wechat-article-parser 0.0.9__tar.gz → 0.0.10__tar.gz

{wechat_article_parser-0.0.9 → wechat_article_parser-0.0.10}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: wechat-article-parser
-Version: 0.0.9
+Version: 0.0.10
 Summary: WeChat MP article parser - extract metadata and content from WeChat public account articles
 License-Expression: MIT
 License-File: LICENSE
@@ -30,6 +30,12 @@ Description-Content-Type: text/markdown
 `8` 对应图片轮播、`10` 对应文字消息，再通过 HTML 或脚本数据区分具体模板。
 类型字段缺失、值未知或对应分支未提取到正文时，会按原有 HTML 特征匹配顺序兜底。
 
+图文正文仅包含视频号视频卡片或微信视频 iframe 时，Markdown 返回“这是一篇纯视频内容。”。
+正文只有付费预览占位标签时，Markdown 返回“这是一篇付费内容，正文需付费后查看。”，
+并保留公开元数据；已有的可见预览正文仍正常提取。
+图文中的公众号推荐卡片按原文顺序转换为 `**公众号名称**：简介`，每张卡片一段；
+名称或简介缺失时，仅输出已有字段。
+
 ## 安装
 
 ```bash
@@ -123,6 +129,8 @@ result = parse(
 
 `is_valid` 为 `True` 的条件：`mp_id`、`mp_name`、`article_id`、`article_msg_id`、`article_idx`、`article_sn`、`article_title`、`article_publish_time` 均不为空/零，且 `article_markdown` 非空或 `article_display_type == "video"`。视频分享允许没有文字正文，仍须满足其他关键字段要求；有效性不表示已获取视频文件。
 
+付费受限页面返回提示且关键元数据完整时，`is_valid` 也为 `True`，这不表示已获取付费正文。
+
 ### 判断文章展示类型
 
 `ArticleDisplayType` 是字符串枚举，既支持枚举比较，也支持语义字符串比较：

{wechat_article_parser-0.0.9 → wechat_article_parser-0.0.10}/README.md

@@ -15,6 +15,12 @@
 `8` 对应图片轮播、`10` 对应文字消息，再通过 HTML 或脚本数据区分具体模板。
 类型字段缺失、值未知或对应分支未提取到正文时，会按原有 HTML 特征匹配顺序兜底。
 
+图文正文仅包含视频号视频卡片或微信视频 iframe 时，Markdown 返回“这是一篇纯视频内容。”。
+正文只有付费预览占位标签时，Markdown 返回“这是一篇付费内容，正文需付费后查看。”，
+并保留公开元数据；已有的可见预览正文仍正常提取。
+图文中的公众号推荐卡片按原文顺序转换为 `**公众号名称**：简介`，每张卡片一段；
+名称或简介缺失时，仅输出已有字段。
+
 ## 安装
 
 ```bash
@@ -108,6 +114,8 @@ result = parse(
 
 `is_valid` 为 `True` 的条件：`mp_id`、`mp_name`、`article_id`、`article_msg_id`、`article_idx`、`article_sn`、`article_title`、`article_publish_time` 均不为空/零，且 `article_markdown` 非空或 `article_display_type == "video"`。视频分享允许没有文字正文，仍须满足其他关键字段要求；有效性不表示已获取视频文件。
 
+付费受限页面返回提示且关键元数据完整时，`is_valid` 也为 `True`，这不表示已获取付费正文。
+
 ### 判断文章展示类型
 
 `ArticleDisplayType` 是字符串枚举，既支持枚举比较，也支持语义字符串比较：

{wechat_article_parser-0.0.9 → wechat_article_parser-0.0.10}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "wechat-article-parser"
-version = "0.0.9"
+version = "0.0.10"
 description = "WeChat MP article parser - extract metadata and content from WeChat public account articles"
 readme = "README.md"
 requires-python = ">=3.10"

{wechat_article_parser-0.0.9 → wechat_article_parser-0.0.10}/src/wechat_article_parser/parser.py

@@ -78,6 +78,21 @@ def _decode_hex_escapes(text: str) -> str:
     return re.sub(r"\\x([0-9a-fA-F]{2})", lambda m: chr(int(m.group(1), 16)), text)
 
 
+def _decode_js_string(text: str) -> str:
+    """解码 JS 字符串转义，保留 HTML 实体、百分号和正文中的空白。"""
+    escapes = {
+        "n": "\n", "r": "\r", "t": "\t", "b": "\b", "f": "\f", "v": "\v",
+        "0": "\0", "\n": "", "\r": "", "\r\n": "",
+    }
+
+    def decode(match: re.Match[str]) -> str:
+        if match[1] or match[2]:
+            return chr(int(match[1] or match[2], 16))
+        return escapes.get(match[3], match[3])
+
+    return re.sub(r"\\(?:x([\da-fA-F]{2})|u([\da-fA-F]{4})|(\r\n|[\s\S]))", decode, text)
+
+
 def _decode_text(text: str, *, preserve_newlines: bool = False) -> str:
     if not text:
         return ""
@@ -265,8 +280,13 @@ def _extract_user_name(scripts: list[Tag]) -> str:
 
 def _extract_cgi_user_name(scripts: list[Tag]) -> str:
     """从当前文章的 cgiDataNew 对象中读取 user_name，排除嵌套账号。"""
+    return _decode_text(_extract_cgi_string(scripts, "user_name"))
+
+
+def _extract_cgi_string(scripts: list[Tag], field: str) -> str:
+    """读取当前 cgiDataNew 的直接字符串属性，不执行表达式或读取嵌套对象。"""
     for script in scripts:
-        if "cgiDataNew" not in script.text or "user_name" not in script.text:
+        if "cgiDataNew" not in script.text or field not in script.text:
             continue
         history: deque[str] = deque(maxlen=5)
         stack: list[str] = []
@@ -289,12 +309,12 @@ def _extract_cgi_user_name(scripts: list[Tag]) -> str:
                     and token in (",", "}")
                     and len(previous) >= 4
                     and previous[-4] in ("{", ",")
-                    and previous[-3].strip("\"'") == "user_name"
+                    and previous[-3].strip("\"'") == field
                     and previous[-2] == ":"
                     and previous[-1].startswith(("'", '"'))
                 ):
-                    value = _decode_text(previous[-1][1:-1])
-                    if value:
+                    value = previous[-1][1:-1]
+                    if value.strip():
                         return value
                 if token in ("{", "["):
                     stack.append(token)
@@ -421,7 +441,26 @@ def _extract_rich_media_content(content_tag: Tag, result: ArticleResult) -> None
     """将 rich_media_content 区块转换为 Markdown。"""
     # prettify 会在行内标签之间插入空白，改变代码和引用链接的内容。
     soup = BeautifulSoup(str(content_tag), "html.parser")
-    seen: set[str] = set()
+    has_video = soup.select_one(
+        'mp-common-videosnap[data-type="video"], iframe.video_iframe, iframe.wx_video_iframe'
+    ) is not None
+
+    # 原生公众号卡片的正文保存在属性中，转换为“名称 + 简介”的独立段落。
+    for profile in soup.find_all("mp-common-profile"):
+        nickname = profile.get("data-nickname", "").strip()
+        signature = profile.get("data-signature", "").strip()
+        if not nickname and not signature:
+            continue
+        paragraph = soup.new_tag("p")
+        if nickname:
+            name = soup.new_tag("strong")
+            name.string = nickname
+            paragraph.append(name)
+        if signature:
+            if nickname:
+                paragraph.append("：")
+            paragraph.append(signature)
+        profile.replace_with(paragraph)
 
     # 处理 <img> 标签
     to_remove = []
@@ -430,17 +469,13 @@ def _extract_rich_media_content(content_tag: Tag, result: ArticleResult) -> None
         if src and src.startswith("http"):
             normalized = _normalize_image_url(src)
             img["src"] = normalized
-            if normalized not in seen:
-                seen.add(normalized)
-                result.images.append(normalized)
         else:
             to_remove.append(img)
     for tag in to_remove:
         tag.decompose()
 
-    # 处理含有 background-image 的 <svg> 标签
-    # 部分文章用嵌套 SVG（外层 SVG > foreignobject > 内层 SVG）承载正文图片，
-    # 需从外向内收集所有图片 URL，一次性替换为多个 <img> 标签
+    # SVG 可通过 background-image 或 foreignObject 内的 <img> 承载正文图片。
+    # 按文档顺序收集两种图片，再统一替换外层 SVG，避免删除内部正文图片。
     to_remove = []
     for svg in soup.find_all("svg", recursive=True):
         # 跳过已被外层 SVG 处理过的嵌套 SVG（已脱离文档树）
@@ -450,11 +485,14 @@ def _extract_rich_media_content(content_tag: Tag, result: ArticleResult) -> None
         if svg.find_parent("svg"):
             continue
 
-        # 收集本 SVG 及所有后代 SVG 中的 background-image 图片
-        all_svgs = [svg] + svg.find_all("svg")
+        # 包含所有嵌套 SVG 和普通图片，保留正文中的重复图片及 alt 文本。
+        image_elements = [svg] + svg.find_all(["svg", "img"])
         img_tags = []
-        for s in all_svgs:
-            style = s.get("style", "")
+        for element in image_elements:
+            if element.name == "img":
+                img_tags.append(soup.new_tag("img", attrs=element.attrs.copy()))
+                continue
+            style = element.get("style", "")
             if "background-image" not in style:
                 continue
             m = re.search(r'url\("([^"]+)"\)', style)
@@ -462,9 +500,6 @@ def _extract_rich_media_content(content_tag: Tag, result: ArticleResult) -> None
                 continue
             normalized = _normalize_image_url(m.group(1))
             img_tags.append(soup.new_tag("img", src=normalized))
-            if normalized not in seen:
-                seen.add(normalized)
-                result.images.append(normalized)
 
         if img_tags:
             svg.replace_with(*img_tags)
@@ -473,30 +508,66 @@ def _extract_rich_media_content(content_tag: Tag, result: ArticleResult) -> None
     for tag in to_remove:
         tag.decompose()
 
+    # 从最终正文收集图片，使列表顺序与 Markdown 一致，并只对列表去重。
+    seen: set[str] = set()
+    for img in soup.find_all("img"):
+        src = img["src"]
+        if src not in seen:
+            seen.add(src)
+            result.images.append(src)
+
     result.article_markdown = _to_markdown(str(soup))
+    if not result.article_markdown.strip():
+        if soup.find("mp-pay-preview-filter") is not None:
+            result.article_markdown = "这是一篇付费内容，正文需付费后查看。"
+        elif has_video:
+            result.article_markdown = "这是一篇纯视频内容。"
 
 
-def _extract_repost_content(content_tag: Tag, result: ArticleResult) -> None:
-    """从转载式文章中提取内容。"""
-    soup = BeautifulSoup(content_tag.prettify(), "html.parser")
+def _extract_repost_content(
+    content_tag: Tag, scripts: list[Tag], result: ArticleResult,
+) -> None:
+    """读取转载附言、cgiDataNew 中的富文本正文和原文链接。"""
+    soup = BeautifulSoup(str(content_tag), "html.parser")
+    parts: list[str] = []
     notice = soup.find("p", id="js_share_notice")
-    if not notice:
-        return
+    if notice:
+        m = re.search(
+            r'''innerHTML\s*=\s*("(?:\\[\s\S]|[^"\\])*"|'(?:\\[\s\S]|[^'\\])*')''',
+            str(notice),
+        )
+        if m:
+            text = _decode_text(_decode_js_string(m[1][1:-1]), preserve_newlines=True)
+            notice_markdown = _to_markdown_plain(f"<p>{text}</p>")
+        else:
+            for script in notice.find_all("script"):
+                script.decompose()
+            notice_markdown = _to_markdown_plain(str(notice))
+        if notice_markdown.strip():
+            parts.append(notice_markdown.strip())
+
+    for field in ("content_noencode", "content"):
+        body = _extract_cgi_string(scripts, field)
+        if body:
+            # 与普通富文本共用图片和 Markdown 处理，避免 data-src 变成空图片链接。
+            body_soup = BeautifulSoup(_decode_js_string(body), "html.parser")
+            _extract_rich_media_content(body_soup, result)
+            if result.article_markdown.strip():
+                parts.append(result.article_markdown.strip())
+                break
 
-    m = re.search(r'innerHTML = "([^"]+)"', str(notice))
-    if not m:
+    if not parts:
         return
 
-    text = _decode_text(m.group(1), preserve_newlines=True)
-    html_content = f"<p>{text}</p>"
-
     share_link = soup.find("span", id="js_share_source")
     if share_link:
         href = share_link.get("data-url")
         if href:
-            html_content += f'<p><a href="{href}">查看原文</a></p>'
+            link = soup.new_tag("a", href=href)
+            link.string = "查看原文"
+            parts.append(_to_markdown_plain(str(link)))
 
-    result.article_markdown = _to_markdown_plain(html_content)
+    result.article_markdown = "\n\n".join(parts)
 
 
 def _extract_plain_text_content(soup: BeautifulSoup, result: ArticleResult) -> None:
@@ -622,7 +693,7 @@ def _try_content_branch(
     if branch == "rich_text":
         _extract_rich_media_content(content, candidate)
     elif branch == "repost":
-        _extract_repost_content(content, candidate)
+        _extract_repost_content(content, scripts, candidate)
     elif branch in ("plain_text", "video"):
         _extract_plain_text_content(soup, candidate)
     elif branch == "swiper":

wechat_article_parser-0.0.10/tests/test_paid_content.py

@@ -0,0 +1,106 @@
+"""付费受限正文的提示、元数据和预览内容回归测试。"""
+
+import pytest
+
+from wechat_article_parser import ArticleDisplayType
+from wechat_article_parser.parser import _parse_html
+
+
+_URL = "https://mp.weixin.qq.com/s/n2q09-I95zwggfbXKAmbuQ"
+_NOTICE = "这是一篇付费内容，正文需付费后查看。"
+_FILTER = (
+    '<p class="js_pay_preview_filter" style="height: 0; overflow: hidden;">'
+    '<mp-pay-preview-filter></mp-pay-preview-filter></p>'
+)
+_METADATA = """
+var biz = "MzkwMTYxMzYwNg==";
+var mid = "2247533326";
+var idx = "5";
+var sn = "5ae1c1d004a9ccbf774610a8be1de3d7";
+window.__allowLoadResFromMp = true;
+var hd_head_img = "https://example.com/avatar";
+var nickname = htmlDecode("真锂研究");
+var oriCreateTime = '1791450000';
+var user_name = "gh_61cfd6274c18";
+"""
+
+
+def _html(body: str, metadata: str = _METADATA, data: str = "") -> str:
+    return (
+        '<meta property="og:title" content="付费文章标题">'
+        f'<div class="rich_media_content" id="js_content">{body}</div>'
+        '<div id="js_fullscreen_layout_padding"></div>'
+        '<script type="text/javascript">var item_show_type = "0";</script>'
+        f'<script type="text/javascript">{metadata}</script>'
+        f'<script type="text/javascript">{data}</script>'
+    )
+
+
+@pytest.mark.parametrize("metadata", [_METADATA, ""])
+def test_paid_placeholder_keeps_metadata_and_does_not_use_unrelated_images(
+    metadata: str,
+) -> None:
+    # 对应真实页面：正文只有付费占位符，脚本另有图片数组。
+    data = """
+    window.cgiDataNew = {
+      is_pay_subscribe: '1' * 1,
+      pay_subscribe_info: {preview_percent: '0' * 1},
+      picture_page_info_list: [{cdn_url: 'https://example.com/unrelated.jpg'}]
+    };
+    """
+    result = _parse_html(_URL, _html(_FILTER, metadata, data))
+
+    assert result.article_markdown == _NOTICE
+    assert result.images == []
+    assert result.article_display_type is ArticleDisplayType.ARTICLE
+    assert result.article_item_show_type == 0
+    assert result.is_valid is bool(metadata)
+    if metadata:
+        assert result.mp_id == 3901613606
+        assert result.mp_name == "真锂研究"
+        assert result.mp_user_name == "gh_61cfd6274c18"
+        assert result.article_msg_id == 2247533326
+        assert result.article_idx == 5
+        assert result.article_sn == "5ae1c1d004a9ccbf774610a8be1de3d7"
+        assert result.article_publish_time == 1791450000
+
+
+@pytest.mark.parametrize("body", [
+    '<p>可见的免费预览正文。</p>',
+    '<img data-src="https://example.com/preview.jpg">',
+])
+def test_paid_filter_does_not_replace_visible_preview(body: str) -> None:
+    result = _parse_html(_URL, _html(body + _FILTER))
+
+    assert result.is_valid
+    assert _NOTICE not in result.article_markdown
+    if "免费预览" in body:
+        assert result.article_markdown == "可见的免费预览正文。"
+    else:
+        assert result.article_markdown == "![](https://example.com/preview.jpg)"
+        assert result.images == ["https://example.com/preview.jpg"]
+
+
+def test_paid_filter_takes_priority_over_empty_video_embed() -> None:
+    body = '<iframe class="wx_video_iframe"></iframe>' + _FILTER
+    result = _parse_html(_URL, _html(body))
+
+    assert result.is_valid
+    assert result.article_markdown == _NOTICE
+
+
+def test_paid_marker_outside_article_does_not_fill_empty_body() -> None:
+    result = _parse_html(_URL, _html("") + _FILTER)
+
+    assert result.article_markdown == ""
+    assert not result.is_valid
+
+
+def test_paid_flag_without_filter_keeps_accessible_body() -> None:
+    result = _parse_html(
+        _URL,
+        _html('<p>已提供的完整正文。</p>', data="window.cgiDataNew = {is_pay_subscribe: '1' * 1};"),
+    )
+
+    assert result.is_valid
+    assert result.article_markdown == "已提供的完整正文。"

{wechat_article_parser-0.0.9 → wechat_article_parser-0.0.10}/tests/test_parser.py

@@ -16,6 +16,18 @@ TEST_URLS = [
     "https://mp.weixin.qq.com/s/ySQdtsRlRmAl_skdc5HQ-A",
     "https://mp.weixin.qq.com/s/MnkArbYQNp3tF29gujMUnQ",
     "https://mp.weixin.qq.com/s/ncPoZnxKdvDRLM_s4DBgmA",
+    # 转载页面的静态附言及脚本正文。
+    "https://mp.weixin.qq.com/s/Pq-JkEjhoIlLvTMzWKVtOw",
+    "https://mp.weixin.qq.com/s/X51M6PfionI6UpX4zsG78Q",
+    # 视频号卡片和微信视频 iframe 的纯视频正文。
+    "https://mp.weixin.qq.com/s/wAyM7yYKDfTH21-XpS9agQ",
+    "https://mp.weixin.qq.com/s/a7P56VOipWNB0UyNaVOZYQ",
+    # 付费受限正文。
+    "https://mp.weixin.qq.com/s/n2q09-I95zwggfbXKAmbuQ",
+    # SVG foreignObject 承载的正文图片。
+    "https://mp.weixin.qq.com/s/3ixeC9whuEgpup2G1yvDZg",
+    # 纯公众号推荐卡片正文。
+    "https://mp.weixin.qq.com/s/PXZuh3fFEi3VXUUI1uBnlw",
 ]
 
 

wechat_article_parser-0.0.10/tests/test_profile_content.py

@@ -0,0 +1,127 @@
+"""公众号推荐卡片转换及正文顺序的离线回归测试。"""
+
+import json
+
+import pytest
+
+from wechat_article_parser import ArticleDisplayType
+from wechat_article_parser.parser import _parse_html
+
+
+_URL = "https://mp.weixin.qq.com/s/PXZuh3fFEi3VXUUI1uBnlw"
+_METADATA = """
+var item_show_type = "0";
+var biz = "MzUyNTgyNjUzMw==";
+var mid = "2247923648";
+var idx = "2";
+var sn = "d744c25ecdf2ac7f361d880dffb0a2ca";
+window.__allowLoadResFromMp = true;
+var hd_head_img = "https://example.com/publisher-avatar";
+var nickname = htmlDecode("药融圈");
+var oriCreateTime = '1791441362';
+"""
+
+
+def _html(body: str) -> str:
+    return (
+        '<meta property="og:title" content="公众号推荐文章">'
+        f'<div class="rich_media_content">{body}</div>'
+        '<div id="js_fullscreen_layout_padding"></div>'
+        f'<script type="text/javascript">{_METADATA}</script>'
+    )
+
+
+def test_profile_only_article_preserves_names_signatures_and_metadata() -> None:
+    body = (
+        '<section><span><mp-common-profile data-nickname="药融圈" '
+        'data-signature="中国医药信息和人脉的链接平台" '
+        'data-headimg="https://example.com/card-avatar"></mp-common-profile></span></section>'
+        '<section><mp-common-profile data-nickname="药通社" '
+        'data-signature="政策、资讯、产业、技术"></mp-common-profile></section>'
+    )
+    result = _parse_html(
+        _URL,
+        _html(body)
+        + '<mp-common-profile data-nickname="页外推荐" data-signature="不属于正文"></mp-common-profile>',
+    )
+
+    assert result.article_markdown == (
+        '**药融圈**：中国医药信息和人脉的链接平台\n\n'
+        '**药通社**：政策、资讯、产业、技术'
+    )
+    assert result.images == []
+    assert result.is_valid
+    assert result.mp_id == 3525826533
+    assert result.mp_name == "药融圈"
+    assert result.article_msg_id == 2247923648
+    assert result.article_idx == 2
+    assert result.article_sn == "d744c25ecdf2ac7f361d880dffb0a2ca"
+    assert result.article_publish_time == 1791441362
+    assert result.article_display_type is ArticleDisplayType.ARTICLE
+    assert result.article_item_show_type == 0
+
+
+def test_profile_mixed_with_text_and_images_preserves_order_without_duplicate_card() -> None:
+    result = _parse_html(_URL, _html(
+        '<p>推荐以下公众号。</p><mp-common-profile data-nickname="药融圈" '
+        'data-signature="医药信息平台">'
+        '<img src="https://example.com/card-avatar"><p>药融圈</p><p>医药信息平台</p>'
+        '</mp-common-profile>'
+        '<p><img data-src="https://mmbiz.qpic.cn/mmbiz_jpg/body/0?wx_fmt=jpeg"></p>'
+        '<p>正文结尾。</p>'
+    ))
+
+    assert result.article_markdown == (
+        '推荐以下公众号。\n\n**药融圈**：医药信息平台\n\n'
+        '![](https://mmbiz.qpic.cn/mmbiz_jpg/body/640)\n\n正文结尾。'
+    )
+    assert result.images == ["https://mmbiz.qpic.cn/mmbiz_jpg/body/640"]
+
+
+@pytest.mark.parametrize(("attributes", "expected"), [
+    ('data-nickname="  药融圈  "', '**药融圈**'),
+    ('data-signature="  医药信息平台  "', '医药信息平台'),
+    ('data-nickname=" " data-signature=" "', ''),
+    ('', ''),
+])
+def test_profile_missing_fields_keeps_available_text_without_empty_placeholders(
+    attributes: str, expected: str,
+) -> None:
+    result = _parse_html(_URL, _html(f'<mp-common-profile {attributes}></mp-common-profile>'))
+
+    assert result.article_markdown == expected
+    assert result.is_valid is bool(expected)
+    assert result.images == []
+
+
+def test_profile_attributes_keep_entities_and_escape_markdown_characters() -> None:
+    result = _parse_html(_URL, _html(
+        '<mp-common-profile data-nickname="药品&amp;研发*圈*" '
+        'data-signature="A &amp; B，进度 100%，保留 %20 和 &lt;tag&gt;。">'
+        '</mp-common-profile>'
+    ))
+
+    assert result.article_markdown == r'**药品&研发\*圈\***：A & B，进度 100%，保留 %20 和 <tag>。'
+
+
+def test_repost_script_body_uses_same_profile_conversion() -> None:
+    body = (
+        '<mp-common-profile data-nickname="药通社" '
+        'data-signature="政策、资讯、产业、技术"></mp-common-profile>'
+    )
+    html = (
+        '<meta property="og:title" content="转载推荐文章">'
+        '<div class="original_page"><span id="js_share_source" '
+        'data-url="https://example.com/original"></span></div>'
+        f'<script type="text/javascript">{_METADATA}</script>'
+        '<script type="text/javascript">window.cgiDataNew = '
+        f'{{content_noencode: {json.dumps(body, ensure_ascii=False)}}};</script>'
+    )
+    result = _parse_html(_URL, html)
+
+    assert result.is_valid
+    assert result.article_markdown == (
+        '**药通社**：政策、资讯、产业、技术\n\n'
+        '[查看原文](https://example.com/original)'
+    )
+    assert result.images == []

wechat_article_parser-0.0.10/tests/test_repost_content.py

@@ -0,0 +1,140 @@
+"""转载页面的静态附言、脚本正文及元数据回归测试。"""
+
+import pytest
+
+from wechat_article_parser import ArticleDisplayType
+from wechat_article_parser.parser import _parse_html
+
+
+_URL = "https://mp.weixin.qq.com/s/Pq-JkEjhoIlLvTMzWKVtOw"
+_SOURCE = "https://mp.weixin.qq.com/s?__biz=original&mid=456&idx=1&sn=source"
+_BODY = """<section>
+<p>转载正文第一段，保留 100% 和 %20。</p>
+<img data-src="https://mmbiz.qpic.cn/mmbiz_png/first/0?wx_fmt=png&amp;from=appmsg">
+<p>正文里的 &lt;tag&gt;、O'Reilly 和反斜杠 \\。</p>
+<pre><code>if ready:
+    print(&quot;&lt;tag&gt;&quot;)
+</code></pre>
+<img data-src="https://mmbiz.qpic.cn/mmbiz_png/second/0?wx_fmt=png">
+<p>转载正文最后一段。</p>
+</section>"""
+
+
+def _html(data: str = "", notice: str = "分享一篇文章。") -> str:
+    # 对应真实页面：正文容器尚未填充，正文和元数据分散在多个脚本。
+    return f"""
+    <meta property="og:title" content="转载文章标题">
+    <div class="original_page">
+      <p id="js_share_notice">{notice}</p>
+      <div id="js_share_headimg"><img src="https://example.com/avatar"></div>
+      <div id="js_content"></div>
+      <span id="js_share_source" data-url="{_SOURCE}">阅读全文</span>
+    </div>
+    <div id="js_fullscreen_layout_padding"></div>
+    <script type="text/javascript">
+      var biz = "MzAxNjI5NzI4MA==" || "";
+      var mid = "2247577355" || "";
+      var idx = "4" || "";
+      var sn = "signature" || "";
+      window.__allowLoadResFromMp = true;
+    </script>
+    <script type="text/javascript">
+      var hd_head_img = "https://example.com/publisher-avatar";
+      var nickname = htmlDecode("新浪医药");
+      var item_show_type = "0";
+      var user_name = "gh_3675b3bc7097";
+    </script>
+    <script type="text/javascript">
+      function setTime() {{ var oriCreateTime = '1791459000'; }}
+    </script>
+    <script type="text/javascript">{data}</script>
+    """
+
+
+def _js_body(body: str, quote: str = "'") -> str:
+    escaped = body.translate(str.maketrans({
+        "\\": "\\\\", "'": "\\'", '"': r"\x22", "<": r"\x3c",
+        ">": r"\x3e", "&": r"\x26", "\n": r"\n",
+    }))
+    return quote + escaped + quote
+
+
+@pytest.mark.parametrize("field", ["content_noencode", "content"])
+@pytest.mark.parametrize("quote", ["'", '"'])
+def test_static_repost_reads_script_body_and_preserves_metadata(
+    field: str, quote: str,
+) -> None:
+    data = f"""
+    window.cgiDataNew = {{
+      reference: {{content_noencode: '嵌套的错误正文', content: '嵌套的错误正文'}},
+      {quote}{field}{quote}: {_js_body(_BODY, quote)},
+      picture_page_info_list: [{{cdn_url: 'https://example.com/wrong-image'}}]
+    }};
+    """
+    result = _parse_html(_URL, _html(data))
+
+    assert result.is_valid
+    assert result.article_display_type is ArticleDisplayType.ARTICLE
+    assert result.article_item_show_type == 0
+    assert result.mp_id == 3016297280
+    assert result.mp_name == "新浪医药"
+    assert result.mp_user_name == "gh_3675b3bc7097"
+    assert result.article_msg_id == 2247577355
+    assert result.article_idx == 4
+    assert result.article_sn == "signature"
+    assert result.article_publish_time == 1791459000
+    assert result.images == [
+        "https://mmbiz.qpic.cn/mmbiz_png/first/640",
+        "https://mmbiz.qpic.cn/mmbiz_png/second/640",
+    ]
+    markdown = result.article_markdown
+    assert markdown.startswith("分享一篇文章。")
+    assert "转载正文第一段，保留 100% 和 %20。" in markdown
+    assert "正文里的 <tag>、O'Reilly 和反斜杠" in markdown
+    assert '```\nif ready:\n    print("<tag>")\n```' in markdown
+    assert "转载正文最后一段。" in markdown
+    assert markdown.endswith(f"[查看原文]({_SOURCE})")
+    assert markdown.count("![") == 2
+    assert "![]()" not in markdown
+    assert "嵌套的错误正文" not in markdown
+    assert "avatar" not in markdown
+    assert "wrong-image" not in markdown
+
+
+def test_static_notice_without_script_body_still_parses() -> None:
+    result = _parse_html(_URL, _html())
+
+    assert result.is_valid
+    assert result.article_markdown == f"分享一篇文章。\n\n[查看原文]({_SOURCE})"
+    assert result.images == []
+
+
+def test_legacy_script_notice_keeps_body_and_newlines() -> None:
+    notice = r'<script>notice.innerHTML = "旧版附言\x0a第二行";</script>'
+    result = _parse_html(_URL, _html(notice=notice))
+
+    assert result.is_valid
+    assert result.article_markdown == f"旧版附言  \n第二行\n\n[查看原文]({_SOURCE})"
+
+
+def test_empty_unencoded_body_falls_back_to_content() -> None:
+    data = "window.cgiDataNew = {content_noencode: '', content: '<p>备用正文</p>'};"
+    result = _parse_html(_URL, _html(data, notice=""))
+
+    assert result.is_valid
+    assert result.article_markdown == f"备用正文\n\n[查看原文]({_SOURCE})"
+
+
+def test_repost_ignores_body_in_comments_strings_and_other_objects() -> None:
+    data = """
+    // window.cgiDataNew = {content_noencode: '注释中的错误正文'};
+    var example = "window.cgiDataNew = {content: '字符串中的错误正文'};";
+    other.window.cgiDataNew = {content_noencode: '其他对象中的错误正文'};
+    window.cgiDataNew = {
+      reference: {content_noencode: '嵌套的错误正文'},
+      content_noencode: loadUntrustedContent()
+    };
+    """
+    result = _parse_html(_URL, _html(data))
+
+    assert result.article_markdown == f"分享一篇文章。\n\n[查看原文]({_SOURCE})"

wechat_article_parser-0.0.10/tests/test_svg_content.py

@@ -0,0 +1,98 @@
+"""SVG 容器承载正文图片的离线回归测试。"""
+
+from wechat_article_parser import ArticleDisplayType, ArticleResult
+from wechat_article_parser.parser import _parse_html
+
+
+_URL = "https://mp.weixin.qq.com/s/3ixeC9whuEgpup2G1yvDZg"
+_METADATA = """
+var item_show_type = "0";
+var biz = "MjM5MjY2MTc5NQ==";
+var mid = "2652403355";
+var idx = "1";
+var sn = "9c582b2554db23b592ed758e95680f0a";
+window.__allowLoadResFromMp = true;
+var hd_head_img = "https://example.com/avatar";
+var nickname = htmlDecode("CFMOTO春风动力");
+var oriCreateTime = '1791454818';
+"""
+
+
+def _parse(body: str) -> ArticleResult:
+    return _parse_html(
+        _URL,
+        '<meta property="og:title" content="SVG 图片正文">'
+        f'<div class="rich_media_content">{body}</div>'
+        '<div id="js_fullscreen_layout_padding"></div>'
+        f'<script type="text/javascript">{_METADATA}</script>',
+    )
+
+
+def test_foreignobject_images_keep_body_order_and_rich_text_metadata() -> None:
+    # 对应真实页面：整个正文由三个 SVG 中的长图组成，没有背景图。
+    body = "".join(
+        '<svg viewBox="0 0 1080 2362"><foreignObject>'
+        f'<span><img data-src="https://mmbiz.qpic.cn/mmbiz_jpg/body-{i}/0?wx_fmt=jpeg"></span>'
+        '</foreignObject></svg>'
+        for i in range(3)
+    )
+    result = _parse(body)
+    urls = [f"https://mmbiz.qpic.cn/mmbiz_jpg/body-{i}/640" for i in range(3)]
+
+    assert result.article_markdown == "".join(f"![]({url})" for url in urls)
+    assert result.images == urls
+    assert result.is_valid
+    assert result.mp_id == 2392661795
+    assert result.mp_name == "CFMOTO春风动力"
+    assert result.article_msg_id == 2652403355
+    assert result.article_idx == 1
+    assert result.article_sn == "9c582b2554db23b592ed758e95680f0a"
+    assert result.article_publish_time == 1791454818
+    assert result.article_display_type is ArticleDisplayType.ARTICLE
+
+
+def test_nested_svg_images_and_backgrounds_keep_source_order() -> None:
+    result = _parse(
+        '<svg style=\'background-image: url("https://example.com/first.jpg")\'>'
+        '<foreignObject><img data-src="https://example.com/second.jpg" alt="第二张">'
+        '<svg style=\'background-image: url("https://example.com/third.jpg")\'>'
+        '<foreignObject><img src="https://example.com/fourth.jpg"></foreignObject>'
+        '</svg></foreignObject></svg>'
+    )
+
+    assert result.article_markdown == "".join([
+        "![](https://example.com/first.jpg)",
+        "![第二张](https://example.com/second.jpg)",
+        "![](https://example.com/third.jpg)",
+        "![](https://example.com/fourth.jpg)",
+    ])
+    assert result.images == [
+        "https://example.com/first.jpg",
+        "https://example.com/second.jpg",
+        "https://example.com/third.jpg",
+        "https://example.com/fourth.jpg",
+    ]
+
+
+def test_repeated_svg_images_remain_in_body_but_image_list_is_unique() -> None:
+    result = _parse(
+        '<img src="https://mmbiz.qpic.cn/mmbiz_jpg/body/0">'
+        '<svg><foreignObject><svg><foreignObject>'
+        '<img data-src="https://mmbiz.qpic.cn/mmbiz_jpg/body/640?wx_fmt=jpeg">'
+        '</foreignObject></svg></foreignObject></svg>'
+    )
+
+    url = "https://mmbiz.qpic.cn/mmbiz_jpg/body/640"
+    assert result.images == [url]
+    assert result.article_markdown.count(f"![]({url})") == 2
+
+
+def test_decorative_svg_and_invalid_images_do_not_pollute_body() -> None:
+    result = _parse(
+        '<p>正文</p><svg><title>装饰分隔线</title><path d="M0 0 L100 0"></path></svg>'
+        '<svg><foreignObject><img><img src="data:image/png;base64,AA==">'
+        '</foreignObject></svg>'
+    )
+
+    assert result.article_markdown == "正文"
+    assert result.images == []

wechat_article_parser-0.0.10/tests/test_video_content.py

@@ -0,0 +1,85 @@
+"""纯视频号卡片正文的离线回归测试。"""
+
+import pytest
+
+from wechat_article_parser import ArticleDisplayType
+from wechat_article_parser.parser import _parse_html
+
+
+_URL = "https://mp.weixin.qq.com/s/wAyM7yYKDfTH21-XpS9agQ"
+_CARD = (
+    '<mp-common-videosnap data-type="video" data-id="video-id" '
+    'data-desc="视频描述" data-url="https://example.com/cover.jpg">'
+    '</mp-common-videosnap>'
+)
+_IFRAME = (
+    '<iframe class="video_iframe rich_pages wx_video_iframe" data-mpvid="wxv_123" '
+    'data-src="https://mp.weixin.qq.com/mp/readtemplate?t=pages/video_player_tmpl'
+    '&amp;action=mpvideo&amp;vid=wxv_123"></iframe>'
+)
+_METADATA = """
+var biz = "MjM5NTAwMzE0MA==";
+var mid = "2908209265";
+var idx = "4";
+var sn = "signature";
+window.__allowLoadResFromMp = true;
+var hd_head_img = "https://example.com/avatar";
+var nickname = htmlDecode("卡车之家");
+var oriCreateTime = '1791462000';
+"""
+
+
+def _html(body: str, metadata: str = _METADATA) -> str:
+    return (
+        '<meta property="og:title" content="后双胎大轻客，江淮星锐EV6">'
+        f'<div class="rich_media_content">{body}</div>'
+        '<script type="text/javascript">var item_show_type = "0";</script>'
+        f'<script type="text/javascript">{metadata}</script>'
+    )
+
+
+@pytest.mark.parametrize("metadata", [_METADATA, ""])
+@pytest.mark.parametrize("video", [_CARD, _IFRAME])
+def test_video_only_body_gets_one_sentence_and_still_requires_metadata(
+    metadata: str, video: str,
+) -> None:
+    result = _parse_html(
+        _URL,
+        _html(f'<section>{video}</section><p><br></p><mp-style-type></mp-style-type>', metadata),
+    )
+
+    assert result.article_markdown == "这是一篇纯视频内容。"
+    assert result.article_display_type is ArticleDisplayType.ARTICLE
+    assert result.article_item_show_type == 0
+    assert result.images == []
+    assert result.is_valid is bool(metadata)
+
+
+@pytest.mark.parametrize("body", [
+    '<p>实际文字正文</p>',
+    '<img data-src="https://example.com/body.jpg">',
+])
+@pytest.mark.parametrize("video", [_CARD, _IFRAME])
+def test_video_with_other_content_keeps_existing_markdown(body: str, video: str) -> None:
+    result = _parse_html(_URL, _html(body + video))
+
+    assert result.is_valid
+    assert "这是一篇纯视频内容。" not in result.article_markdown
+    assert "cover.jpg" not in result.article_markdown
+    if "实际文字正文" in body:
+        assert result.article_markdown == "实际文字正文"
+    else:
+        assert result.images == ["https://example.com/body.jpg"]
+        assert result.article_markdown == "![](https://example.com/body.jpg)"
+
+
+@pytest.mark.parametrize("body", [
+    "",
+    _CARD.replace('data-type="video"', 'data-type="image"'),
+    '<iframe src="https://example.com/unrelated"></iframe>',
+])
+def test_empty_or_non_video_body_does_not_get_video_notice(body: str) -> None:
+    result = _parse_html(_URL, _html(body))
+
+    assert result.article_markdown == ""
+    assert not result.is_valid