feat: add search result video crawling

Douyin.py

@@ -12,6 +12,7 @@ from __future__ import annotations
 
 import argparse
 import json
+import random
 import re
 import socket
 import sys
@@ -19,6 +20,7 @@ import time
 from dataclasses import dataclass
 from pathlib import Path
 from typing import Any
+from urllib.parse import quote
 
 DEFAULT_USER_URL = (
     "https://www.douyin.com/user/"
@@ -29,6 +31,8 @@ DEFAULT_BROWSER_PORT = 9223
 LISTEN_TARGET = "web/aweme/post/"
 RECOMMENDATION_LISTEN_TARGET = "aweme/v2/web/module/feed/"
 SINGLE_VIDEO_LISTEN_TARGET = "web/aweme/detail/"
+SEARCH_LISTEN_TARGET = "aweme/v1/web/general/search/single"
+MAX_FILENAME_BYTES = 240
 INVALID_FILENAME_CHARS = re.compile(r'[\\/:*?"<>|\r\n\t]')
 RECOMMENDATION_URL_PATTERN = re.compile(r"^https?://www\.douyin\.com/?(?:jingxuan)?(?:\?.*)?$")
 CREATOR_URL_PATTERN = re.compile(r"^https?://www\.douyin\.com/user/[^/?#]+(?:\?.*)?$")
@@ -44,11 +48,48 @@ class ResolvedTarget:
     aweme_id: str | None = None
 
 
+@dataclass(frozen=True)
+class ScrollSettings:
+    mode: str = "human"
+    min_wait: float = 2.0
+    max_wait: float = 8.0
+    reverse_scroll_probability: float = 0.2
+    max_runtime: float = 600.0
+    min_scroll: int = 300
+    max_scroll: int = 900
+    min_reverse_scroll: int = 80
+    max_reverse_scroll: int = 250
+
+
+@dataclass(frozen=True)
+class HumanScrollPlan:
+    down_distance: int
+    down_wait: float
+    reverse_distance: int = 0
+    reverse_wait: float = 0.0
+    settle_wait: float = 0.0
+
+
 def sanitize_filename(value: str, fallback: str = "untitled") -> str:
     cleaned = INVALID_FILENAME_CHARS.sub("_", value).strip(" ._")
     return cleaned or fallback
 
 
+def truncate_utf8_bytes(value: str, max_bytes: int) -> str:
+    if len(value.encode("utf-8")) <= max_bytes:
+        return value
+
+    result = ""
+    used = 0
+    for character in value:
+        character_bytes = len(character.encode("utf-8"))
+        if used + character_bytes > max_bytes:
+            break
+        result += character
+        used += character_bytes
+    return result.rstrip(" ._")
+
+
 def is_recommendation_url(value: str) -> bool:
     return bool(RECOMMENDATION_URL_PATTERN.match(value.strip()))
 
@@ -76,6 +117,10 @@ def build_video_page_url(aweme_id: str) -> str:
     return f"https://www.douyin.com/video/{aweme_id}"
 
 
+def build_search_page_url(keyword: str) -> str:
+    return f"https://www.douyin.com/search/{quote(keyword)}?type=general"
+
+
 def parse_target_input(value: str, source: str) -> ResolvedTarget:
     normalized = value.strip()
     if is_recommendation_url(normalized):
@@ -139,6 +184,40 @@ def choose_video_url(url_list: list[str]) -> str:
     raise ValueError("url_list 为空，无法选择视频地址。")
 
 
+def extract_url_list_from_play_addr(play_addr: Any) -> list[str]:
+    if not isinstance(play_addr, dict):
+        return []
+
+    url_list = play_addr.get("url_list") or []
+    if not isinstance(url_list, list):
+        return []
+
+    return [str(url) for url in url_list if str(url).strip()]
+
+
+def extract_video_url_list(video: Any) -> list[str]:
+    if not isinstance(video, dict):
+        return []
+
+    for address_key in ("play_addr", "play_addr_h264", "play_addr_lowbr"):
+        url_list = extract_url_list_from_play_addr(video.get(address_key))
+        if url_list:
+            return url_list
+
+    bit_rate_list = video.get("bit_rate") or []
+    if not isinstance(bit_rate_list, list):
+        return []
+
+    for bit_rate in bit_rate_list:
+        if not isinstance(bit_rate, dict):
+            continue
+        url_list = extract_url_list_from_play_addr(bit_rate.get("play_addr"))
+        if url_list:
+            return url_list
+
+    return []
+
+
 def build_output_path(
     title: str, 
     video_id: str, 
@@ -146,11 +225,20 @@ def build_output_path(
     author_name: str | None = None,
 ) -> Path:
     safe_title = sanitize_filename(title, fallback="untitled")
+    suffix = f"-{video_id}.mp4"
     if author_name:
         safe_author = sanitize_filename(author_name, fallback="unknown")
-        filename = f"[{safe_author}]{safe_title}-{video_id}.mp4"
+        prefix = f"[{safe_author}]"
     else:
-        filename = f"{safe_title}-{video_id}.mp4"
+        prefix = ""
+
+    title_budget = MAX_FILENAME_BYTES - len(prefix.encode("utf-8")) - len(suffix.encode("utf-8"))
+    if title_budget < 1:
+        prefix_budget = MAX_FILENAME_BYTES - len(suffix.encode("utf-8")) - 1
+        prefix = truncate_utf8_bytes(prefix, max(1, prefix_budget))
+        title_budget = MAX_FILENAME_BYTES - len(prefix.encode("utf-8")) - len(suffix.encode("utf-8"))
+
+    filename = f"{prefix}{truncate_utf8_bytes(safe_title, max(1, title_budget))}{suffix}"
     return output_dir / filename
 
 
@@ -200,8 +288,7 @@ def parse_aweme_items(body: Any) -> list[dict[str, str]]:
             continue
 
         video = aweme.get("video") or {}
-        play_addr = video.get("play_addr") or {}
+        url_list = extract_video_url_list(video)
-        url_list = play_addr.get("url_list") or []
         if not url_list:
             continue
 
@@ -219,7 +306,7 @@ def parse_aweme_items(body: Any) -> list[dict[str, str]]:
             {
                 "title": title,
                 "video_id": video_id,
-                "video_url": choose_video_url([str(url) for url in url_list]),
+                "video_url": choose_video_url(url_list),
                 "author_name": author_name,
                 "author_id": author_id,
             }
@@ -244,6 +331,25 @@ def parse_single_aweme_item(body: Any) -> dict[str, str]:
     raise ValueError("接口响应中缺少可下载的单视频数据。")
 
 
+def parse_search_items(body: Any) -> list[dict[str, str]]:
+    if not isinstance(body, dict):
+        raise ValueError("接口响应不是字典，无法解析。")
+
+    data = body.get("data")
+    if not isinstance(data, list):
+        raise ValueError("搜索接口响应中缺少 data。")
+
+    aweme_list = []
+    for entry in data:
+        if not isinstance(entry, dict):
+            continue
+        aweme_info = entry.get("aweme_info")
+        if isinstance(aweme_info, dict):
+            aweme_list.append(aweme_info)
+
+    return parse_aweme_items({"aweme_list": aweme_list})
+
+
 def build_headers(referer: str) -> dict[str, str]:
     return {
         "referer": referer,
@@ -285,7 +391,8 @@ def create_page(chromium_page_cls: Any, chromium_options_cls: Any, browser_port:
 
 def wait_for_aweme_packet(page: Any, timeout: int) -> Any | None:
     try:
-        return page.listen.wait(timeout=timeout)
+        packet = page.listen.wait(timeout=timeout)
+        return packet if packet else None
     except Exception as exc:
         print(f"[WARN] 等待接口数据超时或失败: {exc}")
         return None
@@ -296,6 +403,95 @@ def scroll_to_next_page(page: Any) -> None:
     time.sleep(2)
 
 
+def create_human_scroll_plan(
+    settings: ScrollSettings,
+    random_module: Any = random,
+) -> HumanScrollPlan:
+    down_distance = random_module.randint(settings.min_scroll, settings.max_scroll)
+    down_wait = random_module.uniform(settings.min_wait, settings.max_wait)
+    settle_wait = random_module.uniform(settings.min_wait, settings.max_wait)
+
+    reverse_distance = 0
+    reverse_wait = 0.0
+    if random_module.random() < settings.reverse_scroll_probability:
+        reverse_distance = random_module.randint(
+            settings.min_reverse_scroll,
+            settings.max_reverse_scroll,
+        )
+        reverse_wait = random_module.uniform(1.0, min(3.0, settings.max_wait))
+
+    return HumanScrollPlan(
+        down_distance=down_distance,
+        down_wait=down_wait,
+        reverse_distance=reverse_distance,
+        reverse_wait=reverse_wait,
+        settle_wait=settle_wait,
+    )
+
+
+def run_scroll_step(page: Any, distance: int) -> bool:
+    script = f"""
+const distance = {distance};
+function findMainScrollContainer() {{
+  const preferredSelectors = ['.tKqwmYAX', '.route-scroll-container', '.semi-tabs-content'];
+  for (const selector of preferredSelectors) {{
+    const el = document.querySelector(selector);
+    if (el && el.scrollHeight > el.clientHeight + 20) {{
+      return el;
+    }}
+  }}
+
+  const candidates = Array.from(document.querySelectorAll('*'))
+    .filter((el) => {{
+      const rect = el.getBoundingClientRect();
+      return rect.width > 300
+        && rect.height > 200
+        && el.scrollHeight > el.clientHeight + 20;
+    }})
+    .sort((a, b) => {{
+      const areaA = a.getBoundingClientRect().width * a.getBoundingClientRect().height;
+      const areaB = b.getBoundingClientRect().width * b.getBoundingClientRect().height;
+      return areaB - areaA;
+    }});
+
+  return candidates[0] || null;
+}}
+
+const scrollTarget = findMainScrollContainer();
+if (scrollTarget) {{
+  scrollTarget.scrollBy(0, distance);
+  return true;
+}}
+return false;
+"""
+    scrolled_container = bool(page.run_js(script))
+    if not scrolled_container:
+        page.run_js(f"window.scrollBy(0, {distance});")
+    return scrolled_container
+
+
+def run_human_scroll_sequence(page: Any, plan: HumanScrollPlan) -> None:
+    run_scroll_step(page, plan.down_distance)
+    print(f"[INFO] 向下滚动 {plan.down_distance}px，停留 {plan.down_wait:.1f}s")
+    time.sleep(plan.down_wait)
+
+    if plan.reverse_distance > 0:
+        run_scroll_step(page, -plan.reverse_distance)
+        print(f"[INFO] 小幅回滚 {plan.reverse_distance}px，停留 {plan.reverse_wait:.1f}s")
+        time.sleep(plan.reverse_wait)
+        forward_distance = plan.reverse_distance * 2
+        run_scroll_step(page, forward_distance)
+
+    if plan.settle_wait > 0:
+        print(f"[INFO] 继续停留 {plan.settle_wait:.1f}s")
+        time.sleep(plan.settle_wait)
+
+
+def human_like_scroll(page: Any, settings: ScrollSettings | None = None) -> None:
+    scroll_settings = settings or ScrollSettings()
+    run_human_scroll_sequence(page, create_human_scroll_plan(scroll_settings))
+
+
 def download_video(
     requests_module: Any,
     headers: dict[str, str],
@@ -393,6 +589,7 @@ def collect_recommendations(
     timeout: int,
     output_dir: Path,
     browser_port: int | None,
+    scroll_settings: ScrollSettings | None = None,
 ) -> int:
     requests_module, chromium_page_cls, chromium_options_cls = import_runtime_dependencies()
     headers = build_headers("https://www.douyin.com/")
@@ -408,16 +605,22 @@ def collect_recommendations(
     downloaded = 0
     seen_ids: set[str] = set()
     consecutive_empty = 0
-    max_consecutive_empty = 3
+    max_consecutive_empty = 6
+    settings = scroll_settings or ScrollSettings()
+    started_at = time.monotonic()
 
     while downloaded < max_videos:
+        if settings.max_runtime > 0 and time.monotonic() - started_at >= settings.max_runtime:
+            print("[INFO] 已达到最大运行时间，结束抓取。")
+            break
+
         packet = wait_for_aweme_packet(page, timeout=timeout)
         if packet is None:
             consecutive_empty += 1
             if consecutive_empty >= max_consecutive_empty:
                 print("[INFO] 连续多次未获取到新数据，结束抓取。")
                 break
-            scroll_to_next_page(page)
+            human_like_scroll(page, settings=settings)
             continue
 
         try:
@@ -428,14 +631,14 @@ def collect_recommendations(
             consecutive_empty += 1
             if consecutive_empty >= max_consecutive_empty:
                 break
-            scroll_to_next_page(page)
+            human_like_scroll(page, settings=settings)
             continue
 
         if not items:
             consecutive_empty += 1
             if consecutive_empty >= max_consecutive_empty:
                 break
-            scroll_to_next_page(page)
+            human_like_scroll(page, settings=settings)
             continue
 
         consecutive_empty = 0
@@ -476,7 +679,109 @@ def collect_recommendations(
             if consecutive_empty >= max_consecutive_empty:
                 break
 
-        scroll_to_next_page(page)
+        human_like_scroll(page, settings=settings)
+
+    return downloaded
+
+
+def collect_search_results(
+    keyword: str,
+    max_videos: int,
+    timeout: int,
+    output_dir: Path,
+    browser_port: int | None,
+    scroll_settings: ScrollSettings | None = None,
+) -> int:
+    requests_module, chromium_page_cls, chromium_options_cls = import_runtime_dependencies()
+    search_url = build_search_page_url(keyword)
+    headers = build_headers(search_url)
+    if browser_port is not None:
+        ensure_browser_debug_port_ready(browser_port)
+    page = create_page(chromium_page_cls, chromium_options_cls, browser_port)
+    page.listen.start(SEARCH_LISTEN_TARGET)
+
+    print(f"[INFO] 正在打开抖音搜索页：{keyword}。若出现登录或验证码，请先在浏览器窗口里完成。")
+    page.get(search_url)
+    time.sleep(3)
+
+    downloaded = 0
+    seen_ids: set[str] = set()
+    consecutive_empty = 0
+    max_consecutive_empty = 6
+    settings = scroll_settings or ScrollSettings()
+    started_at = time.monotonic()
+
+    while downloaded < max_videos:
+        if settings.max_runtime > 0 and time.monotonic() - started_at >= settings.max_runtime:
+            print("[INFO] 已达到最大运行时间，结束抓取。")
+            break
+
+        packet = wait_for_aweme_packet(page, timeout=timeout)
+        if packet is None:
+            consecutive_empty += 1
+            if consecutive_empty >= max_consecutive_empty:
+                print("[INFO] 连续多次未获取到新搜索数据，结束抓取。")
+                break
+            human_like_scroll(page, settings=settings)
+            continue
+
+        try:
+            payload = extract_aweme_payload(packet.response)
+            items = parse_search_items(payload)
+        except Exception as exc:
+            print(f"[WARN] 解析搜索接口数据失败: {exc}")
+            consecutive_empty += 1
+            if consecutive_empty >= max_consecutive_empty:
+                break
+            human_like_scroll(page, settings=settings)
+            continue
+
+        if not items:
+            consecutive_empty += 1
+            if consecutive_empty >= max_consecutive_empty:
+                break
+            human_like_scroll(page, settings=settings)
+            continue
+
+        consecutive_empty = 0
+        new_items_in_batch = 0
+
+        for item in items:
+            if item["video_id"] in seen_ids:
+                continue
+
+            if downloaded >= max_videos:
+                break
+
+            seen_ids.add(item["video_id"])
+            output_path = build_output_path(
+                title=item["title"],
+                video_id=item["video_id"],
+                output_dir=output_dir,
+                author_name=item.get("author_name"),
+            )
+
+            try:
+                download_video(
+                    requests_module=requests_module,
+                    headers=headers,
+                    video_url=item["video_url"],
+                    output_path=output_path,
+                )
+            except Exception as exc:
+                print(f"[WARN] 下载失败 {item['video_id']}: {exc}")
+                continue
+
+            downloaded += 1
+            new_items_in_batch += 1
+            print(f"[OK] 已保存: {output_path}")
+
+        if new_items_in_batch == 0:
+            consecutive_empty += 1
+            if consecutive_empty >= max_consecutive_empty:
+                break
+
+        human_like_scroll(page, settings=settings)
 
     return downloaded
 
@@ -554,6 +859,41 @@ def build_parser() -> argparse.ArgumentParser:
         default=50,
         help="推荐流最大抓取数量，默认 50",
     )
+    parser.add_argument(
+        "--search-keyword",
+        default=None,
+        help="搜索关键词；提供后抓取搜索结果页视频",
+    )
+    parser.add_argument(
+        "--scroll-mode",
+        choices=["human"],
+        default="human",
+        help="推荐流滚动模式，默认 human",
+    )
+    parser.add_argument(
+        "--min-wait",
+        type=float,
+        default=2.0,
+        help="推荐流每次滚动后的最短等待秒数，默认 2",
+    )
+    parser.add_argument(
+        "--max-wait",
+        type=float,
+        default=8.0,
+        help="推荐流每次滚动后的最长等待秒数，默认 8",
+    )
+    parser.add_argument(
+        "--reverse-scroll-probability",
+        type=float,
+        default=0.2,
+        help="推荐流小幅回滚概率，取值 0 到 1，默认 0.2",
+    )
+    parser.add_argument(
+        "--max-runtime",
+        type=float,
+        default=600.0,
+        help="推荐流最大运行秒数，默认 600；设置为 0 表示不限制",
+    )
     return parser
 
 
@@ -569,34 +909,61 @@ def main(argv: list[str] | None = None) -> int:
         parser.error("--browser-port 必须大于 0")
     if args.max_videos <= 0:
         parser.error("--max-videos 必须大于 0")
+    if args.min_wait < 0:
+        parser.error("--min-wait 不能小于 0")
+    if args.max_wait < args.min_wait:
+        parser.error("--max-wait 必须大于或等于 --min-wait")
+    if not 0 <= args.reverse_scroll_probability <= 1:
+        parser.error("--reverse-scroll-probability 必须在 0 到 1 之间")
+    if args.max_runtime < 0:
+        parser.error("--max-runtime 不能小于 0")
+
+    scroll_settings = ScrollSettings(
+        mode=args.scroll_mode,
+        min_wait=args.min_wait,
+        max_wait=args.max_wait,
+        reverse_scroll_probability=args.reverse_scroll_probability,
+        max_runtime=args.max_runtime,
+    )
 
     try:
-        target = resolve_cli_target(args.target, browser_port=args.browser_port)
+        if args.search_keyword:
-        if target.kind == "creator":
+            total = collect_search_results(
-            total = collect_videos(
+                keyword=args.search_keyword,
-                user_url=target.value,
-                max_pages=args.pages,
-                timeout=args.timeout,
-                output_dir=Path(args.output_dir),
-                browser_port=args.browser_port,
-                auto_scroll=args.pages > 1,
-            )
-        elif target.kind == "recommendation":
-            total = collect_recommendations(
                 max_videos=args.max_videos,
                 timeout=args.timeout,
                 output_dir=Path(args.output_dir),
                 browser_port=args.browser_port,
-            )
+                scroll_settings=scroll_settings,
-        elif target.kind == "single-video":
-            total = collect_single_video(
-                target=target,
-                timeout=args.timeout,
-                output_dir=Path(args.output_dir),
-                browser_port=args.browser_port,
             )
         else:
-            raise RuntimeError(f"不支持的目标类型: {target.kind}")
+            target = resolve_cli_target(args.target, browser_port=args.browser_port)
+            if target.kind == "creator":
+                total = collect_videos(
+                    user_url=target.value,
+                    max_pages=args.pages,
+                    timeout=args.timeout,
+                    output_dir=Path(args.output_dir),
+                    browser_port=args.browser_port,
+                    auto_scroll=args.pages > 1,
+                )
+            elif target.kind == "recommendation":
+                total = collect_recommendations(
+                    max_videos=args.max_videos,
+                    timeout=args.timeout,
+                    output_dir=Path(args.output_dir),
+                    browser_port=args.browser_port,
+                    scroll_settings=scroll_settings,
+                )
+            elif target.kind == "single-video":
+                total = collect_single_video(
+                    target=target,
+                    timeout=args.timeout,
+                    output_dir=Path(args.output_dir),
+                    browser_port=args.browser_port,
+                )
+            else:
+                raise RuntimeError(f"不支持的目标类型: {target.kind}")
     except RuntimeError as exc:
         print(f"[ERROR] {exc}")
         return 1

login_douyin.py

@@ -8,6 +8,7 @@ import time
 from pathlib import Path
 
 DEFAULT_RECOMMENDATION_URL = "https://www.douyin.com/"
+DEFAULT_USER_URL = DEFAULT_RECOMMENDATION_URL
 
 DEFAULT_CHROME_PATH = "/Applications/Google Chrome.app/Contents/MacOS/Google Chrome"
 DEFAULT_BROWSER_PORT = 9223

test_douyin.py

@@ -54,9 +54,32 @@ class FakeRuntimePage:
         self.url = url
 
     def run_js(self, script):
+        # Allow both old scroll_to_next_page and new human_like_scroll
+        if "window.scrollTo" in script or "window.scrollBy" in script:
+            return
         raise AssertionError(f"unexpected scroll script: {script}")
 
 
+class FakeScrollPage:
+    def __init__(self):
+        self.scripts = []
+
+    def run_js(self, script):
+        self.scripts.append(script)
+
+
+class FakeContainerScrollPage:
+    def __init__(self, container_found=True):
+        self.container_found = container_found
+        self.scripts = []
+
+    def run_js(self, script):
+        self.scripts.append(script)
+        if "findMainScrollContainer" in script:
+            return self.container_found
+        return None
+
+
 class DouyinModuleTests(unittest.TestCase):
     def test_module_can_import_without_optional_runtime_dependencies(self) -> None:
         module = importlib.import_module("Douyin")
@@ -95,6 +118,16 @@ class DouyinModuleTests(unittest.TestCase):
         )
         self.assertEqual(output_path.as_posix(), "video/[测试博主]测试标题-123456.mp4")
 
+    def test_build_output_path_limits_long_filename(self) -> None:
+        module = importlib.import_module("Douyin")
+        output_path = module.build_output_path(
+            title="超长标题" * 100,
+            video_id="7619989983668240802",
+            author_name="超长博主名" * 20,
+        )
+        self.assertLessEqual(len(output_path.name.encode("utf-8")), 240)
+        self.assertTrue(output_path.name.endswith("-7619989983668240802.mp4"))
+
     def test_extract_aweme_payload_uses_dict_body(self) -> None:
         module = importlib.import_module("Douyin")
         response = FakeResponse({"aweme_list": []}, "")
@@ -108,11 +141,79 @@ class DouyinModuleTests(unittest.TestCase):
             {"aweme_list": [{"aweme_id": "1"}]},
         )
 
+    def test_wait_for_aweme_packet_treats_false_listener_result_as_missing(self) -> None:
+        module = importlib.import_module("Douyin")
+        page = mock.MagicMock()
+        page.listen.wait.return_value = False
+        self.assertIsNone(module.wait_for_aweme_packet(page, timeout=10))
+
     def test_build_browser_address_from_port(self) -> None:
         module = importlib.import_module("Douyin")
         self.assertEqual(module.build_browser_address(9223), "127.0.0.1:9223")
         self.assertIsNone(module.build_browser_address(None))
 
+    def test_default_scroll_settings_uses_human_mode(self) -> None:
+        module = importlib.import_module("Douyin")
+        settings = module.ScrollSettings()
+        self.assertEqual(settings.mode, "human")
+        self.assertEqual(settings.min_wait, 2.0)
+        self.assertEqual(settings.max_wait, 8.0)
+        self.assertEqual(settings.reverse_scroll_probability, 0.2)
+
+    def test_create_human_scroll_plan_uses_configured_ranges(self) -> None:
+        module = importlib.import_module("Douyin")
+        settings = module.ScrollSettings(
+            min_wait=2.0,
+            max_wait=4.0,
+            min_scroll=300,
+            max_scroll=900,
+            reverse_scroll_probability=0.0,
+        )
+        plan = module.create_human_scroll_plan(settings, random_module=module.random.Random(7))
+        self.assertGreaterEqual(plan.down_distance, 300)
+        self.assertLessEqual(plan.down_distance, 900)
+        self.assertGreaterEqual(plan.down_wait, 2.0)
+        self.assertLessEqual(plan.down_wait, 4.0)
+        self.assertEqual(plan.reverse_distance, 0)
+
+    def test_create_human_scroll_plan_can_include_reverse_scroll(self) -> None:
+        module = importlib.import_module("Douyin")
+        settings = module.ScrollSettings(reverse_scroll_probability=1.0)
+        plan = module.create_human_scroll_plan(settings, random_module=module.random.Random(3))
+        self.assertGreaterEqual(plan.reverse_distance, 80)
+        self.assertLessEqual(plan.reverse_distance, 250)
+        self.assertGreater(plan.reverse_wait, 0)
+
+    def test_run_human_scroll_sequence_scrolls_down_and_optionally_back_up(self) -> None:
+        module = importlib.import_module("Douyin")
+        page = FakeScrollPage()
+        plan = module.HumanScrollPlan(
+            down_distance=500,
+            down_wait=2.5,
+            reverse_distance=120,
+            reverse_wait=1.0,
+            settle_wait=3.0,
+        )
+        with mock.patch.object(module.time, "sleep") as mocked_sleep:
+            module.run_human_scroll_sequence(page, plan)
+        self.assertIn("window.scrollBy(0, 500);", page.scripts)
+        self.assertIn("window.scrollBy(0, -120);", page.scripts)
+        self.assertIn("window.scrollBy(0, 240);", page.scripts)
+        mocked_sleep.assert_has_calls([mock.call(2.5), mock.call(1.0), mock.call(3.0)])
+
+    def test_run_scroll_step_prefers_main_scroll_container(self) -> None:
+        module = importlib.import_module("Douyin")
+        page = FakeContainerScrollPage(container_found=True)
+        self.assertTrue(module.run_scroll_step(page, 500))
+        self.assertIn("const distance = 500;", page.scripts[-1])
+        self.assertIn("scrollTarget.scrollBy(0, distance);", page.scripts[-1])
+
+    def test_run_scroll_step_falls_back_to_window_when_container_is_missing(self) -> None:
+        module = importlib.import_module("Douyin")
+        page = FakeContainerScrollPage(container_found=False)
+        self.assertFalse(module.run_scroll_step(page, 500))
+        self.assertEqual(page.scripts[-1], "window.scrollBy(0, 500);")
+
     def test_ensure_browser_debug_port_ready_accepts_open_port(self) -> None:
         module = importlib.import_module("Douyin")
         connection = mock.MagicMock()
@@ -291,6 +392,49 @@ class DouyinModuleTests(unittest.TestCase):
         self.assertEqual(items[0]["author_name"], "测试博主")
         self.assertEqual(items[0]["author_id"], "123456789")
 
+    def test_parse_aweme_items_uses_play_addr_h264_when_play_addr_is_missing(self) -> None:
+        module = importlib.import_module("Douyin")
+        payload = {
+            "aweme_list": [
+                {
+                    "aweme_id": "7619989983668240802",
+                    "desc": "推荐视频",
+                    "video": {
+                        "play_addr_h264": {
+                            "url_list": ["https://v26-web.douyinvod.com/example/h264.mp4"]
+                        }
+                    },
+                }
+            ]
+        }
+        items = module.parse_aweme_items(payload)
+        self.assertEqual(len(items), 1)
+        self.assertEqual(items[0]["video_url"], "https://v26-web.douyinvod.com/example/h264.mp4")
+
+    def test_parse_aweme_items_uses_bit_rate_play_addr_when_top_level_addresses_are_missing(self) -> None:
+        module = importlib.import_module("Douyin")
+        payload = {
+            "aweme_list": [
+                {
+                    "aweme_id": "7619989983668240802",
+                    "desc": "推荐视频",
+                    "video": {
+                        "bit_rate": [
+                            {
+                                "format": "mp4",
+                                "play_addr": {
+                                    "url_list": ["https://v11-weba.douyinvod.com/example/bitrate.mp4"]
+                                },
+                            }
+                        ]
+                    },
+                }
+            ]
+        }
+        items = module.parse_aweme_items(payload)
+        self.assertEqual(len(items), 1)
+        self.assertEqual(items[0]["video_url"], "https://v11-weba.douyinvod.com/example/bitrate.mp4")
+
     def test_build_video_page_url_uses_aweme_id(self) -> None:
         module = importlib.import_module("Douyin")
         self.assertEqual(
@@ -298,6 +442,38 @@ class DouyinModuleTests(unittest.TestCase):
             "https://www.douyin.com/video/7619989983668240802",
         )
 
+    def test_build_search_page_url_encodes_keyword(self) -> None:
+        module = importlib.import_module("Douyin")
+        self.assertEqual(
+            module.build_search_page_url("猫咪"),
+            "https://www.douyin.com/search/%E7%8C%AB%E5%92%AA?type=general",
+        )
+
+    def test_parse_search_items_extracts_aweme_info(self) -> None:
+        module = importlib.import_module("Douyin")
+        payload = {
+            "data": [
+                {
+                    "type": 1,
+                    "aweme_info": {
+                        "aweme_id": "7319795133048769829",
+                        "desc": "猫咪视频",
+                        "author": {"nickname": "奶芙芙", "uid": "75478174642"},
+                        "video": {
+                            "play_addr_lowbr": {
+                                "url_list": ["https://v26-web.douyinvod.com/example/search.mp4"]
+                            }
+                        },
+                    },
+                }
+            ]
+        }
+        items = module.parse_search_items(payload)
+        self.assertEqual(len(items), 1)
+        self.assertEqual(items[0]["video_id"], "7319795133048769829")
+        self.assertEqual(items[0]["author_name"], "奶芙芙")
+        self.assertEqual(items[0]["video_url"], "https://v26-web.douyinvod.com/example/search.mp4")
+
     def test_collect_recommendations_downloads_videos_with_author_prefix(self) -> None:
         module = importlib.import_module("Douyin")
         packet = FakePacket(
@@ -321,7 +497,7 @@ class DouyinModuleTests(unittest.TestCase):
         with mock.patch.object(module, "import_runtime_dependencies", return_value=(object(), object(), object())):
             with mock.patch.object(module, "create_page", return_value=page):
                 with mock.patch.object(module, "download_video") as mocked_download:
-                    with mock.patch.object(module, "scroll_to_next_page"):
+                    with mock.patch.object(module, "human_like_scroll"):
                         downloaded = module.collect_recommendations(
                             max_videos=50,
                             timeout=10,
@@ -409,6 +585,49 @@ class DouyinModuleTests(unittest.TestCase):
         args = module.build_parser().parse_args(["--max-videos", "30"])
         self.assertEqual(args.max_videos, 30)
 
+    def test_build_parser_has_human_scroll_arguments(self) -> None:
+        module = importlib.import_module("Douyin")
+        args = module.build_parser().parse_args(
+            [
+                "--scroll-mode",
+                "human",
+                "--min-wait",
+                "3",
+                "--max-wait",
+                "9",
+                "--reverse-scroll-probability",
+                "0.4",
+                "--max-runtime",
+                "600",
+            ]
+        )
+        self.assertEqual(args.scroll_mode, "human")
+        self.assertEqual(args.min_wait, 3)
+        self.assertEqual(args.max_wait, 9)
+        self.assertEqual(args.reverse_scroll_probability, 0.4)
+        self.assertEqual(args.max_runtime, 600)
+
+    def test_build_parser_has_search_keyword_argument(self) -> None:
+        module = importlib.import_module("Douyin")
+        args = module.build_parser().parse_args(["--search-keyword", "猫咪"])
+        self.assertEqual(args.search_keyword, "猫咪")
+
+    def test_main_dispatches_search_flow_for_search_keyword(self) -> None:
+        module = importlib.import_module("Douyin")
+        stdout = io.StringIO()
+        with redirect_stdout(stdout):
+            with mock.patch.object(module, "collect_search_results", return_value=7) as mocked_collect:
+                exit_code = module.main(["--search-keyword", "猫咪"])
+        self.assertEqual(exit_code, 0)
+        mocked_collect.assert_called_once_with(
+            keyword="猫咪",
+            max_videos=50,
+            timeout=10,
+            output_dir=module.Path("video"),
+            browser_port=9223,
+            scroll_settings=module.ScrollSettings(),
+        )
+
     def test_build_parser_defaults_to_zero_argument_current_page_flow(self) -> None:
         module = importlib.import_module("Douyin")
         args = module.build_parser().parse_args([])
@@ -442,6 +661,7 @@ class DouyinModuleTests(unittest.TestCase):
             timeout=10,
             output_dir=module.Path("video"),
             browser_port=9223,
+            scroll_settings=module.ScrollSettings(),
         )
 
     def test_main_without_target_dispatches_current_page_creator_flow(self) -> None:

test_learning_examples.py

@@ -299,6 +299,8 @@ class PlaywrightLearningHelperTests(unittest.TestCase):
                 "title": "Playwright 示例",
                 "video_id": "7619989983668240802",
                 "video_url": "https://v26-web.douyinvod.com/example/single.mp4",
+                "author_name": "unknown",
+                "author_id": "unknown",
             },
         )