How arpan-vs/rust-lstar works
rust-lstar is a Rust library for active DFA/Mealy-machine learning. A caller supplies an input alphabet and a knowledge base for the system under learning; LSTAR builds and updates an observation table using membership queries, constructs a hypothesis automaton, and uses an equivalence-test strategy to search for counterexamples. The library also provides cached, fake, and TCP-backed knowledge bases, several equivalence strategies, automaton playback and DOT import/export. Edges are grounded in the README and shown excerpts or SOURCE INDEX references; unsampled details are not inferred beyond those references.
Main entry points: base.rs Knowledge base · automata.rs Learned automaton · mod.rs Equivalence interface · observation_table.rs Observation table
Components 22
Learning
- LSTAR learner
src/lstar.rs - Observation table
src/observation_table.rs - Membership query
src/query.rs
Equivalence testing
- Equivalence interface
src/eqtest/mod.rs - BDist strategy
src/eqtest/bdist_method.rs - Random walk
src/eqtest/random_walk.rs - W-method
src/eqtest/w_method.rs - Composite tests
src/eqtest/multiple_eqtests.rs
Knowledge bases
- Knowledge base
src/knowledge_base/base.rs - Active query API
src/knowledge_base/active.rs - Fake knowledge base
src/knowledge_base/fake.rs - TCP knowledge base
src/knowledge_base/network.rs - Query cache tree
src/knowledge_base/tree.rs - Query statistics
src/knowledge_base/stats.rs
Automata and symbols
- Learned automaton
src/automata/automata.rs - DOT import/export
src/automata/dot_parser.rs - Automaton states
src/automata/state.rs - Transitions
src/automata/transition.rs - Alphabet letters
src/letter.rs - Input/output words
src/word.rs
Other components
- Library caller
- System under learning
Connections 26
- Library caller → LSTAR learner: starts learning
README.md - Library caller → Knowledge base: provides knowledge base
README.md - LSTAR learner → Observation table: coordinates learning
src/lstar.rs - LSTAR learner → Equivalence interface: tests hypothesis
src/lstar.rs - Observation table → Membership query: creates queries
src/observation_table.rs - Observation table → Knowledge base: resolves queries
src/observation_table.rs - Observation table → Learned automaton: builds hypothesis
src/observation_table.rs - Equivalence interface → BDist strategy: exposes strategy
src/eqtest/mod.rs - Equivalence interface → Random walk: exposes strategy
src/eqtest/mod.rs - Equivalence interface → W-method: exposes strategy
src/eqtest/mod.rs - Equivalence interface → Composite tests: exposes composition
src/eqtest/mod.rs - BDist strategy → Knowledge base: queries system
src/eqtest/bdist_method.rs - Random walk → Knowledge base: queries system
src/eqtest/random_walk.rs - W-method → Knowledge base: queries system
src/eqtest/w_method.rs - Knowledge base → Query cache tree: uses query cache
src/knowledge_base/base.rs - Knowledge base → Query statistics: tracks statistics
src/knowledge_base/base.rs - Fake knowledge base → Knowledge base: implements interface
src/knowledge_base/fake.rs - Fake knowledge base → Learned automaton: uses automaton
src/knowledge_base/fake.rs - TCP knowledge base → Active query API: implements active queries
src/knowledge_base/network.rs - TCP knowledge base → System under learning: sends TCP queries
- Learned automaton → Automaton states: uses states
src/automata/automata.rs - Learned automaton → Transitions: uses transitions
src/automata/automata.rs - Learned automaton → DOT import/export: serializes and parses
src/automata/automata.rs - DOT import/export → Automaton states: builds states
src/automata/dot_parser.rs - DOT import/export → Transitions: builds transitions
src/automata/dot_parser.rs - Learned automaton → Membership query: plays input words
src/automata/automata.rs