File size: 6,488 Bytes
d69fc90
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
#!/usr/bin/env python3
"""End-to-end pipeline: CL-native macro discovery + expansion + classification.

1. For each library, runs SBCL with generate.lisp to produce verified JSONL
2. Merges with macro definitions from Phase 1 extractions
3. Classifies and builds train/val/test splits
"""

from __future__ import annotations

import json
import subprocess
import sys
from pathlib import Path

sys.path.insert(0, str(Path(__file__).parent.parent / "src"))

from cl_macros.classifier import classify_all, quality_report, detect_techniques, detect_category, assess_complexity, assess_capture_risk
from cl_macros.dataset import build_dataset, save_dataset, build_training_record
from cl_macros.ext.library_index import LibraryIndex
from cl_macros.knowledge_base import ALL_EXAMPLES
from cl_macros.schema import (
    TransformationExample,
    Complexity,
    Source,
)

SBCL = "sbcl"
GENERATE_LISP = Path(__file__).parent.parent / "src" / "cl_macros" / "verif" / "generate.lisp"
OUTPUT_DIR = Path("data/generated")


def run_sbcl(lib_name: str, output_path: Path) -> bool:
    """Run SBCL with generate.lisp for one library."""
    result = subprocess.run(
        [
            SBCL, "--noinform", "--non-interactive",
            "--load", str(GENERATE_LISP),
            "--eval", f'(lol-gen:main "{lib_name}" "{output_path}")',
        ],
        capture_output=True, text=True, timeout=300,
    )
    return result.returncode == 0


def load_macro_defs(lib_name: str) -> dict[str, str]:
    """Load macro definitions from Phase 1 extraction files."""
    ext_path = Path("data/extractions") / f"{lib_name}_extractions.jsonl"
    defs = {}
    if ext_path.exists():
        with open(ext_path) as f:
            for line in f:
                rec = json.loads(line.strip())
                name = rec.get("macro_name", "")
                full = rec.get("macro_definition", "")
                if name and full:
                    defs[name.upper()] = full
    return defs


def source_for_lib(lib_name: str) -> Source:
    """Map library name to Source enum."""
    mapping = {
        "alexandria": Source.ALEXANDRIA,
        "serapeum": Source.SERAPEUM,
        "anaphora": Source.ANAPHORA,
        "iterate": Source.ITERATE,
        "trivia": Source.TRIVIA,
        "arrow-macros": Source.ARROW_MACROS,
        "modf": Source.MODF,
        "access": Source.ACCESS,
        "for": Source.FOR,
        "cl-interpol": Source.CL_INTERPOL,
        "screamer": Source.SCREAMER,
        "coalton": Source.COALTON,
        "nhooks": Source.NHOOKS,
        "generic-cl": Source.GENERIC_CL,
    }
    return mapping.get(lib_name, Source.OTHER_LIBRARY)


def build_examples(records: list[dict], lib_name: str) -> list[TransformationExample]:
    """Convert raw SBCL output to TransformationExample objects."""
    macro_defs = load_macro_defs(lib_name)
    examples = []
    seen = set()

    for rec in records:
        if rec.get("status") != "verified":
            continue

        macro_name = rec["macro_name"]
        call_form = rec["call_form"]
        expanded = rec.get("expanded", "")

        # Deduplicate: skip identical (macro_name, expanded) pairs
        key = (macro_name, expanded)
        if key in seen:
            continue
        seen.add(key)

        # Skip trivial expansions (same as input)
        if call_form.strip() == expanded.strip():
            continue

        # Get source macro definition
        macro_def = macro_defs.get(macro_name.upper(), f"(defmacro {macro_name} ...)")

        ex = TransformationExample(
            id=f"{lib_name}-{macro_name}-{len(examples)}",
            before_code=call_form,
            problem_pattern=f"Macro call that should be transformed by {macro_name}",
            macro_definition=macro_def,
            after_expansion=expanded,
            macro_category=None,  # auto-detect
            technique=[],
            source=source_for_lib(lib_name),
            complexity=Complexity.BASIC,  # auto-detect
            library_name=lib_name,
            macro_name=macro_name,
            is_verified=True,
            macroexpand_1_result=expanded,
            formulation="macro-from-usage",
        )
        examples.append(ex)

    return examples


def main():
    import argparse
    ap = argparse.ArgumentParser()
    ap.add_argument("--tier", default="tier1")
    ap.add_argument("--library", default=None)
    args = ap.parse_args()

    idx = LibraryIndex()
    if args.library:
        libs = [idx.get(args.library)]
    else:
        libs = idx.list_libraries(args.tier)

    OUTPUT_DIR.mkdir(parents=True, exist_ok=True)

    all_examples = list(ALL_EXAMPLES)  # preserve hand-curated
    stats = {}

    for lib in libs:
        print(f"\n--- {lib.name} ({lib.tier}) ---")
        output_path = OUTPUT_DIR / f"{lib.name}_generated.jsonl"

        if not output_path.exists():
            print(f"  Running SBCL...")
            if not run_sbcl(lib.name, output_path):
                print(f"  SBCL FAILED for {lib.name}")
                continue

        # Read results
        with open(output_path) as f:
            records = [json.loads(line) for line in f if line.strip()]

        verified = sum(1 for r in records if r.get("status") == "verified")
        errors = sum(1 for r in records if r.get("status") != "verified")
        print(f"  Records: {len(records)} ({verified} verified, {errors} errors)")

        examples = build_examples(records, lib.name)
        print(f"  Valid examples: {len(examples)}")
        all_examples.extend(examples)
        stats[lib.name] = len(examples)

    # Classify
    print(f"\n=== Classification ===")
    print(f"Total examples: {len(all_examples)}")
    classified = classify_all(all_examples)
    report = quality_report(classified)
    print(f"Mean score: {report['mean_score']:.3f}")
    print(f"Categories: {report['category_distribution']}")

    # Build splits
    print(f"\n=== Building splits ===")
    train, val, test = build_dataset(classified, min_quality=0.3)
    print(f"Train: {len(train)}, Val: {len(val)}, Test: {len(test)}")

    split_dir = Path("data/splits")
    paths = save_dataset(train, val, test, split_dir)
    for name, p in paths.items():
        print(f"  {name}: {p}")

    # Stats
    stat_path = OUTPUT_DIR / "generation_stats.json"
    with open(stat_path, "w") as f:
        json.dump({"stats": stats, "report": report}, f, indent=2)
    print(f"\nStats: {stat_path}")


if __name__ == "__main__":
    main()