Execute TTL Cache Operations with Fetch-on-Miss

Implement a pure simulator for a TTL cache. The simulator includes put, get, delete, size, and a deterministic adapter for a loader-based fetch operation.

def execute_ttl_cache(operations: list[list[object]]) -> list[list[object]]:
    ...

Operation formats are:

• ["put", time_ms, key, value, ttl_ms]
• ["get", time_ms, key]
• ["delete", time_ms, key]
• ["size", time_ms]
• ["fetch", time_ms, key, ttl_ms, loader_status, loader_payload]

Keys are strings and cached values are integers. Times are nonnegative integers in nondecreasing order. A positive TTL expires at put_time + ttl_ms, and the key is invalid at that exact time. TTL zero means no expiration.

The exact result for each operation is:

• put: ["ok"]
• get: ["ok", value] for a valid entry, otherwise ["ok", null]
• delete: ["ok", boolean], where the boolean is true when a stored key existed, even if it was expired
• size: ["ok", count], including stored expired entries
• fetch: return a valid cached value without consulting the supplied loader fields; on a miss or expired entry, loader_status is either "ok" or "error". For "ok", replace the entry with loader_payload and the supplied TTL, then return ["ok", value]. For "error", do not modify the cache and return ["error", reason].

Expired entries remain stored until delete or replacement. This makes the specified size behavior observable. The fetch operation represents the outcome of calling a loader rather than accepting a function object, so every input remains JSON-marshalable and deterministic.

Constraints and Error Rules

• 0 <= len(operations) <= 200000
• ttl_ms is nonnegative.
• loader_payload is an integer when loader_status is "ok" and a string reason when it is "error".
• Malformed operations, decreasing timestamps, invalid types, or unknown names must raise ValueError.
• Return values are compared exactly.

Example

Input:
[
  ["put", 0, "u", 1, 5],
  ["get", 4, "u"],
  ["get", 5, "u"],
  ["size", 5],
  ["fetch", 6, "u", 10, "ok", 2],
  ["fetch", 7, "v", 0, "error", "db_down"],
  ["size", 7],
  ["delete", 8, "u"],
  ["size", 8]
]

Output:
[
  ["ok"],
  ["ok", 1],
  ["ok", null],
  ["ok", 1],
  ["ok", 2],
  ["error", "db_down"],
  ["ok", 1],
  ["ok", true],
  ["ok", 0]
]

Hints

• Keep the stored record distinct from the result of checking whether it is currently valid.
• Fetch has a check phase and, only on a miss, a loader-outcome phase.
• Replacing an expired key should update all of its metadata.

Execute TTL Cache Operations with Fetch-on-Miss

Hints

1. Store existence separately from current validity.
2. Fetch consults loader fields only when the current record is missing or expired.

Approach

Each stored record keeps a value and optional absolute expiry. Get checks validity without deleting. Delete and size inspect storage, while fetch returns a valid value or applies the supplied loader outcome on a miss.

• Time complexity: O(1) expected per operation
• Space complexity: O(number of stored keys)