feat(ai-studio): read shopee source image urls

app/image_studio.py

@@ -9,11 +9,12 @@ from dataclasses import dataclass
 from datetime import datetime
 from typing import Iterable, Optional
 
-from . import appconfig, db
+from . import accounts, appconfig, chrome, db, editor
 from .config import make_slug
 
 
 PROJECT_STATUS_ACTIVE = "active"
+ASSET_KIND_ORIGINAL = "original"
 ASSET_STATUS_AVAILABLE = "available"
 ASSET_STATUS_MISSING = "missing"
 ASSET_STATUSES = {ASSET_STATUS_AVAILABLE, ASSET_STATUS_MISSING}
@@ -90,6 +91,10 @@ class ImageStudioSelection:
     updated_at: str
 
 
+class ImageStudioError(RuntimeError):
+    """Raised when the AI image studio service cannot complete an operation."""
+
+
 def _now() -> str:
     return datetime.now().isoformat(timespec="seconds")
 
@@ -141,6 +146,83 @@ def _task_key(project_id):
     return f"image-studio-{int(project_id)}-{uuid.uuid4().hex}"
 
 
+def _db_path(path=None, config=None) -> str:
+    return path or appconfig.db_path(config)
+
+
+def _normalize_item_id(item_id) -> str:
+    text = str(item_id or "").strip()
+    if not text:
+        raise ImageStudioError("商品ID不能为空")
+    return text
+
+
+def _notify_step(callback, step, result="start", detail=None):
+    if callback is None:
+        return
+    payload = {"step": step, "result": result}
+    if detail:
+        payload["detail"] = str(detail)
+    try:
+        callback(payload)
+    except Exception:
+        pass
+
+
+def _is_definitive_logged_out(status):
+    reason = str((status or {}).get("reason") or "").strip()
+    url = str((status or {}).get("url") or "").lower()
+    return reason.startswith("LOGIN_PAGE") or (
+        "accounts.shopee." in url and "/seller/login" in url
+    )
+
+
+def _login_status_detail(status):
+    status = status or {}
+    reason = status.get("reason") or "未知原因"
+    url = status.get("url") or "未知URL"
+    cookie_names = [str(name) for name in (status.get("cookie_names") or []) if name]
+    cookie_text = ",".join(sorted(cookie_names)) if cookie_names else "未读到登录Cookie"
+    return f"原因={reason}，URL={url}，Cookie名称={cookie_text}"
+
+
+def _ensure_account_ready_for_read(account, *, path=None, config=None, login_timeout=8, on_step=None):
+    port = accounts.normalize_debug_port(_get(account, "debug_port"))
+    alias = str(_get(account, "alias") or "").strip()
+    _notify_step(on_step, "ensure_chrome", "start", f"账号 {alias} debug_port={port}")
+    if chrome.is_running(port):
+        chrome_info = {
+            "action": "reused",
+            "reused": True,
+            "launched": False,
+            "debug_port": port,
+        }
+        _notify_step(on_step, "ensure_chrome", "reused", f"账号 {alias} Chrome 已打开")
+    else:
+        try:
+            chrome_info = accounts.launch_for_login(account, path=path, config=config)
+        except Exception as exc:
+            raise ImageStudioError(f"账号 {alias} Chrome 启动失败：{exc}") from exc
+        action = "launched" if chrome_info.get("launched") else "reused"
+        _notify_step(on_step, "ensure_chrome", action, f"账号 {alias} Chrome 已{ '启动' if action == 'launched' else '复用' }")
+
+    _notify_step(on_step, "login_check", "start", f"账号 {alias}")
+    try:
+        status = accounts.detect_login(account, timeout=login_timeout, path=path, config=config)
+    except Exception as exc:
+        status = {"logged_in": False, "reason": f"LOGIN_CHECK_FAILED: {exc}"}
+    if status.get("logged_in"):
+        _notify_step(on_step, "login_check", "success", f"账号 {alias} 已登录")
+        return {**chrome_info, "login_status": status, "login_uncertain": False}
+    if _is_definitive_logged_out(status):
+        detail = _login_status_detail(status)
+        _notify_step(on_step, "login_check", "blocked", detail)
+        raise ImageStudioError(f"账号 {alias} 未登录：{detail}")
+    detail = _login_status_detail(status)
+    _notify_step(on_step, "login_check", "uncertain", f"账号 {alias} 登录状态暂不稳定，继续尝试读取商品：{detail}")
+    return {**chrome_info, "login_status": status, "login_uncertain": True}
+
+
 def _ensure_assets_belong_to_project(database, project_id, asset_ids):
     ids = [int(asset_id) for asset_id in asset_ids if asset_id is not None]
     if not ids:
@@ -370,6 +452,91 @@ def list_assets(project_id, kind=None, include_missing=True, path=None, conn=Non
         return _fetch_all(database, sql, params, ImageStudioAsset)
 
 
+def sync_original_asset_urls(project_id, image_urls, path=None, conn=None):
+    """Store the read-only Shopee main image URL snapshot as remote-only assets."""
+
+    normalized = []
+    seen_urls = set()
+    for position, item in enumerate(image_urls or [], start=1):
+        if isinstance(item, dict):
+            src = str(item.get("src") or "").strip()
+            source_order = int(item.get("index") or position)
+        else:
+            src = str(item or "").strip()
+            source_order = position
+        if not src or src in seen_urls:
+            continue
+        seen_urls.add(src)
+        normalized.append({"src": src, "source_order": source_order})
+
+    now = _now()
+    with _connection(conn, path) as database:
+        with database:
+            existing_rows = database.execute(
+                """
+                SELECT * FROM image_studio_assets
+                WHERE project_id = ? AND kind = ?
+                """,
+                (int(project_id), ASSET_KIND_ORIGINAL),
+            ).fetchall()
+            by_url = {
+                str(row["remote_url"]): row
+                for row in existing_rows
+                if row["remote_url"]
+            }
+            active_ids = set()
+            for item in normalized:
+                row = by_url.get(item["src"])
+                if row is not None:
+                    active_ids.add(int(row["id"]))
+                    database.execute(
+                        """
+                        UPDATE image_studio_assets
+                        SET source_order = ?,
+                            status = ?,
+                            updated_at = ?
+                        WHERE id = ?
+                        """,
+                        (
+                            int(item["source_order"]),
+                            ASSET_STATUS_AVAILABLE,
+                            now,
+                            int(row["id"]),
+                        ),
+                    )
+                else:
+                    cursor = database.execute(
+                        """
+                        INSERT INTO image_studio_assets
+                          (project_id, kind, remote_url, local_path, status,
+                           source_order, created_at, updated_at)
+                        VALUES (?, ?, ?, NULL, ?, ?, ?, ?)
+                        """,
+                        (
+                            int(project_id),
+                            ASSET_KIND_ORIGINAL,
+                            item["src"],
+                            ASSET_STATUS_AVAILABLE,
+                            int(item["source_order"]),
+                            now,
+                            now,
+                        ),
+                    )
+                    active_ids.add(int(cursor.lastrowid))
+            missing_ids = [int(row["id"]) for row in existing_rows if int(row["id"]) not in active_ids]
+            if missing_ids:
+                placeholders = ",".join("?" for _ in missing_ids)
+                database.execute(
+                    f"""
+                    UPDATE image_studio_assets
+                    SET status = ?, updated_at = ?
+                    WHERE id IN ({placeholders})
+                    """,
+                    [ASSET_STATUS_MISSING, now, *missing_ids],
+                )
+        return list_assets(project_id, kind=ASSET_KIND_ORIGINAL, conn=database)
+
+
 def mark_asset_status(asset_id, status, path=None, conn=None):
     if str(status) not in ASSET_STATUSES:
         raise db.DbError("AI工场资产状态必须是 available 或 missing")
@@ -563,3 +730,61 @@ def list_selections(project_id, selection_type=None, path=None, conn=None):
     sql += " ORDER BY selection_type, position"
     with _connection(conn, path) as database:
         return _fetch_all(database, sql, params, ImageStudioSelection)
+
+
+def pull_remote_main_image_urls(
+    account_or_alias,
+    item_id,
+    *,
+    path=None,
+    config=None,
+    login_timeout=8,
+    on_step=None,
+):
+    """Create/open an AI studio project and read Shopee main image URLs without editing."""
+
+    cfg = appconfig.load_config() if config is None else config
+    database_path = _db_path(path, cfg)
+    db.init_db(database_path)
+    item = _normalize_item_id(item_id)
+    try:
+        account = accounts.resolve_account(account_or_alias, path=database_path, config=cfg)
+    except Exception as exc:
+        raise ImageStudioError(f"AI工场账号不可用：{exc}") from exc
+
+    project = create_or_get_project(account, item_id=item, path=database_path)
+    readiness = _ensure_account_ready_for_read(
+        account,
+        path=database_path,
+        config=cfg,
+        login_timeout=login_timeout,
+        on_step=on_step,
+    )
+    cdp = None
+    try:
+        _notify_step(on_step, "open_product", "start", f"商品 {item}")
+        cdp = editor.open_product(account, item, on_step=on_step, bring_to_front=False)
+        _notify_step(on_step, "open_product", "success", f"商品 {item}")
+        _notify_step(on_step, "read_main_images", "start", f"商品 {item}")
+        images = editor.read_product_image_urls(cdp)
+        if not images:
+            raise ImageStudioError("未读取到蝦皮商品主图 URL")
+        assets = sync_original_asset_urls(project.id, images, path=database_path)
+        _notify_step(on_step, "read_main_images", "success", f"读取 {len(images)} 张主图 URL")
+        project = get_project(project.id, path=database_path)
+        return {
+            "project": project,
+            "images": images,
+            "assets": assets,
+            "account": account,
+            "readiness": readiness,
+        }
+    except ImageStudioError:
+        raise
+    except editor.EditorError as exc:
+        raise ImageStudioError(str(exc)) from exc
+    except Exception as exc:
+        raise ImageStudioError(f"读取蝦皮原主图失败：{exc}") from exc
+    finally:
+        if cdp is not None:
+            editor.close_readonly_product(cdp)