hack/tools/scripts/audit_channels.py

#!/usr/bin/env python3
"""Audit: collect every USABLE key from (a) vault keys.json and (b) all
results/*/usable.txt buckets, dedup against keys already in NewAPI, and emit
a per-provider list of candidates that need a channel.

Does NOT create anything. Writes results/channel_audit.json.
"""
import json, re, sys
from pathlib import Path
from collections import defaultdict

sys.path.insert(0, str(Path(__file__).resolve().parent))
import newapi_client as n

H = Path(__file__).resolve().parent / "llm-key-hunter"
VAULT = H / "usable_keys"
RESULTS = H / "results"

# provider label normalisation -> vault dir
def norm(label):
    l = label.lower().replace(" ", "").replace("_", "").replace("-", "")
    m = {
        "minimax": "minimax", "deepseek": "deepseek", "kimi": "kimi",
        "moonshot": "kimi", "siliconflow": "siliconflow", "zhipuai": "zhipu",
        "zhipu": "zhipu", "bigmodel": "zhipu", "volcanoark": "volcanoark",
        "volcano": "volcanoark", "volcengine": "volcanoark", "doubao": "volcanoark",
        "longcat": "longcat", "dashscope": "dashscope", "qwen": "dashscope",
        "freemodel": "freemodel", "xunfei": "xunfei", "iflytek": "xunfei",
        "spark": "xunfei", "codingplan": "codingplan", "opencode": "codingplan",
        "mimo": "mimo", "xiaomimimo": "mimo", "kiro": "kiro",
    }
    return m.get(l)


def load_vault_keys():
    out = defaultdict(list)
    for d in sorted(VAULT.iterdir()):
        kj = d / "keys.json"
        if not d.is_dir() or not kj.exists():
            continue
        try:
            data = json.loads(kj.read_text())
        except Exception:
            continue
        for e in data:
            k = e.get("key")
            if k:
                out[d.name].append((k, e.get("source", "")))
    return out


def parse_usable_line(line):
    line = line.strip()
    if not line or line.startswith("#"):
        return None
    # tab-separated: provider<TAB>key<TAB>url<TAB>detail
    if "\t" in line:
        parts = line.split("\t")
        if len(parts) >= 2:
            return parts[0], parts[1], parts[2] if len(parts) > 2 else ""
    # pipe-separated: Provider|key|url|desc
    if "|" in line:
        parts = line.split("|")
        if len(parts) >= 2:
            return parts[0], parts[1], parts[2] if len(parts) > 2 else ""
    return None


def load_bucket_keys():
    """keys from all results/*/usable.txt that may not be vaulted yet."""
    out = defaultdict(dict)  # vault_dir -> {key: source}
    for f in sorted(RESULTS.glob("*/usable.txt")):
        for line in f.read_text().splitlines():
            r = parse_usable_line(line)
            if not r:
                continue
            label, key, src = r
            vd = norm(label)
            if vd:
                out[vd][key] = src
    return out


def newapi_keys():
    keys = set()
    for ch in n.list_channels():
        try:
            d = n.get_channel(ch["id"])
            if d and d.get("key"):
                keys.add(d["key"])
        except Exception:
            pass
    return keys


def channel_status_summary():
    """count channels by status across all IDs."""
    counts = defaultdict(int)
    disabled = []
    for ch in n.list_channels():
        cid = ch["id"]
        try:
            d = n.get_channel(cid)
        except Exception:
            continue
        st = d.get("status", 0)
        counts[st] += 1
        if st != 1:  # 1=enabled
            disabled.append((cid, d.get("name", ""), st))
    return counts, disabled


def main():
    print("Loading NewAPI channel keys (this scans all channels)...")
    existing = newapi_keys()
    print(f"  {len(existing)} unique keys currently in NewAPI")

    counts, disabled = channel_status_summary()
    print(f"  channel status: {dict(counts)}  (1=enabled, 2=manual-disabled, 3=auto-disabled)")

    vault = load_vault_keys()
    buckets = load_bucket_keys()

    missing = defaultdict(list)  # vault_dir -> [(key,source)]
    for vd, items in vault.items():
        for key, src in items:
            if key not in existing:
                missing[vd].append((key, src))
    # bucket keys not in vault and not in NewAPI
    extra = defaultdict(list)
    vault_keyset = {k for items in vault.values() for k, _ in items}
    for vd, kmap in buckets.items():
        for key, src in kmap.items():
            if key not in vault_keyset and key not in existing:
                extra[vd].append((key, src))

    report = {
        "existing_newapi_keys": len(existing),
        "channel_status": dict(counts),
        "disabled_channels": disabled,
        "vault_not_in_newapi": {k: v for k, v in missing.items()},
        "bucket_not_in_vault_or_newapi": {k: v for k, v in extra.items()},
    }
    out = RESULTS / "channel_audit.json"
    out.write_text(json.dumps(report, indent=2, ensure_ascii=False))

    print("\n=== vault keys WITHOUT a NewAPI channel ===")
    tot = 0
    for vd, items in sorted(missing.items()):
        print(f"  {vd:14} {len(items)}")
        tot += len(items)
    print(f"  TOTAL missing: {tot}")

    print("\n=== bucket usable keys NOT vaulted & NOT in NewAPI ===")
    te = 0
    for vd, items in sorted(extra.items()):
        print(f"  {vd:14} {len(items)}")
        te += len(items)
    print(f"  TOTAL extra: {te}")

    print(f"\n=== disabled/abnormal channels: {len(disabled)} ===")
    for cid, name, st in disabled[:30]:
        print(f"  {cid} {name:30} status={st}")
    if len(disabled) > 30:
        print(f"  ... and {len(disabled)-30} more")
    print(f"\naudit written -> {out}")


if __name__ == "__main__":
    main()