| 1 | # -*- coding: utf-8 -*- |
| 2 | from __future__ import annotations |
| 3 | |
| 4 | import asyncio |
| 5 | import inspect |
| 6 | import os |
| 7 | from datetime import datetime |
| 8 | from pathlib import Path |
| 9 | |
| 10 | from patchright.async_api import Page |
| 11 | from patchright.async_api import Playwright |
| 12 | from patchright.async_api import async_playwright |
| 13 | |
| 14 | from conf import DEBUG_MODE, LOCAL_CHROME_HEADLESS, LOCAL_CHROME_PATH |
| 15 | from uploader.base_video import BaseVideoUploader |
| 16 | from utils.base_social_media import set_init_script |
| 17 | from utils.login_qrcode import build_login_qrcode_path |
| 18 | from utils.login_qrcode import decode_qrcode_from_path |
| 19 | from utils.login_qrcode import print_terminal_qrcode |
| 20 | from utils.login_qrcode import remove_qrcode_file |
| 21 | from utils.login_qrcode import save_data_url_image |
| 22 | from utils.log import xiaohongshu_logger |
| 23 | |
| 24 | XHS_DEFAULT_CREATOR_BASE_URL = "https://creator.xiaohongshu.com" |
| 25 | XHS_CREATOR_BASE_URL_ENV = "SAU_XHS_CREATOR_BASE_URL" |
| 26 | XHS_PUBLISH_SUCCESS_URL_PATTERN = "**/publish/success?**" |
| 27 | XHS_LOGIN_BOX_SELECTOR = "div[class*='login-box']" |
| 28 | XHS_LOGIN_SWITCH_SELECTOR = "img.css-wemwzq" |
| 29 | XIAOHONGSHU_PUBLISH_STRATEGY_IMMEDIATE = "immediate" |
| 30 | XIAOHONGSHU_PUBLISH_STRATEGY_SCHEDULED = "scheduled" |
| 31 | |
| 32 | |
| 33 | def _build_xhs_creator_url(path: str) -> str: |
| 34 | base_url = os.getenv( |
| 35 | XHS_CREATOR_BASE_URL_ENV, |
| 36 | XHS_DEFAULT_CREATOR_BASE_URL, |
| 37 | ).strip().rstrip("/") |
| 38 | if not base_url: |
| 39 | base_url = XHS_DEFAULT_CREATOR_BASE_URL |
| 40 | return f"{base_url}/{path.lstrip('/')}" |
| 41 | |
| 42 | |
| 43 | def _msg(emoji: str, text: str) -> str: |
| 44 | return f"{emoji} {text}" |
| 45 | |
| 46 | |
| 47 | async def _emit_qrcode_callback(qrcode_callback, payload: dict): |
| 48 | if not qrcode_callback: |
| 49 | return |
| 50 | |
| 51 | callback_result = qrcode_callback(payload) |
| 52 | if inspect.isawaitable(callback_result): |
| 53 | await callback_result |
| 54 | |
| 55 | |
| 56 | def _build_login_result( |
| 57 | success: bool, |
| 58 | status: str, |
| 59 | message: str, |
| 60 | account_file: str, |
| 61 | qrcode: dict | None = None, |
| 62 | current_url: str = "", |
| 63 | ) -> dict: |
| 64 | return { |
| 65 | "success": success, |
| 66 | "status": status, |
| 67 | "message": message, |
| 68 | "account_file": str(account_file), |
| 69 | "qrcode": qrcode, |
| 70 | "current_url": current_url, |
| 71 | } |
| 72 | |
| 73 | |
| 74 | async def _open_xhs_qrcode_panel(page: Page) -> None: |
| 75 | login_box = page.locator(XHS_LOGIN_BOX_SELECTOR).first |
| 76 | await login_box.wait_for(state="visible", timeout=30000) |
| 77 | |
| 78 | scan_text = login_box.locator("div:has-text('扫一扫')").first |
| 79 | if await scan_text.count(): |
| 80 | return |
| 81 | |
| 82 | switch_img = login_box.locator(XHS_LOGIN_SWITCH_SELECTOR).first |
| 83 | await switch_img.wait_for(state="visible", timeout=10000) |
| 84 | await switch_img.click() |
| 85 | await login_box.locator("div:has-text('扫一扫')").first.wait_for(state="visible", timeout=10000) |
| 86 | |
| 87 | |
| 88 | async def _find_xhs_qrcode_locator(page: Page): |
| 89 | await _open_xhs_qrcode_panel(page) |
| 90 | |
| 91 | qrcode_img = page.locator('.login-box-container').get_by_text("APP扫一扫登录").filter(visible=True).locator("xpath=..//following-sibling::div//img").nth(0) |
| 92 | |
| 93 | if await qrcode_img.count(): |
| 94 | return qrcode_img |
| 95 | |
| 96 | raise RuntimeError("未在扫一扫登录区域找到小红书二维码图片") |
| 97 | |
| 98 | |
| 99 | async def _extract_xhs_qrcode_src(page: Page) -> str: |
| 100 | qrcode_img = await _find_xhs_qrcode_locator(page) |
| 101 | await qrcode_img.wait_for(state="visible", timeout=30000) |
| 102 | qrcode_src = await qrcode_img.get_attribute("src") |
| 103 | if not qrcode_src: |
| 104 | raise RuntimeError("未获取到小红书登录二维码地址") |
| 105 | return qrcode_src |
| 106 | |
| 107 | |
| 108 | async def _save_xhs_qrcode( |
| 109 | page: Page, |
| 110 | account_file: str, |
| 111 | previous_qrcode_path: Path | None = None, |
| 112 | qrcode_callback=None, |
| 113 | ) -> dict: |
| 114 | qrcode_src = await _extract_xhs_qrcode_src(page) |
| 115 | qrcode_path = build_login_qrcode_path(account_file, suffix="xhs_login_qrcode") |
| 116 | qrcode_img = await _find_xhs_qrcode_locator(page) |
| 117 | |
| 118 | if qrcode_src.startswith("data:image/"): |
| 119 | save_data_url_image(qrcode_src, qrcode_path) |
| 120 | else: |
| 121 | qrcode_path.parent.mkdir(parents=True, exist_ok=True) |
| 122 | await qrcode_img.screenshot(path=str(qrcode_path)) |
| 123 | |
| 124 | if previous_qrcode_path and previous_qrcode_path != qrcode_path: |
| 125 | if remove_qrcode_file(previous_qrcode_path): |
| 126 | xiaohongshu_logger.info(_msg("🧹", f"临时二维码文件已清理: {previous_qrcode_path}")) |
| 127 | |
| 128 | xiaohongshu_logger.info(_msg("🖼️", f"二维码已经准备好啦，已保存到: {qrcode_path}")) |
| 129 | qrcode_content = decode_qrcode_from_path(qrcode_path) |
| 130 | if qrcode_content: |
| 131 | print_terminal_qrcode(qrcode_content, qrcode_path, "小红书APP") |
| 132 | else: |
| 133 | xiaohongshu_logger.warning(_msg("😵", f"终端没法完整显示二维码，请打开 {qrcode_path} 扫码")) |
| 134 | |
| 135 | qrcode_info = { |
| 136 | "image_path": str(qrcode_path), |
| 137 | "image_data_url": qrcode_src, |
| 138 | } |
| 139 | await _emit_qrcode_callback(qrcode_callback, qrcode_info) |
| 140 | return qrcode_info |
| 141 | |
| 142 | |
| 143 | async def _is_xhs_login_completed(page: Page) -> bool: |
| 144 | if page.url.startswith(_build_xhs_creator_url("/login")): |
| 145 | return False |
| 146 | |
| 147 | login_box = page.locator(XHS_LOGIN_BOX_SELECTOR).first |
| 148 | if not await login_box.count(): |
| 149 | return True |
| 150 | |
| 151 | try: |
| 152 | return not await login_box.is_visible() |
| 153 | except Exception: |
| 154 | return True |
| 155 | |
| 156 | |
| 157 | async def cookie_auth(account_file): |
| 158 | if not os.path.exists(account_file): |
| 159 | return False |
| 160 | |
| 161 | async with async_playwright() as playwright: |
| 162 | if LOCAL_CHROME_PATH: |
| 163 | browser = await playwright.chromium.launch(headless=True, executable_path=LOCAL_CHROME_PATH) |
| 164 | else: |
| 165 | browser = await playwright.chromium.launch(headless=True, channel="chromium") |
| 166 | try: |
| 167 | context = await browser.new_context(storage_state=account_file) |
| 168 | context = await set_init_script(context) |
| 169 | page = await context.new_page() |
| 170 | await page.goto( |
| 171 | _build_xhs_creator_url( |
| 172 | "/publish/publish?from=homepage&target=video" |
| 173 | ) |
| 174 | ) |
| 175 | await page.wait_for_timeout(3000) |
| 176 | |
| 177 | if page.url.startswith(_build_xhs_creator_url("/login")): |
| 178 | xiaohongshu_logger.info(_msg("🥹", "cookie 已失效，得重新登录一下")) |
| 179 | return False |
| 180 | |
| 181 | login_box = page.locator(XHS_LOGIN_BOX_SELECTOR).first |
| 182 | if await login_box.count(): |
| 183 | try: |
| 184 | if await login_box.is_visible(): |
| 185 | xiaohongshu_logger.info(_msg("🥹", "页面仍然停留在登录二维码页，按 cookie 失效处理")) |
| 186 | return False |
| 187 | except Exception: |
| 188 | return False |
| 189 | |
| 190 | xiaohongshu_logger.success(_msg("🥳", "cookie 有效")) |
| 191 | return True |
| 192 | except Exception as exc: |
| 193 | xiaohongshu_logger.warning(_msg("😵", f"cookie 校验时出错，按失效处理: {exc}")) |
| 194 | return False |
| 195 | finally: |
| 196 | await browser.close() |
| 197 | |
| 198 | |
| 199 | async def xiaohongshu_setup( |
| 200 | account_file, |
| 201 | handle=False, |
| 202 | return_detail=False, |
| 203 | qrcode_callback=None, |
| 204 | headless: bool = LOCAL_CHROME_HEADLESS, |
| 205 | ): |
| 206 | if not os.path.exists(account_file) or not await cookie_auth(account_file): |
| 207 | if not handle: |
| 208 | result = _build_login_result(False, "cookie_invalid", "cookie文件不存在或已失效", account_file) |
| 209 | return result if return_detail else False |
| 210 | xiaohongshu_logger.info(_msg("🥹", "cookie 失效了，准备打开浏览器重新登录")) |
| 211 | result = await xiaohongshu_cookie_gen( |
| 212 | account_file, |
| 213 | qrcode_callback=qrcode_callback, |
| 214 | headless=headless, |
| 215 | ) |
| 216 | return result if return_detail else result["success"] |
| 217 | |
| 218 | result = _build_login_result(True, "cookie_valid", "cookie有效", account_file) |
| 219 | return result if return_detail else True |
| 220 | |
| 221 | |
| 222 | async def xiaohongshu_cookie_gen( |
| 223 | account_file, |
| 224 | qrcode_callback=None, |
| 225 | poll_interval: int = 3, |
| 226 | max_checks: int = 100, |
| 227 | headless: bool = LOCAL_CHROME_HEADLESS, |
| 228 | ): |
| 229 | if headless: |
| 230 | xiaohongshu_logger.info(_msg("🖼️", "小红书登录将以无头模式运行，小人会输出终端二维码并保存本地二维码图片")) |
| 231 | |
| 232 | account_path = Path(account_file) |
| 233 | account_path.parent.mkdir(parents=True, exist_ok=True) |
| 234 | |
| 235 | async with async_playwright() as playwright: |
| 236 | browser = await playwright.chromium.launch(headless=headless, channel="chromium") |
| 237 | context = await browser.new_context() |
| 238 | context = await set_init_script(context) |
| 239 | qrcode_path = None |
| 240 | qrcode_info = None |
| 241 | result = _build_login_result(False, "failed", "小红书登录失败", account_file) |
| 242 | try: |
| 243 | page = await context.new_page() |
| 244 | await page.goto(_build_xhs_creator_url("/login")) |
| 245 | qrcode_info = await _save_xhs_qrcode(page, account_file, qrcode_callback=qrcode_callback) |
| 246 | qrcode_path = Path(qrcode_info["image_path"]) |
| 247 | xiaohongshu_logger.info(_msg("🧍", "请扫码，小人正在耐心等待登录完成")) |
| 248 | |
| 249 | for _ in range(max_checks): |
| 250 | if await _is_xhs_login_completed(page): |
| 251 | await asyncio.sleep(2) |
| 252 | await context.storage_state(path=account_file) |
| 253 | if await cookie_auth(account_file): |
| 254 | xiaohongshu_logger.success(_msg("🥳", "小红书扫码登录成功，小人开心收工")) |
| 255 | result = _build_login_result(True, "success", "小红书扫码登录成功", account_file, qrcode_info, page.url) |
| 256 | else: |
| 257 | result = _build_login_result( |
| 258 | False, |
| 259 | "cookie_invalid", |
| 260 | "小红书扫码流程结束，但 cookie 校验失败", |
| 261 | account_file, |
| 262 | qrcode_info, |
| 263 | page.url, |
| 264 | ) |
| 265 | return result |
| 266 | |
| 267 | await asyncio.sleep(poll_interval) |
| 268 | |
| 269 | result = _build_login_result( |
| 270 | False, |
| 271 | "timeout", |
| 272 | "等待小红书扫码登录超时", |
| 273 | account_file, |
| 274 | qrcode_info, |
| 275 | page.url, |
| 276 | ) |
| 277 | except Exception as exc: |
| 278 | result = _build_login_result(False, "failed", str(exc), account_file, current_url=page.url if "page" in locals() else "") |
| 279 | finally: |
| 280 | if remove_qrcode_file(qrcode_path): |
| 281 | xiaohongshu_logger.info(_msg("🧹", f"临时二维码文件已清理: {qrcode_path}")) |
| 282 | if not result["success"]: |
| 283 | xiaohongshu_logger.error(_msg("😢", f"登录失败: {result['message']}")) |
| 284 | await context.close() |
| 285 | await browser.close() |
| 286 | return result |
| 287 | |
| 288 | |
| 289 | class XiaoHongShuBaseUploader(BaseVideoUploader): |
| 290 | def __init__( |
| 291 | self, |
| 292 | publish_date: datetime | int, |
| 293 | account_file, |
| 294 | publish_strategy: str = XIAOHONGSHU_PUBLISH_STRATEGY_IMMEDIATE, |
| 295 | debug: bool = DEBUG_MODE, |
| 296 | headless: bool = LOCAL_CHROME_HEADLESS, |
| 297 | ): |
| 298 | self.publish_date = publish_date |
| 299 | self.account_file = str(account_file) |
| 300 | self.publish_strategy = publish_strategy |
| 301 | self.debug = debug |
| 302 | self.date_format = "%Y年%m月%d日 %H:%M" |
| 303 | self.local_executable_path = LOCAL_CHROME_PATH |
| 304 | self.headless = headless |
| 305 | |
| 306 | async def validate_base_args(self): |
| 307 | if not os.path.exists(self.account_file): |
| 308 | raise RuntimeError(f"cookie文件不存在，请先完成小红书登录: {self.account_file}") |
| 309 | if not await cookie_auth(self.account_file): |
| 310 | raise RuntimeError(f"cookie文件已失效，请先完成小红书登录: {self.account_file}") |
| 311 | |
| 312 | if self.publish_strategy not in { |
| 313 | XIAOHONGSHU_PUBLISH_STRATEGY_IMMEDIATE, |
| 314 | XIAOHONGSHU_PUBLISH_STRATEGY_SCHEDULED, |
| 315 | }: |
| 316 | raise ValueError(f"不支持的发布策略: {self.publish_strategy}") |
| 317 | |
| 318 | if self.publish_strategy == XIAOHONGSHU_PUBLISH_STRATEGY_SCHEDULED: |
| 319 | self.publish_date = self.validate_publish_date(self.publish_date) |
| 320 | else: |
| 321 | self.publish_date = 0 |
| 322 | |
| 323 | async def set_schedule_time_xiaohongshu(self, page: Page, publish_date: datetime): |
| 324 | xiaohongshu_logger.info(_msg("🕒", f"小人准备设置定时发布时间: {publish_date.strftime(self.date_format)}")) |
| 325 | await page.locator('.custom-switch-card').filter(has_text="定时发布").locator('.d-switch').click() |
| 326 | await asyncio.sleep(1) |
| 327 | publish_date_hour = publish_date.strftime("%Y-%m-%d %H:%M") |
| 328 | time_input = page.locator('.d-datepicker-input-filter input.d-text') |
| 329 | await time_input.fill(str(publish_date_hour)) |
| 330 | await asyncio.sleep(1) |
| 331 | |
| 332 | async def set_location(self, page: Page, location: str = "青岛市"): |
| 333 | if not location: |
| 334 | return True |
| 335 | |
| 336 | xiaohongshu_logger.info(_msg("📍", f"小人准备设置位置: {location}")) |
| 337 | loc_ele = await page.wait_for_selector('div.d-text.d-select-placeholder.d-text-ellipsis.d-text-nowrap') |
| 338 | await loc_ele.click() |
| 339 | await page.wait_for_timeout(1000) |
| 340 | await page.keyboard.type(location) |
| 341 | dropdown_selector = 'div.d-popover.d-popover-default.d-dropdown.--size-min-width-large' |
| 342 | await page.wait_for_timeout(2000) |
| 343 | try: |
| 344 | await page.wait_for_selector(dropdown_selector, timeout=3000) |
| 345 | except Exception: |
| 346 | xiaohongshu_logger.warning(_msg("😵", "位置下拉列表没按预期出现，小人继续按旧逻辑查找")) |
| 347 | await page.wait_for_timeout(1000) |
| 348 | flexible_xpath = ( |
| 349 | f'//div[contains(@class, "d-popover") and contains(@class, "d-dropdown")]' |
| 350 | f'//div[contains(@class, "d-options-wrapper")]' |
| 351 | f'//div[contains(@class, "d-grid") and contains(@class, "d-options")]' |
| 352 | f'//div[contains(@class, "name") and text()="{location}"]' |
| 353 | ) |
| 354 | await page.wait_for_timeout(3000) |
| 355 | try: |
| 356 | location_option = await page.wait_for_selector( |
| 357 | flexible_xpath, |
| 358 | timeout=3000 |
| 359 | ) |
| 360 | |
| 361 | if not location_option: |
| 362 | location_option = await page.wait_for_selector( |
| 363 | f'//div[contains(@class, "d-popover") and contains(@class, "d-dropdown")]' |
| 364 | f'//div[contains(@class, "d-options-wrapper")]' |
| 365 | f'//div[contains(@class, "d-grid") and contains(@class, "d-options")]' |
| 366 | f'/div[1]//div[contains(@class, "name") and text()="{location}"]', |
| 367 | timeout=2000 |
| 368 | ) |
| 369 | |
| 370 | await location_option.scroll_into_view_if_needed() |
| 371 | await location_option.click() |
| 372 | xiaohongshu_logger.success(_msg("🥳", f"位置已经设置成 {location}")) |
| 373 | return True |
| 374 | except Exception as e: |
| 375 | xiaohongshu_logger.error(_msg("😢", f"设置位置失败: {e}")) |
| 376 | try: |
| 377 | all_options = await page.query_selector_all( |
| 378 | '//div[contains(@class, "d-popover") and contains(@class, "d-dropdown")]' |
| 379 | '//div[contains(@class, "d-options-wrapper")]' |
| 380 | '//div[contains(@class, "d-grid") and contains(@class, "d-options")]' |
| 381 | '/div' |
| 382 | ) |
| 383 | xiaohongshu_logger.debug(_msg("🧍", f"位置下拉里一共找到 {len(all_options)} 个选项")) |
| 384 | for i, option in enumerate(all_options[:3]): |
| 385 | option_text = await option.inner_text() |
| 386 | xiaohongshu_logger.debug(_msg("🧾", f"候选位置 {i + 1}: {option_text.strip()[:50]}")) |
| 387 | except Exception as inner_e: |
| 388 | xiaohongshu_logger.debug(_msg("😵", f"读取位置候选列表失败: {inner_e}")) |
| 389 | return False |
| 390 | |
| 391 | async def fill_title(self, page: Page) -> None: |
| 392 | title_container = page.locator('input[placeholder*="填写标题"]') |
| 393 | await title_container.fill(self.title[:20]) |
| 394 | |
| 395 | async def fill_desc(self, page: Page) -> None: |
| 396 | if not getattr(self, "desc", ""): |
| 397 | return |
| 398 | |
| 399 | desc = page.locator('p[data-placeholder*="输入正文描述"]') |
| 400 | await desc.click() |
| 401 | await page.keyboard.press("Backspace") |
| 402 | await page.keyboard.press("Control+KeyA") |
| 403 | await page.keyboard.press("Delete") |
| 404 | await page.keyboard.type(self.desc) |
| 405 | await page.keyboard.press("Enter") |
| 406 | |
| 407 | async def fill_tags(self, page: Page) -> None: |
| 408 | if not getattr(self, "tags", None): |
| 409 | return |
| 410 | |
| 411 | # 小红书标签上限为 10 个，超过会导致死循环卡住发布 |
| 412 | max_tags = 10 |
| 413 | if len(self.tags) > max_tags: |
| 414 | xiaohongshu_logger.warning( |
| 415 | _msg("🏷️", f"标签数量 {len(self.tags)} 超过小红书上限 {max_tags}，只取前 {max_tags} 个: {self.tags[:max_tags]}") |
| 416 | ) |
| 417 | self.tags = self.tags[:max_tags] |
| 418 | |
| 419 | if not getattr(self, "desc", ""): |
| 420 | desc = page.locator('p[data-placeholder*="输入正文描述"]') |
| 421 | await desc.click() |
| 422 | |
| 423 | for tag in self.tags: # 循环处理所有 tags |
| 424 | # 话题候选下拉框依赖小红书联想接口实时返回，网络抖动/无匹配时会等不到。 |
| 425 | # 标签是可选增强项：等不到候选框就跳过该标签继续，不让整条发布因此失败。 |
| 426 | try: |
| 427 | await page.keyboard.type("#" + tag, delay=30) |
| 428 | await page.locator('#creator-editor-topic-container').wait_for( |
| 429 | state="visible", |
| 430 | timeout=6000 |
| 431 | ) |
| 432 | first_item = page.locator('#creator-editor-topic-container .item').first |
| 433 | await first_item.wait_for(state="visible", timeout=4000) |
| 434 | await first_item.click() |
| 435 | except Exception as exc: |
| 436 | xiaohongshu_logger.warning( |
| 437 | _msg("🏷️", f"话题『{tag}』未出现候选，跳过该标签继续发布: {exc}") |
| 438 | ) |
| 439 | # 清掉已键入但未成词的 "#tag" 文本，避免它残留进正文 |
| 440 | for _ in range(len("#" + tag)): |
| 441 | await page.keyboard.press("Backspace") |
| 442 | continue |
| 443 | |
| 444 | async def fill_meta(self, page: Page) -> None: |
| 445 | await self.fill_title(page) |
| 446 | await self.fill_desc(page) |
| 447 | await self.fill_tags(page) |
| 448 | |
| 449 | async def check_original_declaration(self, page: Page) -> None: |
| 450 | """勾选原创声明（如果页面上有的话）""" |
| 451 | try: |
| 452 | # 小红书的原创声明通常是 checkbox 或 switch 组件 |
| 453 | original_checkbox = page.locator('div.original-declaration checkbox, div.original-declaration input[type="checkbox"], label:has-text("原创") input[type="checkbox"]').first |
| 454 | if await original_checkbox.count() and not await original_checkbox.is_checked(): |
| 455 | await original_checkbox.check() |
| 456 | xiaohongshu_logger.success(_msg("✅", "原创声明已勾选")) |
| 457 | return |
| 458 | |
| 459 | # 尝试通过文本匹配找到原创声明区域并点击 |
| 460 | original_text = page.locator('div:has-text("原创声明"), span:has-text("原创声明"), div:has-text("原创"), label:has-text("原创")').first |
| 461 | if await original_text.count(): |
| 462 | await original_text.click() |
| 463 | xiaohongshu_logger.success(_msg("✅", "原创声明已勾选")) |
| 464 | return |
| 465 | |
| 466 | xiaohongshu_logger.info(_msg("🧾", "未发现原创声明选项，跳过")) |
| 467 | except Exception as exc: |
| 468 | xiaohongshu_logger.warning(_msg("⚠️", f"勾选原创声明时出错，跳过: {exc}")) |
| 469 | |
| 470 | |
| 471 | class XiaoHongShuVideo(XiaoHongShuBaseUploader): |
| 472 | def __init__( |
| 473 | self, |
| 474 | title, |
| 475 | file_path, |
| 476 | tags, |
| 477 | publish_date: datetime | int, |
| 478 | account_file, |
| 479 | thumbnail_path=None, |
| 480 | desc: str | None = None, |
| 481 | publish_strategy: str = XIAOHONGSHU_PUBLISH_STRATEGY_IMMEDIATE, |
| 482 | debug: bool = DEBUG_MODE, |
| 483 | headless: bool = LOCAL_CHROME_HEADLESS, |
| 484 | ): |
| 485 | super().__init__( |
| 486 | publish_date=publish_date, |
| 487 | account_file=account_file, |
| 488 | publish_strategy=publish_strategy, |
| 489 | debug=debug, |
| 490 | headless=headless, |
| 491 | ) |
| 492 | self.title = title |
| 493 | self.file_path = file_path |
| 494 | self.tags = tags or [] |
| 495 | self.thumbnail_path = thumbnail_path |
| 496 | self.desc = desc or "" |
| 497 | |
| 498 | async def validate_upload_args(self): |
| 499 | await self.validate_base_args() |
| 500 | if not self.title or not str(self.title).strip(): |
| 501 | raise ValueError("视频模式下，title 是必须的") |
| 502 | |
| 503 | self.file_path = str(self.validate_video_file(self.file_path)) |
| 504 | if self.thumbnail_path: |
| 505 | self.thumbnail_path = str(self.validate_image_file(self.thumbnail_path)) |
| 506 | |
| 507 | async def handle_upload_error(self, page: Page): |
| 508 | xiaohongshu_logger.warning(_msg("😵", "视频上传摔了一跤，小人马上重新上传")) |
| 509 | await page.locator('div.progress-div [class^="upload-btn-input"]').set_input_files(self.file_path) |
| 510 | |
| 511 | async def set_thumbnail(self, page: Page, thumbnail_path: str): |
| 512 | if not thumbnail_path: |
| 513 | return |
| 514 | |
| 515 | xiaohongshu_logger.info(_msg("🖼️", "小人准备设置封面")) |
| 516 | |
| 517 | cover_plugin_title = page.locator("div.cover-plugin-title").filter(has_text="设置封面") |
| 518 | cover_upload_dialog = cover_plugin_title.locator( |
| 519 | "xpath=ancestor::div[contains(@class, 'cover-plugin-preview')]" |
| 520 | ).locator("div.cover > div.default:visible") |
| 521 | await cover_upload_dialog.wait_for(state="visible", timeout=30000) |
| 522 | |
| 523 | await cover_upload_dialog.click(force=True) |
| 524 | |
| 525 | modal = page.locator("div.d-modal.cover-modal") |
| 526 | await modal.wait_for(state="visible", timeout=30000) |
| 527 | |
| 528 | file_input = modal.locator('input[type="file"][accept*="image"]').first |
| 529 | await file_input.wait_for(state="attached", timeout=10000) |
| 530 | await file_input.set_input_files(thumbnail_path) |
| 531 | await page.wait_for_timeout(2000) |
| 532 | |
| 533 | confirm_button = modal.locator("button.mojito-button").filter(has_text="确定").first |
| 534 | await confirm_button.wait_for(state="visible", timeout=10000) |
| 535 | await confirm_button.click() |
| 536 | |
| 537 | await modal.wait_for(state="hidden", timeout=30000) |
| 538 | xiaohongshu_logger.success(_msg("🥳", "封面已经设置完成")) |
| 539 | |
| 540 | async def upload_video_content(self, page: Page) -> None: |
| 541 | xiaohongshu_logger.info(_msg("🏃", f"小人开始搬运视频: {self.title}.mp4")) |
| 542 | xiaohongshu_logger.info(_msg("🧭", "小人正在赶往视频发布页")) |
| 543 | publish_url = _build_xhs_creator_url( |
| 544 | "/publish/publish?from=homepage&target=video" |
| 545 | ) |
| 546 | await page.goto(publish_url) |
| 547 | await page.wait_for_url(publish_url) |
| 548 | await page.locator("div[class^='upload-content'] input[class='upload-input']").set_input_files(self.file_path) |
| 549 | |
| 550 | while True: |
| 551 | try: |
| 552 | upload_input = await page.wait_for_selector('input.upload-input', timeout=3000) |
| 553 | preview_new = await upload_input.query_selector( |
| 554 | 'xpath=following-sibling::div[contains(@class, "preview-new")]') |
| 555 | if preview_new: |
| 556 | # 获取整个预览区域的文本，更鲁棒地判断上传状态 |
| 557 | all_text = await preview_new.inner_text() |
| 558 | upload_success = any(keyword in all_text for keyword in ['上传成功', '分辨率', '重新上传', '编辑封面', '已上传', '已选择', '100%']) |
| 559 | |
| 560 | if not upload_success: |
| 561 | # 检查是否有特定的状态码或百分比 |
| 562 | stage_elements = await preview_new.query_selector_all('div.stage') |
| 563 | for stage in stage_elements: |
| 564 | text_content = await page.evaluate('(element) => element.textContent', stage) |
| 565 | if '上传成功' in text_content or '分辨率' in text_content: |
| 566 | upload_success = True |
| 567 | break |
| 568 | |
| 569 | if upload_success: |
| 570 | xiaohongshu_logger.success(_msg("🥳", "视频已经传完啦")) |
| 571 | break |
| 572 | |
| 573 | if self.debug: |
| 574 | normalized_text = all_text.strip().replace("\n", " ") |
| 575 | xiaohongshu_logger.debug(_msg("🧍", f"预览区域内容: {normalized_text}")) |
| 576 | xiaohongshu_logger.debug(_msg("🧍", "还没看到上传成功标识，小人继续等一会")) |
| 577 | else: |
| 578 | # 尝试检查标题输入框是否已经出现，如果是，说明已经进入编辑状态 |
| 579 | title_container = page.locator('input[placeholder*="填写标题"]') |
| 580 | if await title_container.count() > 0 and await title_container.is_visible(): |
| 581 | xiaohongshu_logger.success(_msg("🥳", "虽然没看到预览区，但标题框出来了，小人继续")) |
| 582 | break |
| 583 | xiaohongshu_logger.debug(_msg("🧍", "还没拿到预览区域，小人继续等一会")) |
| 584 | except Exception as e: |
| 585 | xiaohongshu_logger.debug(_msg("😵", f"上传状态还没稳定下来，小人继续观察: {e}")) |
| 586 | await asyncio.sleep(2) |
| 587 | |
| 588 | xiaohongshu_logger.info(_msg("✍️", "小人开始填标题、描述和话题")) |
| 589 | await self.fill_meta(page) |
| 590 | |
| 591 | await self.set_thumbnail(page, self.thumbnail_path) |
| 592 | |
| 593 | # await self.set_location(page, "青岛市") |
| 594 | |
| 595 | await self.check_original_declaration(page) |
| 596 | |
| 597 | if self.publish_strategy == XIAOHONGSHU_PUBLISH_STRATEGY_SCHEDULED and self.publish_date != 0: |
| 598 | await self.set_schedule_time_xiaohongshu(page, self.publish_date) |
| 599 | |
| 600 | while True: |
| 601 | try: |
| 602 | if self.publish_strategy == XIAOHONGSHU_PUBLISH_STRATEGY_SCHEDULED: |
| 603 | await page.locator('button:has-text("定时发布")').click() |
| 604 | else: |
| 605 | await page.locator('button:has-text("发布")').click() |
| 606 | await page.wait_for_url( |
| 607 | XHS_PUBLISH_SUCCESS_URL_PATTERN, |
| 608 | timeout=3000 |
| 609 | ) |
| 610 | xiaohongshu_logger.success(_msg("🥳", "视频发布成功，小人开心收工")) |
| 611 | break |
| 612 | except Exception: |
| 613 | xiaohongshu_logger.info(_msg("🏃", "小人正在冲刺发布视频")) |
| 614 | if self.debug: |
| 615 | await page.screenshot(full_page=True) |
| 616 | await asyncio.sleep(0.5) |
| 617 | |
| 618 | async def upload(self, playwright: Playwright) -> None: |
| 619 | xiaohongshu_logger.info(_msg("🧍", "小人先检查 cookie、视频文件、封面和发布时间")) |
| 620 | await self.validate_upload_args() |
| 621 | xiaohongshu_logger.info(_msg("🥳", "上传前检查通过")) |
| 622 | browser = await playwright.chromium.launch(headless=self.headless, channel="chromium") |
| 623 | context = await browser.new_context( |
| 624 | permissions=["geolocation"], |
| 625 | storage_state=self.account_file, |
| 626 | ) |
| 627 | context = await set_init_script(context) |
| 628 | |
| 629 | try: |
| 630 | page = await context.new_page() |
| 631 | await self.upload_video_content(page) |
| 632 | await context.storage_state(path=self.account_file) |
| 633 | xiaohongshu_logger.success(_msg("🥳", "cookie 更新完毕")) |
| 634 | finally: |
| 635 | await context.close() |
| 636 | await browser.close() |
| 637 | |
| 638 | async def xiaohongshu_upload_video(self): |
| 639 | async with async_playwright() as playwright: |
| 640 | await self.upload(playwright) |
| 641 | |
| 642 | async def main(self): |
| 643 | await self.xiaohongshu_upload_video() |
| 644 | |
| 645 | |
| 646 | class XiaoHongShuNote(XiaoHongShuBaseUploader): |
| 647 | def __init__( |
| 648 | self, |
| 649 | image_paths, |
| 650 | note, |
| 651 | tags, |
| 652 | publish_date: datetime | int, |
| 653 | account_file, |
| 654 | title: str | None = None, |
| 655 | desc: str | None = None, |
| 656 | publish_strategy: str = XIAOHONGSHU_PUBLISH_STRATEGY_IMMEDIATE, |
| 657 | debug: bool = DEBUG_MODE, |
| 658 | headless: bool = LOCAL_CHROME_HEADLESS, |
| 659 | ): |
| 660 | super().__init__( |
| 661 | publish_date=publish_date, |
| 662 | account_file=account_file, |
| 663 | publish_strategy=publish_strategy, |
| 664 | debug=debug, |
| 665 | headless=headless, |
| 666 | ) |
| 667 | self.image_paths = image_paths |
| 668 | self.note = note or "" |
| 669 | self.tags = tags or [] |
| 670 | self.desc = desc if desc is not None else self.note |
| 671 | self.title = title or ((self.desc or self.note)[:20] if (self.desc or self.note) else "") |
| 672 | |
| 673 | async def validate_upload_args(self): |
| 674 | await self.validate_base_args() |
| 675 | if not self.image_paths: |
| 676 | raise ValueError("图文模式下，图片是必须的") |
| 677 | if not self.title or not str(self.title).strip(): |
| 678 | raise ValueError("图文模式下，title 是必须的") |
| 679 | |
| 680 | if isinstance(self.image_paths, (str, Path)): |
| 681 | self.image_paths = [self.image_paths] |
| 682 | |
| 683 | normalized_image_paths = [] |
| 684 | for image_path in self.image_paths: |
| 685 | normalized_image_paths.append(str(self.validate_image_file(image_path))) |
| 686 | self.image_paths = normalized_image_paths |
| 687 | |
| 688 | async def upload_note_content(self, page: Page) -> None: |
| 689 | xiaohongshu_logger.info(_msg("🏃", f"小人开始搬运图文，共 {len(self.image_paths)} 张图片")) |
| 690 | xiaohongshu_logger.info(_msg("🧭", "小人正在赶往图文发布页")) |
| 691 | publish_url = _build_xhs_creator_url( |
| 692 | "/publish/publish?from=homepage&target=image" |
| 693 | ) |
| 694 | await page.goto(publish_url) |
| 695 | await page.wait_for_url(publish_url) |
| 696 | |
| 697 | upload_input = page.locator('input[type="file"][accept*="image"]').first |
| 698 | if not await upload_input.count(): |
| 699 | upload_input = page.locator("div[class^='upload-content'] input[class='upload-input']").first |
| 700 | |
| 701 | await upload_input.wait_for(state="attached", timeout=30000) |
| 702 | xiaohongshu_logger.info(_msg("📤", "小人正在上传图片")) |
| 703 | await upload_input.set_input_files(self.image_paths) |
| 704 | |
| 705 | while True: |
| 706 | try: |
| 707 | title_container = page.locator('input[placeholder*="填写标题"]').first |
| 708 | await title_container.wait_for(state="visible", timeout=3000) |
| 709 | xiaohongshu_logger.success(_msg("🥳", "图文素材已经传完，可以开始填写内容了")) |
| 710 | break |
| 711 | except Exception: |
| 712 | xiaohongshu_logger.debug(_msg("🧍", "图文素材还在上传，小人继续等一会")) |
| 713 | await asyncio.sleep(1) |
| 714 | |
| 715 | xiaohongshu_logger.info(_msg("✍️", "小人开始填标题、描述和话题")) |
| 716 | await self.fill_meta(page) |
| 717 | |
| 718 | await self.check_original_declaration(page) |
| 719 | |
| 720 | if self.publish_strategy == XIAOHONGSHU_PUBLISH_STRATEGY_SCHEDULED and self.publish_date != 0: |
| 721 | await self.set_schedule_time_xiaohongshu(page, self.publish_date) |
| 722 | |
| 723 | while True: |
| 724 | try: |
| 725 | if self.publish_strategy == XIAOHONGSHU_PUBLISH_STRATEGY_SCHEDULED: |
| 726 | await page.locator('button:has-text("定时发布")').click() |
| 727 | else: |
| 728 | await page.locator('button:has-text("发布")').click() |
| 729 | await page.wait_for_url( |
| 730 | XHS_PUBLISH_SUCCESS_URL_PATTERN, |
| 731 | timeout=3000 |
| 732 | ) |
| 733 | xiaohongshu_logger.success(_msg("🥳", "图文发布成功，小人开心收工")) |
| 734 | break |
| 735 | except Exception: |
| 736 | xiaohongshu_logger.info(_msg("🏃", "小人正在冲刺发布图文")) |
| 737 | if self.debug: |
| 738 | await page.screenshot(full_page=True) |
| 739 | await asyncio.sleep(0.5) |
| 740 | |
| 741 | async def upload(self, playwright: Playwright) -> None: |
| 742 | xiaohongshu_logger.info(_msg("🧍", "小人先检查 cookie、图片和发布时间")) |
| 743 | await self.validate_upload_args() |
| 744 | xiaohongshu_logger.info(_msg("🥳", "图文上传前检查通过")) |
| 745 | browser = await playwright.chromium.launch(headless=self.headless, channel="chromium") |
| 746 | context = await browser.new_context( |
| 747 | permissions=["geolocation"], |
| 748 | storage_state=self.account_file, |
| 749 | ) |
| 750 | context = await set_init_script(context) |
| 751 | |
| 752 | try: |
| 753 | page = await context.new_page() |
| 754 | await self.upload_note_content(page) |
| 755 | await context.storage_state(path=self.account_file) |
| 756 | xiaohongshu_logger.success(_msg("🥳", "cookie 更新完毕")) |
| 757 | finally: |
| 758 | await context.close() |
| 759 | await browser.close() |
| 760 | |
| 761 | async def xiaohongshu_upload_note(self): |
| 762 | async with async_playwright() as playwright: |
| 763 | await self.upload(playwright) |
| 764 | |
| 765 | async def main(self): |
| 766 | await self.xiaohongshu_upload_note() |
| 767 |