cli-memory-os 0.1.2__tar.gz → 0.1.4__tar.gz

{cli_memory_os-0.1.2/cli_memory_os.egg-info → cli_memory_os-0.1.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: cli-memory-os
-Version: 0.1.2
+Version: 0.1.4
 Summary: CLI-based Personal Knowledge Operating System
 Author-email: Anirudh T <anirudh200584@gmail.com>
 License-Expression: MIT

cli_memory_os-0.1.4/cli/commands/ask.py

@@ -0,0 +1,67 @@
+"""
+Command: memory-os ask
+
+Queries the RAG pipeline with a question.
+"""
+
+import sys
+import sqlite3
+from core.llm import run_hybrid_rag
+from storage.db import init_db
+
+
+def run_and_print_ask(query: str):
+    """Run RAG query pipeline and print the structured result blocks."""
+    rag_res = run_hybrid_rag(query)
+    
+    print("========================================")
+    print("ANSWER")
+    print("========================================")
+    print(rag_res["answer"])
+    print("\n========================================")
+    print("SOURCES")
+    print("========================================")
+    if rag_res.get("sources"):
+        for s in rag_res["sources"]:
+            print(f"- {s}")
+    else:
+        print("None")
+    print("\n========================================")
+    print("REPOSITORIES USED")
+    print("========================================")
+    if rag_res.get("repositories"):
+        for r in rag_res["repositories"]:
+            print(f"- {r}")
+    else:
+        print("None")
+    print("\n========================================")
+    confidence = rag_res.get("confidence", 0.0)
+    print(f"Confidence: {confidence:.2f}")
+    print("========================================")
+
+
+def execute(args):
+    """Run the ask command parser handler."""
+    # Ensure database is initialized
+    try:
+        init_db()
+    except sqlite3.OperationalError as e:
+        print("----------------------------------")
+        print("Workspace has not been initialized.")
+        print("")
+        print("Run:")
+        print("memory-os init")
+        print("----------------------------------")
+        sys.exit(1)
+
+    if not args.question:
+        print("Usage: memory-os ask <question>")
+        sys.exit(1)
+
+    query = " ".join(args.question)
+    
+    try:
+        run_and_print_ask(query)
+    except Exception as e:
+        print(f"❌ Error executing query: {e}")
+        sys.exit(1)

cli_memory_os-0.1.4/cli/commands/sync.py

@@ -0,0 +1,70 @@
+"""
+Command: memory-os sync
+
+Synchronizes data from connected sources (GitHub, Gmail, Notion),
+chunks documents, generates embeddings, and syncs databases.
+"""
+
+import sys
+import time
+import logging
+import sqlite3
+from storage.db import init_db, clear_all
+from core.vector_store import run_reindexing, close_qdrant_client
+from storage.graph import GraphStore
+from connectors.github import sync_github
+from connectors.gmail import sync_gmail
+from connectors.notion import sync_notion
+
+
+def execute(args):
+    """Run the synchronization pipeline."""
+    # Ensure database is initialized
+    try:
+        init_db()
+    except sqlite3.OperationalError as e:
+        print("----------------------------------")
+        print("Workspace has not been initialized.")
+        print("")
+        print("Run:")
+        print("memory-os init")
+        print("----------------------------------")
+        sys.exit(1)
+
+    start_time = time.perf_counter()
+    
+    rebuild = getattr(args, "rebuild", False)
+    source = getattr(args, "source", None)
+    
+    if rebuild:
+        print("Performing full rebuild reset...")
+        clear_all()
+        close_qdrant_client()
+        run_reindexing()
+        GraphStore().clear_graph()
+
+    # Sync based on source selection
+    if source == "github":
+        print("Syncing GitHub source...")
+        sync_github()
+    elif source == "gmail":
+        print("Syncing Gmail source...")
+        sync_gmail()
+    elif source == "notion":
+        print("Syncing Notion source...")
+        sync_notion()
+    else:
+        print("Syncing all sources (GitHub, Gmail, Notion)...")
+        sync_github()
+        sync_gmail()
+        sync_notion()
+
+    print("Running document chunking and vector reindexing...")
+    run_reindexing()
+
+    print("Extracting and syncing knowledge graph relationships...")
+    GraphStore().extract_and_sync_graph()
+
+    duration = time.perf_counter() - start_time
+    print(f"\nSync complete. Total Duration: {duration:.2f}s")
+    logging.getLogger("main").info(f"Sync completed in {duration:.2f}s")

{cli_memory_os-0.1.2 → cli_memory_os-0.1.4}/cli/main.py

@@ -115,7 +115,6 @@ def run_interactive():
         close_qdrant_client,
     )
     from core.embedder import Embedder
-    from core.llm import run_hybrid_rag
     from storage.graph import GraphStore
     from connectors.github import sync_github
     from connectors.gmail import sync_gmail
@@ -283,30 +282,11 @@ def run_interactive():
                 if not arg:
                     print("Usage: ask <question>")
                 else:
-                    rag_res = run_hybrid_rag(arg)
-                    print("========================================")
-                    print("ANSWER")
-                    print("========================================")
-                    print(rag_res["answer"])
-                    print("\n========================================")
-                    print("SOURCES")
-                    print("========================================")
-                    if rag_res["sources"]:
-                        for s in rag_res["sources"]:
-                            print(f"- {s}")
-                    else:
-                        print("None")
-                    print("\n========================================")
-                    print("REPOSITORIES USED")
-                    print("========================================")
-                    if rag_res["repositories"]:
-                        for r in rag_res["repositories"]:
-                            print(f"- {r}")
-                    else:
-                        print("None")
-                    print("\n========================================")
-                    print(f"Confidence: {rag_res['confidence']:.2f}")
-                    print("========================================")
+                    from cli.commands.ask import run_and_print_ask
+                    try:
+                        run_and_print_ask(arg)
+                    except Exception as e:
+                        print(f"❌ Error executing query: {e}")
 
             elif cmd == "repo-info":
                 if not arg:
@@ -441,11 +421,16 @@ def route_command(args):
     elif command == "logs":
         from cli.commands.logs import execute
         execute(args)
+    elif command == "sync":
+        from cli.commands.sync import execute
+        execute(args)
+    elif command == "ask":
+        from cli.commands.ask import execute
+        execute(args)
     else:
-        # For commands not yet implemented, show a helpful message
-        print(f"Command '{command}' is not yet implemented.")
-        print("Run 'memory-os --help' for available commands.")
-        print("Run 'memory-os' (no arguments) for the interactive REPL.")
+        from cli.parser import build_parser
+        build_parser().print_help()
+        sys.exit(1)
 
 
 

{cli_memory_os-0.1.2 → cli_memory_os-0.1.4/cli_memory_os.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: cli-memory-os
-Version: 0.1.2
+Version: 0.1.4
 Summary: CLI-based Personal Knowledge Operating System
 Author-email: Anirudh T <anirudh200584@gmail.com>
 License-Expression: MIT

{cli_memory_os-0.1.2 → cli_memory_os-0.1.4}/cli_memory_os.egg-info/SOURCES.txt

@@ -6,6 +6,7 @@ cli/__init__.py
 cli/main.py
 cli/parser.py
 cli/commands/__init__.py
+cli/commands/ask.py
 cli/commands/benchmark.py
 cli/commands/config_cmd.py
 cli/commands/doctor.py
@@ -19,6 +20,7 @@ cli/commands/restart.py
 cli/commands/start.py
 cli/commands/status.py
 cli/commands/stop.py
+cli/commands/sync.py
 cli/commands/version.py
 cli/commands/workspace.py
 cli_memory_os.egg-info/PKG-INFO
@@ -58,4 +60,5 @@ tests/test_commands_phase2.py
 tests/test_compose_manager.py
 tests/test_first_time_user_experience.py
 tests/test_infrastructure_stabilization.py
+tests/test_sync_ask_cli.py
 tests/test_v1_features.py

{cli_memory_os-0.1.2 → cli_memory_os-0.1.4}/cli_memory_os.egg-info/top_level.txt

@@ -5,4 +5,5 @@ dist
 infrastructure
 logs
 models
+scratch
 storage

{cli_memory_os-0.1.2 → cli_memory_os-0.1.4}/connectors/github.py

@@ -67,23 +67,36 @@ def sync_github():
 
         print("Syncing GitHub...\n")
         
-        # Fetch repositories
-        resp = s.execute(tool_slug="github_list_repositories_for_the_authenticated_user", arguments={})
-        if resp.error or not resp.data:
-            print("No repositories found.")
-            return
-
-        resp_data = resp.data
-        if isinstance(resp_data, dict) and "response_data" in resp_data:
-            resp_data = resp_data["response_data"]
-
+        # Fetch repositories with pagination
         repos = []
-        if isinstance(resp_data, list):
-            repos = resp_data
-        elif isinstance(resp_data, dict):
-            repos = resp_data.get("repositories") or resp_data.get("items") or []
+        page = 1
+        while True:
+            resp = s.execute(
+                tool_slug="github_list_repositories_for_the_authenticated_user",
+                arguments={"per_page": 100, "page": page}
+            )
+            if resp.error or not resp.data:
+                break
+                
+            resp_data = resp.data
+            if isinstance(resp_data, dict) and "response_data" in resp_data:
+                resp_data = resp_data["response_data"]
+                
+            page_repos = []
+            if isinstance(resp_data, list):
+                page_repos = resp_data
+            elif isinstance(resp_data, dict):
+                page_repos = resp_data.get("repositories") or resp_data.get("items") or []
+                
+            if not isinstance(page_repos, list) or not page_repos:
+                break
+                
+            repos.extend(page_repos)
+            if len(page_repos) < 100:
+                break
+            page += 1
 
-        if not isinstance(repos, list) or not repos:
+        if not repos:
             print("No repositories found.")
             return
 

{cli_memory_os-0.1.2 → cli_memory_os-0.1.4}/connectors/gmail.py

@@ -23,19 +23,56 @@ def sync_gmail():
             print("Gmail connection not active.")
             return
 
-        # Fetch emails
-        resp = s.execute(tool_slug="gmail_fetch_emails", arguments={"max_results": 50, "include_payload": True})
-        if not resp or resp.error or not resp.data:
-            print("No emails found.")
-            return
+        # Determine last fetched date from stored emails to optimize retrieval query
+        query = ""
+        try:
+            conn = get_connection()
+            cursor = conn.cursor()
+            cursor.execute("SELECT MAX(received_at) FROM emails WHERE received_at IS NOT NULL AND received_at != ''")
+            row = cursor.fetchone()
+            conn.close()
+            if row and row[0]:
+                import re
+                m = re.match(r"(\d{4})[-/](\d{2})[-/](\d{2})", row[0])
+                if m:
+                    query = f"after:{m.group(1)}/{m.group(2)}/{m.group(3)}"
+        except Exception as e:
+            import logging
+            logging.getLogger("gmail").warning(f"Failed to query last fetched date from SQLite: {e}")
+
+        # Fetch emails with pagination
+        emails = []
+        page_token = None
+        while True:
+            args = {"max_results": 100, "include_payload": True}
+            if query:
+                args["q"] = query
+            if page_token:
+                args["page_token"] = page_token
+
+            resp = s.execute(tool_slug="gmail_fetch_emails", arguments=args)
+            if not resp or resp.error or not resp.data:
+                break
+
+            page_emails = resp.data.get("messages", [])
+            if not page_emails:
+                page_emails = resp.data.get("response_data", {}).get("messages", [])
+
+            if not page_emails:
+                break
+
+            emails.extend(page_emails)
+            page_token = resp.data.get("nextPageToken")
+            if not page_token:
+                break
+
+            # Safety check: avoid infinite loops if user has thousands of new emails
+            if len(emails) >= 500:
+                break
 
-        emails = resp.data.get("messages", [])
         if not emails:
-            # Check response_data structure just in case
-            emails = resp.data.get("response_data", {}).get("messages", [])
-            if not emails:
-                print("No emails found.")
-                return
+            print("No emails found.")
+            return
 
         print(f"Found {len(emails)} emails")
         

{cli_memory_os-0.1.2 → cli_memory_os-0.1.4}/core/vector_store.py

@@ -19,10 +19,10 @@ def get_qdrant_client() -> QdrantClient:
     if _qdrant_client is None:
         url = os.getenv("QDRANT_URL")
         if url:
-            _qdrant_client = QdrantClient(url=url)
+            _qdrant_client = QdrantClient(url=url, timeout=60.0)
         else:
             path = os.getenv("MEMORY_OS_QDRANT_PATH", "qdrant_storage")
-            _qdrant_client = QdrantClient(path=path)
+            _qdrant_client = QdrantClient(path=path, timeout=60.0)
     return _qdrant_client
 
 

{cli_memory_os-0.1.2 → cli_memory_os-0.1.4}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "cli-memory-os"
-version = "0.1.2"
+version = "0.1.4"
 description = "CLI-based Personal Knowledge Operating System"
 readme = "README.md"
 requires-python = ">=3.12"

{cli_memory_os-0.1.2 → cli_memory_os-0.1.4}/tests/test_first_time_user_experience.py

@@ -76,6 +76,7 @@ def test_init_missing_docker(capsys):
     """Test memory-os init when Docker Desktop is not running."""
     with patch("infrastructure.docker.check_docker_installed", return_value=(True, "Docker version 24.0.7")), \
          patch("infrastructure.docker.check_docker_compose_installed", return_value=(True, "Docker Compose version 2.23.3")), \
+         patch("infrastructure.docker.check_docker_daemon", return_value=(False, "Docker Desktop is not running.")), \
          patch("infrastructure.docker.check_docker_running", return_value=False):
          
         with pytest.raises(SystemExit) as excinfo:

cli_memory_os-0.1.4/tests/test_sync_ask_cli.py

@@ -0,0 +1,111 @@
+"""
+Unit and integration tests for sync, ask commands, and the interactive REPL.
+"""
+
+import sys
+import os
+import unittest
+from unittest.mock import patch, MagicMock
+
+# Ensure project root is in path
+sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+
+
+class DummyArgs:
+    def __init__(self, **kwargs):
+        for k, v in kwargs.items():
+            setattr(self, k, v)
+
+
+class TestSyncAskCLI(unittest.TestCase):
+    """Test suite verifying sync and ask CLI command routing and logic."""
+
+    @patch("cli.commands.sync.init_db")
+    @patch("cli.commands.sync.sync_github")
+    @patch("cli.commands.sync.sync_gmail")
+    @patch("cli.commands.sync.sync_notion")
+    @patch("cli.commands.sync.run_reindexing")
+    @patch("cli.commands.sync.GraphStore")
+    def test_sync_command_all_sources(
+        self, mock_graph_store_class, mock_run_reindexing, mock_sync_notion, mock_sync_gmail, mock_sync_github, mock_init_db
+    ):
+        """Test memory-os sync command runs sync for all sources."""
+        mock_graph_inst = MagicMock()
+        mock_graph_store_class.return_value = mock_graph_inst
+
+        from cli.commands.sync import execute as execute_sync
+
+        args = DummyArgs(rebuild=False, source=None)
+        execute_sync(args)
+
+        mock_init_db.assert_called_once()
+        mock_sync_github.assert_called_once()
+        mock_sync_gmail.assert_called_once()
+        mock_sync_notion.assert_called_once()
+        mock_run_reindexing.assert_called_once()
+        mock_graph_inst.extract_and_sync_graph.assert_called_once()
+
+    @patch("cli.commands.sync.init_db")
+    @patch("cli.commands.sync.sync_github")
+    @patch("cli.commands.sync.run_reindexing")
+    @patch("cli.commands.sync.GraphStore")
+    def test_sync_command_single_source(
+        self, mock_graph_store_class, mock_run_reindexing, mock_sync_github, mock_init_db
+    ):
+        """Test memory-os sync command can sync only a single source."""
+        mock_graph_inst = MagicMock()
+        mock_graph_store_class.return_value = mock_graph_inst
+
+        from cli.commands.sync import execute as execute_sync
+
+        args = DummyArgs(rebuild=False, source="github")
+        execute_sync(args)
+
+        mock_init_db.assert_called_once()
+        mock_sync_github.assert_called_once()
+        mock_run_reindexing.assert_called_once()
+        mock_graph_inst.extract_and_sync_graph.assert_called_once()
+
+    @patch("cli.commands.ask.init_db")
+    @patch("cli.commands.ask.run_hybrid_rag")
+    def test_ask_command(self, mock_run_hybrid_rag, mock_init_db):
+        """Test memory-os ask command query execution."""
+        mock_run_hybrid_rag.return_value = {
+            "answer": "Python is used in core services.",
+            "sources": ["source_1.txt"],
+            "repositories": ["repo-a"],
+            "confidence": 0.95
+        }
+
+        from cli.commands.ask import execute as execute_ask
+
+        args = DummyArgs(question=["What", "projects", "use", "Python?"])
+        execute_ask(args)
+
+        mock_init_db.assert_called_once()
+        mock_run_hybrid_rag.assert_called_once_with("What projects use Python?")
+
+    @patch("storage.db.init_db")
+    @patch("core.embedder.Embedder")
+    @patch("cli.main.input")
+    @patch("cli.commands.ask.run_and_print_ask")
+    def test_interactive_repl_ask(self, mock_run_and_print_ask, mock_input, mock_embedder_class, mock_init_db):
+        """Test interactive REPL triggers the ask command routing for natural language query."""
+        # Mock REPL sequence: 1. A question, 2. 'exit' to terminate loop
+        mock_input.side_effect = ["What projects use Python?", "exit"]
+        
+        # Mock embedder instantiation
+        mock_embedder_inst = MagicMock()
+        mock_embedder_class.return_value = mock_embedder_inst
+
+        from cli.main import run_interactive
+
+        with patch("core.vector_store.close_qdrant_client"), patch("storage.graph.GraphStore"):
+            run_interactive()
+
+        # Verify that natural language query gets routed to ask and calls run_and_print_ask
+        mock_run_and_print_ask.assert_called_once_with("What projects use Python?")
+
+
+if __name__ == "__main__":
+    unittest.main()