Commit Graph
44 Commits
Author SHA1 Message Date
KoalaTrapLord 2af1d58104 fix: write ML JSONL to commander-ai-lab/results/ (not Forge CWD)
LabCli passed relative 'results' to enableMlLogging(), which resolved
to Forge-Repo/forge-gui/results/ because the Java subprocess CWD is
forgeWorkDir. Now uses outputPath's parent directory (absolute) so
JSONL training data goes to commander-ai-lab/results/ where the
Python ML pipeline and Training page expect it.
2026-04-05 17:29:21 +00:00
KoalaTrapLord 547a84d8c2 fix: DecisionExtractor regex patterns to match actual Forge verbose output
Forge verbose log uses different verb forms and prefixes than the
patterns expected:
  - Turn: 'Turn: Turn 5 (Ai(1)' not 'Turn 5 (Ai(1)'
  - Cast: 'cast' (no s) not 'casts', 'Add To Stack:' prefix, no period
  - Land: 'played' not 'plays', 'Land:' prefix, card ID suffix '(76)'
  - Draw: 'drew' variant, card ID suffix

Updated all four patterns to match both old and new Forge output
formats. This is the final fix for the training data pipeline —
DecisionExtractor will now find cast/land/draw/attack events in
the Forge game log and emit JSONL training snapshots.
2026-04-05 17:24:34 +00:00
KoalaTrapLord 71b86b74eb fix: compile error in BatchRunner deck name log line 2026-04-05 16:55:38 +00:00
KoalaTrapLord d00093897a fix: read Name= from .dck header for Forge deck lookup
Forge's sim mode looks up decks by the Name= field inside the .dck
file metadata header, NOT by filename. For example:
  Filename: Counter_Intelligence.dck
  Header:   Name=Counter Intelligence
  Forge expects: 'Counter Intelligence' (with space)

BatchRunner now reads the Name= field from the .dck file and passes
that to Forge. Falls back to filename with underscores→spaces if
the Name= field can't be read.
2026-04-05 16:54:38 +00:00
KoalaTrapLord 0b01f1c467 fix: pass bare deck names to Forge sim (not full paths)
Forge's sim mode looks up decks by NAME in its own
%APPDATA%/Forge/decks/commander/ directory. Passing a full path like
D:\...\precon-decks\Counterpunch.dck causes 'Could not load deck'.

Now strips the path and .dck extension so Forge sees just the bare
name (e.g. 'Counterpunch') and finds it in its deck folder.

Precon .dck files must be copied to %APPDATA%/Forge/decks/commander/
for Forge to find them.
2026-04-05 16:47:54 +00:00
KoalaTrapLord 8995d743af fix: remove !quiet guard from ML decision logging in BatchRunner
The condition 'mlLoggingEnabled && !quiet && mlDecisionLogger != null'
silently skipped ML training data extraction whenever quiet mode was
active. DecisionExtractor works with the same log lines regardless of
quiet mode, so the guard was unnecessary and prevented JSONL generation.

Now: mlLoggingEnabled && mlDecisionLogger != null — ML logging runs in
all modes as long as --ml-log is passed.
2026-04-05 16:19:56 +00:00
KoalaTrapLord 14ff543135 fix: DecisionLogger JSONL filename prefix — ml-decisions-sim-* for dataset_builder discovery
The Java DecisionLogger was writing files as ml-decisions-{batchId}.jsonl
but the Python dataset_builder.py scans for ml-decisions-sim-*.jsonl to
categorize them as 'forge' source training data.

This one-line fix adds the 'sim-' prefix so the Forge batch heuristic
path now feeds directly into the ML training pipeline:

  BatchRunner → DecisionExtractor → DecisionLogger
    → results/ml-decisions-sim-{batchId}.jsonl
    → dataset_builder.py discovers as 'forge' source
    → encode → train

The full chain was already wired (Issue #83 Phase 1 infrastructure),
this naming mismatch was the only thing preventing it from working.

Refs: Issue #83 Phase 1
2026-04-05 16:04:12 +00:00
KoalaTrapLord a23fdc55a4 feat: PolicyControllerAi integration — Forge AI decision interception via policy server
## New File
- PolicyControllerAi.java (840 lines): Extends Forge's PlayerControllerAi
  to intercept AI decisions at chooseSpellAbilityToPlay() and route them
  through the Python policy server. Falls back to Forge built-in AI on
  any failure. Inner PolicyAiController overrides AiController with full
  legal action gathering via ComputerUtilAbility.

## Wiring
- GameSession.java: Import + Javadoc for PolicyControllerAi decision flow.
  Logs decision pipeline when policy server connects. End-of-game reward
  submission via PolicyControllerAi lifecycle.
- BatchRunner.java: New setPolicyServer(url) method. Appends
  --policy midrange --policyServer <url> to Forge subprocess command when
  policy server URL is configured.

## Path Updates (Forge-Repo rename)
- README.md: Forge link → KoalaTrapLord/Forge-Repo, build path updated
- .env.example: FORGE_JAR_PATH and FORGE_WORKING_DIR → Forge-Repo
- lab_api.py: Auto-detect paths try Forge-Repo first, forge-repo as legacy

## MacroActionExecutor
- Added public isRemovalCard(), isDrawCard(), isRampCard() wrappers so
  PolicyControllerAi can consult the same card-role databases.

Refs: Issue #83 Phase 2.4
2026-04-05 14:47:15 +00:00
rgriffithandClaude Opus 4.6 d308e8f413 fix: AI land-only bug — enforce 1-land/turn, populate legal moves, tune policy
Addresses the three root causes in commander-ai-lab-land-only-bug-analysis.md
plus several collateral fixes discovered along the way.

P0 — Land-per-turn enforcement (engine.py, deepseek_engine.py, turn_manager.py)
  * Guard _play_land() against double plays by tracking
    _lands_played_this_turn vs _land_plays_this_turn_limit on each player
  * Reset per-turn counters at the start of each player's turn
  * Ramp cards that grant additional plays can bump the limit

P0 — Legal move pipeline (new legal_move_generator.py, game_state.py)
  * New generate_legal_moves() emits the full macro-action set
    (play_land, cast_creature, cast_removal, cast_ramp, cast_draw,
    cast_commander, attack, block, pass_priority) based on phase,
    active player, land-drop state, mana, and commander tax
  * get_legal_moves() delegates to the new generator instead of the
    pass-only stub that blocked the turn_manager/policy-server path
  * apply_move() gains handlers for play_land, cast_spell,
    cast_commander, and declare_attacker — spells resolve to
    battlefield or graveyard, stats update correctly

P1 — Policy exploration (LearnedPolicy.java, scope.py)
  * Default greedy=false so the neural policy samples its
    distribution instead of always taking argmax
  * Raise ppo_entropy_coeff 0.01 → 0.04 to keep the action
    distribution flat during PPO self-play and prevent the
    pass/land collapse

Collateral fixes pulled in during the same session:
  * lan-client/precon-loader.js: await enrichAllZonesWithScryfall()
    so opening-hand cards have typeLine/cmc before the AI's first turn
  * lan-client/ui.js: closeZoneModal auto-shuffles the library
    whenever the zone modal closes
  * routes/game.py: BatchState() was missing total_games/threads,
    crashing POST /api/game/start
  * routes/deckgen.py: dead code after raise left headers undefined,
    breaking the Anthropic API path

All sim/phase/integration tests pass (650+).

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-04-05 10:34:37 -04:00
KoalaTrapLord 74d24d606c fix: wire PolicyClient into real Forge IPC path via Priority interception
Forge's sim mode does not honour --policy/--policyServer CLI flags, so
the Python brain was never called during live games. AI only played lands
because Forge ran its built-in AI for every decision.

Fix: when parseForgeLine() detects a Priority event for an AI seat, the
session now immediately calls policyClient.decide() with the current
state snapshot and POSTs the result to /api/policy/override before Forge
reads the next line. This mirrors what handleHumanTurn() already does
for seat 0. The synthetic fallback path is unchanged.
2026-04-03 17:42:01 -04:00
KoalaTrapLord d98e6331e9 fix: point all deck lookups to precon-decks directory
- start-lab.bat: Set FORGE_DECKS_DIR to D:\ForgeCommander\commander-ai-lab\precon-decks
- BatchRunner.java: resolveForgeDecksPath() now checks PRECON_DECKS_DIR env var
  first, then falls back to the hardcoded precon-decks path, before trying
  %APPDATA%\Forge\decks\commander. Also pass full absolute path to Forge -d flag.
- .env.example: document PRECON_DECKS_DIR variable
2026-04-01 23:10:32 -04:00
KoalaTrapLord 126684bb94 feat(bridge): Phase 2.3 — wire GameSession to real Forge IPC subprocess (issue #83)
Replaces synthetic stub loop with a live Forge process:
- Spawns Forge with --policy + --policyServer flags so Forge calls
  back to /api/policy/decide at each AI decision point
- Reads Forge stdout line-by-line and parses zone/phase/life updates
  via the same regex patterns used in BatchRunner
- Syncs full board state (battlefield, hand, graveyard, exile,
  command zone, stack) on every Forge state-change line
- Routes human input from humanActionQueue via a side-channel
  HTTP POST to /api/policy/override on the policy server
- Handles activePlayer + priorityPlayer correctly for 3-4 player pods
- Falls back to synthetic stub loop if Forge subprocess fails to start
- Watchdog thread kills stalled Forge process after 10 min
2026-03-29 04:31:49 -04:00
KoalaTrapLord da8de77704 feat(bridge): add buildStateVector() emitting state_vector float[] in snapshots (#83 Step 1)
Adds a buildStateVector(int seat) method to GameSession that produces
a float[29] global scalar vector matching STATE_DIMS layout in
ml/config/scope.py:

  Per-player block (14 features × 2 players = 28):
    [0]  life_total           / 40.0
    [1]  commander_damage     / 21.0  (stub 0.0 until real Forge tracks it)
    [2]  mana_available       / 20.0  (stub 0.0 until mana pool wired)
    [3]  commander_tax        / 10.0
    [4]  cards_in_hand        / 15.0
    [5]  cards_in_graveyard   / 100.0
    [6]  creatures_on_battlefield / 30.0
    [7]  total_power_on_board / 100.0  (stub = creature count until P/T data)
    [8]  is_active_player     (0 or 1)
    [9]  phase one-hot[0]  main1
    [10] phase one-hot[1]  combat
    [11] phase one-hot[2]  main2
    [12] phase one-hot[3]  end
    [13] (padding / reserved)

  Global:
    [28] turn_number / max_turns (25)

buildStateSnapshot() now includes "state_vector": float[] so that
PolicyClient.decide(JsonObject) sends a machine-readable vector the
Python /api/policy/decide endpoint can pass directly to the encoder
without re-parsing the nested human-readable fields.

Stub fields (mana_available, commander_damage, total_power_on_board)
are zero until real Forge board state is wired in Step 4.
2026-03-29 04:24:55 -04:00
KoalaTrapLord f851df6dbf fix: add getBaseUrl(), PolicyDecision record, and decide(String,int) to PolicyClient
GameSession.java references these APIs which were missing from
PolicyClient, causing 3 compilation errors. Added:
- getBaseUrl() getter
- PolicyDecision record (action, confidence)
- decide(String, int) overload that delegates to decide(JsonObject)
2026-03-24 22:46:45 -04:00
KoalaTrapLord b2d1078ca9 fix: disable quiet mode when ML logging is enabled in createWorker
DecisionExtractor requires verbose Forge output to reconstruct board
states. When quiet=true, Forge runs with -q which strips the verbose
log lines, causing 0 decisions to be extracted. This change passes
quiet=false to BatchRunner when mlLogEnabled is true.
2026-03-24 18:45:36 -04:00
KoalaTrapLord f51ff61568 feat(83): Phase 3 — Performance & Scale (#83) (#101)
* feat(83): Phase 3.1 — persistent JVM worker pool in MultiThreadBatchRunner (#83)

* feat(83): Phase 3.2 — async Forge episode generator with producer-consumer pipeline (#83)

* feat(83): Phase 3.3 — deprecate self_play.py synthetic mode for training (#83)

Mark self_play.py synthetic training as deprecated per Phase 3.3:
- Add deprecation notice to module docstring
- Add import warnings and _TRAINING_DEPRECATION_MSG constant
- Add DeprecationWarning in collect_rollouts()
- Synthetic sim retained for local testing/unit tests only
- scope.py already defaults forge_weight=1.0, ppo_weight=0.0

* feat(83): Phase 3.3 — update ml/training/__init__.py exports (#83)
2026-03-24 10:43:51 -04:00
KoalaTrapLord 39bc36afef rename: update JsonExporter.java javadoc example from batch- to ml-decision- 2026-03-23 23:10:44 -04:00
KoalaTrapLord dffd89a01c rename: update LabCli.java default output from batch-latest to ml-decision-latest 2026-03-23 23:09:51 -04:00
KoalaTrapLord 411fa41063 feat(83): Phase 2 — Live Forge-to-Policy-Server IPC (#100)
* feat(83): add PolicyClient for live Forge-to-policy-server IPC (#83)

Phase 2.1: Bridge between Forge decision points and Python policy
server for online training. Calls /api/policy/decide endpoint,
collects (state, action, reward) tuples in real-time, and flushes
them to the server for online PPO updates.

Falls back to Forge built-in AI if server is unavailable.

* feat(83): add /api/policy/* routes for live Forge IPC (#83)

Phase 2.2: Policy decision routes for real-time Forge IPC.

- POST /api/policy/decide — full game state inference with log_prob/value
- POST /api/policy/collect — online learning tuple batch collection
- POST /api/policy/reward — end-of-game reward propagation
- GET /api/policy/health — readiness check for Java PolicyClient
- GET /api/policy/stats — session statistics

Includes Forge phase mapping and PlayerZoneState with full zone contents.

* feat(83): integrate PolicyClient into GameSession for bidirectional sync (#83)
2026-03-23 22:08:37 -04:00
KoalaTrapLord 7e0aba78ba fix: add ML logging support to MultiThreadBatchRunner 2026-03-23 19:01:43 -04:00
KoalaTrapLord 7036c9568f fix: enable ML decision logging in multi-thread batch path 2026-03-23 18:58:14 -04:00
KoalaTrapLord b72d5fba83 fix(#92): bump READ_TIMEOUT_MS from 5s to 300s in PolicyServerClient.java 2026-03-22 23:44:47 -04:00
KoalaTrapLord f8b13703b8 fix: CAST_PATTERN now matches both 'cast' and 'casts' for Forge compatibility 2026-03-20 15:04:37 -04:00
KoalaTrapLord ff6040b53e Remove stray closing brace that would break compilation 2026-03-20 13:31:50 -04:00
KoalaTrapLord db7bc62d7e Fix commander damage tracking, add PLAYER_DAMAGE_SOURCE_PATTERN and diagnostic logging 2026-03-20 13:28:54 -04:00
KoalaTrapLord fbffe81846 feat: auto-detect commander card names from .dck files for commander cast tracking 2026-03-20 12:48:48 -04:00
KoalaTrapLord a2fbc30b2b fix: resolve forge_dir to forge-gui sibling when res/ not found in forge-gui-desktop 2026-03-20 11:22:44 -04:00
KoalaTrapLord 93a421b90a fix: make colons optional in all regex patterns to match Forge output format (no colons after Game Outcome, Life, Add To Stack, Land, Zone Change, etc.); fix VERBOSE_TURN_PATTERN for 'Turn Turn N AiX' format; fix LIFE_CHANGE_PATTERN to use space instead of > separator 2026-03-20 09:26:02 -04:00
KoalaTrapLord ee95f26bc4 fix: restore double-backslash escaping in Ai regex patterns (fixes compilation errors) 2026-03-20 09:03:55 -04:00
KoalaTrapLord 2ee574f678 diag: add [COMBAT-LINE] logging to identify Forge verbose output format for combat stats parsing 2026-03-20 08:15:03 -04:00
KoalaTrapLord 473a2bdcc7 fix: make Ai regex patterns support both Ai(N) and AiN formats for Forge output compatibility 2026-03-19 23:29:53 -04:00
KoalaTrapLord a6e50ee0f0 feat: aggregate combat stats in StatsAggregator 2026-03-19 22:45:05 -04:00
KoalaTrapLord e2a6a75fd9 feat: add combat stat fields to DeckSummary schema 2026-03-19 22:39:51 -04:00
KoalaTrapLord 030bdf2b81 fix: CAST_PATTERN and LAND_PLAY_PATTERN regex to match Forge verbose log prefixes
- CAST_PATTERN: Add optional (?:Add To Stack:\s*)? prefix to match Forge's verbose log format
- LAND_PLAY_PATTERN: Add optional (?:Land:\s*)? prefix to match Forge's verbose log format
- Add [PARSE] diagnostic logging after combat stats parsing to confirm spells/lands/cmdr stats per seat

Root cause: Forge verbose output prefixes cast lines with "Add To Stack: " and land lines with "Land: "
but the regexes expected Ai(N) at the start of the line, causing 0 matches for all combat stats.
2026-03-19 22:06:17 -04:00
KoalaTrapLord 4e8a999ebd feat: add Unity/Godot WebSocket server (Phase 1 stub)
- pom.xml: add Java-WebSocket 1.5.6 dependency
- bridge/GameWsServer.java: WS server on ws://localhost:7654
- bridge/GameSession.java: Phase 1 stub game loop with human-input queue
- cli/LabCli.java: add ws-server subcommand via picocli
2026-03-18 00:02:18 -04:00
KoalaTrapLord 2f117eedcb fix: Java 17 compat - inline ternary in this() call instead of local var 2026-03-16 20:34:05 +00:00
KoalaTrapLord 8f8abb79c6 fix: match actual Forge verbose log format - cast/played/Life:/Zone Change: patterns 2026-03-16 20:25:35 +00:00
KoalaTrapLord 78898f6da0 fix: regex patterns match Forge verbose output, single-winner resolution, 300s clock for 4p 2026-03-16 17:34:12 +00:00
KoalaTrapLord 69242a8f6e fix: remove duplicate numPlayers declaration in parseForgeOutput 2026-03-16 17:13:48 +00:00
KoalaTrapLord 2b6ccac3ea fix: hardcoded 3-element arrays in BatchRunner causing empty stats for 4-player games 2026-03-16 17:09:57 +00:00
KoalaTrapLord 94bb369866 feat: add --deck4/--commander4 support for 4-player games in LabCli + BatchRunner 2026-03-16 02:28:06 +00:00
KoalaTrapLord a0adfc37e8 v24: Resolve GitHub Issues #1-#5
Issue #1 — Winner's life total now captured:
  - Parse verbose Forge log for life-total-change events
  - Track last-known life per seat via 3 regex patterns
  - Winner's final life set from log data (not left at default 40)
  - Losers who died to mill/poison retain their actual life total

Issue #2 — Win condition classification formalized:
  - New WinCondition enum (schema/WinCondition.java) with 8 categories
  - Centralized Forge loss-reason-string → enum mapping (regex-based)
  - UNKNOWN fallback with logging for unrecognized loss reasons
  - All classification now goes through WinCondition.classify()

Issue #3 — JVM startup overhead reduced:
  - Added -XX:+UseSerialGC (lighter GC for batch processes)
  - Added -XX:TieredStopAtLevel=1 (faster JIT, less compilation)
  - ForgeWorker stub class for future persistent batch mode
  - Estimated reduction: ~2-3s → ~1-1.5s per game startup

Issue #4 — Single-threaded throughput optimization:
  - New CLI flags: --ai-simplified, --ai-think-time, --benchmark
  - JVM flags passed through to Forge subprocess
  - Benchmark mode: run 1 game before batch, report wall/Forge/overhead times
  - API support: aiSimplified, aiThinkTimeMs in StartRequest

Issue #5 — Threading, progress, and backpressure:
  - --threads 0 or -1 auto-detects optimal count from CPU topology
  - detectOptimalThreads() estimates physical cores (excludes HT)
  - ProgressCallback interface for real-time sims/sec reporting
  - Per-game progress emitted as [PROGRESS] lines, parsed by lab_api.py
  - --max-queue flag for graceful backpressure (limits concurrent JVMs)
  - Final throughput summary printed at batch completion
  - StatusResponse now includes simsPerSec field

Also: .gitignore restored, lab_api.py updated with all new API fields
2026-03-12 01:26:36 +00:00
KoalaTrapLord a5fe8cb3fa v23: React + TypeScript SPA migration
Complete frontend rewrite from vanilla HTML/CSS/JS to React 19 + TypeScript + Tailwind CSS v4.

Frontend (frontend/commander-ai-lab-ui/):
- Vite + React 19 + TypeScript + Tailwind CSS v4 + React Router
- 7 pages: BatchSim, Collection, Decks, AutoGen, Simulator, Coach, Training
- Reusable components: Spinner, StatusBadge, CardImage, ManaSymbols, EmptyState
- Full typed API client layer (8 modules matching all backend endpoints)
- Recharts for mana curves, win rate charts, progress bars
- Lucide icons throughout
- Custom dark theme with MTG mana color tokens
- Code splitting (recharts + react in separate chunks)

Backend (lab_api.py):
- SPA-aware static file serving with catch-all for client-side routing
- Fallback to legacy ui/ folder if React dist not built
- Updated startup log with SPA route list

Build system:
- .gitignore added (was missing - removes pyc, db, node_modules, etc.)
- build-full-installer-v2.py: v23 with Node.js check + React build step
- Vite build config with manual chunks for optimal loading

Also includes: V3 deck generator fixes, coach improvements, DeepSeek sim modules
2026-03-12 01:12:02 +00:00
KoalaTrapLord e6b5e0cedb Initial commit 2026-03-10 12:13:56 -04:00