] = set()
+                if tail_label not in nodes_by_label:
+                    nodes_by_label[tail_label] = set()
+                nodes_by_label[head_label].add(head)
+                nodes_by_label[tail_label].add(tail)
+                rel_type, rel_name = self._clean_relation(relation)
+                if rel_type not in relations_by_type:
+                    relations_by_type[rel_type] = []
+                relations_by_type[rel_type].append({
+                    'head': head,
+                    'tail': tail,
+                    'head_label': head_label,
+                    'tail_label': tail_label,
+                    'rel_name': rel_name
+                })
+                valid_triples.append({
+                    'head': head,
+                    'relation': relation,
+                    'tail': tail,
+                    'head_type': head_type,
+                    'tail_type': tail_type
+                })
+            info(f"读取节点: {sum(len(nodes) for nodes in nodes_by_label.values())} 个")
+            info(f"读取关系: {sum(len(rels) for rels in relations_by_type.values())} 条")
+            return nodes_by_label, relations_by_type, valid_triples
+        except Exception as e:
+            error(f"读取三元组失败: {str(e)}")
+            raise RuntimeError(f"读取三元组失败: {str(e)}")
+
+    def _create_node(self, label: str, nodes: Set[str], document_id: str, knowledge_base_id: str, userid: str) -> int:
+        """创建节点，包含 document_id、knowledge_base_id 和 userid 属性"""
+        count = 0
+        for node_name in nodes:
+            query = (
+                f"MATCH (n:{label} {{name: $name, document_id: $doc_id, "
+                f"knowledge_base_id: $kb_id, userid: $userid}}) RETURN n"
+            )
+            try:
+                if self.g.run(query, name=node_name, doc_id=document_id,
+                            kb_id=knowledge_base_id, userid=userid).data():
+                    continue
+                node = Node(
+                    label,
+                    name=node_name,
+                    document_id=document_id,
+                    knowledge_base_id=knowledge_base_id,
+                    userid=userid
+                )
+                self.g.create(node)
+                count += 1
+                debug(f"创建节点: {label} - {node_name} (document_id: {document_id}, "
+                      f"knowledge_base_id: {knowledge_base_id}, userid: {userid})")
+            except Exception as e:
+                error(f"创建节点失败: {label} - {node_name}, 错误: {str(e)}")
+        info(f"创建 {label} 节点: {count}/{len(nodes)} 个")
+        return count
+
+    def _create_relationship(self, rel_type: str, relations: List[Dict], document_id: str, knowledge_base_id: str, userid: str) -> int:
+        """创建关系，包含 document_id、knowledge_base_id 和 userid 属性"""
+        count = 0
+        total = len(relations)
+        seen_edges = set()
+        for rel in relations:
+            head, tail, head_label, tail_label, rel_name = (
+                rel['head'], rel['tail'], rel['head_label'], rel['tail_label'], rel['rel_name']
+            )
+            edge_key = f"{head_label}:{head}###{tail_label}:{tail}###{rel_type}"
+            if edge_key in seen_edges:
+                continue
+            seen_edges.add(edge_key)
+            query = (
+                f"MATCH (p:{head_label} {{name: $head, document_id: $doc_id, "
+                f"knowledge_base_id: $kb_id, userid: $userid}}), "
+                f"(q:{tail_label} {{name: $tail, document_id: $doc_id, "
+                f"knowledge_base_id: $kb_id, userid: $userid}}) "
+                f"CREATE (p)-[r:{rel_type} {{name: $rel_name, document_id: $doc_id, "
+                f"knowledge_base_id: $kb_id, userid: $userid}}]->(q)"
+            )
+            try:
+                self.g.run(query, head=head, tail=tail, rel_name=rel_name,
+                          doc_id=document_id, kb_id=knowledge_base_id,
+                          userid=userid)
+                count += 1
+                debug(f"创建关系: {head} -[{rel_type}]-> {tail} (document_id: {document_id}, "
+                      f"knowledge_base_id: {knowledge_base_id}, userid: {userid})")
+            except Exception as e:
+                error(f"创建关系失败: {query}, 错误: {str(e)}")
+        info(f"创建 {rel_type} 关系: {count}/{total} 条")
+        return count
+
+    async def _get_embeddings(self, texts: List[str]) -> List[List[float]]:
+        """调用嵌入服务获取文本的向量，带缓存"""
+        try:
+            uncached_texts = [text for text in texts if text not in EMBED_CACHE]
+            if uncached_texts:
+                async with aiohttp.ClientSession() as session:
+                    async with session.post(
+                        "http://localhost:9998/v1/embeddings",
+                        headers={"Content-Type": "application/json"},
+                        json={"input": uncached_texts}
+                    ) as response:
+                        if response.status != 200:
+                            error(f"嵌入服务调用失败，状态码: {response.status}")
+                            raise RuntimeError(f"嵌入服务调用失败: {response.status}")
+                        result = await response.json()
+                        if result.get("object") != "list" or not result.get("data"):
+                            error(f"嵌入服务响应格式错误: {result}")
+                            raise RuntimeError("嵌入服务响应格式错误")
+                        embeddings = [item["embedding"] for item in result["data"]]
+                        for text, embedding in zip(uncached_texts, embeddings):
+                            EMBED_CACHE[text] = np.array(embedding) / np.linalg.norm(embedding)
+                        debug(f"成功获取 {len(embeddings)} 个新嵌入向量，缓存大小: {len(EMBED_CACHE)}")
+            return [EMBED_CACHE[text] for text in texts]
+        except Exception as e:
+            error(f"嵌入服务调用失败: {str(e)}")
+            raise RuntimeError(f"嵌入服务调用失败: {str(e)}")
+
+    async def handle_connection(self, action: str, params: Dict = None) -> Dict:
+        """处理 Neo4j 数据库操作"""
+        if not params:
+            params = {}
+        try:
+            debug(f"处理 Neo4j 操作: action={action}, params={params}")
+            collection_name = "neo4j"  # 为了与 MilvusConnection 的返回格式一致
+            document_id = params.get("document_id", "")
+            if not self._initialized:
+                error("Neo4j 服务未启动")
+                return {
+                    "status": "error",
+                    "message": "Neo4j 服务未启动",
+                    "collection_name": collection_name,
+                    "document_id": document_id,
+                    "status_code": 400
+                }
+            if action == "initialize":
+                return {
+                    "status": "success",
+                    "message": "Neo4j 服务已初始化",
+                    "collection_name": collection_name,
+                    "document_id": document_id,
+                    "status_code": 200
+                }
+            elif action == "insert_triples":
+                triples = params.get("triples", [])
+                document_id = params.get("document_id", "")
+                knowledge_base_id = params.get("knowledge_base_id", "")
+                userid = params.get("userid", "")
+                if not triples or not document_id or not knowledge_base_id or not userid:
+                    return {
+                        "status": "error",
+                        "message": "triples、document_id、knowledge_base_id 和 userid 不能为空",
+                        "collection_name": collection_name,
+                        "document_id": document_id,
+                        "status_code": 400
+                    }
+                return self._insert_triples(triples, document_id, knowledge_base_id, userid)
+            elif action == "delete_document":
+                document_id = params.get("document_id", "")
+                if not document_id:
+                    return {
+                        "status": "error",
+                        "message": "document_id 不能为空",
+                        "collection_name": collection_name,
+                        "document_id": document_id,
+                        "status_code": 400
+                    }
+                return self._delete_document(document_id)
+            elif action == "delete_knowledge_base":
+                userid = params.get("userid", "")
+                knowledge_base_id = params.get("knowledge_base_id", "")
+                if not userid or not knowledge_base_id:
+                    return {
+                        "status": "error",
+                        "message": "userid 和 knowledge_base_id 不能为空",
+                        "collection_name": collection_name,
+                        "document_id": document_id,
+                        "status_code": 400
+                    }
+                return self._delete_knowledge_base(userid, knowledge_base_id)
+            elif action == "match_triplets":
+                query = params.get("query", "")
+                query_entities = params.get("query_entities", [])
+                userid = params.get("userid", "")
+                knowledge_base_id = params.get("knowledge_base_id", "")
+                if not query or not query_entities or not userid or not knowledge_base_id:
+                    return {
+                        "status": "error",
+                        "message": "query、query_entities、userid 和 knowledge_base_id 不能为空",
+                        "collection_name": collection_name,
+                        "document_id": document_id,
+                        "status_code": 400
+                    }
+                return await self._match_triplets(query, query_entities, userid, knowledge_base_id)
+            else:
+                return {
+                    "status": "error",
+                    "message": f"未知的 action: {action}",
+                    "collection_name": collection_name,
+                    "document_id": document_id,
+                    "status_code": 400
+                }
+        except Exception as e:
+            error(f"处理 Neo4j 操作失败: action={action}, 错误: {str(e)}")
+            return {
+                "status": "error",
+                "message": f"服务器错误: {str(e)}",
+                "collection_name": collection_name,
+                "document_id": document_id,
+                "status_code": 400
+            }
+
+    def _insert_triples(self, triples: List[Dict], document_id: str, knowledge_base_id: str, userid: str) -> Dict:
+        """插入三元组到 Neo4j"""
+        try:
+            if not triples:
+                return {
+                    "status": "success",
+                    "message": "没有三元组需要插入",
+                    "nodes_created": 0,
+                    "rels_created": 0,
+                    "collection_name": "neo4j",
+                    "document_id": document_id,
+                    "status_code": 200
+                }
+            nodes_by_label, relations_by_type, _ = self._read_nodes(triples)
+            total_nodes = 0
+            for label, nodes in nodes_by_label.items():
+                total_nodes += self._create_node(label, nodes, document_id, knowledge_base_id, userid)
+            total_rels = 0
+            for rel_type, relations in relations_by_type.items():
+                total_rels += self._create_relationship(rel_type, relations, document_id, knowledge_base_id, userid)
+            info(f"成功插入 {total_nodes} 个节点和 {total_rels} 个关系")
+            return {
+                "status": "success",
+                "message": f"成功插入 {total_nodes} 个节点和 {total_rels} 个关系",
+                "nodes_created": total_nodes,
+                "rels_created": total_rels,
+                "collection_name": "neo4j",
+                "document_id": document_id,
+                "status_code": 200
+            }
+        except Exception as e:
+            error(f"插入三元组失败: {str(e)}")
+            return {
+                "status": "error",
+                "message": f"插入三元组失败: {str(e)}",
+                "collection_name": "neo4j",
+                "document_id": document_id,
+                "status_code": 400
+            }
+
+    def _delete_document(self, document_id: str) -> Dict:
+        """删除 Neo4j 中指定文档的数据"""
+        try:
+            query = """
+            MATCH (n {document_id: $document_id})
+            OPTIONAL MATCH (n)-[r {document_id: $document_id}]->()
+            WITH collect(r) AS rels, collect(n) AS nodes
+            FOREACH (r IN rels | DELETE r)
+            FOREACH (n IN nodes | DELETE n)
+            RETURN size(nodes) AS node_count, size(rels) AS rel_count, [r IN rels | type(r)] AS rel_types
+            """
+            result = self.g.run(query, document_id=document_id).data()
+            nodes_deleted = result[0]['node_count'] if result else 0
+            rels_deleted = result[0]['rel_count'] if result else 0
+            rel_types = result[0]['rel_types'] if result else []
+            info(f"成功删除 document_id={document_id} 的 {nodes_deleted} 个节点和 {rels_deleted} 个关系，关系类型: {rel_types}")
+            return {
+                "status": "success",
+                "message": f"成功删除 {nodes_deleted} 个节点和 {rels_deleted} 个关系",
+                "nodes_deleted": nodes_deleted,
+                "rels_deleted": rels_deleted,
+                "collection_name": "neo4j",
+                "document_id": document_id,
+                "status_code": 200
+            }
+        except Exception as e:
+            error(f"删除 document_id={document_id} 的 Neo4j 数据失败: {str(e)}")
+            return {
+                "status": "error",
+                "message": f"删除 Neo4j 数据失败: {str(e)}",
+                "collection_name": "neo4j",
+                "document_id": document_id,
+                "status_code": 400
+            }
+
+    def _delete_knowledge_base(self, userid: str, knowledge_base_id: str) -> Dict:
+        """删除 Neo4j 中指定用户和知识库的数据"""
+        try:
+            query = """
+            MATCH (n {userid: $userid, knowledge_base_id: $knowledge_base_id})
+            OPTIONAL MATCH (n)-[r {userid: $userid, knowledge_base_id: $knowledge_base_id}]->()
+            WITH collect(r) AS rels, collect(n) AS nodes
+            FOREACH (r IN rels | DELETE r)
+            FOREACH (n IN nodes | DELETE n)
+            RETURN size(nodes) AS node_count, size(rels) AS rel_count, [r IN rels | type(r)] AS rel_types
+            """
+            result = self.g.run(query, userid=userid, knowledge_base_id=knowledge_base_id).data()
+            nodes_deleted = result[0]['node_count'] if result else 0
+            rels_deleted = result[0]['rel_count'] if result else 0
+            rel_types = result[0]['rel_types'] if result else []
+            info(f"成功删除 userid={userid}, knowledge_base_id={knowledge_base_id} 的 {nodes_deleted} 个节点和 {rels_deleted} 个关系")
+            return {
+                "status": "success",
+                "message": f"成功删除 {nodes_deleted} 个节点和 {rels_deleted} 个关系",
+                "nodes_deleted": nodes_deleted,
+                "rels_deleted": rels_deleted,
+                "collection_name": "neo4j",
+                "document_id": "",
+                "status_code": 200
+            }
+        except Exception as e:
+            error(f"删除 Neo4j 数据失败: {str(e)}")
+            return {
+                "status": "error",
+                "message": f"删除 Neo4j 数据失败: {str(e)}",
+                "collection_name": "neo4j",
+                "document_id": "",
+                "status_code": 400
+            }
+
+    async def _match_triplets(self, query: str, query_entities: List[str], userid: str, knowledge_base_id: str) -> Dict:
+        """匹配查询实体与 Neo4j 中的三元组"""
+        start_time = time.time()
+        timing_stats = {}
+        matched_triplets = []
+        ENTITY_SIMILARITY_THRESHOLD = 0.8
+        try:
+            debug(f"已连接到 Neo4j: {self.neo4j_uri}")
+            timing_stats["neo4j_connect"] = time.time() - start_time
+
+            matched_names = set()
+            entity_match_start = time.time()
+            for entity in query_entities:
+                normalized_entity = entity.lower().strip()
+                query = """
+                MATCH (n {userid: $userid, knowledge_base_id: $knowledge_base_id})
+                WHERE toLower(n.name) CONTAINS $entity
+                   OR apoc.text.levenshteinSimilarity(toLower(n.name), $entity) > 0.7
+                RETURN n.name, apoc.text.levenshteinSimilarity(toLower(n.name), $entity) AS sim
+                ORDER BY sim DESC
+                LIMIT 100
+                """
+                try:
+                    results = self.g.run(query, userid=userid, knowledge_base_id=knowledge_base_id, entity=normalized_entity).data()
+                    for record in results:
+                        matched_names.add(record['n.name'])
+                        debug(f"实体 {entity} 匹配节点: {record['n.name']} (Levenshtein 相似度: {record['sim']:.2f})")
+                except Exception as e:
+                    debug(f"模糊匹配实体 {entity} 失败: {str(e)}")
+                    continue
+            timing_stats["entity_match"] = time.time() - entity_match_start
+            debug(f"实体匹配耗时: {timing_stats['entity_match']:.3f} 秒")
+
+            triplets = []
+            if matched_names:
+                triplet_query_start = time.time()
+                query = """
+                MATCH (h {userid: $userid, knowledge_base_id: $knowledge_base_id})-[r {userid: $userid, knowledge_base_id: $knowledge_base_id}]->(t {userid: $userid, knowledge_base_id: $knowledge_base_id})
+                WHERE h.name IN $matched_names OR t.name IN $matched_names
+                RETURN h.name AS head, r.name AS type, t.name AS tail
+                LIMIT 100
+                """
+                try:
+                    results = self.g.run(query, userid=userid, knowledge_base_id=knowledge_base_id, matched_names=list(matched_names)).data()
+                    seen = set()
+                    for record in results:
+                        head, type_, tail = record['head'], record['type'], record['tail']
+                        triplet_key = (head.lower(), type_.lower(), tail.lower())
+                        if triplet_key not in seen:
+                            seen.add(triplet_key)
+                            triplets.append({
+                                'head': head,
+                                'type': type_,
+                                'tail': tail,
+                                'head_type': '',
+                                'tail_type': ''
+                            })
+                    debug(f"从 Neo4j 加载三元组: knowledge_base_id={knowledge_base_id}, 数量={len(triplets)}")
+                except Exception as e:
+                    error(f"检索三元组失败: knowledge_base_id={knowledge_base_id}, 错误: {str(e)}")
+                    return {
+                        "status": "error",
+                        "message": f"检索三元组失败: {str(e)}",
+                        "triplets": [],
+                        "timing": timing_stats,
+                        "collection_name": "neo4j",
+                        "document_id": "",
+                        "status_code": 400
+                    }
+                timing_stats["triplet_query"] = time.time() - triplet_query_start
+                debug(f"Neo4j 三元组查询耗时: {timing_stats['triplet_query']:.3f} 秒")
+
+            if not triplets:
+                debug(f"知识库 knowledge_base_id={knowledge_base_id} 无匹配三元组")
+                return {
+                    "status": "success",
+                    "message": f"知识库 knowledge_base_id={knowledge_base_id} 无匹配三元组",
+                    "triplets": [],
+                    "timing": timing_stats,
+                    "collection_name": "neo4j",
+                    "document_id": "",
+                    "status_code": 200
+                }
+
+            embedding_start = time.time()
+            texts_to_embed = query_entities + [t['head'] for t in triplets] + [t['tail'] for t in triplets]
+            embeddings = await self._get_embeddings(texts_to_embed)
+            entity_vectors = {entity: embeddings[i] for i, entity in enumerate(query_entities)}
+            head_vectors = {t['head']: embeddings[len(query_entities) + i] for i, t in enumerate(triplets)}
+            tail_vectors = {t['tail']: embeddings[len(query_entities) + len(triplets) + i] for i, t in enumerate(triplets)}
+            debug(f"成功获取 {len(embeddings)} 个嵌入向量（{len(query_entities)} entities + {len(triplets)} heads + {len(triplets)} tails）")
+            timing_stats["embedding"] = time.time() - embedding_start
+            debug(f"嵌入向量生成耗时: {timing_stats['embedding']:.3f} 秒")
+
+            similarity_start = time.time()
+            for entity in query_entities:
+                entity_vec = entity_vectors[entity]
+                for d_triplet in triplets:
+                    d_head_vec = head_vectors[d_triplet['head']]
+                    d_tail_vec = tail_vectors[d_triplet['tail']]
+                    head_similarity = 1 - cosine(entity_vec, d_head_vec)
+                    tail_similarity = 1 - cosine(entity_vec, d_tail_vec)
+                    if head_similarity >= ENTITY_SIMILARITY_THRESHOLD or tail_similarity >= ENTITY_SIMILARITY_THRESHOLD:
+                        matched_triplets.append(d_triplet)
+                        debug(f"匹配三元组: {d_triplet['head']} - {d_triplet['type']} - {d_triplet['tail']} "
+                              f"(entity={entity}, head_sim={head_similarity:.2f}, tail_sim={tail_similarity:.2f})")
+            timing_stats["similarity"] = time.time() - similarity_start
+            debug(f"相似度计算耗时: {timing_stats['similarity']:.3f} 秒")
+
+            unique_matched = []
+            seen = set()
+            for t in matched_triplets:
+                identifier = (t['head'].lower(), t['type'].lower(), t['tail'].lower())
+                if identifier not in seen:
+                    seen.add(identifier)
+                    unique_matched.append(t)
+
+            timing_stats["total_time"] = time.time() - start_time
+            info(f"找到 {len(unique_matched)} 个匹配的三元组，总耗时: {timing_stats['total_time']:.3f} 秒")
+            return {
+                "status": "success",
+                "message": f"找到 {len(unique_matched)} 个匹配的三元组",
+                "triplets": unique_matched,
+                "timing": timing_stats,
+                "collection_name": "neo4j",
+                "document_id": "",
+                "status_code": 200
+            }
+
+        except Exception as e:
+            error(f"匹配三元组失败: {str(e)}")
+            timing_stats["total_time"] = time.time() - start_time
+            return {
+                "status": "error",
+                "message": f"匹配三元组失败: {str(e)}",
+                "triplets": [],
+                "timing": timing_stats,
+                "collection_name": "neo4j",
+                "document_id": "",
+                "status_code": 400
+            }
+
+connection_register('Neo4j', Neo4jConnection)
+info("Neo4jconnection registered")

llmengine/neo4j.py

@@ -0,0 +1,355 @@
+import argparse
+import os
+from aiohttp import web
+from llmengine.db_neo4j import Neo4jConnection
+from llmengine.base_db import connection_register, get_connection_class
+from appPublic.registerfunction import RegisterFunction
+from appPublic.log import debug, error, info
+from ahserver.serverenv import ServerEnv
+from ahserver.webapp import webserver
+import json
+
+helptext = """Neo4j Connection Service API:
+
+1. Initialize Endpoint:
+path: /v1/initialize
+method: POST
+headers: {"Content-Type": "application/json"}
+data: {}
+response:
+- Success: HTTP 200, {
+    "status": "success",
+    "message": "Neo4j 服务已初始化",
+    "collection_name": "neo4j",
+    "document_id": "",
+    "status_code": 200
+}
+- Error: HTTP 400, {
+    "status": "error",
+    "message": "<error message>",
+    "collection_name": "neo4j",
+    "document_id": "",
+    "status_code": 400
+}
+
+2. Insert Triples Endpoint:
+path: /v1/inserttriples
+method: POST
+headers: {"Content-Type": "application/json"}
+data: {
+    "triples": [
+        {"head": "entity1", "head_type": "Person", "type": "related_to", "tail": "entity2", "tail_type": "Organization"},
+        ...
+    ],
+    "document_id": "uuid",
+    "knowledge_base_id": "kb123",
+    "userid": "user123"
+}
+response:
+- Success: HTTP 200, {
+    "status": "success",
+    "message": "成功插入 <nodes> 个节点和 <rels> 个关系",
+    "nodes_created": <int>,
+    "rels_created": <int>,
+    "collection_name": "neo4j",
+    "document_id": "<uuid>",
+    "status_code": 200
+}
+- Error: HTTP 400, {
+    "status": "error",
+    "message": "<error message>",
+    "collection_name": "neo4j",
+    "document_id": "<uuid>",
+    "status_code": 400
+}
+
+3. Delete Document Endpoint:
+path: /v1/deletedocument
+method: POST
+headers: {"Content-Type": "application/json"}
+data: {
+    "document_id": "uuid"
+}
+response:
+- Success: HTTP 200, {
+    "status": "success",
+    "message": "成功删除 <nodes> 个节点和 <rels> 个关系",
+    "nodes_deleted": <int>,
+    "rels_deleted": <int>,
+    "collection_name": "neo4j",
+    "document_id": "<uuid>",
+    "status_code": 200
+}
+- Error: HTTP 400, {
+    "status": "error",
+    "message": "<error message>",
+    "collection_name": "neo4j",
+    "document_id": "<uuid>",
+    "status_code": 400
+}
+
+4. Delete Knowledge Base Endpoint:
+path: /v1/deleteknowledgebase
+method: POST
+headers: {"Content-Type": "application/json"}
+data: {
+    "userid": "user123",
+    "knowledge_base_id": "kb123"
+}
+response:
+- Success: HTTP 200, {
+    "status": "success",
+    "message": "成功删除 <nodes> 个节点和 <rels> 个关系",
+    "nodes_deleted": <int>,
+    "rels_deleted": <int>,
+    "collection_name": "neo4j",
+    "document_id": "",
+    "status_code": 200
+}
+- Error: HTTP 400, {
+    "status": "error",
+    "message": "<error message>",
+    "collection_name": "neo4j",
+    "document_id": "",
+    "status_code": 400
+}
+
+5. Match Triplets Endpoint:
+path: /v1/matchtriplets
+method: POST
+headers: {"Content-Type": "application/json"}
+data: {
+    "query": "query text",
+    "query_entities": ["entity1", "entity2"],
+    "userid": "user123",
+    "knowledge_base_id": "kb123"
+}
+response:
+- Success: HTTP 200, {
+    "status": "success",
+    "message": "找到 <count> 个匹配的三元组",
+    "triplets": [
+        {"head": "entity1", "type": "related_to", "tail": "entity2", "head_type": "", "tail_type": ""},
+        ...
+    ],
+    "timing": {
+        "neo4j_connect": <float>,
+        "entity_match": <float>,
+        "triplet_query": <float>,
+        "embedding": <float>,
+        "similarity": <float>,
+        "total_time": <float>
+    },
+    "collection_name": "neo4j",
+    "document_id": "",
+    "status_code": 200
+}
+- Error: HTTP 400, {
+    "status": "error",
+    "message": "<error message>",
+    "triplets": [],
+    "timing": {},
+    "collection_name": "neo4j",
+    "document_id": "",
+    "status_code": 400
+}
+
+6. Connection Endpoint:
+path: /v1/connection
+method: POST
+headers: {"Content-Type": "application/json"}
+data: {
+    "action": "initialize|insert_triples|delete_document|delete_knowledge_base|match_triplets",
+    "params": {...}
+}
+response:
+- Success: HTTP 200, {"status": "success", ...}
+- Error: HTTP 400, {"status": "error", "message": "<error message>", "collection_name": "neo4j", "document_id": "", "status_code": 400}
+
+7. Docs Endpoint:
+path: /docs
+method: GET
+response: This help text
+"""
+
+def init():
+    rf = RegisterFunction()
+    rf.register('initialize', initialize)
+    rf.register('inserttriples', insert_triples)
+    rf.register('deletedocument', delete_document)
+    rf.register('deleteknowledgebase', delete_knowledge_base)
+    rf.register('matchtriplets', match_triplets)
+    rf.register('connection', handle_connection)
+    rf.register('docs', docs)
+
+async def docs(request, params_kw, *params, **kw):
+    return web.Response(text=helptext, content_type='text/plain')
+
+async def initialize(request, params_kw, *params, **kw):
+    debug(f'Received initialize params: {params_kw=}')
+    se = ServerEnv()
+    engine = se.engine
+    debug(f'Engine: {engine}')
+    if engine is None or not isinstance(engine, Neo4jConnection):
+        error("Neo4jConnection not initialized")
+        return web.json_response({
+            "status": "error",
+            "message": "Neo4j 服务未启动",
+            "collection_name": "neo4j",
+            "document_id": "",
+            "status_code": 400
+        }, dumps=lambda obj: json.dumps(obj, ensure_ascii=False), status=400)
+    try:
+        result = await engine.handle_connection("initialize", params_kw)
+        debug(f'Initialize result: {result=}')
+        status = 200 if result.get("status") == "success" else 400
+        return web.json_response(result, dumps=lambda obj: json.dumps(obj, ensure_ascii=False), status=status)
+    except Exception as e:
+        error(f'初始化失败: {str(e)}')
+        return web.json_response({
+            "status": "error",
+            "message": str(e),
+            "collection_name": "neo4j",
+            "document_id": "",
+            "status_code": 400
+        }, dumps=lambda obj: json.dumps(obj, ensure_ascii=False), status=400)
+
+async def insert_triples(request, params_kw, *params, **kw):
+    debug(f'Received insert_triples params: {params_kw=}')
+    se = ServerEnv()
+    engine = se.engine
+    try:
+        required_fields = ['triples', 'document_id', 'knowledge_base_id', 'userid']
+        missing_fields = [field for field in required_fields if field not in params_kw or not params_kw[field]]
+        if missing_fields:
+            raise ValueError(f"缺少必填字段: {', '.join(missing_fields)}")
+        result = await engine.handle_connection("insert_triples", params_kw)
+        debug(f'Insert triples result: {result=}')
+        status = 200 if result.get("status") == "success" else 400
+        return web.json_response(result, dumps=lambda obj: json.dumps(obj, ensure_ascii=False), status=status)
+    except Exception as e:
+        error(f'插入三元组失败: {str(e)}')
+        return web.json_response({
+            "status": "error",
+            "message": str(e),
+            "collection_name": "neo4j",
+            "document_id": params_kw.get("document_id", ""),
+            "status_code": 400
+        }, dumps=lambda obj: json.dumps(obj, ensure_ascii=False), status=400)
+
+async def delete_document(request, params_kw, *params, **kw):
+    debug(f'Received delete_document params: {params_kw=}')
+    se = ServerEnv()
+    engine = se.engine
+    try:
+        if not params_kw.get('document_id'):
+            raise ValueError("document_id 不能为空")
+        result = await engine.handle_connection("delete_document", params_kw)
+        debug(f'Delete document result: {result=}')
+        status = 200 if result.get("status") == "success" else 400
+        return web.json_response(result, dumps=lambda obj: json.dumps(obj, ensure_ascii=False), status=status)
+    except Exception as e:
+        error(f'删除文档失败: {str(e)}')
+        return web.json_response({
+            "status": "error",
+            "message": str(e),
+            "collection_name": "neo4j",
+            "document_id": params_kw.get("document_id", ""),
+            "status_code": 400
+        }, dumps=lambda obj: json.dumps(obj, ensure_ascii=False), status=400)
+
+async def delete_knowledge_base(request, params_kw, *params, **kw):
+    debug(f'Received delete_knowledge_base params: {params_kw=}')
+    se = ServerEnv()
+    engine = se.engine
+    try:
+        required_fields = ['userid', 'knowledge_base_id']
+        missing_fields = [field for field in required_fields if field not in params_kw or not params_kw[field]]
+        if missing_fields:
+            raise ValueError(f"缺少必填字段: {', '.join(missing_fields)}")
+        result = await engine.handle_connection("delete_knowledge_base", params_kw)
+        debug(f'Delete knowledge base result: {result=}')
+        status = 200 if result.get("status") == "success" else 400
+        return web.json_response(result, dumps=lambda obj: json.dumps(obj, ensure_ascii=False), status=status)
+    except Exception as e:
+        error(f'删除知识库失败: {str(e)}')
+        return web.json_response({
+            "status": "error",
+            "message": str(e),
+            "collection_name": "neo4j",
+            "document_id": "",
+            "status_code": 400
+        }, dumps=lambda obj: json.dumps(obj, ensure_ascii=False), status=400)
+
+async def match_triplets(request, params_kw, *params, **kw):
+    debug(f'Received match_triplets params: {params_kw=}')
+    se = ServerEnv()
+    engine = se.engine
+    try:
+        required_fields = ['query', 'query_entities', 'userid', 'knowledge_base_id']
+        missing_fields = [field for field in required_fields if field not in params_kw or not params_kw[field]]
+        if missing_fields:
+            raise ValueError(f"缺少必填字段: {', '.join(missing_fields)}")
+        result = await engine.handle_connection("match_triplets", params_kw)
+        debug(f'Match triplets result: {result=}')
+        status = 200 if result.get("status") == "success" else 400
+        return web.json_response(result, dumps=lambda obj: json.dumps(obj, ensure_ascii=False), status=status)
+    except Exception as e:
+        error(f'匹配三元组失败: {str(e)}')
+        return web.json_response({
+            "status": "error",
+            "message": str(e),
+            "triplets": [],
+            "timing": {},
+            "collection_name": "neo4j",
+            "document_id": "",
+            "status_code": 400
+        }, dumps=lambda obj: json.dumps(obj, ensure_ascii=False), status=400)
+
+async def handle_connection(request, params_kw, *params, **kw):
+    debug(f'Received connection params: {params_kw=}')
+    se = ServerEnv()
+    engine = se.engine
+    try:
+        data = await request.json()
+        action = data.get('action')
+        if not action:
+            debug(f'action 未提供')
+            return web.json_response({
+                "status": "error",
+                "message": "action 参数未提供",
+                "collection_name": "neo4j",
+                "document_id": "",
+                "status_code": 400
+            }, dumps=lambda obj: json.dumps(obj, ensure_ascii=False), status=400)
+        result = await engine.handle_connection(action, data.get('params', {}))
+        debug(f'Connection result: {result=}')
+        status = 200 if result.get("status") == "success" else 400
+        return web.json_response(result, dumps=lambda obj: json.dumps(obj, ensure_ascii=False), status=status)
+    except Exception as e:
+        error(f'处理连接操作失败: {str(e)}')
+        return web.json_response({
+            "status": "error",
+            "message": str(e),
+            "collection_name": "neo4j",
+            "document_id": "",
+            "status_code": 400
+        }, dumps=lambda obj: json.dumps(obj, ensure_ascii=False), status=400)
+
+def main():
+    parser = argparse.ArgumentParser(prog="Neo4j Connection Service")
+    parser.add_argument('-w', '--workdir', help="Working directory")
+    parser.add_argument('-p', '--port', default='8885', help="Port to run the server on")
+    parser.add_argument('connection_path', help="Connection class path (e.g., Neo4j)")
+    args = parser.parse_args()
+    debug(f"Arguments: {args}")
+    Klass = get_connection_class(args.connection_path)
+    se = ServerEnv()
+    se.engine = Klass()
+    workdir = args.workdir or os.getcwd()
+    port = args.port
+    debug(f'{args=}')
+    webserver(init, workdir, port)
+
+if __name__ == '__main__':
+    main()

test/bgererank/=1.3.3

@@ -1,82 +0,0 @@
-Looking in indexes: https://pypi.tuna.tsinghua.edu.cn/simple
-Requirement already satisfied: FlagEmbedding in /share/vllm-0.8.5/lib/python3.10/site-packages (1.3.5)
-Requirement already satisfied: torch>=1.6.0 in /share/vllm-0.8.5/lib/python3.10/site-packages (from FlagEmbedding) (2.6.0)
-Requirement already satisfied: transformers>=4.44.2 in /share/vllm-0.8.5/lib/python3.10/site-packages (from FlagEmbedding) (4.51.3)
-Requirement already satisfied: datasets>=2.19.0 in /share/vllm-0.8.5/lib/python3.10/site-packages (from FlagEmbedding) (3.6.0)
-Requirement already satisfied: accelerate>=0.20.1 in /share/vllm-0.8.5/lib/python3.10/site-packages (from FlagEmbedding) (1.7.0)
-Requirement already satisfied: sentence_transformers in /share/vllm-0.8.5/lib/python3.10/site-packages (from FlagEmbedding) (4.1.0)
-Requirement already satisfied: peft in /share/vllm-0.8.5/lib/python3.10/site-packages (from FlagEmbedding) (0.16.0)
-Requirement already satisfied: ir-datasets in /share/vllm-0.8.5/lib/python3.10/site-packages (from FlagEmbedding) (0.5.11)
-Requirement already satisfied: sentencepiece in /share/vllm-0.8.5/lib/python3.10/site-packages (from FlagEmbedding) (0.2.0)
-Requirement already satisfied: protobuf in /share/vllm-0.8.5/lib/python3.10/site-packages (from FlagEmbedding) (4.25.7)
-Requirement already satisfied: psutil in /share/vllm-0.8.5/lib/python3.10/site-packages (from accelerate>=0.20.1->FlagEmbedding) (7.0.0)
-Requirement already satisfied: numpy<3.0.0,>=1.17 in /share/vllm-0.8.5/lib/python3.10/site-packages (from accelerate>=0.20.1->FlagEmbedding) (2.2.5)
-Requirement already satisfied: safetensors>=0.4.3 in /share/vllm-0.8.5/lib/python3.10/site-packages (from accelerate>=0.20.1->FlagEmbedding) (0.5.3)
-Requirement already satisfied: packaging>=20.0 in /share/vllm-0.8.5/lib/python3.10/site-packages (from accelerate>=0.20.1->FlagEmbedding) (24.2)
-Requirement already satisfied: pyyaml in /share/vllm-0.8.5/lib/python3.10/site-packages (from accelerate>=0.20.1->FlagEmbedding) (6.0.2)
-Requirement already satisfied: huggingface-hub>=0.21.0 in /share/vllm-0.8.5/lib/python3.10/site-packages (from accelerate>=0.20.1->FlagEmbedding) (0.30.2)
-Requirement already satisfied: pyarrow>=15.0.0 in /share/vllm-0.8.5/lib/python3.10/site-packages (from datasets>=2.19.0->FlagEmbedding) (20.0.0)
-Requirement already satisfied: fsspec[http]<=2025.3.0,>=2023.1.0 in /share/vllm-0.8.5/lib/python3.10/site-packages (from datasets>=2.19.0->FlagEmbedding) (2025.3.0)
-Requirement already satisfied: xxhash in /share/vllm-0.8.5/lib/python3.10/site-packages (from datasets>=2.19.0->FlagEmbedding) (3.5.0)
-Requirement already satisfied: multiprocess<0.70.17 in /share/vllm-0.8.5/lib/python3.10/site-packages (from datasets>=2.19.0->FlagEmbedding) (0.70.16)
-Requirement already satisfied: tqdm>=4.66.3 in /share/vllm-0.8.5/lib/python3.10/site-packages (from datasets>=2.19.0->FlagEmbedding) (4.67.1)
-Requirement already satisfied: pandas in /share/vllm-0.8.5/lib/python3.10/site-packages (from datasets>=2.19.0->FlagEmbedding) (2.3.0)
-Requirement already satisfied: dill<0.3.9,>=0.3.0 in /share/vllm-0.8.5/lib/python3.10/site-packages (from datasets>=2.19.0->FlagEmbedding) (0.3.8)
-Requirement already satisfied: filelock in /share/vllm-0.8.5/lib/python3.10/site-packages (from datasets>=2.19.0->FlagEmbedding) (3.18.0)
-Requirement already satisfied: requests>=2.32.2 in /share/vllm-0.8.5/lib/python3.10/site-packages (from datasets>=2.19.0->FlagEmbedding) (2.32.3)
-Requirement already satisfied: nvidia-cusparselt-cu12==0.6.2 in /share/vllm-0.8.5/lib/python3.10/site-packages (from torch>=1.6.0->FlagEmbedding) (0.6.2)
-Requirement already satisfied: sympy==1.13.1 in /share/vllm-0.8.5/lib/python3.10/site-packages (from torch>=1.6.0->FlagEmbedding) (1.13.1)
-Requirement already satisfied: nvidia-cuda-nvrtc-cu12==12.4.127 in /share/vllm-0.8.5/lib/python3.10/site-packages (from torch>=1.6.0->FlagEmbedding) (12.4.127)
-Requirement already satisfied: triton==3.2.0 in /share/vllm-0.8.5/lib/python3.10/site-packages (from torch>=1.6.0->FlagEmbedding) (3.2.0)
-Requirement already satisfied: jinja2 in /share/vllm-0.8.5/lib/python3.10/site-packages (from torch>=1.6.0->FlagEmbedding) (3.1.6)
-Requirement already satisfied: nvidia-cusolver-cu12==11.6.1.9 in /share/vllm-0.8.5/lib/python3.10/site-packages (from torch>=1.6.0->FlagEmbedding) (11.6.1.9)
-Requirement already satisfied: typing-extensions>=4.10.0 in /share/vllm-0.8.5/lib/python3.10/site-packages (from torch>=1.6.0->FlagEmbedding) (4.13.2)
-Requirement already satisfied: nvidia-nvtx-cu12==12.4.127 in /share/vllm-0.8.5/lib/python3.10/site-packages (from torch>=1.6.0->FlagEmbedding) (12.4.127)
-Requirement already satisfied: nvidia-nccl-cu12==2.21.5 in /share/vllm-0.8.5/lib/python3.10/site-packages (from torch>=1.6.0->FlagEmbedding) (2.21.5)
-Requirement already satisfied: nvidia-cuda-cupti-cu12==12.4.127 in /share/vllm-0.8.5/lib/python3.10/site-packages (from torch>=1.6.0->FlagEmbedding) (12.4.127)
-Requirement already satisfied: nvidia-cuda-runtime-cu12==12.4.127 in /share/vllm-0.8.5/lib/python3.10/site-packages (from torch>=1.6.0->FlagEmbedding) (12.4.127)
-Requirement already satisfied: nvidia-cufft-cu12==11.2.1.3 in /share/vllm-0.8.5/lib/python3.10/site-packages (from torch>=1.6.0->FlagEmbedding) (11.2.1.3)
-Requirement already satisfied: nvidia-cublas-cu12==12.4.5.8 in /share/vllm-0.8.5/lib/python3.10/site-packages (from torch>=1.6.0->FlagEmbedding) (12.4.5.8)
-Requirement already satisfied: nvidia-curand-cu12==10.3.5.147 in /share/vllm-0.8.5/lib/python3.10/site-packages (from torch>=1.6.0->FlagEmbedding) (10.3.5.147)
-Requirement already satisfied: nvidia-nvjitlink-cu12==12.4.127 in /share/vllm-0.8.5/lib/python3.10/site-packages (from torch>=1.6.0->FlagEmbedding) (12.4.127)
-Requirement already satisfied: nvidia-cudnn-cu12==9.1.0.70 in /share/vllm-0.8.5/lib/python3.10/site-packages (from torch>=1.6.0->FlagEmbedding) (9.1.0.70)
-Requirement already satisfied: networkx in /share/vllm-0.8.5/lib/python3.10/site-packages (from torch>=1.6.0->FlagEmbedding) (3.4.2)
-Requirement already satisfied: nvidia-cusparse-cu12==12.3.1.170 in /share/vllm-0.8.5/lib/python3.10/site-packages (from torch>=1.6.0->FlagEmbedding) (12.3.1.170)
-Requirement already satisfied: mpmath<1.4,>=1.1.0 in /share/vllm-0.8.5/lib/python3.10/site-packages (from sympy==1.13.1->torch>=1.6.0->FlagEmbedding) (1.3.0)
-Requirement already satisfied: tokenizers<0.22,>=0.21 in /share/vllm-0.8.5/lib/python3.10/site-packages (from transformers>=4.44.2->FlagEmbedding) (0.21.1)
-Requirement already satisfied: regex!=2019.12.17 in /share/vllm-0.8.5/lib/python3.10/site-packages (from transformers>=4.44.2->FlagEmbedding) (2024.11.6)
-Requirement already satisfied: zlib-state>=0.1.3 in /share/vllm-0.8.5/lib/python3.10/site-packages (from ir-datasets->FlagEmbedding) (0.1.9)
-Requirement already satisfied: trec-car-tools>=2.5.4 in /share/vllm-0.8.5/lib/python3.10/site-packages (from ir-datasets->FlagEmbedding) (2.6)
-Requirement already satisfied: lz4>=3.1.10 in /share/vllm-0.8.5/lib/python3.10/site-packages (from ir-datasets->FlagEmbedding) (4.4.4)
-Requirement already satisfied: ijson>=3.1.3 in /share/vllm-0.8.5/lib/python3.10/site-packages (from ir-datasets->FlagEmbedding) (3.4.0)
-Requirement already satisfied: warc3-wet>=0.2.3 in /share/vllm-0.8.5/lib/python3.10/site-packages (from ir-datasets->FlagEmbedding) (0.2.5)
-Requirement already satisfied: inscriptis>=2.2.0 in /share/vllm-0.8.5/lib/python3.10/site-packages (from ir-datasets->FlagEmbedding) (2.6.0)
-Requirement already satisfied: beautifulsoup4>=4.4.1 in /share/vllm-0.8.5/lib/python3.10/site-packages (from ir-datasets->FlagEmbedding) (4.13.4)
-Requirement already satisfied: lxml>=4.5.2 in /share/vllm-0.8.5/lib/python3.10/site-packages (from ir-datasets->FlagEmbedding) (4.9.4)
-Requirement already satisfied: warc3-wet-clueweb09>=0.2.5 in /share/vllm-0.8.5/lib/python3.10/site-packages (from ir-datasets->FlagEmbedding) (0.2.5)
-Requirement already satisfied: unlzw3>=0.2.1 in /share/vllm-0.8.5/lib/python3.10/site-packages (from ir-datasets->FlagEmbedding) (0.2.3)
-Requirement already satisfied: scikit-learn in /share/vllm-0.8.5/lib/python3.10/site-packages (from sentence_transformers->FlagEmbedding) (1.7.0)
-Requirement already satisfied: Pillow in /share/vllm-0.8.5/lib/python3.10/site-packages (from sentence_transformers->FlagEmbedding) (11.2.1)
-Requirement already satisfied: scipy in /share/vllm-0.8.5/lib/python3.10/site-packages (from sentence_transformers->FlagEmbedding) (1.15.2)
-Requirement already satisfied: soupsieve>1.2 in /share/vllm-0.8.5/lib/python3.10/site-packages (from beautifulsoup4>=4.4.1->ir-datasets->FlagEmbedding) (2.7)
-Requirement already satisfied: aiohttp!=4.0.0a0,!=4.0.0a1 in /share/vllm-0.8.5/lib/python3.10/site-packages (from fsspec[http]<=2025.3.0,>=2023.1.0->datasets>=2.19.0->FlagEmbedding) (3.10.10)
-Requirement already satisfied: idna<4,>=2.5 in /share/vllm-0.8.5/lib/python3.10/site-packages (from requests>=2.32.2->datasets>=2.19.0->FlagEmbedding) (3.10)
-Requirement already satisfied: charset-normalizer<4,>=2 in /share/vllm-0.8.5/lib/python3.10/site-packages (from requests>=2.32.2->datasets>=2.19.0->FlagEmbedding) (3.4.1)
-Requirement already satisfied: urllib3<3,>=1.21.1 in /share/vllm-0.8.5/lib/python3.10/site-packages (from requests>=2.32.2->datasets>=2.19.0->FlagEmbedding) (2.4.0)
-Requirement already satisfied: certifi>=2017.4.17 in /share/vllm-0.8.5/lib/python3.10/site-packages (from requests>=2.32.2->datasets>=2.19.0->FlagEmbedding) (2025.4.26)
-Requirement already satisfied: cbor>=1.0.0 in /share/vllm-0.8.5/lib/python3.10/site-packages (from trec-car-tools>=2.5.4->ir-datasets->FlagEmbedding) (1.0.0)
-Requirement already satisfied: MarkupSafe>=2.0 in /share/vllm-0.8.5/lib/python3.10/site-packages (from jinja2->torch>=1.6.0->FlagEmbedding) (3.0.2)
-Requirement already satisfied: python-dateutil>=2.8.2 in /share/vllm-0.8.5/lib/python3.10/site-packages (from pandas->datasets>=2.19.0->FlagEmbedding) (2.9.0.post0)
-Requirement already satisfied: tzdata>=2022.7 in /share/vllm-0.8.5/lib/python3.10/site-packages (from pandas->datasets>=2.19.0->FlagEmbedding) (2025.2)
-Requirement already satisfied: pytz>=2020.1 in /share/vllm-0.8.5/lib/python3.10/site-packages (from pandas->datasets>=2.19.0->FlagEmbedding) (2025.2)
-Requirement already satisfied: threadpoolctl>=3.1.0 in /share/vllm-0.8.5/lib/python3.10/site-packages (from scikit-learn->sentence_transformers->FlagEmbedding) (3.6.0)
-Requirement already satisfied: joblib>=1.2.0 in /share/vllm-0.8.5/lib/python3.10/site-packages (from scikit-learn->sentence_transformers->FlagEmbedding) (1.5.1)
-Requirement already satisfied: async-timeout<5.0,>=4.0 in /share/vllm-0.8.5/lib/python3.10/site-packages (from aiohttp!=4.0.0a0,!=4.0.0a1->fsspec[http]<=2025.3.0,>=2023.1.0->datasets>=2.19.0->FlagEmbedding) (4.0.3)
-Requirement already satisfied: aiosignal>=1.1.2 in /share/vllm-0.8.5/lib/python3.10/site-packages (from aiohttp!=4.0.0a0,!=4.0.0a1->fsspec[http]<=2025.3.0,>=2023.1.0->datasets>=2.19.0->FlagEmbedding) (1.3.2)
-Requirement already satisfied: yarl<2.0,>=1.12.0 in /share/vllm-0.8.5/lib/python3.10/site-packages (from aiohttp!=4.0.0a0,!=4.0.0a1->fsspec[http]<=2025.3.0,>=2023.1.0->datasets>=2.19.0->FlagEmbedding) (1.20.0)
-Requirement already satisfied: multidict<7.0,>=4.5 in /share/vllm-0.8.5/lib/python3.10/site-packages (from aiohttp!=4.0.0a0,!=4.0.0a1->fsspec[http]<=2025.3.0,>=2023.1.0->datasets>=2.19.0->FlagEmbedding) (6.4.3)
-Requirement already satisfied: aiohappyeyeballs>=2.3.0 in /share/vllm-0.8.5/lib/python3.10/site-packages (from aiohttp!=4.0.0a0,!=4.0.0a1->fsspec[http]<=2025.3.0,>=2023.1.0->datasets>=2.19.0->FlagEmbedding) (2.6.1)
-Requirement already satisfied: attrs>=17.3.0 in /share/vllm-0.8.5/lib/python3.10/site-packages (from aiohttp!=4.0.0a0,!=4.0.0a1->fsspec[http]<=2025.3.0,>=2023.1.0->datasets>=2.19.0->FlagEmbedding) (25.3.0)
-Requirement already satisfied: frozenlist>=1.1.1 in /share/vllm-0.8.5/lib/python3.10/site-packages (from aiohttp!=4.0.0a0,!=4.0.0a1->fsspec[http]<=2025.3.0,>=2023.1.0->datasets>=2.19.0->FlagEmbedding) (1.6.0)
-Requirement already satisfied: six>=1.5 in /share/vllm-0.8.5/lib/python3.10/site-packages (from python-dateutil>=2.8.2->pandas->datasets>=2.19.0->FlagEmbedding) (1.17.0)
-Requirement already satisfied: propcache>=0.2.1 in /share/vllm-0.8.5/lib/python3.10/site-packages (from yarl<2.0,>=1.12.0->aiohttp!=4.0.0a0,!=4.0.0a1->fsspec[http]<=2025.3.0,>=2023.1.0->datasets>=2.19.0->FlagEmbedding) (0.3.1)

test/bgererank/entities.service

@@ -1,14 +0,0 @@
-[Unit]
-Wants=systemd-networkd.service
-
-[Service]
-WorkingDirectory=/share/run/entities
-ExecStart=/share/run/entities/start.sh
-ExecStop=/share/run/entities/stop.sh
-StandardOutput=append:/var/log/entities/entities.log
-StandardError=append:/var/log/entities/entities.log
-SyslogIdentifier=entities
-
-[Install]
-WantedBy=multi-user.target
-

test/bgererank/start.sh

@@ -1,3 +0,0 @@
-#!/bin/bash
-
-/share/vllm-0.8.5/bin/python -m llmengine.bgererank -p 8887 /share/models/BAAI/bge-reranker-v2-m3

test/bgererank/stop.sh

@@ -1,10 +0,0 @@
-#!/bin/bash
-
-# 查找并终止运行在端口 8887 上的进程
-pid=$(lsof -t -i:8887)
-if [ -n "$pid" ]; then
-    echo "终止进程: $pid"
-    kill -9 $pid
-else
-    echo "未找到运行在端口 8887 上的进程"
-fi

test/neo4j/conf/config.json

@@ -1,5 +1,10 @@
 {
     "filesroot": "$[workdir]$/files",
+    "neo4j": {
+        "uri": "bolt://10.18.34.18:7687",
+        "user": "neo4j",
+        "password": "261229..wmh"
+    },
     "logger": {
         "name": "llmengine",
         "levelname": "info",
@@ -11,7 +16,7 @@
         ],
         "client_max_size": 10000,
         "host": "0.0.0.0",
-        "port": 8887,
+        "port": 8885,
         "coding": "utf-8",
         "indexes": [
             "index.html",
@@ -19,11 +24,35 @@
         ],
         "startswiths": [
             {
-                "leading": "/v1/bgererank",
+                "leading": "/idfile",
-                "registerfunction": "bgererank"
+                "registerfunction": "idfile"
             },
             {
-                "leading": "/v1/docs",
+                "leading": "/v1/connection",
+                "registerfunction": "connection"
+            },
+            {
+                "leading": "/v1/initialize",
+                "registerfunction": "initialize"
+            },
+            {
+                "leading": "/v1/inserttriples",
+                "registerfunction": "inserttriples"
+            },
+            {
+                "leading": "/v1/deletedocument",
+                "registerfunction": "deletedocument"
+            },
+            {
+                "leading": "/v1/deleteknowledgebase",
+                "registerfunction": "deleteknowledgebase"
+            },
+            {
+                "leading": "/v1/matchtriplets",
+                "registerfunction": "matchtriplets"
+            },
+            {
+                "leading": "/docs",
                 "registerfunction": "docs"
             }
         ],
@@ -44,4 +73,4 @@
             "url": "redis://127.0.0.1:6379"
         }
     }
-}
+}

test/neo4j/neo4j.service

@@ -0,0 +1,15 @@
+[Unit]
+Wants=systemd-networkd.service
+
+[Service]
+Type=forking
+WorkingDirectory=/share/run/neo4j
+ExecStart=/share/run/neo4j/start.sh
+ExecStop=/share/run/neo4j/stop.sh
+StandardOutput=append:/var/log/neo4j/neo4j.log
+StandardError=append:/var/log/neo4j/neo4j.log
+SyslogIdentifier=/share/run/neo4j
+TimeoutStartSec=300
+
+[Install]
+WantedBy=multi-user.target

test/neo4j/start.sh

@@ -0,0 +1,2 @@
+#!/bin/bash
+CUDA_VISIBLE_DEVICES=7 /share/vllm-0.8.5/bin/python -m llmengine.neo4j -p 8885 Neo4j &

test/neo4j/stop.sh

@@ -0,0 +1,12 @@
+#!/bin/bash
+
+PORT=8885
+PID=$(lsof -t -i:$PORT)
+
+if [ -n "$PID" ]; then
+    echo "找到端口 $PORT 的进程: PID=$PID"
+    kill -9 $PID
+    echo "已终止端口 $PORT 的进程"
+else
+    echo "未找到端口 $PORT 的进程"
+fi