OSIRIS
Object Storage, Indexing & Retrieval Intelligent System
OSIRIS is a Rust-native object storage and retrieval layer built on top of HeroDB, providing structured storage with metadata, field indexing, and search capabilities.
Features
- Object Storage: Store structured objects with metadata (title, tags, MIME type, timestamps)
- Namespace Management: Organize objects into isolated namespaces
- Field Indexing: Fast filtering by tags and metadata fields
- Text Search: Simple keyword-based search across object content
- CLI Interface: Command-line tools for object management and search
- 9P Filesystem: Mount OSIRIS as a filesystem (future)
Quick Start
Prerequisites
Start HeroDB:
cd /path/to/herodb
cargo run --release -- --dir ./data --admin-secret mysecret --port 6379
Installation
cd /path/to/osiris
cargo build --release
Initialize
# Create configuration
mkdir -p ~/.config/osiris
cat > ~/.config/osiris/config.toml <<EOF
[herodb]
url = "redis://localhost:6379"
[namespaces.notes]
db_id = 1
EOF
# Initialize OSIRIS
./target/release/osiris init --herodb redis://localhost:6379
# Create a namespace
./target/release/osiris ns create notes
Usage
# Add objects
./target/release/osiris put notes/my-note.md ./my-note.md --tags topic=rust,project=osiris
# Get objects
./target/release/osiris get notes/my-note.md
# Search
./target/release/osiris find --ns notes --filter topic=rust
./target/release/osiris find "retrieval" --ns notes
# Delete objects
./target/release/osiris del notes/my-note.md
# Show statistics
./target/release/osiris stats --ns notes
Architecture
HeroDB (unmodified)
│
├── KV store + encryption
└── RESP protocol
↑
│
└── OSIRIS
├── store/ – object schema + persistence
├── index/ – field index & keyword scanning
├── retrieve/ – query planner + filtering
├── interfaces/ – CLI, 9P
└── config/ – namespaces + settings
Data Model
Objects are stored with metadata:
- ID: Unique identifier (UUID or user-assigned)
- Namespace: Logical grouping (e.g., "notes", "calendar")
- Title: Optional human-readable title
- MIME Type: Content type
- Tags: Key-value pairs for categorization
- Timestamps: Created and updated times
- Text Content: Optional plain text content
Keyspace Design
meta:<id> → serialized OsirisObject
field:<field>:<val> → Set of IDs (for equality filtering)
scan:index → list of IDs for text scan
Example:
field:tag:project=osiris → {note_1, note_2}
field:mime:text/markdown → {note_1, note_3}
Future Enhancements
- Content-addressable deduplication
- Vector embeddings for semantic search
- Relation graphs
- Full-text search with Tantivy
- 9P filesystem interface
License
See LICENSE file.
Description
Languages
Rust
97.8%
V
2.2%