gnosisllm-knowledge 0.2.0__tar.gz → 0.4.0__tar.gz

{gnosisllm_knowledge-0.2.0 → gnosisllm_knowledge-0.4.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: gnosisllm-knowledge
-Version: 0.2.0
+Version: 0.4.0
 Summary: Enterprise-grade knowledge loading, indexing, and search for Python
 License: MIT
 Keywords: knowledge-base,rag,semantic-search,vector-search,opensearch,llm,embeddings,enterprise
@@ -41,10 +41,12 @@ Enterprise-grade knowledge loading, indexing, and semantic search library for Py
 
 - **Semantic Search**: Vector-based similarity search using OpenAI embeddings
 - **Hybrid Search**: Combine semantic and keyword (BM25) search for best results
+- **Agentic Search**: AI-powered search with reasoning and natural language answers
+- **Agentic Memory**: Conversational memory with automatic fact extraction
 - **Multiple Loaders**: Load content from websites, sitemaps, and files
 - **Intelligent Chunking**: Sentence-aware text splitting with configurable overlap
 - **OpenSearch Backend**: Production-ready with k-NN vector search
-- **Multi-Tenancy**: Built-in support for account and collection isolation
+- **Multi-Tenancy**: Index isolation for complete tenant separation (tenant-agnostic library)
 - **Event-Driven**: Observer pattern for progress tracking and monitoring
 - **SOLID Architecture**: Clean, maintainable, and extensible codebase
 
@@ -142,14 +144,15 @@ gnosisllm-knowledge load <URL> [OPTIONS]
 
 Options:
   --type         Source type: website, sitemap (auto-detects)
-  --index        Target index name (default: knowledge)
-  --account-id   Multi-tenant account ID
+  --index        Target index name (e.g., knowledge-tenant-123)
   --collection-id Collection grouping ID
   --batch-size   Documents per batch (default: 100)
   --max-urls     Max URLs from sitemap (default: 1000)
   --dry-run      Preview without indexing
 ```
 
+Multi-tenancy is achieved through index isolation. Use `--index` with tenant-specific names (e.g., `--index knowledge-tenant-123`).
+
 ### Search
 
 Search indexed content with multiple modes:
@@ -159,14 +162,15 @@ gnosisllm-knowledge search <QUERY> [OPTIONS]
 
 Options:
   --mode         Search mode: semantic, keyword, hybrid, agentic
-  --index        Index to search (default: knowledge)
+  --index        Index to search (e.g., knowledge-tenant-123)
   --limit        Max results (default: 5)
-  --account-id   Filter by account
   --collection-ids Filter by collections (comma-separated)
   --json         Output as JSON for scripting
   --interactive  Interactive search session
 ```
 
+Multi-tenancy is achieved through index isolation. Use `--index` with tenant-specific names.
+
 ## Architecture
 
 ```
@@ -317,22 +321,81 @@ agent_body = {
 
 ## Multi-Tenancy
 
+This library is **tenant-agnostic**. Multi-tenancy is achieved through **index isolation** - each tenant gets their own OpenSearch index.
+
 ```python
-# Load with tenant isolation
+# The calling application (e.g., API) constructs tenant-specific index names
+index_name = f"knowledge-{account_id}"
+
+# Create Knowledge instance for the tenant
+knowledge = Knowledge.from_opensearch(
+    host="localhost",
+    port=9200,
+    index_prefix=index_name,  # knowledge-tenant-123
+)
+
+# Load content to tenant's isolated index
 await knowledge.load(
     source="https://docs.example.com/sitemap.xml",
-    account_id="tenant-123",
     collection_id="docs",
 )
 
-# Search within tenant
+# Search within tenant's index (no account_id filter needed)
 results = await knowledge.search(
     "query",
-    account_id="tenant-123",
     collection_ids=["docs"],
 )
 ```
 
+**Note**: For audit purposes, you can store `account_id` in document metadata:
+```python
+await knowledge.load(
+    source="https://docs.example.com/sitemap.xml",
+    document_defaults={"metadata": {"account_id": "tenant-123"}},
+)
+```
+
+## Agentic Memory
+
+Conversational memory with automatic fact extraction using OpenSearch's ML Memory plugin.
+
+```bash
+# Setup memory connectors
+gnosisllm-knowledge memory setup --openai-key sk-...
+
+# Create container and store conversations
+gnosisllm-knowledge memory container create my-memory
+gnosisllm-knowledge memory store <container-id> --file messages.json --user-id alice
+gnosisllm-knowledge memory recall <container-id> "user preferences" --user-id alice
+```
+
+```python
+from gnosisllm_knowledge import Memory, MemoryStrategy, StrategyConfig, Message
+
+memory = Memory.from_env()
+
+# Create container with strategies
+container = await memory.create_container(
+    name="agent-memory",
+    strategies=[
+        StrategyConfig(type=MemoryStrategy.SEMANTIC, namespace=["user_id"]),
+    ],
+)
+
+# Store conversation with fact extraction
+await memory.store(
+    container_id=container.id,
+    messages=[Message(role="user", content="I prefer dark mode")],
+    user_id="alice",
+    infer=True,
+)
+
+# Recall memories
+result = await memory.recall(container.id, "preferences", user_id="alice")
+```
+
+See [docs/memory.md](docs/memory.md) for full documentation.
+
 ## Event Tracking
 
 ```python

{gnosisllm_knowledge-0.2.0 → gnosisllm_knowledge-0.4.0}/README.md

@@ -6,10 +6,12 @@ Enterprise-grade knowledge loading, indexing, and semantic search library for Py
 
 - **Semantic Search**: Vector-based similarity search using OpenAI embeddings
 - **Hybrid Search**: Combine semantic and keyword (BM25) search for best results
+- **Agentic Search**: AI-powered search with reasoning and natural language answers
+- **Agentic Memory**: Conversational memory with automatic fact extraction
 - **Multiple Loaders**: Load content from websites, sitemaps, and files
 - **Intelligent Chunking**: Sentence-aware text splitting with configurable overlap
 - **OpenSearch Backend**: Production-ready with k-NN vector search
-- **Multi-Tenancy**: Built-in support for account and collection isolation
+- **Multi-Tenancy**: Index isolation for complete tenant separation (tenant-agnostic library)
 - **Event-Driven**: Observer pattern for progress tracking and monitoring
 - **SOLID Architecture**: Clean, maintainable, and extensible codebase
 
@@ -107,14 +109,15 @@ gnosisllm-knowledge load <URL> [OPTIONS]
 
 Options:
   --type         Source type: website, sitemap (auto-detects)
-  --index        Target index name (default: knowledge)
-  --account-id   Multi-tenant account ID
+  --index        Target index name (e.g., knowledge-tenant-123)
   --collection-id Collection grouping ID
   --batch-size   Documents per batch (default: 100)
   --max-urls     Max URLs from sitemap (default: 1000)
   --dry-run      Preview without indexing
 ```
 
+Multi-tenancy is achieved through index isolation. Use `--index` with tenant-specific names (e.g., `--index knowledge-tenant-123`).
+
 ### Search
 
 Search indexed content with multiple modes:
@@ -124,14 +127,15 @@ gnosisllm-knowledge search <QUERY> [OPTIONS]
 
 Options:
   --mode         Search mode: semantic, keyword, hybrid, agentic
-  --index        Index to search (default: knowledge)
+  --index        Index to search (e.g., knowledge-tenant-123)
   --limit        Max results (default: 5)
-  --account-id   Filter by account
   --collection-ids Filter by collections (comma-separated)
   --json         Output as JSON for scripting
   --interactive  Interactive search session
 ```
 
+Multi-tenancy is achieved through index isolation. Use `--index` with tenant-specific names.
+
 ## Architecture
 
 ```
@@ -282,22 +286,81 @@ agent_body = {
 
 ## Multi-Tenancy
 
+This library is **tenant-agnostic**. Multi-tenancy is achieved through **index isolation** - each tenant gets their own OpenSearch index.
+
 ```python
-# Load with tenant isolation
+# The calling application (e.g., API) constructs tenant-specific index names
+index_name = f"knowledge-{account_id}"
+
+# Create Knowledge instance for the tenant
+knowledge = Knowledge.from_opensearch(
+    host="localhost",
+    port=9200,
+    index_prefix=index_name,  # knowledge-tenant-123
+)
+
+# Load content to tenant's isolated index
 await knowledge.load(
     source="https://docs.example.com/sitemap.xml",
-    account_id="tenant-123",
     collection_id="docs",
 )
 
-# Search within tenant
+# Search within tenant's index (no account_id filter needed)
 results = await knowledge.search(
     "query",
-    account_id="tenant-123",
     collection_ids=["docs"],
 )
 ```
 
+**Note**: For audit purposes, you can store `account_id` in document metadata:
+```python
+await knowledge.load(
+    source="https://docs.example.com/sitemap.xml",
+    document_defaults={"metadata": {"account_id": "tenant-123"}},
+)
+```
+
+## Agentic Memory
+
+Conversational memory with automatic fact extraction using OpenSearch's ML Memory plugin.
+
+```bash
+# Setup memory connectors
+gnosisllm-knowledge memory setup --openai-key sk-...
+
+# Create container and store conversations
+gnosisllm-knowledge memory container create my-memory
+gnosisllm-knowledge memory store <container-id> --file messages.json --user-id alice
+gnosisllm-knowledge memory recall <container-id> "user preferences" --user-id alice
+```
+
+```python
+from gnosisllm_knowledge import Memory, MemoryStrategy, StrategyConfig, Message
+
+memory = Memory.from_env()
+
+# Create container with strategies
+container = await memory.create_container(
+    name="agent-memory",
+    strategies=[
+        StrategyConfig(type=MemoryStrategy.SEMANTIC, namespace=["user_id"]),
+    ],
+)
+
+# Store conversation with fact extraction
+await memory.store(
+    container_id=container.id,
+    messages=[Message(role="user", content="I prefer dark mode")],
+    user_id="alice",
+    infer=True,
+)
+
+# Recall memories
+result = await memory.recall(container.id, "preferences", user_id="alice")
+```
+
+See [docs/memory.md](docs/memory.md) for full documentation.
+
 ## Event Tracking
 
 ```python

{gnosisllm_knowledge-0.2.0 → gnosisllm_knowledge-0.4.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "gnosisllm-knowledge"
-version = "0.2.0"
+version = "0.4.0"
 description = "Enterprise-grade knowledge loading, indexing, and search for Python"
 authors = [
     {name = "David Marsa", email = "david.marsa@neomanex.com"},

{gnosisllm_knowledge-0.2.0 → gnosisllm_knowledge-0.4.0}/src/gnosisllm_knowledge/__init__.py

@@ -38,7 +38,7 @@ Features:
     - SOLID principles throughout
 """
 
-from gnosisllm_knowledge.api import Knowledge
+from gnosisllm_knowledge.api import Knowledge, Memory
 from gnosisllm_knowledge.backends import (
     AgenticSearchFallback,
     MemoryIndexer,
@@ -51,6 +51,20 @@ from gnosisllm_knowledge.backends import (
 )
 from gnosisllm_knowledge.chunking import FixedSizeChunker, SentenceChunker
 from gnosisllm_knowledge.core.domain.document import Document, DocumentStatus, TextChunk
+from gnosisllm_knowledge.core.domain.memory import (
+    ContainerConfig,
+    ContainerInfo,
+    HistoryEntry,
+    MemoryEntry,
+    MemoryStats,
+    MemoryStrategy,
+    MemoryType,
+    Message,
+    Namespace,
+    RecallResult,
+    SessionInfo,
+    StrategyConfig,
+)
 from gnosisllm_knowledge.core.domain.result import (
     BatchResult,
     IndexResult,
@@ -58,9 +72,9 @@ from gnosisllm_knowledge.core.domain.result import (
     ValidationResult,
 )
 from gnosisllm_knowledge.core.domain.search import (
-    AgentType,
     AgenticSearchQuery,
     AgenticSearchResult,
+    AgentType,
     ReasoningStep,
     SearchMode,
     SearchQuery,
@@ -72,15 +86,27 @@ from gnosisllm_knowledge.core.exceptions import (
     AgenticSearchError,
     ConfigurationError,
     ConnectionError,
+    ContainerExistsError,
+    ContainerNotFoundError,
     IndexError,
+    InferenceError,
+    InferenceTimeoutError,
     KnowledgeError,
     LoadError,
+    MemoryConfigurationError,
+    MemoryError,
     SearchError,
+    SessionNotFoundError,
+)
+from gnosisllm_knowledge.core.streaming import (
+    BatchCollector,
+    BoundedQueue,
+    PipelineConfig,
 )
 from gnosisllm_knowledge.fetchers import (
     HTTPContentFetcher,
-    NeoreaderContentFetcher,
     NeoreaderConfig,
+    NeoreaderContentFetcher,
 )
 from gnosisllm_knowledge.loaders import (
     LoaderFactory,
@@ -95,58 +121,84 @@ from gnosisllm_knowledge.services import (
 __version__ = "0.2.0"
 
 __all__ = [
-    # Main API
-    "Knowledge",
-    # Domain Models
-    "Document",
-    "DocumentStatus",
-    "TextChunk",
-    "SearchQuery",
-    "SearchResult",
-    "SearchResultItem",
-    "SearchMode",
     "AgentType",
+    "AgenticSearchError",
+    "AgenticSearchFallback",
     "AgenticSearchQuery",
     "AgenticSearchResult",
-    "ReasoningStep",
-    "LoadResult",
-    "IndexResult",
+    "BatchCollector",
     "BatchResult",
-    "ValidationResult",
+    "BoundedQueue",
+    "ConfigurationError",
+    "ConnectionError",
+    "ContainerConfig",
+    "ContainerExistsError",
+    "ContainerInfo",
+    "ContainerNotFoundError",
+    # Domain Models
+    "Document",
+    "DocumentStatus",
     # Events
     "Event",
-    "EventType",
     "EventEmitter",
+    "EventType",
+    "FixedSizeChunker",
+    # Fetchers
+    "HTTPContentFetcher",
+    "HistoryEntry",
+    "IndexError",
+    "IndexResult",
+    "InferenceError",
+    "InferenceTimeoutError",
+    # Main API
+    "Knowledge",
     # Exceptions
     "KnowledgeError",
-    "ConfigurationError",
-    "ConnectionError",
+    # Services
+    "KnowledgeIndexingService",
+    "KnowledgeSearchService",
     "LoadError",
-    "IndexError",
-    "SearchError",
-    "AgenticSearchError",
+    "LoadResult",
     # Loaders
     "LoaderFactory",
-    "WebsiteLoader",
-    "SitemapLoader",
-    # Fetchers
-    "HTTPContentFetcher",
-    "NeoreaderContentFetcher",
+    "Memory",
+    "MemoryConfigurationError",
+    "MemoryEntry",
+    # Memory Exceptions
+    "MemoryError",
+    # Memory Backend (for testing)
+    "MemoryIndexer",
+    "MemorySearcher",
+    "MemoryStats",
+    # Memory Domain Models
+    "MemoryStrategy",
+    "MemoryType",
+    "Message",
+    "Namespace",
     "NeoreaderConfig",
-    # Chunkers
-    "SentenceChunker",
-    "FixedSizeChunker",
+    "NeoreaderContentFetcher",
+    "OpenSearchAgenticSearcher",
     # OpenSearch Backend
     "OpenSearchConfig",
     "OpenSearchIndexer",
     "OpenSearchKnowledgeSearcher",
     "OpenSearchSetupAdapter",
-    "OpenSearchAgenticSearcher",
-    "AgenticSearchFallback",
-    # Memory Backend (for testing)
-    "MemoryIndexer",
-    "MemorySearcher",
-    # Services
-    "KnowledgeIndexingService",
-    "KnowledgeSearchService",
+    # Streaming Pipeline
+    "PipelineConfig",
+    "ReasoningStep",
+    "RecallResult",
+    "SearchError",
+    "SearchMode",
+    "SearchQuery",
+    "SearchResult",
+    "SearchResultItem",
+    # Chunkers
+    "SentenceChunker",
+    "SessionInfo",
+    "SessionNotFoundError",
+    "SitemapLoader",
+    "StrategyConfig",
+    "TextChunk",
+    "ValidationResult",
+    "WebsiteLoader",
 ]

gnosisllm_knowledge-0.4.0/src/gnosisllm_knowledge/api/__init__.py

@@ -0,0 +1,6 @@
+"""High-level API for knowledge and memory operations."""
+
+from gnosisllm_knowledge.api.knowledge import Knowledge
+from gnosisllm_knowledge.api.memory import Memory
+
+__all__ = ["Knowledge", "Memory"]