feat: implement KEYS command

[ithustle]

Summary

• Add KEYS command (CMD 0x0D) that lists all non-expired keys in numbered format (1) key1\n2) key2\n)
• Dedicated send_keys_reply() tags the response frame with CMD_KEYS so the client correctly handles empty lists ((empty list)) vs key listings (raw text, no quotes)
• Server handler iterates hash table with safe next pointer caching, lazy expiration via check_and_expire(), and dynamic buffer with realloc growth (max ~65500 bytes)
• 4 integration tests: empty store, multiple keys, expired key exclusion, key after DEL

Files changed (11)

• command_defs.h — #define CMD_KEYS 0x0D
• command_parser.h/c — construct_keys_command()
• command_registry.h/c — send_keys_reply()
• server_command_handlers.h/c — handle_keys_command() + registration
• client_command_handlers.h/c — cmd_keys(), command_table[], cmd_unknown() filter, CMD_KEYS response branch
• README.md — KEYS added to Server commands table
• test_integration.c — 4 new tests

Test plan

• make -f Makefile.fkvs build — compiles without new warnings
• ctest --output-on-failure — all 3 test suites pass (39 tests total)

Test project /Users/ithustle/dev/server/fkvs
Start 1: CounterTest
1/3 Test #1: CounterTest ...................... Passed 0.00 sec
Start 2: StringUtilsTest
2/3 Test #2: StringUtilsTest .................. Passed 0.00 sec
Start 3: IntegrationTest
3/3 Test #3: IntegrationTest .................. Passed 18.77 sec

100% tests passed, 0 tests failed out of 3

Total Test time (real) = 18.77 sec

[AlexJuca]

@ithustle I will have a look at these during this week.

[AlexJuca]

@ithustle Did you test this implementation?

[ithustle]

@ithustle Did you test this implementation?

@AlexJuca yes I did. In fact, I’m using this feature on fkvs locally

[AlexJuca]

@ithustle Did you test this implementation?

@AlexJuca yes I did. In fact, I’m using this feature on fkvs locally

Cool, I found and fixed a few issues while reviewing the PR.

• The branch was out of date with main, which made the PR unmergeable.
• KEYS accepted malformed frames instead of validating the exact command shape.
• The CLI treated any command starting with KEYS as valid, for example KEYSPACE.
• Large KEYS responses could silently truncate instead of returning an error.
• Sanitizers found a use-after-free when KEYS expired keys while iterating the store.

These are the fixes I applied:

• Merged the latest main.
• Added strict server-side validation for KEYS frames.
• Tightened CLI command matching so only KEYS is accepted.
• Changed oversized KEYS responses to return an error instead of partial output.
• Fixed expiry deletion order to avoid using a freed key pointer.
• Added regression tests for malformed KEYS frames and oversized output.

[AlexJuca]

@ithustle I've added several improvements to harden the implement. Thanks a bunch for the preliminary implementation. 💪🏾

Merging.