From 84f3d6e284574f9c892feb42bdcbe6586cf9c9ce Mon Sep 17 00:00:00 2001 From: yumoqing Date: Fri, 24 Jul 2026 13:24:20 +0800 Subject: [PATCH] init: RAG business module --- __init__.py | 9 + conf/config.json | 12 + documents.json | 25 + engine_configs.json | 26 + init.py | 551 ++++++++++++++++++ knowledge_bases.json | 26 + models/document_chunks.json | 30 + models/documents.json | 37 ++ models/engine_configs.json | 33 ++ models/entities.json | 35 ++ models/entity_relations.json | 29 + models/knowledge_bases.json | 32 + models/media_tags.json | 28 + models/subscriptions.json | 34 ++ models/tags.json | 24 + models/usage_logs.json | 30 + setup.py | 12 + subscriptions.json | 28 + tags.json | 22 + wwwroot/documents_list/index.ui | 65 +++ wwwroot/engine_configs_list/index.ui | 148 +++++ wwwroot/knowledge_bases_list/create_kb.dspy | 19 + .../delete_tree_item.dspy | 7 + wwwroot/knowledge_bases_list/detail.ui | 92 +++ .../knowledge_bases_list/get_tree_data.dspy | 17 + wwwroot/knowledge_bases_list/index.ui | 43 ++ wwwroot/knowledge_bases_list/new_kb_form.ui | 30 + .../knowledge_bases_list/new_tree_item.dspy | 15 + .../update_tree_item.dspy | 10 + wwwroot/knowledge_bases_list/upload.dspy | 32 + wwwroot/knowledge_bases_list/upload.js | 26 + wwwroot/subscriptions_list/index.ui | 121 ++++ 32 files changed, 1648 insertions(+) create mode 100644 __init__.py create mode 100644 conf/config.json create mode 100644 documents.json create mode 100644 engine_configs.json create mode 100644 init.py create mode 100644 knowledge_bases.json create mode 100644 models/document_chunks.json create mode 100644 models/documents.json create mode 100644 models/engine_configs.json create mode 100644 models/entities.json create mode 100644 models/entity_relations.json create mode 100644 models/knowledge_bases.json create mode 100644 models/media_tags.json create mode 100644 models/subscriptions.json create mode 100644 models/tags.json create mode 100644 models/usage_logs.json create mode 100644 setup.py create mode 100644 subscriptions.json create mode 100644 tags.json create mode 100644 wwwroot/documents_list/index.ui create mode 100644 wwwroot/engine_configs_list/index.ui create mode 100644 wwwroot/knowledge_bases_list/create_kb.dspy create mode 100644 wwwroot/knowledge_bases_list/delete_tree_item.dspy create mode 100644 wwwroot/knowledge_bases_list/detail.ui create mode 100644 wwwroot/knowledge_bases_list/get_tree_data.dspy create mode 100644 wwwroot/knowledge_bases_list/index.ui create mode 100644 wwwroot/knowledge_bases_list/new_kb_form.ui create mode 100644 wwwroot/knowledge_bases_list/new_tree_item.dspy create mode 100644 wwwroot/knowledge_bases_list/update_tree_item.dspy create mode 100644 wwwroot/knowledge_bases_list/upload.dspy create mode 100644 wwwroot/knowledge_bases_list/upload.js create mode 100644 wwwroot/subscriptions_list/index.ui diff --git a/__init__.py b/__init__.py new file mode 100644 index 0000000..8e7eac3 --- /dev/null +++ b/__init__.py @@ -0,0 +1,9 @@ +# -*- coding:utf-8 -*- +""" +RAG业务模块 - 知识库、文档、引擎配置、订阅管理 +作为独立模块被ragserver或其他应用引用 +""" + +from .init import init_rag_module + +__all__ = ['init_rag_module'] diff --git a/conf/config.json b/conf/config.json new file mode 100644 index 0000000..68ec277 --- /dev/null +++ b/conf/config.json @@ -0,0 +1,12 @@ +{ + "module_name": "rag", + "registerfunction": { + "kb_list": "rag", + "engines": "rag", + "doc_upload": "rag", + "doc_delete": "rag", + "dir_create": "rag", + "dir_delete": "rag", + "dir_list": "rag" + } +} diff --git a/documents.json b/documents.json new file mode 100644 index 0000000..937f7d1 --- /dev/null +++ b/documents.json @@ -0,0 +1,25 @@ +{ + "tblname": "documents", + "params": { + "title": "文档管理", + "browserfields": [ + {"field": "file_name", "title": "文件名", "width": "25%"}, + {"field": "file_type", "title": "类型", "width": "10%"}, + {"field": "file_size", "title": "大小", "width": "10%"}, + {"field": "chunk_count", "title": "分片数", "width": "8%"}, + {"field": "status", "title": "状态", "width": "10%"}, + {"field": "kb_id", "title": "知识库", "width": "17%"}, + {"field": "created_at", "title": "上传时间", "width": "20%"} + ], + "editfields": [ + {"field": "file_name", "uitype": "Text", "required": true}, + {"field": "kb_id", "uitype": "Text", "required": true}, + {"field": "file_type", "uitype": "Text"} + ], + "searchfields": ["file_name"], + "sort": "created_at desc", + "data_filter": { + "org_id": "{{userorgid}}" + } + } +} diff --git a/engine_configs.json b/engine_configs.json new file mode 100644 index 0000000..c338fb2 --- /dev/null +++ b/engine_configs.json @@ -0,0 +1,26 @@ +{ + "tblname": "engine_configs", + "params": { + "title": "引擎配置", + "browserfields": [ + {"field": "engine_type", "title": "引擎类型", "width": "15%"}, + {"field": "engine_name", "title": "名称", "width": "15%"}, + {"field": "endpoint_url", "title": "服务地址", "width": "25%"}, + {"field": "model_name", "title": "模型", "width": "15%"}, + {"field": "is_default", "title": "默认", "width": "8%"}, + {"field": "status", "title": "状态", "width": "10%"}, + {"field": "org_id", "title": "机构", "width": "12%"} + ], + "editfields": [ + {"field": "engine_type", "uitype": "Text", "required": true}, + {"field": "engine_name", "uitype": "Text", "required": true}, + {"field": "endpoint_url", "uitype": "Text"}, + {"field": "api_key", "uitype": "Text"}, + {"field": "model_name", "uitype": "Text"}, + {"field": "is_default", "uitype": "Text"}, + {"field": "config_json", "uitype": "Text"} + ], + "searchfields": ["engine_type", "engine_name", "endpoint_url"], + "sort": "engine_type, priority desc" + } +} diff --git a/init.py b/init.py new file mode 100644 index 0000000..e7f3fda --- /dev/null +++ b/init.py @@ -0,0 +1,551 @@ +# -*- coding:utf-8 -*- +""" +RagServer DSPY Handlers - RAG 核心业务逻辑 + 文件上传/删除 +""" +from traceback import format_exc +from ahserver.serverenv import ServerEnv +from appPublic.registerfunction import RegisterFunction +from appPublic.log import debug, exception +from sqlor.dbpools import get_sor_context +import json, os, uuid, time + + +async def status_handler(request, params_kw, *args, **kwargs): + return json.dumps({ + "service": "ragserver", "version": "0.1.0", + "endpoints": ["/api/status", "/api/kb/list", "/api/doc/upload", + "/api/doc/delete", "/api/search", "/api/engines", + "/api/dir/create", "/api/dir/delete", "/api/dir/list", + "/api/tag/create", "/api/tag/list", "/api/tag/delete", + "/api/tag/assign", "/api/tag/unassign", "/api/tag/media_tags", + "/api/tag/search"] + }, indent=2, ensure_ascii=False) + + +async def kb_list_handler(request, params_kw, *args, **kwargs): + env = request._run_ns + try: + userorgid = await env.get_userorgid() + async with get_sor_context(env, 'rag') as sor: + recs = await sor.R("knowledge_bases", {}) + cards = [] + for r in recs: + doc_recs = await sor.R("documents", {"kb_id": r.id}) + doc_count = len(doc_recs) + total_size = r.total_size or 0 + if total_size >= 1073741824: + size_str = f"{total_size/1073741824:.1f}GB" + elif total_size >= 1048576: + size_str = f"{total_size/1048576:.0f}MB" + elif total_size >= 1024: + size_str = f"{total_size/1024:.0f}KB" + else: + size_str = f"{total_size}B" + engine = r.embedding_engine or "CLIP" + card = {"widgettype":"VBox","options":{"cwidth":16,"cheight":12,"bgcolor":"#f0f7ff","padding":"16px","css":"card clickable","border":"1px solid #d0e4f7"},"subwidgets":[ + {"widgettype":"Text","options":{"text":"📚 " + str(r.name),"cfontsize":16,"fontWeight":"bold"}}, + {"widgettype":"Text","options":{"text":str(engine)+" · 1024维","cfontsize":12,"color":"#888"}}, + {"widgettype":"HBox","options":{"spacing":"12px"},"subwidgets":[ + {"widgettype":"Text","options":{"text":"📄 "+str(doc_count)+"文档","cfontsize":12,"color":"#666"}}, + {"widgettype":"Text","options":{"text":"💾 "+size_str,"cfontsize":12,"color":"#666"}}]}], + "binds":[{"wid":"self","event":"click","actiontype":"urlwidget","target":"app.rag_main_content","mode":"replace","options":{"url":"/knowledge_bases_list/detail.ui","params":{"kb_id":str(r.id),"kb_name":str(r.name)}}}]} + cards.append(card) + if not cards: + cards.append({"widgettype":"Text","options":{"text":"暂无知识库","color":"#aaa","cfontsize":14,"halign":"center"}}) + result = {"widgettype":"HBox","options":{"width":"100%","spacing":"12px","wrap":True},"subwidgets":cards} + return json.dumps(result, ensure_ascii=False) + except Exception as e: + exception(f"kb_list: {e}, {format_exc()}") + return json.dumps({"error": str(e)}) + + +async def engines_handler(request, params_kw, *args, **kwargs): + env = request._run_ns + try: + userorgid = await env.get_userorgid() + async with get_sor_context(env, 'rag') as sor: + sql = "SELECT * FROM engine_configs WHERE status='active' AND (org_id IS NULL OR org_id=${org_id}$) ORDER BY engine_type, priority DESC" + recs = await sor.sqlExe(sql, {"org_id": userorgid}) + rows = [dict(r) for r in recs] + return json.dumps({"status": "SUCCEEDED", "data": {"rows": rows, "total": len(rows)}}, ensure_ascii=False, default=str) + except Exception as e: + exception(f"engines: {e}, {format_exc()}") + return json.dumps({"error": str(e)}) + + +async def search_handler(request, params_kw, *args, **kwargs): + env = request._run_ns + try: + query = params_kw.get("query", "") + kb_id = params_kw.get("kb_id", "") + top_k = int(params_kw.get("top_k", 5)) + if not query: + return json.dumps({"error": "query required"}) + try: + from pipeline import search as pipeline_search + result = pipeline_search(query, pipeline_name="kg-rag-standard", + collection=kb_id or "knowledge", + graph_name=kb_id or "knowledge", + top_k=top_k, llm_func=None) + return json.dumps(result, ensure_ascii=False) + except ImportError: + return json.dumps({"status": "FALLBACK", "message": "pipeline not available"}) + except Exception as e: + exception(f"search: {e}, {format_exc()}") + return json.dumps({"error": str(e)}) + + +async def doc_upload_handler(request, params_kw, *args, **kwargs): + """文件上传 → 保存 → DB记录 → 触发RAG入库""" + env = request._run_ns + try: + userorgid = await env.get_userorgid() + kb_id = params_kw.get("kb_id", "") + file_name = params_kw.get("file_name", "upload.bin") + if not kb_id: + return json.dumps({"error": "kb_id required"}) + + # Read raw file data from request body + file_data = await request.read() + + if not file_data: + return json.dumps({"error": "no file data"}) + + # Save file + doc_id = uuid.uuid4().hex[:16] + ext = os.path.splitext(file_name)[1] or ".bin" + saved_name = f"{doc_id}{ext}" + files_dir = os.path.join(os.path.dirname(os.path.abspath(__file__)), "files") + os.makedirs(files_dir, exist_ok=True) + file_path = os.path.join(files_dir, saved_name) + with open(file_path, "wb") as f: + f.write(file_data) + + file_size = len(file_data) + file_type = _detect_file_type(file_name, "application/octet-stream") + + # Create document record + async with get_sor_context(env, 'rag') as sor: + await sor.sqlExe( + "INSERT INTO documents (id, kb_id, file_name, file_type, file_size, file_path, mime_type, status, org_id, created_at, updated_at) " + "VALUES (${id}$, ${kb_id}$, ${file_name}$, ${file_type}$, ${file_size}$, ${file_path}$, ${mime_type}$, 'pending', ${org_id}$, NOW(), NOW())", + {"id": doc_id, "kb_id": kb_id, "file_name": file_name, "file_type": file_type, + "file_size": file_size, "file_path": "/idfile/files/" + saved_name, + "mime_type": "application/octet-stream", "org_id": userorgid}) + + # Update KB stats + await sor.sqlExe( + "UPDATE knowledge_bases SET doc_count=doc_count+1, total_size=total_size+${size}$ WHERE id=${kb_id}$", + {"size": file_size, "kb_id": kb_id}) + + # Trigger async ingest for text-based files + ingest_result = None + if file_type == "text": + try: + text = file_data.decode("utf-8", errors="replace") + from pipeline import ingest as pipeline_ingest + ingest_result = pipeline_ingest( + text, pipeline_name="kg-rag-standard", + collection=kb_id, graph_name=kb_id, llm_func=None) + # Update status to done + async with get_sor_context(env, 'rag') as sor: + await sor.sqlExe( + "UPDATE documents SET status='done', chunk_count=${chunks}$ WHERE id=${id}$", + {"chunks": ingest_result.get("chunks", 0) if ingest_result else 0, "id": doc_id}) + if ingest_result: + await sor.sqlExe( + "UPDATE knowledge_bases SET chunk_count=chunk_count+${n}$ WHERE id=${kb_id}$", + {"n": ingest_result.get("chunks", 0), "kb_id": kb_id}) + except Exception as e: + exception(f"async ingest failed: {e}") + + return json.dumps({ + "status": "SUCCEEDED", + "doc_id": doc_id, + "file_name": file_name, + "file_size": file_size, + "file_type": file_type, + "ingest": ingest_result + }, ensure_ascii=False, default=str) + + except Exception as e: + exception(f"doc_upload: {e}, {format_exc()}") + return json.dumps({"error": str(e)}) + + +async def doc_delete_handler(request, params_kw, *args, **kwargs): + """文件删除 → 清理VDB → 清理图 → 清理DB → 删文件""" + env = request._run_ns + try: + doc_id = params_kw.get("doc_id", "") + if not doc_id: + return json.dumps({"error": "doc_id required"}) + + async with get_sor_context(env, 'rag') as sor: + # Get document info + recs = await sor.R("documents", {"id": doc_id}) + if not recs: + return json.dumps({"error": "document not found"}) + doc = recs[0] + + # Get chunks to clean VDB + chunks = await sor.R("document_chunks", {"doc_id": doc_id}) + + # Delete from VDB + if chunks: + vector_ids = [c.vector_id for c in chunks if c.vector_id] + if vector_ids: + try: + _call_vdb("/v1/delete", {"colname": doc.kb_id, "ids": vector_ids}) + except Exception as e: + exception(f"vdb delete failed: {e}") + + # Delete from graph + entities = await sor.R("entities", {"kb_id": doc.kb_id}) + if entities: + try: + _call_graph("/api/graph/save", {"graph": doc.kb_id}) + except Exception as e: + exception(f"graph cleanup failed: {e}") + + # Delete DB records + await sor.sqlExe("DELETE FROM document_chunks WHERE doc_id=${id}$", {"id": doc_id}) + await sor.sqlExe("DELETE FROM entities WHERE kb_id=${kb_id}$", {"kb_id": doc.kb_id}) + await sor.sqlExe("DELETE FROM entity_relations WHERE kb_id=${kb_id}$", {"kb_id": doc.kb_id}) + await sor.sqlExe("DELETE FROM documents WHERE id=${id}$", {"id": doc_id}) + + # Update KB stats + await sor.sqlExe( + "UPDATE knowledge_bases SET doc_count=GREATEST(doc_count-1,0), total_size=GREATEST(total_size-${size}$,0), chunk_count=GREATEST(chunk_count-${n}$,0) WHERE id=${kb_id}$", + {"size": doc.file_size, "n": len(chunks), "kb_id": doc.kb_id}) + + # Delete file from disk + file_path = doc.file_path + if file_path and file_path.startswith("/idfile/"): + real_path = os.path.join(os.path.dirname(os.path.abspath(__file__)), "files", + os.path.basename(file_path)) + if os.path.exists(real_path): + os.remove(real_path) + + return json.dumps({"status": "SUCCEEDED", "doc_id": doc_id, "chunks_deleted": len(chunks)}) + + except Exception as e: + exception(f"doc_delete: {e}, {format_exc()}") + return json.dumps({"error": str(e)}) + + +def _detect_file_type(name, mime): + """Detect file type from name and MIME""" + ext = os.path.splitext(name)[1].lower() + if ext in ('.txt', '.md', '.json', '.csv', '.xml', '.html', '.py', '.js', '.css', '.yaml', '.yml'): + return "text" + if ext in ('.jpg', '.jpeg', '.png', '.gif', '.bmp', '.webp', '.svg'): + return "image" + if ext in ('.mp3', '.wav', '.flac', '.ogg', '.m4a', '.aac'): + return "audio" + if ext in ('.mp4', '.avi', '.mov', '.mkv', '.webm'): + return "video" + if ext == '.pdf': + return "text" + return "other" + + +def _call_vdb(path, data, timeout=10): + """Call VDB service""" + import urllib.request + url = f"http://localhost:8886{path}" + req = urllib.request.Request(url, data=json.dumps(data).encode(), + headers={"Content-Type": "application/json"}) + with urllib.request.urlopen(req, timeout=timeout) as resp: + return json.loads(resp.read()) + + +def _call_graph(path, data, timeout=10): + """Call Graph service""" + import urllib.request + url = f"http://localhost:9092{path}" + req = urllib.request.Request(url, data=json.dumps(data).encode(), + headers={"Content-Type": "application/json"}) + with urllib.request.urlopen(req, timeout=timeout) as resp: + return json.loads(resp.read()) + + +async def dir_create_handler(request, params_kw, *args, **kwargs): + """创建目录""" + env = request._run_ns + try: + kb_id = params_kw.get("kb_id", "") + parent_id = params_kw.get("parent_id", "") + dir_name = params_kw.get("dir_name", "") + if not kb_id or not dir_name: + return json.dumps({"error": "kb_id and dir_name required"}) + async with get_sor_context(env, 'rag') as sor: + dir_id = uuid.uuid4().hex[:16] + await sor.sqlExe( + "INSERT INTO document_chunks (id, doc_id, kb_id, chunk_index, chunk_type, content, description, created_at) " + "VALUES (${id}$, '', ${kb_id}$, 0, 'directory', ${name}$, ${parent}$, NOW())", + {"id": dir_id, "kb_id": kb_id, "name": dir_name, "parent": parent_id}) + return json.dumps({"status": "SUCCEEDED", "dir_id": dir_id}) + except Exception as e: + exception(f"dir_create: {e}") + return json.dumps({"error": str(e)}) + + +async def dir_delete_handler(request, params_kw, *args, **kwargs): + """删除目录/文件""" + env = request._run_ns + try: + item_id = params_kw.get("item_id", "") + if not item_id: + return json.dumps({"error": "item_id required"}) + async with get_sor_context(env, 'rag') as sor: + await sor.sqlExe("DELETE FROM document_chunks WHERE id=${id}$ OR doc_id=${id}$", {"id": item_id}) + return json.dumps({"status": "SUCCEEDED"}) + except Exception as e: + exception(f"dir_delete: {e}") + return json.dumps({"error": str(e)}) + + +async def dir_list_handler(request, params_kw, *args, **kwargs): + """列出知识库的目录树""" + env = request._run_ns + try: + kb_id = params_kw.get("kb_id", "") + if not kb_id: + return json.dumps({"error": "kb_id required"}) + async with get_sor_context(env, 'rag') as sor: + # Get directories (chunk_type='directory') + dirs = await sor.sqlExe( + "SELECT id, content as label, description as parent_id FROM document_chunks WHERE kb_id=${kb_id}$ AND chunk_type='directory'", + {"kb_id": kb_id}) + # Get files (documents table) + docs = await sor.sqlExe( + "SELECT id, file_name as label, '' as parent_id FROM documents WHERE kb_id=${kb_id}$", + {"kb_id": kb_id}) + items = [] + for d in dirs: + items.append({"id": d.id, "label": d.label, "parent_id": d.parent_id or "", "type": "dir"}) + for d in docs: + items.append({"id": d.id, "label": d.label, "parent_id": "", "type": "file"}) + return json.dumps({"status": "SUCCEEDED", "items": items}) + except Exception as e: + exception(f"dir_list: {e}") + return json.dumps({"error": str(e)}) + + +async def tag_create_handler(request, params_kw, *args, **kwargs): + """创建标签""" + env = request._run_ns + try: + kb_id = params_kw.get("kb_id", "") + name = params_kw.get("name", "").strip() + color = params_kw.get("color", "#3b82f6") + if not kb_id or not name: + return json.dumps({"error": "kb_id and name required"}) + userorgid = await env.get_userorgid() + async with get_sor_context(env, 'rag') as sor: + tag_id = uuid.uuid4().hex[:16] + await sor.sqlExe( + "INSERT INTO tags (id, kb_id, name, color, org_id, created_at) " + "VALUES (${id}$, ${kb_id}$, ${name}$, ${color}$, ${org_id}$, NOW())", + {"id": tag_id, "kb_id": kb_id, "name": name, "color": color, "org_id": userorgid}) + return json.dumps({"status": "SUCCEEDED", "tag_id": tag_id, "name": name, "color": color}) + except Exception as e: + exception(f"tag_create: {e}, {format_exc()}") + return json.dumps({"error": str(e)}) + + +async def tag_list_handler(request, params_kw, *args, **kwargs): + """列出知识库的所有标签""" + env = request._run_ns + try: + kb_id = params_kw.get("kb_id", "") + if not kb_id: + return json.dumps({"error": "kb_id required"}) + userorgid = await env.get_userorgid() + async with get_sor_context(env, 'rag') as sor: + recs = await sor.R("tags", {"kb_id": kb_id, "org_id": userorgid}) + tags = [{"id": r.id, "name": r.name, "color": r.color, "created_at": str(r.created_at)} for r in recs] + return json.dumps({"status": "SUCCEEDED", "tags": tags}, ensure_ascii=False, default=str) + except Exception as e: + exception(f"tag_list: {e}, {format_exc()}") + return json.dumps({"error": str(e)}) + + +async def tag_delete_handler(request, params_kw, *args, **kwargs): + """删除标签（级联删除关联）""" + env = request._run_ns + try: + tag_id = params_kw.get("tag_id", "") + if not tag_id: + return json.dumps({"error": "tag_id required"}) + async with get_sor_context(env, 'rag') as sor: + await sor.sqlExe("DELETE FROM media_tags WHERE tag_id=${id}$", {"id": tag_id}) + await sor.sqlExe("DELETE FROM tags WHERE id=${id}$", {"id": tag_id}) + return json.dumps({"status": "SUCCEEDED", "tag_id": tag_id}) + except Exception as e: + exception(f"tag_delete: {e}, {format_exc()}") + return json.dumps({"error": str(e)}) + + +async def tag_assign_handler(request, params_kw, *args, **kwargs): + """给媒体/人脸/声纹打标签""" + env = request._run_ns + try: + kb_id = params_kw.get("kb_id", "") + media_type = params_kw.get("media_type", "") # document / face / voice + media_id = params_kw.get("media_id", "") + tag_id = params_kw.get("tag_id", "") + if not all([kb_id, media_type, media_id, tag_id]): + return json.dumps({"error": "kb_id, media_type, media_id, tag_id required"}) + if media_type not in ("document", "face", "voice"): + return json.dumps({"error": "media_type must be document/face/voice"}) + async with get_sor_context(env, 'rag') as sor: + mt_id = uuid.uuid4().hex[:16] + await sor.sqlExe( + "INSERT INTO media_tags (id, kb_id, media_type, media_id, tag_id, created_at) " + "VALUES (${id}$, ${kb_id}$, ${type}$, ${mid}$, ${tid}$, NOW())", + {"id": mt_id, "kb_id": kb_id, "type": media_type, "mid": media_id, "tid": tag_id}) + return json.dumps({"status": "SUCCEEDED", "media_tag_id": mt_id}) + except Exception as e: + exception(f"tag_assign: {e}, {format_exc()}") + return json.dumps({"error": str(e)}) + + +async def tag_unassign_handler(request, params_kw, *args, **kwargs): + """取消标签关联""" + env = request._run_ns + try: + media_type = params_kw.get("media_type", "") + media_id = params_kw.get("media_id", "") + tag_id = params_kw.get("tag_id", "") + if not all([media_type, media_id, tag_id]): + return json.dumps({"error": "media_type, media_id, tag_id required"}) + async with get_sor_context(env, 'rag') as sor: + await sor.sqlExe( + "DELETE FROM media_tags WHERE media_type=${type}$ AND media_id=${mid}$ AND tag_id=${tid}$", + {"type": media_type, "mid": media_id, "tid": tag_id}) + return json.dumps({"status": "SUCCEEDED"}) + except Exception as e: + exception(f"tag_unassign: {e}, {format_exc()}") + return json.dumps({"error": str(e)}) + + +async def tag_media_tags_handler(request, params_kw, *args, **kwargs): + """查询某媒体的所有标签""" + env = request._run_ns + try: + media_type = params_kw.get("media_type", "") + media_id = params_kw.get("media_id", "") + if not all([media_type, media_id]): + return json.dumps({"error": "media_type and media_id required"}) + async with get_sor_context(env, 'rag') as sor: + recs = await sor.sqlExe( + "SELECT t.id, t.name, t.color FROM media_tags mt " + "JOIN tags t ON mt.tag_id=t.id " + "WHERE mt.media_type=${type}$ AND mt.media_id=${mid}$", + {"type": media_type, "mid": media_id}) + tags = [{"id": r.id, "name": r.name, "color": r.color} for r in recs] + return json.dumps({"status": "SUCCEEDED", "tags": tags}) + except Exception as e: + exception(f"tag_media_tags: {e}, {format_exc()}") + return json.dumps({"error": str(e)}) + + +async def tag_search_handler(request, params_kw, *args, **kwargs): + """组合标签检索：按 tag_ids 过滤，再语义检索""" + env = request._run_ns + try: + query = params_kw.get("query", "") + kb_id = params_kw.get("kb_id", "") + tag_ids_str = params_kw.get("tag_ids", "") # comma-separated tag IDs + top_k = int(params_kw.get("top_k", 5)) + match_mode = params_kw.get("match_mode", "any") # any / all + if not kb_id: + return json.dumps({"error": "kb_id required"}) + async with get_sor_context(env, 'rag') as sor: + if tag_ids_str: + tag_ids = [t.strip() for t in tag_ids_str.split(",") if t.strip()] + media_ids_by_tag = [] + for tid in tag_ids: + recs = await sor.sqlExe( + "SELECT media_type, media_id FROM media_tags WHERE kb_id=${kb_id}$ AND tag_id=${tid}$", + {"kb_id": kb_id, "tid": tid}) + mids = {(r.media_type, r.media_id) for r in recs} + media_ids_by_tag.append(mids) + if match_mode == "all": + matched = media_ids_by_tag[0] + for s in media_ids_by_tag[1:]: + matched = matched & s + else: + matched = set() + for s in media_ids_by_tag: + matched |= s + if not matched: + return json.dumps({"status": "SUCCEEDED", "results": [], "message": "no media match tags"}) + doc_ids = [mid for mt, mid in matched if mt == "document"] + face_ids = [mid for mt, mid in matched if mt == "face"] + voice_ids = [mid for mt, mid in matched if mt == "voice"] + results = [] + if doc_ids: + docs = await sor.sqlExe( + "SELECT id, file_name, file_type, file_size, status, created_at FROM documents WHERE id IN (${ids}$)", + {"ids": doc_ids}) + for d in docs: + results.append({"type": "document", "id": d.id, "name": d.file_name, "file_type": d.file_type, "size": d.file_size, "status": d.status, "created_at": str(d.created_at)}) + if face_ids: + faces = await sor.sqlExe( + "SELECT id, name, description, face_embedding_id, created_at FROM entities WHERE id IN (${ids}$) AND entity_type='person'", + {"ids": face_ids}) + for f in faces: + results.append({"type": "face", "id": f.id, "name": f.name, "description": f.description, "created_at": str(f.created_at)}) + if voice_ids: + voices = await sor.sqlExe( + "SELECT id, name, description, voice_embedding_id, created_at FROM entities WHERE id IN (${ids}$) AND entity_type='voice'", + {"ids": voice_ids}) + for v in voices: + results.append({"type": "voice", "id": v.id, "name": v.name, "description": v.description, "created_at": str(v.created_at)}) + if query: + import urllib.request + tag_filtered_docs = [r for r in results if r["type"] == "document"] + vdb_docs = [] + for doc in tag_filtered_docs: + chunks = await sor.sqlExe( + "SELECT content FROM document_chunks WHERE doc_id=${id}$ LIMIT 3", + {"id": doc["id"]}) + for c in chunks: + vdb_docs.append({"doc_id": doc["id"], "content": c.content}) + results = {"documents": [{"id": r["id"], "name": r["name"], "file_type": r["file_type"]} for r in tag_filtered_docs], + "faces": [r for r in results if r["type"] == "face"], + "voices": [r for r in results if r["type"] == "voice"], + "chunks": vdb_docs, + "query": query, + "tag_ids": tag_ids, + "match_mode": match_mode} + return json.dumps({"status": "SUCCEEDED", "results": results, "tag_ids": tag_ids, "match_mode": match_mode}, ensure_ascii=False, default=str) + else: + return json.dumps({"status": "SUCCEEDED", "results": [], "message": "no tag_ids provided"}) + except Exception as e: + exception(f"tag_search: {e}, {format_exc()}") + return json.dumps({"error": str(e)}) + + +def init_ragserver(): + env = ServerEnv() + rf = RegisterFunction() + rf.register("status", status_handler) + rf.register("kb_list", kb_list_handler) + rf.register("engines", engines_handler) + rf.register("search", search_handler) + rf.register("doc_upload", doc_upload_handler) + rf.register("doc_delete", doc_delete_handler) + rf.register("dir_create", dir_create_handler) + rf.register("dir_delete", dir_delete_handler) + rf.register("dir_list", dir_list_handler) + rf.register("tag_create", tag_create_handler) + rf.register("tag_list", tag_list_handler) + rf.register("tag_delete", tag_delete_handler) + rf.register("tag_assign", tag_assign_handler) + rf.register("tag_unassign", tag_unassign_handler) + rf.register("tag_media_tags", tag_media_tags_handler) + rf.register("tag_search", tag_search_handler) diff --git a/knowledge_bases.json b/knowledge_bases.json new file mode 100644 index 0000000..99482d5 --- /dev/null +++ b/knowledge_bases.json @@ -0,0 +1,26 @@ +{ + "tblname": "knowledge_bases", + "params": { + "title": "知识库管理", + "browserfields": [ + {"field": "name", "title": "名称", "width": "20%"}, + {"field": "description", "title": "描述", "width": "30%"}, + {"field": "doc_count", "title": "文档数", "width": "10%"}, + {"field": "total_size", "title": "大小", "width": "10%"}, + {"field": "embedding_engine", "title": "向量引擎", "width": "15%"}, + {"field": "created_at", "title": "创建时间", "width": "15%"} + ], + "editfields": [ + {"field": "name", "uitype": "Text", "required": true}, + {"field": "description", "uitype": "Text"}, + {"field": "embedding_engine", "uitype": "Text", "default": "clip-vith14"}, + {"field": "vdb_collection", "uitype": "Text"}, + {"field": "graph_name", "uitype": "Text"} + ], + "searchfields": ["name", "description"], + "sort": "created_at desc", + "data_filter": { + "org_id": "{{userorgid}}" + } + } +} diff --git a/models/document_chunks.json b/models/document_chunks.json new file mode 100644 index 0000000..c545831 --- /dev/null +++ b/models/document_chunks.json @@ -0,0 +1,30 @@ +{ + "summary": [ + { + "name": "document_chunks", + "title": "文档分片", + "primary": ["id"] + } + ], + "fields": [ + {"name": "id", "title": "ID", "type": "str", "length": 32}, + {"name": "doc_id", "title": "文档ID", "type": "str", "length": 32}, + {"name": "kb_id", "title": "知识库ID", "type": "str", "length": 32}, + {"name": "chunk_index", "title": "分片序号", "type": "int", "default": 0}, + {"name": "chunk_type", "title": "分片类型", "type": "str", "length": 32}, + {"name": "content", "title": "文本内容", "type": "text", "length": 65535}, + {"name": "description", "title": "描述", "type": "text", "length": 65535}, + {"name": "vector_id", "title": "向量库ID", "type": "str", "length": 128}, + {"name": "start_offset", "title": "起始偏移", "type": "int", "default": 0}, + {"name": "end_offset", "title": "结束偏移", "type": "int", "default": 0}, + {"name": "time_start", "title": "时间起始(秒)", "type": "float", "length": 10, "dec": 3}, + {"name": "time_end", "title": "时间结束(秒)", "type": "float", "length": 10, "dec": 3}, + {"name": "metadata", "title": "元数据JSON", "type": "text", "length": 2000}, + {"name": "created_at", "title": "创建时间", "type": "datetime"} + ], + "indexes": [ + {"name": "idx_chunk_doc", "idxtype": "index", "idxfields": ["doc_id"]}, + {"name": "idx_chunk_kb", "idxtype": "index", "idxfields": ["kb_id"]}, + {"name": "idx_chunk_vector", "idxtype": "index", "idxfields": ["vector_id"]} + ] +} diff --git a/models/documents.json b/models/documents.json new file mode 100644 index 0000000..ff04880 --- /dev/null +++ b/models/documents.json @@ -0,0 +1,37 @@ +{ + "summary": [ + { + "name": "documents", + "title": "文档", + "primary": ["id"] + } + ], + "fields": [ + {"name": "id", "title": "ID", "type": "str", "length": 32}, + {"name": "kb_id", "title": "知识库ID", "type": "str", "length": 32}, + {"name": "file_name", "title": "文件名", "type": "str", "length": 255}, + {"name": "file_type", "title": "文件类型", "type": "str", "length": 32}, + {"name": "file_size", "title": "文件大小", "type": "int", "default": 0}, + {"name": "file_path", "title": "文件路径", "type": "str", "length": 500}, + {"name": "mime_type", "title": "MIME类型", "type": "str", "length": 128}, + {"name": "chunk_count", "title": "分片数", "type": "int", "default": 0}, + {"name": "embed_dim", "title": "向量维度", "type": "int", "default": 1024}, + {"name": "metadata", "title": "元数据JSON", "type": "text", "length": 4000}, + {"name": "status", "title": "状态", "type": "str", "length": 16, "default": "pending"}, + {"name": "error_msg", "title": "错误信息", "type": "text", "length": 2000}, + {"name": "org_id", "title": "所属机构", "type": "str", "length": 32}, + {"name": "created_by", "title": "创建人", "type": "str", "length": 32}, + {"name": "created_at", "title": "创建时间", "type": "datetime"}, + {"name": "updated_at", "title": "更新时间", "type": "datetime"} + ], + "codes": [ + {"field": "file_type", "table": "appcodes_kv", "valuefield": "k", "textfield": "v", "cond": "parentid='file_type'"}, + {"field": "status", "table": "appcodes_kv", "valuefield": "k", "textfield": "v", "cond": "parentid='doc_status'"}, + {"field": "kb_id", "table": "knowledge_bases", "valuefield": "id", "textfield": "name"} + ], + "indexes": [ + {"name": "idx_doc_kb", "idxtype": "index", "idxfields": ["kb_id"]}, + {"name": "idx_doc_org", "idxtype": "index", "idxfields": ["org_id"]}, + {"name": "idx_doc_status", "idxtype": "index", "idxfields": ["status"]} + ] +} diff --git a/models/engine_configs.json b/models/engine_configs.json new file mode 100644 index 0000000..be5dfe7 --- /dev/null +++ b/models/engine_configs.json @@ -0,0 +1,33 @@ +{ + "summary": [ + { + "name": "engine_configs", + "title": "引擎配置", + "primary": ["id"] + } + ], + "fields": [ + {"name": "id", "title": "ID", "type": "str", "length": 32}, + {"name": "org_id", "title": "机构ID", "type": "str", "length": 32, "nullable": "yes"}, + {"name": "engine_type", "title": "引擎类型", "type": "str", "length": 32}, + {"name": "engine_name", "title": "引擎名称", "type": "str", "length": 64}, + {"name": "endpoint_url", "title": "服务地址", "type": "str", "length": 255}, + {"name": "api_key", "title": "API密钥", "type": "str", "length": 255}, + {"name": "model_name", "title": "模型名称", "type": "str", "length": 128}, + {"name": "config_json", "title": "扩展配置JSON", "type": "text", "length": 4000}, + {"name": "is_default", "title": "是否默认", "type": "int", "default": 0}, + {"name": "priority", "title": "优先级", "type": "int", "default": 0}, + {"name": "status", "title": "状态", "type": "str", "length": 16, "default": "active"}, + {"name": "health_check_at", "title": "最近健康检查", "type": "datetime"}, + {"name": "created_at", "title": "创建时间", "type": "datetime"}, + {"name": "updated_at", "title": "更新时间", "type": "datetime"} + ], + "codes": [ + {"field": "engine_type", "table": "appcodes_kv", "valuefield": "k", "textfield": "v", "cond": "parentid='engine_type'"}, + {"field": "status", "table": "appcodes_kv", "valuefield": "k", "textfield": "v", "cond": "parentid='engine_status'"} + ], + "indexes": [ + {"name": "idx_eng_type", "idxtype": "index", "idxfields": ["engine_type", "is_default"]}, + {"name": "idx_eng_org", "idxtype": "index", "idxfields": ["org_id", "engine_type"]} + ] +} diff --git a/models/entities.json b/models/entities.json new file mode 100644 index 0000000..68b896a --- /dev/null +++ b/models/entities.json @@ -0,0 +1,35 @@ +{ + "summary": [ + { + "name": "entities", + "title": "实体", + "primary": ["id"] + } + ], + "fields": [ + {"name": "id", "title": "ID", "type": "str", "length": 32}, + {"name": "kb_id", "title": "知识库ID", "type": "str", "length": 32}, + {"name": "name", "title": "实体名", "type": "str", "length": 255}, + {"name": "entity_type", "title": "实体类型", "type": "str", "length": 64}, + {"name": "description", "title": "描述", "type": "text", "length": 2000}, + {"name": "graph_node_id", "title": "图节点ID", "type": "str", "length": 128}, + {"name": "source_doc_ids", "title": "来源文档ID列表", "type": "text", "length": 2000}, + {"name": "face_embedding_id", "title": "人脸特征ID", "type": "str", "length": 128}, + {"name": "voice_embedding_id", "title": "声纹特征ID", "type": "str", "length": 128}, + {"name": "metadata", "title": "元数据JSON", "type": "text", "length": 2000}, + {"name": "org_id", "title": "所属机构", "type": "str", "length": 32}, + {"name": "created_at", "title": "创建时间", "type": "datetime"}, + {"name": "updated_at", "title": "更新时间", "type": "datetime"} + ], + "codes": [ + {"field": "entity_type", "table": "appcodes_kv", "valuefield": "k", "textfield": "v", "cond": "parentid='entity_type'"}, + {"field": "kb_id", "table": "knowledge_bases", "valuefield": "id", "textfield": "name"} + ], + "indexes": [ + {"name": "idx_entity_kb", "idxtype": "index", "idxfields": ["kb_id"]}, + {"name": "idx_entity_name", "idxtype": "index", "idxfields": ["kb_id", "name"]}, + {"name": "idx_entity_type", "idxtype": "index", "idxfields": ["kb_id", "entity_type"]}, + {"name": "idx_entity_face", "idxtype": "index", "idxfields": ["face_embedding_id"]}, + {"name": "idx_entity_voice", "idxtype": "index", "idxfields": ["voice_embedding_id"]} + ] +} diff --git a/models/entity_relations.json b/models/entity_relations.json new file mode 100644 index 0000000..243b31e --- /dev/null +++ b/models/entity_relations.json @@ -0,0 +1,29 @@ +{ + "summary": [ + { + "name": "entity_relations", + "title": "实体关系", + "primary": ["id"] + } + ], + "fields": [ + {"name": "id", "title": "ID", "type": "str", "length": 32}, + {"name": "kb_id", "title": "知识库ID", "type": "str", "length": 32}, + {"name": "subject_id", "title": "主体实体ID", "type": "str", "length": 32}, + {"name": "object_id", "title": "客体实体ID", "type": "str", "length": 32}, + {"name": "relation_type", "title": "关系类型", "type": "str", "length": 64}, + {"name": "description", "title": "关系描述", "type": "text", "length": 2000}, + {"name": "graph_edge_id", "title": "图边ID", "type": "str", "length": 128}, + {"name": "source_chunk_ids", "title": "来源分片ID列表", "type": "text", "length": 2000}, + {"name": "org_id", "title": "所属机构", "type": "str", "length": 32}, + {"name": "created_at", "title": "创建时间", "type": "datetime"} + ], + "codes": [ + {"field": "kb_id", "table": "knowledge_bases", "valuefield": "id", "textfield": "name"} + ], + "indexes": [ + {"name": "idx_rel_kb", "idxtype": "index", "idxfields": ["kb_id"]}, + {"name": "idx_rel_subject", "idxtype": "index", "idxfields": ["subject_id"]}, + {"name": "idx_rel_object", "idxtype": "index", "idxfields": ["object_id"]} + ] +} diff --git a/models/knowledge_bases.json b/models/knowledge_bases.json new file mode 100644 index 0000000..0ea0967 --- /dev/null +++ b/models/knowledge_bases.json @@ -0,0 +1,32 @@ +{ + "summary": [ + { + "name": "knowledge_bases", + "title": "知识库", + "primary": ["id"] + } + ], + "fields": [ + {"name": "id", "title": "ID", "type": "str", "length": 32}, + {"name": "name", "title": "名称", "type": "str", "length": 100}, + {"name": "description", "title": "描述", "type": "text", "length": 2000}, + {"name": "org_id", "title": "所属机构", "type": "str", "length": 32}, + {"name": "embedding_engine", "title": "向量化引擎", "type": "str", "length": 64, "default": "clip-vith14"}, + {"name": "vdb_collection", "title": "向量库集合名", "type": "str", "length": 128}, + {"name": "graph_name", "title": "图名称", "type": "str", "length": 128}, + {"name": "doc_count", "title": "文档数", "type": "int", "default": 0}, + {"name": "chunk_count", "title": "分片数", "type": "int", "default": 0}, + {"name": "total_size", "title": "总大小(字节)", "type": "int", "default": 0}, + {"name": "status", "title": "状态", "type": "str", "length": 16, "default": "active"}, + {"name": "created_by", "title": "创建人", "type": "str", "length": 32}, + {"name": "created_at", "title": "创建时间", "type": "datetime"}, + {"name": "updated_at", "title": "更新时间", "type": "datetime"} + ], + "codes": [ + {"field": "status", "table": "appcodes_kv", "valuefield": "k", "textfield": "v", "cond": "parentid='kb_status'"} + ], + "indexes": [ + {"name": "idx_kb_org", "idxtype": "index", "idxfields": ["org_id"]}, + {"name": "idx_kb_name", "idxtype": "unique", "idxfields": ["org_id", "name"]} + ] +} diff --git a/models/media_tags.json b/models/media_tags.json new file mode 100644 index 0000000..1e34090 --- /dev/null +++ b/models/media_tags.json @@ -0,0 +1,28 @@ +{ + "summary": [ + { + "name": "media_tags", + "title": "媒体标签关联", + "primary": ["id"] + } + ], + "fields": [ + {"name": "id", "title": "ID", "type": "str", "length": 32}, + {"name": "kb_id", "title": "知识库ID", "type": "str", "length": 32}, + {"name": "media_type", "title": "媒体类型", "type": "str", "length": 32}, + {"name": "media_id", "title": "媒体ID", "type": "str", "length": 32}, + {"name": "tag_id", "title": "标签ID", "type": "str", "length": 32}, + {"name": "created_at", "title": "创建时间", "type": "datetime"} + ], + "codes": [ + {"field": "media_type", "table": "appcodes_kv", "valuefield": "k", "textfield": "v", "cond": "parentid='media_type'"}, + {"field": "tag_id", "table": "tags", "valuefield": "id", "textfield": "name"}, + {"field": "kb_id", "table": "knowledge_bases", "valuefield": "id", "textfield": "name"} + ], + "indexes": [ + {"name": "idx_mt_media", "idxtype": "index", "idxfields": ["media_type", "media_id"]}, + {"name": "idx_mt_tag", "idxtype": "index", "idxfields": ["tag_id"]}, + {"name": "idx_mt_kb", "idxtype": "index", "idxfields": ["kb_id"]}, + {"name": "idx_mt_unique", "idxtype": "unique", "idxfields": ["media_type", "media_id", "tag_id"]} + ] +} diff --git a/models/subscriptions.json b/models/subscriptions.json new file mode 100644 index 0000000..bec6ac9 --- /dev/null +++ b/models/subscriptions.json @@ -0,0 +1,34 @@ +{ + "summary": [ + { + "name": "subscriptions", + "title": "租户订阅", + "primary": ["id"] + } + ], + "fields": [ + {"name": "id", "title": "ID", "type": "str", "length": 32}, + {"name": "org_id", "title": "机构ID", "type": "str", "length": 32}, + {"name": "plan_name", "title": "套餐名称", "type": "str", "length": 64}, + {"name": "disk_quota_bytes", "title": "磁盘配额(字节)", "type": "int", "default": 1073741824}, + {"name": "disk_used_bytes", "title": "已用磁盘(字节)", "type": "int", "default": 0}, + {"name": "doc_quota", "title": "文档数配额", "type": "int", "default": 1000}, + {"name": "doc_used", "title": "已用文档数", "type": "int", "default": 0}, + {"name": "kb_quota", "title": "知识库数配额", "type": "int", "default": 10}, + {"name": "kb_used", "title": "已用知识库数", "type": "int", "default": 0}, + {"name": "api_call_quota", "title": "API调用配额(月)", "type": "int", "default": 10000}, + {"name": "api_call_used", "title": "已用API调用(月)", "type": "int", "default": 0}, + {"name": "start_date", "title": "开始日期", "type": "date"}, + {"name": "end_date", "title": "结束日期", "type": "date"}, + {"name": "status", "title": "状态", "type": "str", "length": 16, "default": "active"}, + {"name": "created_at", "title": "创建时间", "type": "datetime"}, + {"name": "updated_at", "title": "更新时间", "type": "datetime"} + ], + "codes": [ + {"field": "status", "table": "appcodes_kv", "valuefield": "k", "textfield": "v", "cond": "parentid='sub_status'"} + ], + "indexes": [ + {"name": "idx_sub_org", "idxtype": "unique", "idxfields": ["org_id"]}, + {"name": "idx_sub_status", "idxtype": "index", "idxfields": ["status"]} + ] +} diff --git a/models/tags.json b/models/tags.json new file mode 100644 index 0000000..3953f1b --- /dev/null +++ b/models/tags.json @@ -0,0 +1,24 @@ +{ + "summary": [ + { + "name": "tags", + "title": "标签", + "primary": ["id"] + } + ], + "fields": [ + {"name": "id", "title": "ID", "type": "str", "length": 32}, + {"name": "kb_id", "title": "知识库ID", "type": "str", "length": 32}, + {"name": "name", "title": "标签名", "type": "str", "length": 64}, + {"name": "color", "title": "颜色", "type": "str", "length": 16, "default": "#3b82f6"}, + {"name": "org_id", "title": "所属机构", "type": "str", "length": 32}, + {"name": "created_at", "title": "创建时间", "type": "datetime"} + ], + "codes": [ + {"field": "kb_id", "table": "knowledge_bases", "valuefield": "id", "textfield": "name"} + ], + "indexes": [ + {"name": "idx_tag_kb", "idxtype": "index", "idxfields": ["kb_id"]}, + {"name": "idx_tag_name", "idxtype": "unique", "idxfields": ["kb_id", "name"]} + ] +} diff --git a/models/usage_logs.json b/models/usage_logs.json new file mode 100644 index 0000000..4b28e7d --- /dev/null +++ b/models/usage_logs.json @@ -0,0 +1,30 @@ +{ + "summary": [ + { + "name": "usage_logs", + "title": "用量日志", + "primary": ["id"] + } + ], + "fields": [ + {"name": "id", "title": "ID", "type": "str", "length": 32}, + {"name": "org_id", "title": "机构ID", "type": "str", "length": 32}, + {"name": "kb_id", "title": "知识库ID", "type": "str", "length": 32}, + {"name": "operation", "title": "操作类型", "type": "str", "length": 32}, + {"name": "bytes_delta", "title": "磁盘变化(字节)", "type": "int", "default": 0}, + {"name": "bytes_total", "title": "累计磁盘", "type": "int", "default": 0}, + {"name": "api_calls", "title": "API调用数", "type": "int", "default": 1}, + {"name": "engine_type", "title": "引擎类型", "type": "str", "length": 32}, + {"name": "tokens_used", "title": "Token消耗", "type": "int", "default": 0}, + {"name": "cost_estimate", "title": "预估费用", "type": "float", "length": 20, "dec": 4}, + {"name": "detail", "title": "详情JSON", "type": "text", "length": 2000}, + {"name": "created_at", "title": "创建时间", "type": "datetime"} + ], + "codes": [ + {"field": "operation", "table": "appcodes_kv", "valuefield": "k", "textfield": "v", "cond": "parentid='usage_op'"} + ], + "indexes": [ + {"name": "idx_usage_org", "idxtype": "index", "idxfields": ["org_id", "created_at"]}, + {"name": "idx_usage_kb", "idxtype": "index", "idxfields": ["kb_id"]} + ] +} diff --git a/setup.py b/setup.py new file mode 100644 index 0000000..546fad2 --- /dev/null +++ b/setup.py @@ -0,0 +1,12 @@ +from setuptools import setup, find_packages + +setup( + name='rag', + version='0.1.0', + packages=find_packages(), + include_package_data=True, + package_data={ + 'rag': ['wwwroot/**/*', 'models/*.json', 'conf/*.json'], + }, + install_requires=[], +) diff --git a/subscriptions.json b/subscriptions.json new file mode 100644 index 0000000..53f7ae5 --- /dev/null +++ b/subscriptions.json @@ -0,0 +1,28 @@ +{ + "tblname": "subscriptions", + "params": { + "title": "订阅管理", + "browserfields": [ + {"field": "org_id", "title": "机构", "width": "15%"}, + {"field": "plan_name", "title": "套餐", "width": "15%"}, + {"field": "disk_quota_bytes", "title": "磁盘配额", "width": "12%"}, + {"field": "disk_used_bytes", "title": "已用磁盘", "width": "12%"}, + {"field": "doc_used", "title": "文档数", "width": "10%"}, + {"field": "start_date", "title": "开始", "width": "12%"}, + {"field": "end_date", "title": "结束", "width": "12%"}, + {"field": "status", "title": "状态", "width": "12%"} + ], + "editfields": [ + {"field": "org_id", "uitype": "Text", "required": true}, + {"field": "plan_name", "uitype": "Text", "required": true}, + {"field": "disk_quota_bytes", "uitype": "Text", "default": "1073741824"}, + {"field": "doc_quota", "uitype": "Text"}, + {"field": "kb_quota", "uitype": "Text"}, + {"field": "api_call_quota", "uitype": "Text"}, + {"field": "start_date", "uitype": "Text"}, + {"field": "end_date", "uitype": "Text"} + ], + "searchfields": ["org_id", "plan_name"], + "sort": "created_at desc" + } +} diff --git a/tags.json b/tags.json new file mode 100644 index 0000000..acdf5dd --- /dev/null +++ b/tags.json @@ -0,0 +1,22 @@ +{ + "tblname": "tags", + "params": { + "title": "标签管理", + "browserfields": [ + {"field": "name", "title": "标签名", "width": "25%"}, + {"field": "color", "title": "颜色", "width": "15%"}, + {"field": "kb_id", "title": "知识库", "width": "30%"}, + {"field": "created_at", "title": "创建时间", "width": "30%"} + ], + "editfields": [ + {"field": "name", "uitype": "Text", "required": true, "label": "标签名"}, + {"field": "kb_id", "uitype": "Text", "required": true, "label": "知识库ID"}, + {"field": "color", "uitype": "Text", "default": "#3b82f6", "label": "颜色(#hex)"} + ], + "searchfields": ["name"], + "sort": "created_at desc", + "data_filter": { + "org_id": "{{userorgid}}" + } + } +} diff --git a/wwwroot/documents_list/index.ui b/wwwroot/documents_list/index.ui new file mode 100644 index 0000000..daf666a --- /dev/null +++ b/wwwroot/documents_list/index.ui @@ -0,0 +1,65 @@ +{ + "widgettype": "VBox", + "options": { + "cheight": 40, + "width": "100%", + "padding": "16px", + "spacing": "12px" + }, + "subwidgets": [ + { + "widgettype": "HBox", + "options": { + "width": "100%", + "justifyContent": "space-between", + "alignItems": "center" + }, + "subwidgets": [ + { + "widgettype": "VBox", + "options": {"spacing": "4px"}, + "subwidgets": [ + {"widgettype": "Text", "options": {"text": "文档管理", "cfontsize": 22, "fontWeight": "bold", "color": "#333"}}, + {"widgettype": "Text", "options": {"text": "上传文本、图片、音频、视频文件", "cfontsize": 14, "color": "#888"}} + ] + }, + { + "widgettype": "Button", + "options": { + "label": "📁 上传文件", + "bgcolor": "#50b86c", + "color": "#fff" + } + } + ] + }, + { + "widgettype": "VBox", + "options": { + "width": "100%", + "cheight": 8, + "bgcolor": "#f8f9fa", + "padding": "24px", + "css": "card", + "border": "2px dashed #d0d0d0", + "alignItems": "center", + "justifyContent": "center" + }, + "subwidgets": [ + {"widgettype": "Text", "options": {"text": "📂", "cfontsize": 32}}, + {"widgettype": "Text", "options": {"text": "拖拽文件到此处上传", "cfontsize": 16, "color": "#888"}}, + {"widgettype": "Text", "options": {"text": "支持: TXT, PDF, JPG, PNG, MP3, WAV, MP4, AVI", "cfontsize": 12, "color": "#aaa"}}, + {"widgettype": "Text", "options": {"text": "单个文件最大 500MB · 自动向量化入库", "cfontsize": 12, "color": "#aaa"}} + ] + }, + { + "widgettype": "Text", + "options": { + "text": "暂无文档", + "color": "#aaa", + "cfontsize": 14, + "halign": "center" + } + } + ] +} diff --git a/wwwroot/engine_configs_list/index.ui b/wwwroot/engine_configs_list/index.ui new file mode 100644 index 0000000..f976585 --- /dev/null +++ b/wwwroot/engine_configs_list/index.ui @@ -0,0 +1,148 @@ +{ + "widgettype": "VBox", + "options": { + "cheight": 40, + "width": "100%", + "padding": "16px", + "spacing": "16px" + }, + "subwidgets": [ + { + "widgettype": "Text", + "options": { + "text": "引擎配置", + "cfontsize": 22, + "fontWeight": "bold", + "color": "#333" + } + }, + { + "widgettype": "Text", + "options": { + "text": "管理多媒体RAG的各层引擎，支持租户级自定义覆盖", + "color": "#888", + "cfontsize": 14 + } + }, + { + "widgettype": "HBox", + "options": { + "width": "100%", + "spacing": "12px", + "wrap": true + }, + "subwidgets": [ + { + "widgettype": "VBox", + "options": { + "cwidth": 20, + "cheight": 10, + "bgcolor": "#f8f9fa", + "padding": "12px", + "css": "card" + }, + "subwidgets": [ + {"widgettype": "Text", "options": {"text": "🔤 向量化引擎", "cfontsize": 16, "fontWeight": "bold"}}, + {"widgettype": "Text", "options": {"text": "CLIP ViT-H-14", "cfontsize": 14, "color": "#4a90d9"}}, + {"widgettype": "Text", "options": {"text": "1024维 · 图文跨模态", "cfontsize": 12, "color": "#888"}}, + {"widgettype": "Text", "options": {"text": "状态: ✅ 运行中", "cfontsize": 12, "color": "#50b86c"}}, + {"widgettype": "Text", "options": {"text": "端口: 9086", "cfontsize": 12, "color": "#aaa"}} + ] + }, + { + "widgettype": "VBox", + "options": { + "cwidth": 20, + "cheight": 10, + "bgcolor": "#f8f9fa", + "padding": "12px", + "css": "card" + }, + "subwidgets": [ + {"widgettype": "Text", "options": {"text": "🗄️ 向量数据库", "cfontsize": 16, "fontWeight": "bold"}}, + {"widgettype": "Text", "options": {"text": "Milvus Lite", "cfontsize": 14, "color": "#4a90d9"}}, + {"widgettype": "Text", "options": {"text": "COSINE度量 · HNSW索引", "cfontsize": 12, "color": "#888"}}, + {"widgettype": "Text", "options": {"text": "状态: ✅ 运行中", "cfontsize": 12, "color": "#50b86c"}}, + {"widgettype": "Text", "options": {"text": "端口: 8886", "cfontsize": 12, "color": "#aaa"}} + ] + }, + { + "widgettype": "VBox", + "options": { + "cwidth": 20, + "cheight": 10, + "bgcolor": "#f8f9fa", + "padding": "12px", + "css": "card" + }, + "subwidgets": [ + {"widgettype": "Text", "options": {"text": "📊 重排引擎", "cfontsize": 16, "fontWeight": "bold"}}, + {"widgettype": "Text", "options": {"text": "BGE Reranker v2-m3", "cfontsize": 14, "color": "#4a90d9"}}, + {"widgettype": "Text", "options": {"text": "Cross-encoder精排", "cfontsize": 12, "color": "#888"}}, + {"widgettype": "Text", "options": {"text": "状态: ✅ 运行中", "cfontsize": 12, "color": "#50b86c"}}, + {"widgettype": "Text", "options": {"text": "端口: 9090", "cfontsize": 12, "color": "#aaa"}} + ] + }, + { + "widgettype": "VBox", + "options": { + "cwidth": 20, + "cheight": 10, + "bgcolor": "#f8f9fa", + "padding": "12px", + "css": "card" + }, + "subwidgets": [ + {"widgettype": "Text", "options": {"text": "🕸️ 知识图谱", "cfontsize": 16, "fontWeight": "bold"}}, + {"widgettype": "Text", "options": {"text": "NetworkX", "cfontsize": 14, "color": "#4a90d9"}}, + {"widgettype": "Text", "options": {"text": "实体关系 · 邻居查询", "cfontsize": 12, "color": "#888"}}, + {"widgettype": "Text", "options": {"text": "状态: ✅ 运行中", "cfontsize": 12, "color": "#50b86c"}}, + {"widgettype": "Text", "options": {"text": "端口: 9092", "cfontsize": 12, "color": "#aaa"}} + ] + }, + { + "widgettype": "VBox", + "options": { + "cwidth": 20, + "cheight": 10, + "bgcolor": "#f8f9fa", + "padding": "12px", + "css": "card" + }, + "subwidgets": [ + {"widgettype": "Text", "options": {"text": "🏷️ 实体识别", "cfontsize": 16, "fontWeight": "bold"}}, + {"widgettype": "Text", "options": {"text": "GLiNER Multitask", "cfontsize": 14, "color": "#4a90d9"}}, + {"widgettype": "Text", "options": {"text": "零样本NER · 中文支持", "cfontsize": 12, "color": "#888"}}, + {"widgettype": "Text", "options": {"text": "状态: ✅ 运行中", "cfontsize": 12, "color": "#50b86c"}}, + {"widgettype": "Text", "options": {"text": "端口: 9093", "cfontsize": 12, "color": "#aaa"}} + ] + }, + { + "widgettype": "VBox", + "options": { + "cwidth": 20, + "cheight": 10, + "bgcolor": "#f8f9fa", + "padding": "12px", + "css": "card" + }, + "subwidgets": [ + {"widgettype": "Text", "options": {"text": "👤 人脸识别", "cfontsize": 16, "fontWeight": "bold"}}, + {"widgettype": "Text", "options": {"text": "InsightFace buffalo_l", "cfontsize": 14, "color": "#4a90d9"}}, + {"widgettype": "Text", "options": {"text": "检测+识别 · 512维", "cfontsize": 12, "color": "#888"}}, + {"widgettype": "Text", "options": {"text": "状态: ✅ 运行中", "cfontsize": 12, "color": "#50b86c"}}, + {"widgettype": "Text", "options": {"text": "端口: 9091", "cfontsize": 12, "color": "#aaa"}} + ] + } + ] + }, + { + "widgettype": "Text", + "options": { + "text": "💡 引擎默认全局配置，租户可在订阅设置中覆盖为私有实例", + "color": "#aaa", + "cfontsize": 12 + } + } + ] +} diff --git a/wwwroot/knowledge_bases_list/create_kb.dspy b/wwwroot/knowledge_bases_list/create_kb.dspy new file mode 100644 index 0000000..8b70633 --- /dev/null +++ b/wwwroot/knowledge_bases_list/create_kb.dspy @@ -0,0 +1,19 @@ +ns = params_kw.copy() +db = DBPools() +dbname = get_module_dbname('rag') +async with db.sqlorContext(dbname) as sor: + import uuid + env = request._run_ns + userorgid = await env.get_userorgid() + kb_id = uuid.uuid4().hex[:12] + name = ns.get("name", "") + desc = ns.get("description", "") + await sor.sqlExe( + "INSERT INTO knowledge_bases (id, name, description, org_id, embedding_engine, vdb_collection, doc_count, total_size, chunk_count, status, created_at) " + "VALUES (${id}$, ${name}$, ${desc}$, ${org_id}$, 'CLIP ViT-H-14', 'rag_collection', 0, 0, 0, 'active', NOW())", + {"id": kb_id, "name": name, "desc": desc, "org_id": userorgid}) + return { + "widgettype": "urlwidget", + "options": {"url": entire_url('/knowledge_bases_list/index.ui')} + } +return {"error": "failed"} diff --git a/wwwroot/knowledge_bases_list/delete_tree_item.dspy b/wwwroot/knowledge_bases_list/delete_tree_item.dspy new file mode 100644 index 0000000..6319d8f --- /dev/null +++ b/wwwroot/knowledge_bases_list/delete_tree_item.dspy @@ -0,0 +1,7 @@ +ns = params_kw.copy() +db = DBPools() +dbname = get_module_dbname('rag') +async with db.sqlorContext(dbname) as sor: + await sor.sqlExe("DELETE FROM document_chunks WHERE id=${id}$", {"id": ns.get("id","")}) + return {"widgettype": "Message", "options": {"user_data": {"id": ns.get("id","")}}} +return {"error": "failed"} diff --git a/wwwroot/knowledge_bases_list/detail.ui b/wwwroot/knowledge_bases_list/detail.ui new file mode 100644 index 0000000..fdb81d2 --- /dev/null +++ b/wwwroot/knowledge_bases_list/detail.ui @@ -0,0 +1,92 @@ +{ + "widgettype": "VBox", + "options": {"cheight": 40, "width": "100%", "padding": "0", "spacing": 0}, + "subwidgets": [ + { + "widgettype": "HBox", + "options": {"width": "100%", "bgcolor": "#f8f9fa", "padding": "10px 16px", "alignItems": "center", "borderBottom": "1px solid #e0e0e0"}, + "subwidgets": [ + {"widgettype": "Text", "options": {"text": "← 知识库列表", "css": "clickable", "cfontsize": 14, "color": "#4a90d9"}, "binds": [ + {"wid": "self", "event": "click", "actiontype": "urlwidget", "target": "app.rag_main_content", "mode": "replace", + "options": {"url": "{{entire_url('/knowledge_bases_list/index.ui')}}"}} + ]}, + {"widgettype": "Text", "options": {"text": " / ", "cfontsize": 14, "color": "#ccc"}}, + {"widgettype": "Text", "options": {"text": "📚 {{params_kw.kb_name or '知识库'}}", "cfontsize": 16, "fontWeight": "bold", "color": "#333"}}, + {"widgettype": "Text", "options": {"text": "", "css": "filler"}} + ] + }, + { + "widgettype": "HBox", + "options": {"css": "filler", "width": "100%", "spacing": 0}, + "subwidgets": [ + { + "widgettype": "VBox", + "options": {"cwidth": 16, "bgcolor": "#fafafa", "borderRight": "1px solid #e0e0e0", "spacing": 0}, + "subwidgets": [ + { + "widgettype": "HBox", + "options": {"padding": "10px 12px", "bgcolor": "#f0f0f0", "borderBottom": "1px solid #e0e0e0"}, + "subwidgets": [ + {"widgettype": "Text", "options": {"text": "📁 目录结构", "cfontsize": 13, "fontWeight": "bold", "color": "#666"}} + ] + }, + { + "widgettype": "Tree", + "options": { + "title": "目录", + "id": "dir_tree", + "parentField": "parentid", + "idField": "id", + "textField": "label", + "dataurl": "{{entire_url('./get_tree_data.dspy')}}?kb_id={{params_kw.kb_id}}", + "newdata_params": {"kb_id": "{{params_kw.kb_id}}"}, + "editable": { + "fields": [ + {"name": "name", "title": "名称", "type": "str", "length": 255, "uitype": "str", "label": "名称"} + ], + "add_url": "{{entire_url('./new_tree_item.dspy')}}", + "update_url": "{{entire_url('./update_tree_item.dspy')}}", + "delete_url": "{{entire_url('./delete_tree_item.dspy')}}" + } + }, + "binds": [ + { + "wid": "self", + "event": "selected", + "actiontype": "script", + "script": "window._rag_current_folder = event.params.id ? event.params.id : ''" + } + ] + } + ] + }, + { + "widgettype": "VBox", + "options": {"id": "detail_content", "css": "filler", "padding": "16px", "spacing": "12px"}, + "subwidgets": [ + { + "widgettype": "VBox", + "options": {"id": "drop_zone", "width": "100%", "cheight": 8, "bgcolor": "#f8f9fa", "padding": "16px", "css": "card", "border": "2px dashed #3b82f6", "alignItems": "center", "justifyContent": "center"}, + "subwidgets": [ + {"widgettype": "Text", "options": {"text": "📂 点击选择文件上传", "cfontsize": 15, "color": "#888"}}, + {"widgettype": "Text", "options": {"text": "支持 TXT/PDF/JPG/PNG/MP3/WAV/MP4", "cfontsize": 12, "color": "#aaa"}}, + { + "widgettype": "Button", + "options": {"label": "选择文件", "bgcolor": "#3b82f6", "color": "#fff", "marginTop": "8px"}, + "binds": [{ + "wid": "self", + "event": "click", + "actiontype": "script", + "script": "var i=document.createElement('input');i.type='file';i.multiple=true;i.onchange=function(){var u=new URL(window.location.href);var kb=u.searchParams.get('kb_id')||'';var fd=window._rag_current_folder||'';for(var j=0;j