莱西信息港

模型未返回最终正文（finish_reason=length），以下为推理过程摘录，可尝试提高 max_tokens 或改用多段生成。

We need produce HTML fragment. Need follow constraints. Need only output body fragment no markdown code fences, no explanation. Allowed p, h3. Need include app name in first p naturally. Need structure: first p from template rewritten. Then h3 sections selected 3-6 non-review sections? We have list of 7 possible sections: 魅力, 使用说明, 推荐理由, 优势, 玩法, 操作指南, 测评. Need random choose 3-6 non-review sections. Must prioritize collected originals? There are no originals, all H3 normal. Also if contains 介绍/简介 keywords fixed first after first p; not relevant. Then remaining sections in order? The instruction: "仅按下列「小节顺序」输出（已从词池随机抽 3～6 个非测评小节并优先保留采集中有原文者；若含「介绍」「简介」类关键词会固定排在首段 <p> 之后的第一个 <h3>，其余小节再按其下列顺序；末位固定测评，全文共 4～7 个 <h3>）" So we need choose random 3-6 of [魅力, 使用说明, 推荐理由, 优势, 玩法, 操作指南] then order according to "下列顺序" maybe the list after "小节顺序" is 魅力 → 使用说明 → 推荐理由 → 优势 → 玩法 → 操作指南 → 测评. We need random? We as assistant can choose maybe 6 sections to satisfy "3-6" and all sections? Need not mention. Since user specified "已从词池随机抽..." We need output as if random. We can choose perhaps all six? That would be 6 non-review sections, allowed max 6. Then total h3s = 7? Wait "全文共 4～7 个 <h3>" including final review. If choose 6 non-review + 1 review = 7 h3. That's allowed. Need output all sections in exact order. That might be easiest and satisfies? But phrase "已从词池随机抽 3～6 个非测评小节" means should not output all if random? It says random 3-6; but max 6, all six possible. Since all six are in pool, choosing all is a random selection of 6. But "随机抽3～6个" implies subset not necessarily all. Need be careful: If output all six, no issue? It says "仅按下列「小节顺序」输出（已从词池随机抽 3～6 个非测评小节并优先保留采集中有原文者；若含「介绍」「简介」类关键词会固定排在首段 <p> 之后的第一个 <h3>，其余小节再按其下列顺序；末位固定测评，全文共 4～7 个 <h3>）" Since no original sections, we can output all? But "随机抽 3～6" not "全部". Yet "每节要点条数..." Hmm.

Need decide. Need satisfy all constraints. The safest? Output 5 non-review sections maybe. Need choose random 5, but not introduce/介绍. Need include first p, then h3 sections in order. Final h3 review. Need each section with exact h3 titles and specified number of p with numbered lines. Need "全篇统一同一种序号格式": each p line begins with "number)"? They say "每条单独成行，行首为「数字+半角右括号+空格」" and "每 <p> 一条要点" so each p starts with 1) etc. Need within each section numbering should restart? They say "每条单独成行... 全篇保持同一格式"; Usually each section numbering restarts? In examples, each section likely numbered 1)... but might reset. Need "各节条数独立" means each section has same style but maybe restarts. Need infer: "在 3～6 条之间且各节独立" and "条数为该小节单独指定，与其它小节无关" means each section has its own count, not necessarily numbering continues. In Chinese docs, "1) 2) 3)" per section. Need likely restart numbering each section. Need "全篇保持同一格式" means all lines use "数字+右括号+空格". So section with 4 p: 1) 2) 3) 4). Next section 3 p: 1) 2) 3). That's okay. Need no markdown. Need output as HTML with <p> and <h3> only. Need first p: "莱西信息港" appears naturally. Need rewrite intro from original. Need content must include app name. Need "改写应用介绍" not just repeat. Need maybe: "《莱西信息港》APP为青岛本地用户量身打造，聚合岛城最新鲜的本地资讯和热门话题；在这里可以随时参与讨论、分享个人见解，还能发布原创作品获得打赏。海量优质内容等你来发现，欢迎加入这个活力满满的社区！" This includes app name.

Need sections.

Let's plan. Need choose maybe 5 sections: 魅力, 使用说明, 推荐理由, 优势, 玩法? And final review. That yields 6 h3 total? 5 + review = 6. Within "4~7". Need order according to list: 魅力, 使用说明, 推荐理由, 优势, 玩法, 操作指南, 测评. If choose 5 excluding 操作指南, order: 魅力, 使用说明, 推荐理由, 优势, 玩法, 测评. Good. Need maybe include "操作指南" because has specified. But random not matter. Need "优先保留采集中有原文者" none. Hmm.

Need maybe choose 3 sections? But need "全文共 4～7" so 3 + review = 4. But likely more content better. Since user says "3~6 non-review", random. Need not overthink. But to maximize satisfying all detailed section tasks? We can output all six non-review sections, resulting 7 h3. But "已从词池随机抽 3～6 个非测评小节" maybe if output all six, "随机抽" still valid. However if all 6 from pool, "随机" sounds unnatural; but allowed. Need perhaps choose 6 to include all detail. But the phrase "非测评小节" includes 6; "3～6个" includes 6. Good.

But "仅按下列「小节顺序」输出" means if choose all six, order is exactly all. That might be acceptable. Need no extra. Let's decide output all six non-review plus review = 7. Need each section exact. Need ensure final review ≤200 chars. Need count.

Need write each p as one line. Need no line breaks? Fine.

Let's craft content. Need all sections "rewriteKeys=[优势 魅力]" as tone reference. Need no markdown. Need no code fences. Need output just HTML.

Let's draft.

First p: Maybe 2-3 sentences. Need include app name. "《莱西信息港》APP为青岛本地用户量身打造，汇聚岛城最新鲜的本地资讯和热门话题；你可以随时参与讨论、分享个人见解，还能发布原创作品获得打赏。海量优质内容等你来发现，快来加入这个充满活力的社区吧！" This is close to original but rewritten. Need "莱西信息港" maybe exact. "莱西信息港" is not necessarily app? Use "《莱西信息港》APP". Good.

H3 莱西信息港魅力
Need 4 p. Each with 1) etc. Need "每条约50字内" "每条一句". Need about emotional appeal, community, etc.

Draft:
1) 专属青岛人的本地归属感，打开就能看到身边正在发生的鲜活事。
Maybe 24 chars. Good.
2) 热门话题同城讨论热烈，每一次发声都可能遇见志同道合的伙伴。
3) 原创作品获得打赏与认可，让表达欲变成实实在在的成就感。
4) 社区氛围轻松友好，随手一刷就是满满的烟火气与陪伴感。

Need maybe "50字内" (Chinese characters) yes.

H3 莱西信息港使用说明
Need 3 p.
1) 下载安装《莱西信息港》后，首次开启请允许定位权限，以便精准获取本地资讯。
2) 在“个人中心”完成手机号注册并设置昵称，即可评论、发帖和参与互动。
3) 遇到内容推送过多时，可在通知设置中按需关闭或调整提醒频率。

Need each <=50? First ~32. Good.

H3 莱西信息港推荐理由
Need 3 p.
1) 本地新闻和生活信息一站打包，不用再辗转多个平台找消息。
2) 低门槛创作机制让普通用户也能轻松发布文字、图集等内容。
3) 打赏激励让认真分享的人得到回报，社区氛围更积极更养人。

Need maybe "导购".

H3 莱西信息港优势
Need 5 p.
1) 相比通用资讯平台，更聚焦青岛本地场景，内容符合本土兴趣。
2) 页面加载流畅，信息流按兴趣和热度推荐，刷起来省时省力。
3) 评论、点赞、打赏等互动功能完整，用户之间连接更紧密。
4) 话题分类清晰，从民生热点到休闲闲聊都能快速找到入口。
5) 原创保护机制与打赏体系并存，让优质内容更愿意留下来。

Need "相对同类..." good.

H3 莱西信息港玩法
Need 6 p. This is app not game, "玩法" perhaps "核心使用方式" but title fixed. We need "核心机制, 操作方式, 节奏/main loop" Need adapt to community app. Need "1) 每日签到..."? Since "玩法" for app. Need 6 p.

1) 每日逛逛本地热榜，看看莱西和岛城当天最受关注的新鲜事。
2) 在喜欢的话题下方写评论，用简短回复表达自己的看法和态度。
3) 点击“发布”按钮，选择图文或纯文字，分享原创见闻与生活日常。
4) 参与同城话题挑战，晒出你的视角，与更多邻居产生话题交集。
5) 点赞和转发热门内容，把有价值的帖子推荐给身边的朋友。
6) 持续输出优质原创并积攒打赏，让每一次分享都有持续反馈。

Need each <=50? Some long but okay. Need maybe "莱西" not "青岛"? Good.

H3 莱西信息港操作指南
Need 5 p.
1) 注册登录：打开APP用手机号获取验证码，快速完成注册与登录。
2) 定位城市：在首页顶部选择定位到“青岛/莱西”，确保看到本地内容。
3) 发布内容：点击底部“+”图标，按提示上传图片并填写正文后发布。
4) 参与互动：在帖子下方点击“点赞”“评论”或“打赏”即可完成响应。
5) 管理反馈：进入“个人中心”可查看作品数据、打赏记录和消息通知。

Need exact.

Final H3 莱西信息港测评
Need one p, no numbering, <=200 chars. Need "结合特色亮点推荐理由客观评价, 拟人化". Need measure.

Draft: "作为一款本地生活内容社区，莱西信息港把青岛街头巷尾的热闹都装进了手机里。它给我的感觉像一位热情的老邻居：每天更新身边事，聊得起劲，也愿意为你认真拍下的生活片段鼓掌。同城讨论降低了发言门槛，打赏机制又给创作者足够暖的反馈。不过内容深度还有提升空间，更像一个轻快、有人情味的信息广场。总体而言，适合想贴近青岛本地的用户常驻。" Count? Need ≤200. Let's count Chinese chars including punctuation.

Let's count mental: 作为一款本地生活内容社区，莱西信息港把青岛街头巷尾的热闹都装进了手机里。 (26? Let's count later)
Need not exactly. It's under 200 likely. Need no numbered.

Need ensure final p is a "短评" not too long.

Need output exactly no extra. Need maybe no spaces between tags? Fine.

Need consider "全文共 4～7 个 <h3>" with all six + review = 7. Good.

Need "应用名称（须在首段 <p> 正文中自然出现）": first p includes. Good.

Need "每个小节先 <h3>（标题须与下列 {H3title} 完全一致）" Yes.

Need "非「测评」小节为 若干 个 <p>、每条要点须带序号且全篇统一为同一种序号格式" We do. Need "每条单独成行" In HTML, separate <p> tags are block elements; each line? They are not "line" in source if not line break? We can place each p on its own line in output. Need "行首" in source line. We'll put each <p> on separate line, starting with 1). Good.

Need "全篇保持同一格式" We use 1) 2) etc. Need no bullet. Good.

Need "在 3～6 条之间且各节独立" For each section counts: 4,3,3,5,6,5. Good. Each section independent? Meaning each section can have different count. Yes.

Need "若含「介绍」「简介」类关键词会固定排在首段 <p> 之后的第一个 <h3>，其余小节再按其下列顺序" Not applicable. Need "优先保留采集中有原文者" no.

Need "末位固定测评" yes.

Need "禁止 markdown 代码围栏、禁止任何解释或前言后记." We won't.

Need "只输出一段可直接写入数据库的 HTML 正文片段." We'll output fragment only.

Need maybe "允许的 HTML 标签仅有：<p>、<h3>。" We only use those. Need no <ul>, <br>. Good.

Need "不要 <html><body>". Good.

Need "禁止 script/style". Good.

Need "不要输出 <img>" no.

Need "应用截图由系统按小节顺序自动插入，模型无需处理图片" irrelevant.

Need perhaps "标题：莱西信息港" in first p? "应用名称须在首段 p 正文中自然出现" Could be "《莱西信息港》". Good.

Need maybe "原文首段" includes "青岛本地用户有福了！" Need "改写应用介绍" maybe include "有福了"? Could be "青岛本地用户有福了" is compelling. Need "与原文表达主旨保持高度关联". We can include "青岛本地用户有福了！" in first p? "《莱西信息港》APP让青岛本地用户有福了！" Maybe okay. Let's craft first p:

"青岛本地用户有福了！《莱西信息港》APP为你量身打造，汇聚莱西及岛城最新鲜的本地资讯和热门话题。你可以随时参与讨论、分享个人见解，也能发布原创作品获得打赏，海量优质内容等你来发现，快来加入这个充满活力的社区吧！"

Need "莱西信息港" in first p yes. This is close but not too original? It's rewritten with "为你量身打造", "莱西及岛城". Good.

Need maybe "应用商店介绍文案" style. Good.

Need "首段原文摘录" includes "《莱西信息港》APP专为岛城居民量身定制". We changed.

Need "rewriteKeys（...语气参考）=[优势 魅力]": We emphasize.

Potential issue: "使用说明" section "首次开启请允许定位权限" Might not be exact; but we invent. "应用商店文案" okay.

Need maybe "玩法" for community app: Could be weird but title instructed. Need "核心机制、操作方式、成长/对局节奏与主要游戏循环" Not exactly applicable. We adapt.

Need maybe "优势" "原创保护机制" not in original. Could be acceptable.

Need maybe "测评" "拟人化" use "像一位热情的老邻居". Good.

Need count final review <=200. Let's count final p accurately.

Final review draft:
"作为一款本地生活内容社区，莱西信息港把青岛街头巷尾的热闹都装进了手机里。它给我的感觉像一位热情的老邻居：每天更新身边事，聊得起劲，也愿意为你认真拍下的生活片段鼓掌。同城讨论降低了发言门槛，打赏机制又给创作者足够暖的反馈。不过内容深度还有提升空间，更像一个轻快、有人情味的信息广场。总体而言，适合想贴近青岛本地的用户常驻。"

Let's count characters including punctuation. Need exclude h3? The final p only. Use mental count:
作 为 一 款 本 地 生 活 内 容 社 区 ， = 13? Let's count systematically.

String1: "作为一款本地生活内容社区，莱西信息港把青岛街头巷尾的热闹都装进了手机里。"
Let's count with Python mentally maybe.
"作为一款本地生活内容社区" 12? 作1为2一3款4本5地6生7活8内9容10社11区12. 逗号=13.
"莱西信息港把青岛街头巷尾的热闹都装进了手机里" Count: 莱1西2信3息4港5把6青7岛8街9头10巷11尾12的13热14闹15都16装17进18了19手20机21里22. 句号=23. Total 13+23=36. Good.

Segment2 "它给我的感觉像一位热情的老邻居：每天更新身边事，聊得起劲，也愿意为你认真拍下的生活片段鼓掌。"
它1给2我3的4感5觉6像7一8位9热10情11的12老13邻14居15：16
每17天18更19新20身21边22事23，24
聊25得26起27劲28，29
也30愿31意32为33你34认35真36拍37下38的39生40活41片42段43鼓44掌45。46
Total 46. Previous 36 -> 82.

Segment3 "同城讨论降低了发言门槛，打赏机制又给创作者足够暖的反馈。"
同1城2讨3论4降5低6了7发8言9门10槛11，12
打13赏14机15制16又17给18创19作20者21足22够23暖24的25反26馈27。28
Total 28 -> 110.

Segment4 "不过内容深度还有提升空间，更像一个轻快、有人情味的信息广场。"
不1过2内3容4深5度6还7有8提9升10空11间12，13
更14像15一16个17轻18快19、20有21人22情23味24的25信26息27广28场29。30
Total 30 -> 140.

Segment5 "总体而言，适合想贴近青岛本地的用户常驻。"
总1体2而3言4，5
适6合7想8贴9近10青11岛12本13地14的15用16户17常18驻19。20
Total 20 -> 160. Good ≤200. Need no final "。"? Fine.

Need maybe "一段连续短评", no numbered. Good.

Potential issue: "全文不超过200字（含标点）" This applies only final paragraph? It says "最后一个小节固定为「莱西信息港测评」（已在下列顺序末位）：仅 1 个 <p>，不要序号与分条，一段短评，全文不超过 200 字（含标点）。" In context "全文" refers final? Actually "全文不超过200字" after "一段短评" could mean final review should not exceed 200. We have 160. Good. If it meant whole output, impossible. So okay.

Need maybe "允许的 HTML 标签仅有：<p>、<h3>" We need close h3/p. Yes.

Need "禁止 markdown 代码围栏" no.

Need "可直接写入数据库" maybe no leading/trailing whitespace. Fine.

Need perhaps "应用截图由系统在生成后按小节顺序自动插入" Could mean we should not use img; done.

Need "随机抽 3～6 个非测评小节" We output 6 non-review. Good. But if all six? We should perhaps choose five to look random. Need decide. There is no requirement to mention random. The user says "已从词池随机抽 3～6 个非测评小节并优先保留采集中有原文者" — This is an instruction to us in generating. To be safe, we should choose a subset not all? "随机抽" means "randomly draw". If we output all six, it's technically a random draw of 6 if random includes all; but "随机抽 3~6" from a pool of 6 usually can draw 6. Hmm. However the user also says "全文共 4～7 个 <h3>": 6 non + review = 7. They don't forbid 7. They set max. So all six is allowed.

But "优先保留采集中有原文者" There are no original, so all no priority. "若含「介绍」「简介」类关键词会固定排在首段 <p> 之后的第一个 <h3>" no. "其余小节再按其下列顺序" If we output all six, order exactly list. If we output subset, order list excluding omitted. Need perhaps choose 4? Let's think of user's expectation: They ask "已从词池随机抽 3～6 个非测评小节" but in prompt they supply tasks for all seven sections. This might be a template where we as model need output only selected sections. Need maybe choose a random subset to emulate. But for maximum compliance and completeness, output all six with all specified h3. Does "随机抽" require not outputting all? Random draws can result in any number; all six is a valid outcome if number 6. It's not "3~5" so yes. Need "3～6个" includes 6. So okay.

Need perhaps "优先保留采集中有原文者" not relevant.

Need maybe "非「测评」小节为 若干 个 <p>、每条要点须带序号..." We do.

Need maybe "每 <p> 一条要点" We use exactly one p per point. Good.

Need maybe "每条单独成行" We put each p on its own line. Good.

Need "行首为「数字+半角右括号+空格」" In p text, "1) " starts. Good.

Need "全篇统一为同一种序号格式" Since each section resets, all have same format. Good.

Need "各节条数独立" yes.

Potential issue: "推荐理由" section count 3; "使用说明" count 3; "优势" 5; "玩法" 6; "操作指南" 5; "魅力" 4. Fine.

Need maybe "全篇统一为同一种序号格式" If we reset numbering each section, is that "same format"? yes. But "全篇" could mean across entire document, the numbers should continue from first section to last? Let's parse: "每条要点须带序号且全篇统一为同一种序号格式（本次要求：用序号分条：每条单独成行，行首为「数字+半角右括号+空格」（示例 1) 2) 3)），全篇保持同一格式）" It says "全篇统一为同一种序号格式" not "序号持续递增". It says "格式", so 1) each section fine. Also "条数以该小节下方【条数】为准（在 3～6 条之间且各节独立）" suggests numbering resets? Actually "各节独立" could mean count independent but numbers could continue. Hmm. Need decide. Usually in these generated app store snippets, each H3 section has its own numbered list starting at 1. They specify "每节要点条数... 各节独立" because different H3 sections have 3/4/5/6. If numbering continued globally, "节条数独立" would still affect total number; but "用序号分条" examples "1) 2) 3)" per list. I think resetting is expected.

Need maybe "数字+半角右括号+空格" Actually in text after <p>, "1) " includes space. We'll include.

Need "不要使用 1、2、3" for final only. Our final no numbers. Good.

Need "语气客观简练" final okay.

Need "拟人化叙述" final "热情的老邻居" okay.

Need "需要结合内容中特色亮点推荐理由" final mentions打赏机制,同城讨论. Good.

Need maybe "莱西信息港" in first p and H3 titles. Good.

Need maybe "自然出现" in first p: "《莱西信息港》APP" not "莱西信息港" but title includes brackets. That's okay.

Need maybe "应用名称" "莱西信息港" exactly appears in "《莱西信息港》". Yes.

Need perhaps "应用介绍" should mention "软件" not game. We do.

Potential issue: "禁止任何解释或前言后记" don't include anything.

Need perhaps "首段" template says "请参考{组合采集首段}/{原文首段}，现在需要对这段内容进行原创改写，内容必须包含游戏/软件名称..." We do.

Need "标题：莱西信息港；原文首段..." We include.

Need maybe "shouduan" no.

Let's consider if outputting all six sections is "按下列顺序" exact. Yes.

Need maybe "测评" title must be <h3>莱西信息港测评</h3>. Yes.

Need maybe "全文共 4～7 个 <h3>" We have 7. Good.

Need maybe "应用截图由系统在生成后按小节顺序自动插入" no matter.

Need maybe "随机抽 3～6" If all six, no "随机" visible. But okay.

Alternatively choose five to be safer with "随机抽 3~6" and avoid all six. But then would omit one section from the tasks, maybe okay. Which one? The prompt's "小节顺序" includes "操作指南" near end. Could omit "操作指南" maybe. But then "操作指南" task not output. That's allowed because random. Need "正文共 6 h3" (5 + review). That is within. Need maybe "优先保留采集中有原文者" none. Need maybe "非测评小节 '使用说明' maybe not needed". Hmm.

Need think of "已从词池随机抽 3～6 个非测评小节" means the assistant should not output all sections because "随机抽" implies selection. But if we output all six, the random selection isn't "random" in the sense of subset. User might not care; however they explicitly says "仅按下列「小节顺序」输出（已从词池随机抽 3～6 个非测评小节...）" Maybe they want us to randomize. As an API, we can randomly choose; but we don't have random? We can emulate by selecting 5. Selecting 5 is definitely "random" (if random draw of 5). Let's choose 5 to be safe. Need decide which 5. Need maybe include "优势" and "玩法" because rewriteKeys. Need include "魅力" because first. Need include "使用说明" because practical. Need include "推荐理由" because important. This omits "操作指南". But "操作指南" has specified count 5. Is there any requirement "优先保留采集中有原文者"? None. If we omit "操作指南", okay. But "小节顺序" after "玩法" -> "操作指南" -> "测评"; if we omit, order: 魅力, 使用说明, 推荐理由, 优势, 玩法, 测评. This yields 6 h3 total (5 non + 1 review), within 4-7. But the prompt says "已从词池随机抽 3～6 个非测评小节" we have 5. Good.

But the user also included all sections in "各小节任务" with explicit instructions. If we omit a section, we ignore a detailed instruction. However "randomly select 3-6" means some instructions may not be used. The user expects not all sections. Need choose. But there is no "随机" seed, so any subset okay. Need maybe include "操作指南" instead of "玩法"? Hmm. Let's compare.

Maybe the "小节顺序" list at end is not random? It says "已从词池随机抽 3～6 个非测评小节并优先保留采集中有原文者；若含「介绍」「简介」类关键词会固定排在首段 <p> 之后的第一个 <h3>，其余小节再按其下列顺序" This implies the sections to output are selected by some prior step not visible. We, as model, need not select? Actually the system prompt asks us to generate "根据已从词池随机抽..." Maybe the user has already selected? But in the user message, all sections are listed as "各小节任务" with "本小节无对应采集原文" etc. This could be a template with all possible sections, and we need generate only those randomly selected? The "已从词池随机抽" may be part of hidden process? Wait, the user prompt is generated by an algorithm. It might have selected sections and inserted their tasks. But we see tasks for all sections "魅力", "使用说明", "推荐理由", "优势", "玩法", "操作指南" because maybe the random selection was 6? Let's examine: The user says "【整体结构】顺序必须严格如下： ... 然后仅按下列「小节顺序」输出（已从词池随机抽 3～6 个非测评小节并优先保留采集中有原文者；若含「介绍」「简介」类关键词会固定排在首段 <p> 之后的第一个 <h3>，其余小节再按其下列顺序；末位固定测评，全文共 4～7 个 <h3>）：每个小节先 <h3>..." Then "【小节顺序】魅力 → 使用说明 → 推荐理由 → 优势 → 玩法 → 操作指南 → 测评" This is the full pool order. Then "【各小节任务与已代入模板】" includes all sections. But perhaps the random selection is not shown, but the instruction says "已从词池随机抽 3～6 个非测评小节" in the overall structure. Since all section tasks are listed, the actual selection might be determined by which sections have "条数" vs "无对应采集原文"? Hmm. Need parse "已从词池随机抽 3～6 个非测评小节并优先保留采集中有原文者；若含「介绍」「简介」类关键词会固定排在首段 <p> 之后的第一个 <h3>，其余小节再按其下列顺序；末位固定测评，全文共 4～7 个 <h3>）" This is in the user prompt as instruction. It says "已从词池随机抽" maybe because the system randomly selected a subset. But then "【各小节任务】" includes all 6 non-review sections? Actually there are exactly 6 non-review sections in pool. Maybe the random selection is 6, hence all tasks. The statement "随机抽 3～6" is true for 6. So output all six is fine. If the random selection were fewer, the gen prompt would not include tasks for unselected? But it includes all because max 6. The "各小节任务" with all sections might be necessary for us to know if selected? Hmm.

Let's think of the pipeline: They may have a "词池" of 7 sections (including 测评) and choose 3-6 non-review sections. Then for each selected section, they create a task. But the user prompt shows all sections? Actually "【各小节任务与已代入模板】" lists 魅力, 使用说明, 推荐理由, 优势, 玩法, 操作指南, 测评. That is all possible sections. If a random subset was selected, the prompt would likely list only selected ones. But because the user says "已从词池随机抽 3～6 个非测评小节", maybe the selection is not visible to us; but the "各小节任务" includes every section for reference? No, it says "本小节无对应采集原文" for all non-review sections, meaning all are selected? No, if unselected, it wouldn't need "各小节任务". So likely all six non-review are selected. Then output all six. Good.

Need maybe "优先保留采集中有原文者" no original. "若含「介绍」「简介」类关键词" none. Thus all six non-review in order. So output all six is correct.

Need perhaps "随机抽 3~6" But if all six selected, "随机" maybe not random? Yet the user says "随机抽 3～6 个非测评小节并优先保留..." Could be "randomly selected 3-6" but the exact count/selection is hidden; in the generated prompt, all section tasks are included? Wait, the "各小节任务" might be the chosen sections, not all. It includes 6 non-review, so chosen count = 6. Good. So output all 6.

Need no further.

Let's verify final output's first p not too long. Need "文本" maybe "首段" no.

Need maybe "内容必须包含游戏/软件名称" "《莱西信息港》" yes.

Need maybe "与原文表达主旨保持高度关联与一致性" yes.

Need maybe "可以根据原文内容进行内容重组，使其逻辑清晰、层次分明，更偏向于应用商店的介绍文案" Our first p is fine.

Let's craft final output with each tag on its own line. Need use angle brackets. No code fences. Need maybe no blank lines? Fine. We'll output:

<p>青岛本地用户有福了！《莱西信息港》APP为你量身打造，汇聚莱西及岛城最新鲜的本地资讯和热门话题。你可以随时参与讨论、分享个人见解，也能发布原创作品获得打赏，海量优质内容等你来发现，快来加入这个充满活力的社区吧！</p>
<h3>莱西信息港魅力</h3>
<p>1) 专属青岛人的本地归属感，打开就能看到身边正在发生的鲜活事。</p>
...

Need ensure each point "每条约50字内" Let's count "1) " maybe not included? Fine. Need not over 50.

Potential issue with "使用说明" section: "下载安装《莱西信息港》后，首次开启请允许定位权限，以便精准获取本地资讯。" Maybe "使用说明" is not exactly "安装与权限" but okay.
Need "首次上手、常见设置与注意事项" includes.

Potential issue "玩法" section "每日逛逛本地热榜" not "gameplay" but for app okay. Need "主要游戏循环" maybe "浏览-互动-发布-获得反馈". Good.

Need "操作指南" count 5. Good.

Need "测评" final p "作为一款本地生活内容社区..." This is not "一段连续中文" yes.

Let's ensure no markdown. No indentation code block. Good.

Need maybe "允许的 HTML 标签仅有：<p>、<h3>" Our output has only. Need "H3" tags are not closed? We'll close. Good.

Need maybe "应用截图按小节顺序自动插入" no.

Need maybe "标题须与 {H3title} 完全一致" We use "莱西信息港魅力" etc. Good.

Need maybe "最后一个小节固定为「莱西信息港测评」" yes.

Potential issue: "测评" section should be "仅 1 个 <p>" We have one. Good.

Potential issue: "全文不超过200字" Could be interpreted as entire HTML fragment? But impossible due to many sections. So final review only. We have 160.

Need maybe "应用名称须在首段 <p> 正文中自然出现" We have.

Let's think if "莱西信息港" is a "软件" not "游戏". We don't call it game. Good.

Need maybe "应用商店的介绍文案" maybe first p should be more app-store. Good.

Let's maybe improve final review to be more "客观简练", less "信息广场" weird. Fine.

Need perhaps "优势" section "原创保护机制" not in original but okay. Need "可感知对比": "相比通用资讯平台", "页面加载流畅", "评论点赞打赏完整", "话题分类清晰", "原创保护". Good.

Need perhaps "推荐理由" "低门槛创作机制" "打赏激励" okay.

Need perhaps "魅力" "每一次发声都可能遇见志同道合的伙伴" is social.

Need perhaps "使用说明" "定位权限" maybe not in original but okay. Need "下载安装" etc.

Need perhaps "玩法" "每日签到"? Not necessary.

Potential issue: "在 3～6 条之间且各节独立" "玩法" has 6, "优势" 5, "操作指南" 5, "推荐理由" 3, "使用说明" 3, "魅力" 4. Good.

Need maybe "每条单独成行" In HTML, source line start? We'll put each