Configurable Expense Reimbursement Rule Evaluator

Problem: Configurable Expense Reimbursement Rule Evaluator Design and implement a rule evaluator for reviewing company reimbursement requests. The system receives an expense request, expense, and a list of JSON-configured review rules, rules. It must evaluate the request against those rules and return the decision, matched rule, and explanation. Requirements Rules must be represented as JSON. Adding or changing a rule must not require adding or modifying a Python/JavaScript rule class. New, updated, or removed rules must be possible through configuration changes only; the evaluator's core logic must remain unchanged. Support the following comparison operators: eq, ne gt, gte, lt, lte in Support logical composition: all: every child condition matches any: at least one child condition matches not: the child condition does not match Each rule contains: id priority (lower values run first) when condition expression action: approve, reject, or manual_review reason Evaluate rules in priority order. Return when the first terminal matching rule (approve or reject) is found. Return manual_review if no rule matches. Expense fields may be nested, such as employee.department. Conditions must support dot-path lookup. Example Expense: { "amount": 180, "currency": "USD", "category": "meal", "employee": { "department": "Engineering", "level": "IC" } } Rules: [ { "id": "reject-large-meal", "priority": 1, "when": { "all": [ {"field": "category", "op": "eq", "value": "meal"}, {"field": "amount", "op": "gt", "value": 200} ] }, "action": "reject", "reason": "Meal expenses above $200 are not reimbursable." }, { "id": "approve-engineering-small-expense", "priority": 2, "when": { "all": [ {"field": "employee.department", "op": "eq", "value": "Engineering"}, {"field": "amount", "op": "lte", "value": 500} ] }, "action": "approve", "reason": "Pre-approved Engineering expense." } ] Expected result: { "action": "approve", "matched_rule_id": "approve-engineering-small-expense", "reason": "Pre-approved Engineering expense." } Describe the JSON rule schema, the evaluation engine, invalid-rule handling, and how you would extend the system with new operators or action types. Example Input expense = {"amount": 180, "category": "meal", "employee": {"department": "Engineering"}}; rules = [{"id":"reject-large-meal","priority":1,"when":{"all":[{"field":"category","op":"eq","value":"meal"},{"field":"amount","op":"gt","value":200}]},"action":"reject","reason":"Too large"},{"id":"approve-eng","priority":2,"when":{"field":"employee.department","op":"eq","value":"Engineering"},"action":"approve","reason":"Engineering policy"}] Output {"action":"approve","matched_rule_id":"approve-eng","reason":"Engineering policy"}