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
|
#!/usr/bin/env python3
from __future__ import annotations
import argparse
import json
import sqlite3
from pathlib import Path
from typing import Any
import numpy as np
from sentence_transformers import SentenceTransformer
import os
from pathlib import Path
DB_PATH = Path("cheat.db")
LOCAL_CACHE_DIR = Path("models/hf")
os.environ.setdefault("HF_HOME", str(LOCAL_CACHE_DIR.resolve()))
os.environ.setdefault("SENTENCE_TRANSFORMERS_HOME", str(LOCAL_CACHE_DIR.resolve()))
def deserialize_embedding(blob: bytes, dim: int) -> np.ndarray:
vec = np.frombuffer(blob, dtype=np.float32)
if vec.shape[0] != dim:
raise ValueError(f"Embedding length mismatch: expected {dim}, got {vec.shape[0]}")
return vec
def load_index(conn: sqlite3.Connection) -> tuple[list[dict[str, Any]], np.ndarray, str]:
rows = conn.execute("""
SELECT
c.id,
c.command,
c.explanation,
c.intent_json,
c.alternatives_json,
c.requires_json,
c.packages_json,
c.tags_json,
c.platform_json,
c.shell_json,
c.safety,
e.model_name,
e.embedding_blob,
e.embedding_dim
FROM cards c
JOIN card_embeddings e ON c.id = e.card_id
ORDER BY c.id
""").fetchall()
if not rows:
raise RuntimeError("No indexed cards found. Run build_index.py first.")
cards: list[dict[str, Any]] = []
vectors: list[np.ndarray] = []
model_name: str | None = None
for row in rows:
(
card_id,
command,
explanation,
intent_json,
alternatives_json,
requires_json,
packages_json,
tags_json,
platform_json,
shell_json,
safety,
row_model_name,
embedding_blob,
embedding_dim,
) = row
if model_name is None:
model_name = row_model_name
elif model_name != row_model_name:
raise RuntimeError("Mixed embedding models found in the index.")
cards.append({
"id": card_id,
"command": command,
"explanation": explanation,
"intent": json.loads(intent_json),
"alternatives": json.loads(alternatives_json),
"requires": json.loads(requires_json),
"packages": json.loads(packages_json),
"tags": json.loads(tags_json),
"platform": json.loads(platform_json),
"shell": json.loads(shell_json),
"safety": safety,
})
vectors.append(deserialize_embedding(embedding_blob, embedding_dim))
matrix = np.vstack(vectors)
return cards, matrix, model_name
def search(
query: str,
top_k: int = 5,
) -> list[dict[str, Any]]:
conn = sqlite3.connect(DB_PATH)
try:
cards, matrix, model_name = load_index(conn)
finally:
conn.close()
model = SentenceTransformer(
model_name,
cache_folder=str(LOCAL_CACHE_DIR.resolve()),
local_files_only=True,
)
qvec = model.encode([query], normalize_embeddings=True, convert_to_numpy=True)[0]
scores = matrix @ qvec
top_indices = np.argsort(scores)[::-1][:top_k]
results: list[dict[str, Any]] = []
for idx in top_indices:
card = dict(cards[idx])
card["score"] = float(scores[idx])
results.append(card)
return results
def main() -> None:
parser = argparse.ArgumentParser(description="Query the local command card index.")
parser.add_argument("query", type=str, help="Natural language query")
parser.add_argument("--top-k", type=int, default=5, help="Number of results to return")
args = parser.parse_args()
results = search(args.query, top_k=args.top_k)
for i, result in enumerate(results, start=1):
print(f"[{i}] score={result['score']:.4f} id={result['id']}")
print(f" command: {result['command']}")
print(f" explanation: {result['explanation']}")
if result["alternatives"]:
print(f" alternatives: {', '.join(result['alternatives'])}")
print(f" intent: {', '.join(result['intent'][:3])}")
print()
if __name__ == "__main__":
main()
|
