cite-agent 1.3.5__py3-none-any.whl → 1.3.7__py3-none-any.whl

cite_agent/project_detector.py

@@ -0,0 +1,148 @@
+#!/usr/bin/env python3
+"""
+Generic project detection - works with ANY IDE/project type
+Not RStudio-specific - detects R, Python, Node, Jupyter, etc.
+"""
+
+import os
+from pathlib import Path
+from typing import Dict, List, Optional, Any
+import glob
+
+
+class ProjectDetector:
+    """Detects project type and provides context"""
+    
+    def __init__(self, working_dir: Optional[str] = None):
+        self.working_dir = Path(working_dir or os.getcwd())
+    
+    def detect_project(self) -> Optional[Dict[str, Any]]:
+        """
+        Detect what kind of project user is working in
+        Returns project info or None if not in a project
+        """
+        project_info = {
+            "type": None,
+            "name": None,
+            "recent_files": [],
+            "description": None
+        }
+        
+        # Check for R project (.Rproj file OR 2+ .R files)
+        rproj_files = list(self.working_dir.glob("*.Rproj"))
+        r_files = list(self.working_dir.glob("*.R")) + list(self.working_dir.glob("*.Rmd"))
+        
+        if rproj_files:
+            project_info["type"] = "R"
+            project_info["name"] = rproj_files[0].stem
+            project_info["recent_files"] = self._get_recent_files([".R", ".Rmd", ".qmd"])
+            project_info["description"] = f"R/RStudio project: {project_info['name']}"
+            return project_info
+        elif len(r_files) >= 2:
+            # R project without .Rproj file
+            project_info["type"] = "R"
+            project_info["name"] = self.working_dir.name
+            project_info["recent_files"] = self._get_recent_files([".R", ".Rmd", ".qmd"])
+            project_info["description"] = f"R project: {project_info['name']}"
+            return project_info
+        
+        # Check for Python project
+        if (self.working_dir / "pyproject.toml").exists() or \
+           (self.working_dir / "setup.py").exists() or \
+           (self.working_dir / "requirements.txt").exists():
+            project_info["type"] = "Python"
+            project_info["name"] = self.working_dir.name
+            project_info["recent_files"] = self._get_recent_files([".py", ".ipynb"])
+            project_info["description"] = f"Python project: {project_info['name']}"
+            return project_info
+        
+        # Check for Node.js project
+        if (self.working_dir / "package.json").exists():
+            project_info["type"] = "Node"
+            project_info["name"] = self.working_dir.name
+            project_info["recent_files"] = self._get_recent_files([".js", ".ts", ".jsx", ".tsx"])
+            project_info["description"] = f"Node.js project: {project_info['name']}"
+            return project_info
+        
+        # Check for Jupyter/Data Science directory
+        ipynb_files = list(self.working_dir.glob("*.ipynb"))
+        if len(ipynb_files) >= 2:  # 2+ notebooks = likely data science project
+            project_info["type"] = "Jupyter"
+            project_info["name"] = self.working_dir.name
+            project_info["recent_files"] = self._get_recent_files([".ipynb", ".py", ".csv"])
+            project_info["description"] = f"Jupyter/Data Science project: {project_info['name']}"
+            return project_info
+        
+        # Check for Git repository
+        if (self.working_dir / ".git").exists():
+            project_info["type"] = "Git"
+            project_info["name"] = self.working_dir.name
+            # Get recent files of any code type
+            project_info["recent_files"] = self._get_recent_files([".py", ".js", ".R", ".java", ".cpp", ".rs"])
+            project_info["description"] = f"Git repository: {project_info['name']}"
+            return project_info
+        
+        # Not in a recognized project
+        return None
+    
+    def _get_recent_files(self, extensions: List[str], limit: int = 5) -> List[Dict[str, Any]]:
+        """Get recently modified files with given extensions"""
+        files = []
+        
+        for ext in extensions:
+            for filepath in self.working_dir.glob(f"**/*{ext}"):
+                if filepath.is_file() and not any(part.startswith('.') for part in filepath.parts):
+                    try:
+                        mtime = filepath.stat().st_mtime
+                        files.append({
+                            "name": filepath.name,
+                            "path": str(filepath),
+                            "relative_path": str(filepath.relative_to(self.working_dir)),
+                            "modified": mtime,
+                            "extension": ext
+                        })
+                    except:
+                        pass
+        
+        # Sort by modification time, newest first
+        files.sort(key=lambda f: f["modified"], reverse=True)
+        return files[:limit]
+    
+    def format_project_banner(self, project_info: Dict[str, Any]) -> str:
+        """Format project info as a nice banner"""
+        if not project_info or not project_info["type"]:
+            return ""
+        
+        icon_map = {
+            "R": "📊",
+            "Python": "🐍",
+            "Node": "📦",
+            "Jupyter": "📓",
+            "Git": "📂"
+        }
+        
+        icon = icon_map.get(project_info["type"], "📁")
+        
+        banner = f"\n{icon} {project_info['description']}\n"
+        
+        if project_info["recent_files"]:
+            banner += "📄 Recent files:\n"
+            for f in project_info["recent_files"][:3]:
+                banner += f"   • {f['relative_path']}\n"
+        
+        return banner
+    
+    def get_project_context_for_llm(self, project_info: Dict[str, Any]) -> str:
+        """Get project context to add to LLM prompts"""
+        if not project_info or not project_info["type"]:
+            return ""
+        
+        context = f"User is working in a {project_info['type']} project: {project_info['name']}\n"
+        
+        if project_info["recent_files"]:
+            context += "Recent files:\n"
+            for f in project_info["recent_files"][:5]:
+                context += f"- {f['relative_path']}\n"
+        
+        return context
+

{cite_agent-1.3.5.dist-info → cite_agent-1.3.7.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: cite-agent
-Version: 1.3.5
+Version: 1.3.7
 Summary: Terminal AI assistant for academic research with citation verification
 Home-page: https://github.com/Spectating101/cite-agent
 Author: Cite-Agent Team

cite_agent-1.3.7.dist-info/RECORD

@@ -0,0 +1,31 @@
+cite_agent/__init__.py,sha256=wAXV2v8nNOmIAd0rh8196ItBl9hHWBVOBl5Re4VB77I,1645
+cite_agent/__main__.py,sha256=6x3lltwG-iZHeQbN12rwvdkPDfd2Rmdk71tOOaC89Mw,179
+cite_agent/__version__.py,sha256=9peaXOar2qezOPJEKG6cD_A0aaXrzdVN8h-v6fBoBEk,22
+cite_agent/account_client.py,sha256=yLuzhIJoIZuXHXGbaVMzDxRATQwcy-wiaLnUrDuwUhI,5725
+cite_agent/agent_backend_only.py,sha256=H4DH4hmKhT0T3rQLAb2xnnJVjxl3pOZaljL9r6JndFY,6314
+cite_agent/ascii_plotting.py,sha256=lk8BaECs6fmjtp4iH12G09-frlRehAN7HLhHt2crers,8570
+cite_agent/auth.py,sha256=YtoGXKwcLkZQbop37iYYL9BzRWBRPlt_D9p71VGViS4,9833
+cite_agent/backend_only_client.py,sha256=WqLF8x7aXTro2Q3ehqKMsdCg53s6fNk9Hy86bGxqmmw,2561
+cite_agent/cli.py,sha256=QO4hmHOeiW_8gxCjos1zk7NV4-joQiLc9LNsv7zCr70,35931
+cite_agent/cli_conversational.py,sha256=RAmgRNRyB8gQ8QLvWU-Tt23j2lmA34rQNT5F3_7SOq0,11141
+cite_agent/cli_enhanced.py,sha256=EAaSw9qtiYRWUXF6_05T19GCXlz9cCSz6n41ASnXIPc,7407
+cite_agent/cli_workflow.py,sha256=4oS_jW9D8ylovXbEFdsyLQONt4o0xxR4Xatfcc4tnBs,11641
+cite_agent/dashboard.py,sha256=VGV5XQU1PnqvTsxfKMcue3j2ri_nvm9Be6O5aVays_w,10502
+cite_agent/enhanced_ai_agent.py,sha256=wyuQu50mZo6jMYZqFD8Bqjk55dFFOBBR28AixchftXY,187083
+cite_agent/project_detector.py,sha256=fPl5cLTy_oyufqrQ7RJ5IRVdofZoPqDRaQXW6tRtBJc,6086
+cite_agent/rate_limiter.py,sha256=-0fXx8Tl4zVB4O28n9ojU2weRo-FBF1cJo9Z5jC2LxQ,10908
+cite_agent/session_manager.py,sha256=B0MXSOsXdhO3DlvTG7S8x6pmGlYEDvIZ-o8TZM23niQ,9444
+cite_agent/setup_config.py,sha256=3m2e3gw0srEWA0OygdRo64r-8HK5ohyXfct0c__CF3s,16817
+cite_agent/streaming_ui.py,sha256=N6TWOo7GVQ_Ynfw73JCfrdGcLIU-PwbS3GbsHQHegmg,7810
+cite_agent/telemetry.py,sha256=55kXdHvI24ZsEkbFtihcjIfJt2oiSXcEpLzTxQ3KCdQ,2916
+cite_agent/ui.py,sha256=r1OAeY3NSeqhAjJYmEBH9CaennBuibFAz1Mur6YF80E,6134
+cite_agent/updater.py,sha256=udoAAN4gBKAvKDV7JTh2FJO_jIhNk9bby4x6n188MEY,8458
+cite_agent/web_search.py,sha256=FZCuNO7MAITiOIbpPbJyt2bzbXPzQla-9amJpnMpW_4,6520
+cite_agent/workflow.py,sha256=a0YC0Mzz4or1C5t2gZcuJBQ0uMOZrooaI8eLu2kkI0k,15086
+cite_agent/workflow_integration.py,sha256=A9ua0DN5pRtuU0cAwrUTGvqt2SXKhEHQbrHx16EGnDM,10910
+cite_agent-1.3.7.dist-info/licenses/LICENSE,sha256=XJkyO4IymhSUniN1ENY6lLrL2729gn_rbRlFK6_Hi9M,1074
+cite_agent-1.3.7.dist-info/METADATA,sha256=YRov18tZiDcm3tnPS7nIPoD8Ruq1wFghllHcqy6mTKc,12231
+cite_agent-1.3.7.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+cite_agent-1.3.7.dist-info/entry_points.txt,sha256=bJ0u28nFIxQKH1PWQ2ak4PV-FAjhoxTC7YADEdDenFw,83
+cite_agent-1.3.7.dist-info/top_level.txt,sha256=NNfD8pxDZzBK8tjDIpCs2BW9Va-OQ5qUFbEx0SgmyIE,11
+cite_agent-1.3.7.dist-info/RECORD,,

{cite_agent-1.3.5.dist-info → cite_agent-1.3.7.dist-info}/top_level.txt

@@ -1,2 +1 @@
 cite_agent
-src

cite_agent-1.3.5.dist-info/RECORD

@@ -1,56 +0,0 @@
-cite_agent/__init__.py,sha256=wAXV2v8nNOmIAd0rh8196ItBl9hHWBVOBl5Re4VB77I,1645
-cite_agent/__main__.py,sha256=6x3lltwG-iZHeQbN12rwvdkPDfd2Rmdk71tOOaC89Mw,179
-cite_agent/__version__.py,sha256=tdqvkGH0OryRjjXzO3HS5DyYol-VTO9fC8m43nB2PgI,22
-cite_agent/account_client.py,sha256=yLuzhIJoIZuXHXGbaVMzDxRATQwcy-wiaLnUrDuwUhI,5725
-cite_agent/agent_backend_only.py,sha256=H4DH4hmKhT0T3rQLAb2xnnJVjxl3pOZaljL9r6JndFY,6314
-cite_agent/ascii_plotting.py,sha256=lk8BaECs6fmjtp4iH12G09-frlRehAN7HLhHt2crers,8570
-cite_agent/auth.py,sha256=YtoGXKwcLkZQbop37iYYL9BzRWBRPlt_D9p71VGViS4,9833
-cite_agent/backend_only_client.py,sha256=WqLF8x7aXTro2Q3ehqKMsdCg53s6fNk9Hy86bGxqmmw,2561
-cite_agent/cli.py,sha256=f0x19N4MB85XgfNtOPEtovLQO_qzTHsli5IpG8wra30,35029
-cite_agent/cli_conversational.py,sha256=RAmgRNRyB8gQ8QLvWU-Tt23j2lmA34rQNT5F3_7SOq0,11141
-cite_agent/cli_enhanced.py,sha256=EAaSw9qtiYRWUXF6_05T19GCXlz9cCSz6n41ASnXIPc,7407
-cite_agent/cli_workflow.py,sha256=4oS_jW9D8ylovXbEFdsyLQONt4o0xxR4Xatfcc4tnBs,11641
-cite_agent/dashboard.py,sha256=VGV5XQU1PnqvTsxfKMcue3j2ri_nvm9Be6O5aVays_w,10502
-cite_agent/enhanced_ai_agent.py,sha256=ha7cps216cNV59yLnQe2-K8Af9wXmSL7TokT3TX-JuY,167857
-cite_agent/rate_limiter.py,sha256=-0fXx8Tl4zVB4O28n9ojU2weRo-FBF1cJo9Z5jC2LxQ,10908
-cite_agent/session_manager.py,sha256=B0MXSOsXdhO3DlvTG7S8x6pmGlYEDvIZ-o8TZM23niQ,9444
-cite_agent/setup_config.py,sha256=3m2e3gw0srEWA0OygdRo64r-8HK5ohyXfct0c__CF3s,16817
-cite_agent/streaming_ui.py,sha256=N6TWOo7GVQ_Ynfw73JCfrdGcLIU-PwbS3GbsHQHegmg,7810
-cite_agent/telemetry.py,sha256=55kXdHvI24ZsEkbFtihcjIfJt2oiSXcEpLzTxQ3KCdQ,2916
-cite_agent/ui.py,sha256=r1OAeY3NSeqhAjJYmEBH9CaennBuibFAz1Mur6YF80E,6134
-cite_agent/updater.py,sha256=udoAAN4gBKAvKDV7JTh2FJO_jIhNk9bby4x6n188MEY,8458
-cite_agent/web_search.py,sha256=FZCuNO7MAITiOIbpPbJyt2bzbXPzQla-9amJpnMpW_4,6520
-cite_agent/workflow.py,sha256=a0YC0Mzz4or1C5t2gZcuJBQ0uMOZrooaI8eLu2kkI0k,15086
-cite_agent/workflow_integration.py,sha256=A9ua0DN5pRtuU0cAwrUTGvqt2SXKhEHQbrHx16EGnDM,10910
-cite_agent-1.3.5.dist-info/licenses/LICENSE,sha256=XJkyO4IymhSUniN1ENY6lLrL2729gn_rbRlFK6_Hi9M,1074
-src/__init__.py,sha256=0eEpjRfjRjOTilP66y-AbGNslBsVYr_clE-bZUzsX7s,40
-src/services/__init__.py,sha256=pTGLCH_84mz4nGtYMwQES5w-LzoSulUtx_uuNM6r-LA,4257
-src/services/simple_enhanced_main.py,sha256=IJoOplCqcVUg3GvN_BRyAhpGrLm_WEPy2jmHcNCY6R0,9257
-src/services/auth_service/__init__.py,sha256=VVFfBUr_GMJuxVH_553D2PZmZ9vhHeab9_qiJEf-g6Q,38
-src/services/auth_service/auth_manager.py,sha256=MJdWFE36R_htoyBbjgGSTSx2Py61sTM3lhBjXBZ4Bog,873
-src/services/graph/__init__.py,sha256=jheRQ-x652RZ68fKyUqUNGXmTAJsp5URVMhlOauFRO0,29
-src/services/graph/knowledge_graph.py,sha256=ips2IpVpxDFkdPku4XKgZNRnoR2NjZqZk3xbIArJaaM,7348
-src/services/llm_service/__init__.py,sha256=eNAsQpJtVXpJENb-gHtpKzWpncnHHAMB05EI48wrugQ,122
-src/services/llm_service/llm_manager.py,sha256=6o5KN-3wJ0hT8PS9hPMpTGS6G9SlleSzYsXZQRjj_vI,21027
-src/services/paper_service/__init__.py,sha256=0ONhTf_3H81l5y6EqHMRZd5dCXLAXDa-gbYwge84zKA,142
-src/services/paper_service/openalex.py,sha256=pPhPcHMK2gQJCUVPB4ujE8xya0UqUvfcN95cy5ooP68,8801
-src/services/performance_service/__init__.py,sha256=48bYfW4pzf-FG9644kTnNwGyD1tJJ7tVn3cD3r_ZAbk,65
-src/services/performance_service/rust_performance.py,sha256=n-FzJ98XslmpUAkmmuaunYDTPz-9ZY-qL4oWAoBAaoA,15558
-src/services/research_service/__init__.py,sha256=ZCBzSUdstHqwMmJ1x0kJK4PkRlv9OrSOEFeQFoVM-7M,813
-src/services/research_service/chatbot.py,sha256=12pVAoe_fd2RXi6_cP-fxfRnWyJStsyn8znVu5cy9qo,91153
-src/services/research_service/citation_manager.py,sha256=vzyVivBS0_9IiFE-wOH9hiLiC-fpHmiaZpR1084DenE,16586
-src/services/research_service/context_manager.py,sha256=FGbeylLWKvgoA5fElyiqg5IhnMBIZ-t3w0oDHN4Zy1E,61332
-src/services/research_service/conversation_manager.py,sha256=-rdzURzu-SiqozyeQLid5a5lS-KzIqGDozdE8BG-DTs,22854
-src/services/research_service/critical_paper_detector.py,sha256=gc3oZHB8RqDhxFqJx21NoKLcHmmqHXRo0eXY-AL5KSc,21941
-src/services/research_service/enhanced_research.py,sha256=5B8zZjJ2iSLEgnjfyDKow5x_MLRANLJdMbLmmPR5Lc0,4268
-src/services/research_service/enhanced_synthesizer.py,sha256=puJg2C10KXryCMPkec-chC4rxbIJdFFswo7w4rbaXkc,16603
-src/services/research_service/query_generator.py,sha256=LcFTGsewE6l2LRgUI2E6fXAcpy4vaYaUFFfZhI_WlYU,30707
-src/services/research_service/synthesizer.py,sha256=lCcu37PWhWVNphHKaJJDIC-JQ5OINAN7OJ7iV9BWAvM,52557
-src/services/search_service/__init__.py,sha256=UZFXdd7r6wietQ2kESXEyGffdfBbpghquecQde7auF4,137
-src/services/search_service/indexer.py,sha256=u3-uwdAfmahWWsdebDF9i8XIyp7YtUMIHzlmBLBnPPM,7252
-src/services/search_service/search_engine.py,sha256=S9HqQ_mk-8W4d4MUOgBbEGQGV29-eSuceSFvVb4Xk-k,12500
-cite_agent-1.3.5.dist-info/METADATA,sha256=qCT90yGGiD7au0geLi4DZYVpOH2i7bWWNJb1tLZHryw,12231
-cite_agent-1.3.5.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-cite_agent-1.3.5.dist-info/entry_points.txt,sha256=bJ0u28nFIxQKH1PWQ2ak4PV-FAjhoxTC7YADEdDenFw,83
-cite_agent-1.3.5.dist-info/top_level.txt,sha256=TgOFqJTIy8vDZuOoYA2QgagkqZtfhM5Acvt_IsWzAKo,15
-cite_agent-1.3.5.dist-info/RECORD,,

src/__init__.py

@@ -1 +0,0 @@
-"""Project root package initializer."""

src/services/__init__.py

@@ -1,132 +0,0 @@
-"""
-Main services package for AI services layer
-Provides unified access to all service components
-"""
-
-# Import all major service classes for easy access
-from .llm_service.llm_manager import LLMManager
-from .research_service.enhanced_research import EnhancedResearchService  
-from .context_manager.advanced_context import AdvancedContextManager
-from .tool_framework.tool_manager import ToolManager
-from .auth_service.auth_manager import auth_manager
-
-# Service registry for dependency injection
-SERVICE_REGISTRY = {}
-
-def register_service(name: str, service_instance):
-    """Register a service instance in the global registry"""
-    SERVICE_REGISTRY[name] = service_instance
-
-def get_service(name: str):
-    """Get a service instance from the registry"""
-    return SERVICE_REGISTRY.get(name)
-
-def initialize_services(config: dict = None):
-    """Initialize all core services with configuration"""
-    config = config or {}
-    
-    # Initialize services (with minimal config for testing)
-    services = {}
-    
-    try:
-        # LLM Manager (needs redis_url but we'll handle gracefully)
-        redis_url = config.get('redis_url', 'redis://localhost:6379')
-        llm_manager = LLMManager(redis_url=redis_url)
-        services['llm_manager'] = llm_manager
-        register_service('llm_manager', llm_manager)
-    except Exception as e:
-        # Graceful fallback for testing
-        print(f"LLM Manager initialization skipped: {e}")
-    
-    try:
-        # Research Service
-        research_service = EnhancedResearchService()
-        services['research_service'] = research_service
-        register_service('research_service', research_service)
-    except Exception as e:
-        print(f"Research Service initialization skipped: {e}")
-    
-    try:
-        # Context Manager  
-        context_manager = AdvancedContextManager()
-        services['context_manager'] = context_manager
-        register_service('context_manager', context_manager)
-    except Exception as e:
-        print(f"Context Manager initialization skipped: {e}")
-    
-    try:
-        # Tool Manager
-        tool_manager = ToolManager()
-        services['tool_manager'] = tool_manager
-        register_service('tool_manager', tool_manager)
-    except Exception as e:
-        print(f"Tool Manager initialization skipped: {e}")
-    
-    return services
-
-class ServiceLayer:
-    """Unified service layer for easy access to all AI services"""
-    
-    def __init__(self, config: dict = None):
-        self.config = config or {}
-        self.services = {}
-        self._initialized = False
-    
-    def initialize(self):
-        """Initialize all services"""
-        if self._initialized:
-            return
-            
-        self.services = initialize_services(self.config)
-        self._initialized = True
-    
-    @property
-    def llm_manager(self) -> LLMManager:
-        """Get LLM Manager service"""
-        return self.services.get('llm_manager')
-    
-    @property 
-    def research_service(self) -> EnhancedResearchService:
-        """Get Research service"""
-        return self.services.get('research_service')
-    
-    @property
-    def context_manager(self) -> AdvancedContextManager:
-        """Get Context Manager service"""
-        return self.services.get('context_manager')
-    
-    @property
-    def tool_manager(self) -> ToolManager:
-        """Get Tool Manager service"""
-        return self.services.get('tool_manager')
-    
-    def get_health_status(self) -> dict:
-        """Get health status of all services"""
-        status = {
-            "services_initialized": self._initialized,
-            "total_services": len(self.services),
-            "available_services": list(self.services.keys())
-        }
-        return status
-
-# Global service layer instance
-_service_layer = None
-
-def get_service_layer(config: dict = None) -> ServiceLayer:
-    """Get the global service layer instance"""
-    global _service_layer
-    if _service_layer is None:
-        _service_layer = ServiceLayer(config)
-    return _service_layer
-
-# Export key classes and functions
-__all__ = [
-    'LLMManager',
-    'EnhancedResearchService', 
-    'AdvancedContextManager',
-    'ToolManager',
-    'ServiceLayer',
-    'get_service_layer',
-    'initialize_services',
-    'auth_manager'
-]

src/services/auth_service/__init__.py

@@ -1,3 +0,0 @@
-"""
-Authentication service package
-"""

src/services/auth_service/auth_manager.py

@@ -1,33 +0,0 @@
-"""
-Basic authentication manager for testing
-"""
-from typing import Dict, Any, Optional
-
-
-class AuthManager:
-    """Basic auth manager for testing purposes"""
-    
-    def __init__(self):
-        self.test_user = {
-            "id": "test_user_123",
-            "username": "test_user",
-            "email": "test@example.com"
-        }
-    
-    async def get_current_user(self) -> Dict[str, Any]:
-        """Return test user for testing"""
-        return self.test_user
-    
-    def verify_token(self, token: str) -> Optional[Dict[str, Any]]:
-        """Verify token (test implementation)"""
-        if token == "test_token":
-            return self.test_user
-        return None
-    
-    def create_token(self, user_data: Dict[str, Any]) -> str:
-        """Create token (test implementation)"""
-        return "test_token"
-
-
-# Global instance
-auth_manager = AuthManager()

src/services/graph/__init__.py

@@ -1 +0,0 @@
-"""Graph service package."""

src/services/graph/knowledge_graph.py

@@ -1,194 +0,0 @@
-"""Lightweight async knowledge graph implementation used by the research synthesizer.
-
-The production design originally assumed an external graph database, but the launch-ready
-runtime needs a dependable in-process implementation that works without external services.
-This module provides a minimal yet functional directed multigraph using in-memory storage.
-
-The implementation focuses on the operations exercised by ``ResearchSynthesizer``:
-
-* ``upsert_entity`` – register/update an entity node with typed metadata
-* ``upsert_relationship`` – connect two entities with rich relationship properties
-* ``get_entity`` / ``get_relationships`` – helper APIs for diagnostics and future features
-
-Data is persisted in memory and optionally mirrored to a JSON file on disk so the graph can
-survive multiple sessions during local development.  All public methods are ``async`` to keep
-parity with the historical interface and to allow easy replacement with an external graph
-backend in the future.
-"""
-
-from __future__ import annotations
-
-import asyncio
-import json
-from dataclasses import dataclass, field
-from pathlib import Path
-from typing import Any, Dict, List, Optional, Tuple
-
-__all__ = ["KnowledgeGraph", "GraphEntity", "GraphRelationship"]
-
-
-@dataclass
-class GraphEntity:
-    """Represents a node in the knowledge graph."""
-
-    entity_id: str
-    entity_type: str
-    properties: Dict[str, Any] = field(default_factory=dict)
-
-    def to_dict(self) -> Dict[str, Any]:
-        return {
-            "id": self.entity_id,
-            "type": self.entity_type,
-            "properties": self.properties,
-        }
-
-
-@dataclass
-class GraphRelationship:
-    """Represents a directed, typed relationship between two entities."""
-
-    rel_type: str
-    source_id: str
-    target_id: str
-    properties: Dict[str, Any] = field(default_factory=dict)
-
-    def to_dict(self) -> Dict[str, Any]:
-        return {
-            "type": self.rel_type,
-            "source": self.source_id,
-            "target": self.target_id,
-            "properties": self.properties,
-        }
-
-
-class KnowledgeGraph:
-    """A simple async-safe in-memory knowledge graph."""
-
-    def __init__(self, *, persistence_path: Optional[Path] = None) -> None:
-        self._entities: Dict[str, GraphEntity] = {}
-        # Adjacency list keyed by (source_id, rel_type) -> list[target_id, props]
-        self._relationships: List[GraphRelationship] = []
-        self._lock = asyncio.Lock()
-        self._persistence_path = persistence_path
-        if self._persistence_path:
-            self._load_from_disk()
-
-    # ------------------------------------------------------------------
-    # Persistence helpers
-    # ------------------------------------------------------------------
-    def _load_from_disk(self) -> None:
-        if not self._persistence_path or not self._persistence_path.exists():
-            return
-        try:
-            payload = json.loads(self._persistence_path.read_text())
-        except Exception:
-            return
-
-        for entity in payload.get("entities", []):
-            graph_entity = GraphEntity(
-                entity_id=entity["id"],
-                entity_type=entity.get("type", "Unknown"),
-                properties=entity.get("properties", {}),
-            )
-            self._entities[graph_entity.entity_id] = graph_entity
-
-        for rel in payload.get("relationships", []):
-            graph_rel = GraphRelationship(
-                rel_type=rel.get("type", "related_to"),
-                source_id=rel.get("source"),
-                target_id=rel.get("target"),
-                properties=rel.get("properties", {}),
-            )
-            self._relationships.append(graph_rel)
-
-    def _persist(self) -> None:
-        if not self._persistence_path:
-            return
-        data = {
-            "entities": [entity.to_dict() for entity in self._entities.values()],
-            "relationships": [rel.to_dict() for rel in self._relationships],
-        }
-        try:
-            self._persistence_path.parent.mkdir(parents=True, exist_ok=True)
-            self._persistence_path.write_text(json.dumps(data, indent=2, sort_keys=True))
-        except Exception:
-            # Persistence failures should never stop the conversation flow
-            pass
-
-    # ------------------------------------------------------------------
-    # Public API
-    # ------------------------------------------------------------------
-    async def upsert_entity(self, entity_type: str, properties: Dict[str, Any]) -> str:
-        """Create or update an entity.
-
-        Args:
-            entity_type: Semantic type (e.g., "Paper", "Author").
-            properties: Arbitrary metadata. ``properties['id']`` is optional; when missing
-                a deterministic identifier is derived from ``properties['external_id']`` or
-                a hash of the payload.
-        Returns:
-            The entity identifier stored in the graph.
-        """
-
-        async with self._lock:
-            entity_id = _determine_entity_id(entity_type, properties)
-            entity = self._entities.get(entity_id)
-            if entity:
-                entity.properties.update(properties)
-            else:
-                entity = GraphEntity(entity_id=entity_id, entity_type=entity_type, properties=properties)
-                self._entities[entity_id] = entity
-            self._persist()
-            return entity_id
-
-    async def upsert_relationship(
-        self,
-        rel_type: str,
-        source_id: str,
-        target_id: str,
-        properties: Optional[Dict[str, Any]] = None,
-    ) -> Tuple[str, str, str]:
-        """Create or update a directed relationship between two entities."""
-
-        properties = properties or {}
-        async with self._lock:
-            relationship = GraphRelationship(
-                rel_type=rel_type,
-                source_id=source_id,
-                target_id=target_id,
-                properties=properties,
-            )
-            self._relationships.append(relationship)
-            self._persist()
-            return (relationship.rel_type, relationship.source_id, relationship.target_id)
-
-    async def get_entity(self, entity_id: str) -> Optional[GraphEntity]:
-        async with self._lock:
-            return self._entities.get(entity_id)
-
-    async def get_relationships(self, entity_id: str) -> List[GraphRelationship]:
-        async with self._lock:
-            return [rel for rel in self._relationships if rel.source_id == entity_id or rel.target_id == entity_id]
-
-    async def stats(self) -> Dict[str, Any]:
-        async with self._lock:
-            return {
-                "entities": len(self._entities),
-                "relationships": len(self._relationships),
-            }
-
-
-def _determine_entity_id(entity_type: str, properties: Dict[str, Any]) -> str:
-    """Best-effort deterministic identifier for an entity."""
-
-    # Preferred explicit IDs
-    for key in ("id", "external_id", "paper_id", "author_id", "identifier"):
-        value = properties.get(key)
-        if value:
-            return str(value)
-
-    # Fall back to hashed representation (order-stable via JSON dumps)
-    import hashlib
-
-    payload = json.dumps({"type": entity_type, "properties": properties}, sort_keys=True)
-    return f"{entity_type}:{hashlib.md5(payload.encode('utf-8')).hexdigest()}"

src/services/llm_service/__init__.py

@@ -1,5 +0,0 @@
-"""LLM service package exposing the unified LLMManager."""
-
-from .llm_manager import LLMManager
-
-__all__ = ["LLMManager"]