momentry_core/scripts/update_qdrant_uuid.py

#!/usr/bin/env python3
"""批量更新 Qdrant collection 中的 file_uuid (舊→新)"""

import json
import subprocess
import sys

QDRANT_URL = "http://localhost:6333"

# UUID mapping: 舊 → 新
UUID_MAP = {
    "aeed71342a899fe4b4c57b7d41bcb692": [
        "bd80fec92b0b6963d177a2c55bf713e2",
    ],
}

# Collections to process
COLLECTIONS = [
    "momentry_dev_v1",
    "momentry_dev_stories",
    "momentry_dev_voice",
    "momentry_dev_rule1_v2",
    "momentry_dev_faces",
    "sentence_story",
    "sentence_summary",
]


def qdrant_get(path: str) -> dict:
    res = subprocess.run(
        ["curl", "-s", "-X", "GET", f"{QDRANT_URL}{path}"],
        capture_output=True, text=True
    )
    return json.loads(res.stdout) if res.stdout.strip() else {}


def qdrant_post(path: str, body: dict) -> dict:
    tmp = "/tmp/qdrant_post.json"
    with open(tmp, "w") as f:
        json.dump(body, f)
    res = subprocess.run(
        ["curl", "-s", "-X", "POST", f"{QDRANT_URL}{path}",
         "-H", "Content-Type: application/json", "-d", f"@{tmp}"],
        capture_output=True, text=True
    )
    return json.loads(res.stdout) if res.stdout.strip() else {}


def qdrant_put(path: str, body: dict) -> dict:
    tmp = "/tmp/qdrant_update.json"
    with open(tmp, "w") as f:
        json.dump(body, f)
    res = subprocess.run(
        ["curl", "-s", "-X", "PUT", f"{QDRANT_URL}{path}",
         "-H", "Content-Type: application/json", "-d", f"@{tmp}"],
        capture_output=True, text=True
    )
    return json.loads(res.stdout) if res.stdout.strip() else {}


def scroll_all(collection: str, filter_old: dict) -> list:
    """Scroll all matching points from a collection"""
    points = []
    offset = None
    while True:
        body = {
            "limit": 1000,
            "with_payload": True,
            "with_vector": True,
            "filter": filter_old,
        }
        if offset:
            body["offset"] = offset
        result = qdrant_post(f"/collections/{collection}/points/scroll", body)
        batch = result.get("result", {}).get("points", [])
        points.extend(batch)
        next_offset = result.get("result", {}).get("next_page_offset")
        if next_offset is None:
            break
        offset = next_offset
    return points


def update_points(collection: str, points: list, old_uuid: str, new_uuid: str):
    """Update file_uuid in payload for the given points"""
    if not points:
        return 0

    updated = []
    for p in points:
        pl = p.get("payload", {})
        # Check both 'uuid' and 'file_uuid' fields
        changed = False
        if pl.get("uuid") == old_uuid:
            pl["uuid"] = new_uuid
            changed = True
        if pl.get("file_uuid") == old_uuid:
            pl["file_uuid"] = new_uuid
            changed = True
        if changed:
            updated.append({
                "id": p["id"],
                "vector": p["vector"],
                "payload": pl,
            })

    if not updated:
        return 0

    # Update in batches of 500
    total = len(updated)
    for i in range(0, total, 500):
        batch = updated[i:i+500]
        result = qdrant_put(
            f"/collections/{collection}/points?wait=true",
            {"points": batch}
        )
        if result.get("status") != "ok":
            print(f"    Error at {i}: {result}")
            return i
    return total


def main():
    for collection in COLLECTIONS:
        # Check if collection exists
        info = qdrant_get(f"/collections/{collection}")
        if "result" not in info:
            continue

        for old_uuid, new_uuids in UUID_MAP.items():
            for new_uuid in new_uuids:
                # Scroll all points with this old UUID
                filter_body = {
                    "must": [
                        {"should": [
                            {"key": "uuid", "match": {"value": old_uuid}},
                            {"key": "file_uuid", "match": {"value": old_uuid}},
                        ]}
                    ]
                }
                points = scroll_all(collection, filter_body)
                if not points:
                    continue

                print(f"{collection}: {len(points)} points with UUID {old_uuid[:8]}...")
                updated = update_points(collection, points, old_uuid, new_uuid)
                print(f"  → {updated} points updated to {new_uuid[:8]}...")

    # Verify
    print("\n=== Verification ===")
    for collection in COLLECTIONS:
        for old_uuid, new_uuids in UUID_MAP.items():
            for what, uuid in [("old", old_uuid), ("new", new_uuids[0])]:
                filter_body = {
                    "must": [
                        {"should": [
                            {"key": "uuid", "match": {"value": uuid}},
                            {"key": "file_uuid", "match": {"value": uuid}},
                        ]}
                    ]
                }
                result = qdrant_post(
                    f"/collections/{collection}/points/count",
                    {"filter": filter_body}
                )
                cnt = result.get("result", {}).get("count", 0)
                if cnt > 0:
                    print(f"  {collection}: {cnt} points with {what} UUID")
    print("✅ Done")


if __name__ == "__main__":
    main()