src/curriculum/runner.py

"""
CurriculumRunner — Vitalis FSI

The childhood listening phase.
Vitalis ingests curated audio, builds hypervector representations,
and feeds the DreamEngine for consolidation.

No external APIs. No cloud. Fully sovereign.
Audio files live locally. Processing is local.
"""
import os
import time
import numpy as np
from pathlib import Path
from typing import List, Tuple

from src.audio_ear.feature_extractor import extract_features
from src.hdc_encoder.encoder import encode
from src.dream_engine.helix_memory import HelixMemory
from src.dream_engine.consolidator import DreamEngine
from src.ide_kernel.ledger import ProjectLedger


class CurriculumRunner:
    SEGMENT_SECONDS = 30
    DREAM_EVERY_N   = 10

    def __init__(
        self,
        audio_dir: str,
        helix_path: Path = None,
        workspace: str = None,
    ):
        self.audio_dir   = Path(audio_dir)
        self.helix_path  = helix_path or (
            Path.home() / ".vitalis_workspace" / "helix_memory.pkl"
        )
        self.workspace   = workspace or os.getcwd()
        self.helix       = HelixMemory(self.helix_path)
        self.dreamer     = DreamEngine(self.helix, buffer_max=500)
        self.ledger      = ProjectLedger(self.workspace)
        self._processed  = 0
        self._start_time = None

    def _discover_audio(self) -> List[Path]:
        """Find all wav files in the audio directory."""
        if not self.audio_dir.exists():
            print(f"[CURRICULUM] Audio dir not found: {self.audio_dir}")
            print(f"[CURRICULUM] Creating directory. Add .wav files to begin.")
            self.audio_dir.mkdir(parents=True, exist_ok=True)
            return []
        files = list(self.audio_dir.rglob("*.wav"))
        print(f"[CURRICULUM] Discovered {len(files)} audio files.")
        return files

    def _process_file(self, wav_path: Path, label: str = "unlabeled") -> bool:
        """Process one wav file into a hypervector and ingest."""
        try:
            mfcc, prosody = extract_features(wav_path)
            hv = encode(mfcc, prosody)
            self.dreamer.ingest(hv, meta={
                "source":    str(wav_path.name),
                "label":     label,
                "timestamp": time.time(),
            })
            return True
        except Exception as e:
            print(f"[CURRICULUM] Error processing {wav_path.name}: {e}")
            return False

    def run(self, total_hours: float = 12.0) -> dict:
        """
        Run the full curriculum.
        Processes audio files in a loop until total_hours elapsed.
        Dreams periodically based on buffer pressure.
        """
        files = self._discover_audio()
        if not files:
            print("[CURRICULUM] No audio files found. "
                  f"Add .wav files to {self.audio_dir} and rerun.")
            return {"status": "no_audio", "processed": 0}

        self._start_time = time.time()
        elapsed_hours    = 0.0
        file_index       = 0
        success_count    = 0

        print(f"[CURRICULUM] Starting {total_hours}h curriculum "
              f"with {len(files)} files.")

        while elapsed_hours < total_hours:
            wav = files[file_index % len(files)]

            # Infer label from parent directory name
            label = wav.parent.name

            success = self._process_file(wav, label)
            if success:
                success_count   += 1
                self._processed += 1

            # Log to ledger
            self.ledger.update_state(
                f"curriculum_segment_{self._processed}",
                f"Completed — {wav.name}"
            )

            # Dream when buffer pressure is high
            if self._processed % self.DREAM_EVERY_N == 0:
                self.dreamer.dream(force=True)
                print(f"[CURRICULUM] {self._processed} segments processed. "
                      f"Elapsed: {elapsed_hours:.2f}h")

            file_index   += 1
            elapsed_hours = (time.time() - self._start_time) / 3600

        # Final dream
        self.dreamer.dream(force=True)

        result = {
            "status":        "complete",
            "processed":     self._processed,
            "successful":    success_count,
            "helix_codes":   len(self.helix.entries),
            "elapsed_hours": round(elapsed_hours, 3),
        }
        print(f"[CURRICULUM] Complete. {result}")
        return result


def run_curriculum(
    audio_dir: str,
    helix_path: Path = None,
    total_hours: float = 12.0,
) -> dict:
    """Convenience entry point."""
    runner = CurriculumRunner(
        audio_dir=audio_dir,
        helix_path=helix_path,
    )
    return runner.run(total_hours=total_hours)