Initial commit

2026-09-04 14:58:42 +08:00
commit 439cad87d9
4601 changed files with 29440 additions and 0 deletions
@@ -0,0 +1,107 @@
+"""Build the combined behavioral and format Profile for one model."""
+
+from __future__ import annotations
+
+import json
+from pathlib import Path
+import subprocess
+import sys
+
+from ..paths import (
+    FINAL_PROFILE_ROOT,
+    MODEL_PREFERENCE_PROFILE_ROOT,
+    PROJECT_ROOT,
+    model_profile_path,
+)
+
+
+MODEL_PROFILE_MODULE = (
+    "scripts.static_compile.profile_generation.model_preference.run_profile"
+)
+FORMAT_PROFILE_MODULE = (
+    "scripts.static_compile.profile_generation.format_preference.run_format_preference"
+)
+
+
+def run_stage(label: str, command: list[str]) -> None:
+    print(f"\n{'=' * 72}\n{label}\n{'=' * 72}", flush=True)
+    subprocess.run(command, check=True, cwd=PROJECT_ROOT)
+
+
+def profile_paths(model_identifier: str) -> tuple[Path, Path]:
+    """Return the behavioral source Profile and final combined Profile paths."""
+
+    upstream = model_profile_path(MODEL_PREFERENCE_PROFILE_ROOT, model_identifier)
+    final = model_profile_path(FINAL_PROFILE_ROOT, model_identifier)
+    return upstream, final
+
+
+def profile_is_reusable(profile_path: Path, model_identifier: str) -> bool:
+    """Return whether a complete combined Profile matches the requested model."""
+
+    try:
+        profile = json.loads(profile_path.read_text(encoding="utf-8"))
+    except (OSError, json.JSONDecodeError):
+        return False
+    model = profile.get("model")
+    return bool(
+        isinstance(model, dict)
+        and model.get("id") == model_identifier.strip("/")
+        and model.get("profile_status") == "complete"
+        and isinstance(profile.get("behavioral_profile"), dict)
+        and isinstance(profile.get("format_preference"), dict)
+    )
+
+
+def generate_profile(
+    model_identifier: str,
+    *,
+    refresh_model_preference: bool = False,
+) -> Path:
+    """Run behavioral profiling followed by format profiling."""
+
+    model_id = model_identifier.strip("/")
+    upstream_profile_path, final_profile_path = profile_paths(model_id)
+
+    model_command = [sys.executable, "-m", MODEL_PROFILE_MODULE, model_id]
+    if refresh_model_preference:
+        model_command.append("--refresh")
+    run_stage(
+        "1/2 Model-preference profiling: collecting and evaluating behavior",
+        model_command,
+    )
+    run_stage(
+        "2/2 Format-preference profiling: measuring output-format sensitivity",
+        [
+            sys.executable,
+            "-m",
+            FORMAT_PROFILE_MODULE,
+            "--model",
+            model_id,
+            "--profile_path",
+            str(final_profile_path),
+            "--base_profile_path",
+            str(upstream_profile_path),
+        ],
+    )
+    if not profile_is_reusable(final_profile_path, model_id):
+        raise RuntimeError(
+            "profile generation did not produce a complete combined Profile at "
+            f"{final_profile_path}"
+        )
+    return final_profile_path
+
+
+def ensure_profile(model_identifier: str, *, refresh: bool = False) -> tuple[Path, bool]:
+    """Reuse the final Profile when present, otherwise generate it."""
+
+    _, final_profile_path = profile_paths(model_identifier)
+    if profile_is_reusable(final_profile_path, model_identifier) and not refresh:
+        return final_profile_path, False
+    return (
+        generate_profile(
+            model_identifier,
+            refresh_model_preference=refresh,
+        ),
+        True,
+    )