From AST to rows: the query engine·14 min

Execute a single hop

0 / 1

# grade this stage (from your course directory)
npx theinference test --stage execute-single-hop --watch

Run a one-segment MATCH end to end: anchor, extend by one relationship, then project RETURN. Pick the anchor node pattern (the start node, which carries the label/property here), seed bindings via the indexes, then for the single segment call neighborEdges filtered by the segment's type and direction — keeping each extended binding whose neighbor also satisfies the target node pattern.

Your task

Implement matchPath + a minimal runQuery

Implement matchPath(store, path) for a one-segment pattern, then enough of runQuery to project RETURN m.title AS title. The fixture runs Kevin Bacon's ACTED_IN OUT movies — three titles.

src/executor.ts

// seed: candidates = findNodes(label, key, value) when label+prop present.
// extend: for each binding b, for each {rel, node} of neighborEdges(cur, type, dir),
//         if nodeMatches(node, targetPat) -> new binding {..b.vars, target: node}.
// project: resolve each ReturnItem (variable / variable.prop) to a value.

Process command

your_program.sh query {fixture}/fixtures/data/nodes.json {fixture}/fixtures/data/relationships.json "MATCH (bacon:Person {name:'Kevin Bacon'})-[:ACTED_IN]->(m:Movie) RETURN m.title AS title"
# -> [{"title":"A Few Good Men"},{"title":"Apollo 13"},{"title":"Frost/Nixon"}]

How it's graded

Process mode, stdout_json_match. No ORDER BY, so the rows are graded as a canonical set — sort them by their stable key before printing (the given cli.ts does this for you when there's no ORDER BY). All three movies must appear, no more.

Bindings, the execution model WHERE filter + RETURN projection