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。
我们需要输出符合要求的HTML片段。需要仔细遵循要求。用户要求：只输出一段可直接写入数据库的 HTML 正文片段。禁止markdown代码围栏、禁止解释。允许的标签只有<p>、<h3>。不能有img。不要html body外壳，不要script/style。
我们需要生成从首段<p>开始，然后按顺序小节。需要根据给出的结构。
整体结构：
1. first element is <p>：依据首段模板改写应用介绍，包含应用名称“炙热江湖”，与原文高度关联。我们需要原创改写首段内容，不是照抄。应该是一段话，可能多个句子。
2. Then only following small sections in order: 游戏介绍 → 亮点 → 优势 → 魅力 → 玩法 → 操作指南 → 测评. But note: 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>. However user also gave "整体结构" with "然后仅按下列「小节顺序」输出" listing all seven? Let's parse.
Original user says: "【整体结构】顺序必须严格如下：
1) 第一个元素是一个 <p>...
2) 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>（标题须与下列 {H3title} 完全一致），非「测评」小节为 **若干** 个 <p>、每条要点须带序号... 【小节顺序】游戏介绍 → 亮点 → 优势 → 魅力 → 玩法 → 操作指南 → 测评"
Then in the detailed tasks, it includes all seven sections, each with specific number of <p> and titles. So likely we need output all seven sections? But the parenthetical says randomly selected 3-6 non-evaluation sections and preferentially retain ones with collected original; if contains "介绍"/"简介" keyword fixed first after first p, other sections in order; final evaluation, total 4-7 h3. But then the "各小节任务与已代入模板" includes all seven sections, each with fixed number of points. It says "本小节无对应采集原文，按「原创H3-普通」撰写" for each, and then for evaluation too. There is no selection instruction to choose subset. Hmm.
Need interpret. The system likely has a template with random selection of 3-6 non-eval sections. But in the user-provided prompt, they actually inserted all sections? Let's read opening: "【整体结构】顺序必须严格如下： ... 2) 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>（标题须与下列 {H3title} 完全一致），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**（本次要求：用序号分条：每条单独成行，行首为「数字+英文句点+空格」（示例 1. 2. 3.），全篇保持同一格式）；**每节要点条数以该小节下方【条数】为准**（在 3～6 条之间且各节独立）。
3) **最后一个小节固定为「炙热江湖测评」**（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）。
【小节顺序】游戏介绍 → 亮点 → 优势 → 魅力 → 玩法 → 操作指南 → 测评"
That "仅按下列「小节顺序」输出" but then the structural constraints "已从词池随机抽 3～6 个非测评小节" indicates maybe the actual task list contains only selected sections? But in this prompt, it lists all seven sections in "各小节任务与已代入模板". Actually the "词池" may include only non-eval sections; but the user has injected all seven? Let's see: "各小节任务与已代入模板" includes every section. It might be that the system assembled a prompt by looping over all available section types, but the "randomly drawn" note remains generic? The parenthetical 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" but then lists "游戏介绍 → 亮点 → 优势 → 魅力 → 玩法 → 操作指南 → 测评" as the order. It explicitly says "全文共 4～7 个 h3" with final eval, so if all seven sections are output, that's 7 h3, within range 4-7. So it might be allowed to output all 7. But does "仅按下列「小节顺序」输出" mean output only those that appear in the following list? The list includes all. However "已从词池随机抽 3～6 个非测评小节" contradicts "all seven" because non-eval are six; max 6, so outputting all six non-eval + eval = 7, exactly max. So possible: random selection yielded six non-eval sections? Actually "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" — since all non-eval sections have no original, but the user has "优先保留采集中有原文者" maybe none, so randomly selected 3-6 among non-eval. But the prompt includes all six? Wait the "各小节任务与已代入模板" includes all six non-eval sections, each with explicit fixed number 5,3,3,3,4,6. That strongly suggests output all six, because they're all specified. But the parenthetical "已从词池随机抽 3～6 个非测评小节" could be a leftover from a meta-prompt; however "本次要求" says "用序号分条" etc. Hmm.
Need decide. The safest? The user explicitly says "顺序必须严格如下" and then includes all sections in "小节顺序"; then in the detailed task for every section, including all. If we omit sections, we'd violate "各小节任务" which include all. Also total h3 would be 7, within 4-7. So output all seven h3 sections. Need not worry about random selection. The instruction "仅按下列「小节顺序」输出" can be read as all the listed sections. Also "末位固定测评" final. So we should output all.
Need ensure first <p> is a rewritten intro, not from template exactly. It must include game name. The first paragraph 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，与原文的表达主旨保持高度关联与一致性。可以根据原文的内容进行内容重组，使其逻辑清晰、层次分明，更偏向于一个应用商店的介绍文案。标题：炙热江湖；原文首段：炙热江湖免广告版是一款充满无限可能的武侠角色扮演游戏。在这个广阔的武侠世界中，你可以自由选择自己的道路，无论是独行江湖、创立门派，还是投身朝廷，都能找到属于自己的精彩。游戏中的NPC会随着时间成长，与他们的互动将触发各种随机事件，让每一次冒险都充满惊喜。天机阁每年发布的天地榜、龙凤榜等榜单，以及比武大会和比武招亲等活动，让江湖生活更加丰富多彩。快来下载炙热江湖免广告版，开启你的武侠传奇吧！"
So first p should be an original rewrite of this paragraph, mentioning "炙热江湖" maybe not "免广告版" because app name is 炙热江湖. Need include in natural first paragraph. It can be around 2-3 sentences. Need not use numbering in first p. It should be a paragraph.
Then h3 sections. Each h3 title exactly: <h3>炙热江湖游戏介绍</h3> etc. For each non-review section, it says "随后 5 个 <p>" etc. Each p is one point, with sequence number at beginning. "每条单独成行" in the output HTML: each <p> contains "1. ..." and separate p. Need use the same format throughout: "数字+英文句点+空格" (1. 2. 3.). Important: "全篇统一为同一种序号格式" — so across sections, numbering? It says "每条单独成行，行首为「数字+英文句点+空格」（示例 1. 2. 3.），全篇保持同一格式". Does that mean numbering must restart per section or continue globally? Usually in such structured generated content, each section's points enumerated from 1. within each h3. The phrase "各节要点条数以该小节下方【条数】为准" and "各节独立" suggests per section independent. It says "每节要点条数以该小节下方【条数】为准（在 3～6 条之间且各节独立）" So numbering can restart at 1 in each section. But "全篇统一为同一种序号格式" not necessarily globally continuous. We should use 1., 2., ... within each section. For sections with 5,3,3,3,4,6, restart. Good.
Need handle "操作指南" and "玩法" rewriteKeys as tone reference? Not important.
Need ensure final <p> for 测评 is one continuous paragraph, no numbering, max 200 chars. Need count Chinese characters carefully. We need output a <h3>炙热江湖测评</h3> followed by one <p> with a short review. Need not exceed 200 chars including punctuation. Need maybe 150-200. We'll craft.
Need think about content. Let's create all sections.
First paragraph: Use "炙热江湖" multiple times? Need natural. We can write: "《炙热江湖》免广告版是一款充满无限可能的武侠角色扮演游戏，玩家在这片辽阔江湖中既可独行闯荡，也能创立门派或投身朝廷。游戏里的NPC会随时间成长，互动触发随机事件，配合天机阁的天地榜、龙凤榜以及比武大会、比武招亲等玩法，让每次冒险都充满新鲜感。快来下载炙热江湖免广告版，开启属于你的武侠传奇吧！" This closely follows original but rewritten. It's a p. No h before? First element is p.
Section 游戏介绍: h3 then 5 <p> with numbered points. Need each point about type, setting, main gameplay. Must be within ~50 chars per point. Need count? Not strictly but "每条约50字内". We'll keep concise.
Possible:
1. 炙热江湖是一款以自由武侠世界为舞台的角色扮演游戏。
2. 玩家可独行江湖、创立门派或投身朝廷，选择多样。
3. 游戏中的NPC会随时间成长，互动触发随机事件。
4. 天机阁定期发布天地榜、龙凤榜等榜单，丰富玩法。
5. 比武大会与比武招亲等活动让江湖生活更加精彩。
Each roughly 25-35 chars. Good.
Section 亮点: 3 points. Need distinctive highlights:
1. 免广告干扰，让玩家全身心投入武侠冒险。
2. NPC会持续成长，交互与随机事件令探索充满惊喜。
3. 多样江湖身份与榜单活动带来极高的自由度和可玩性。
Need "免广告" is from original title "免广告版". Good.
Section 优势: 3 points. "相对同类在玩法、画面、系统、优化等方面的长处，宜写可感知对比." Since no original beyond text, we can infer:
1. 相比同类武侠RPG，炙热江湖提供更自由的成长路线与身份选择。
2. NPC成长与随机事件系统让每次开局都拥有不同体验。
3. 无广告打扰，配合丰富的大会与榜单内容，节奏更沉浸。
Need plausible.
Section 魅力: 3 points. Emotional appeal:
1. 可以按照自己的心意书写江湖传奇，代入感强烈。
2. 与NPC的随机邂逅和成长羁绊让江湖世界充满温度。
3. 从无名小卒到名震天下，每一步成长都让人陶醉。
Section 玩法: 4 points. Core mechanics/loop:
1. 玩家可自由探索地图，接取任务并触发随机关卡事件。
2. 通过修炼武学、结交侠客和扩张势力来提升实力。
3. 参加比武大会、比武招亲等活动，赢取荣誉与奖励。
4. 天地榜、龙凤榜等排名系统持续提供目标与挑战。
Section 操作指南: 6 points. Concrete steps and interface instructions. Need "可照做". Need be plausible for mobile? We can write:
1. 进入游戏后先创建角色，选择身份背景与初始门派。
2. 在主界面点击“江湖”按钮可进入野外地图探索。
3. 与NPC对话时留意选项，可触发随机事件或接受任务。
4. 通过“武学”界面装备和升级已学到的功法技能。
5. 打开“天机阁”可查看天地榜、龙凤榜并领取奖励。
6. 在“活动”界面参与比武大会或比武招亲即可挑战。
Need ensure each is concise. Good.
测评: Need one <p> continuous, no numbering, <=200 chars. Need objective. Could write:
"炙热江湖用自由的路线、鲜活的NPC和密集的榜单活动，勾勒出一个很有生气的武侠世界。免广告设计让体验更加沉浸，随机事件与人物成长则让每次闯荡都有新鲜感。虽然画面不是极致，但玩法和代入感足够打动喜欢武侠的玩家。如果你向往快意恩仇、由自己掌握命运的江湖，这款游戏值得一试。"
Let's count roughly: Chinese chars plus punctuation. Let's count accurately.
String: "炙热江湖用自由的路线、鲜活的NPC和密集的榜单活动，勾勒出一个很有生气的武侠世界。免广告设计让体验更加沉浸，随机事件与人物成长则让每次闯荡都有新鲜感。虽然画面不是极致，但玩法和代入感足够打动喜欢武侠的玩家。如果你向往快意恩仇、由自己掌握命运的江湖，这款游戏值得一试。"
Let's count:
炙热江湖用自由的路线(9? "炙1 热2 江3 湖4 用5 自6 由7 的8 路9 线10" Actually "炙热江湖" 4, "用"1=5, "自由"2=7, "的"1=8, "路线"2=10, "、"=11, "鲜活"2=13, "的"=14, "NPC" letters count? Usually Chinese word count includes letters? We can approximate. "NPC" is 3 characters. Then "和" 1, "密集"2, "的"1, "榜单"2, "活动"2, "，"1, "勾勒"2, "出"1, "一个"2, "很有"2, "生气"2, "的"1, "武侠"2, "世界"2, "。"1. This is getting long. Need not be exact but must be ≤200. It likely is around 150. Let's calculate more systematically later.
Need ensure "全篇不超过200字" for evaluation only? "全文不超过200字" in item 3 says "一段短评，全文不超过200字" Actually "全文" refers to the evaluation paragraph? It says "末位固定测评... 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）" Thus evaluation text <=200. We'll be well under.
Need ensure output is only HTML fragment, no code fences. Should not include "```html". So final answer exactly fragment.
Need consider line breaks and indentation? Fine. We can include line breaks between elements. But "禁止 markdown 代码围栏" no problem. "允许的 HTML 标签仅有 <p>、<h3>" — we will use exactly those. Need no extra tags. <h3> and <p> on separate lines. No bullet tags. Good.
Need think if first paragraph should include app name naturally. We'll include "炙热江湖" and "炙热江湖免广告版" maybe. The app name is "炙热江湖" but original first paragraph uses "炙热江湖免广告版". Since application name requirement says "应用名称（须在首段 <p> 正文中自然出现）：炙热江湖". So first p should contain "炙热江湖". We can mention both "《炙热江湖》" and "免广告版" but no angle brackets? Text can include Chinese title quotes. Fine.
Need ensure "炙热江湖游戏介绍" is first h3 after first p. Yes.
Let's double-check exact titles:
- <h3>炙热江湖游戏介绍</h3>
- <h3>炙热江湖亮点</h3>
- <h3>炙热江湖优势</h3>
- <h3>炙热江湖魅力</h3>
- <h3>炙热江湖玩法</h3>
- <h3>炙热江湖操作指南</h3>
- <h3>炙热江湖测评</h3>
No extra text.
Need ensure "非「测评」小节为 **若干** 个 <p>、每条要点须带序号" — yes.
Need ensure "每节要点条数以该小节下方【条数】为准" — we follow.
Potential issue: "游戏介绍" section has 5 points; "亮点" has 3; "优势" 3; "魅力" 3; "玩法" 4; "操作指南" 6. Total non-eval h3 = 6, total h3 = 7. The parenthetical allowed 4-7 h3. Good.
Need ensure "顺序必须严格如下" — intro p then game intro, etc. Good.
Now craft content more elegantly. Need avoid "应用截图" etc.
Let's refine first paragraph and sections to be cohesive, with original and app store tone.
First p:
"《炙热江湖》免广告版是一款充满无限可能的武侠角色扮演游戏。你可以在广袤武林中自由选择道路，独行江湖、创立门派或投身朝廷，书写专属传奇；NPC会随时间成长并触发随机事件，天机阁的天地榜、龙凤榜及比武大会等活动更为冒险增添无限精彩。"
This includes "炙热江湖" and is a rewrite. Maybe "免广告版" is included. Good. Need not too long. It's okay.
Need note: "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... 更偏向于一个应用商店的介绍文案。" So yes.
Now sections. Need use numbered p lines. Could be:
<section>
<h3>炙热江湖游戏介绍</h3>
<p>1. 炙热江湖是一款以自由武侠世界为背景的角色扮演游戏。</p>
<p>2. 玩家可选择独行江湖、创立门派或投身朝廷，玩法自由。</p>
<p>3. 游戏中的NPC会随时间成长，互动可触发各种随机事件。</p>
<p>4. 天机阁定期发布天地榜、龙凤榜等榜单，内容十分丰富。</p>
<p>5. 比武大会与比武招亲等活动让江湖生活更加多彩。</p>
</section>
Need each <p> line starts with "1. " etc. Good.
Section 亮点:
<p>1. 免广告设计让玩家可以无干扰地沉浸于江湖历练。</p>
<p>2. NPC成长与随机事件令每次探索都充满意外惊喜。</p>
<p>3. 多元身份和榜单比武带来极强的长期可玩性。</p>
Need "免广告" from original. Good.
Section 优势:
<p>1. 自由路线与多线身份让玩家拥有远超同类的选择空间。</p>
<p>2. 天机阁榜单和江湖活动持续更新，保持玩法新鲜感。</p>
<p>3. 无广告打扰，搭配丰富NPC交互，营造沉浸式体验。</p>
Potential "更新" not in original, but fine? Better avoid "更新" unknown. "持续带来目标" perhaps. "天机阁榜单和江湖活动提供了清晰目标与丰富奖励" Good.
Section 魅力:
<p>1. 每个人都能按心意走出独特江湖路，代入感极强。</p>
<p>2. NPC会成长、会回应，让这个虚拟世界更有人情味。</p>
<p>3. 从初出茅庐到名动江湖，成长的成就感令人沉浸。</p>
Section 玩法:
<p>1. 自由探索地图，通过对话和选项触发随机事件与任务。</p>
<p>2. 修炼武学、结交侠客、发展门派，逐步提升自身实力。</p>
<p>3. 参加比武大会和比武招亲，争夺丰厚奖励与江湖声望。</p>
<p>4. 借助天地榜、龙凤榜等排名系统，不断挑战更高目标。</p>
Section 操作指南:
Need "6 条" steps. Use "点击" etc. But the game might be mobile? "免广告版" maybe Android game. We can infer standard. Need "可照做" and clear.
<p>1. 进入游戏后先创建角色，并选择想要的身份背景。</p>
<p>2. 在主界面点击“江湖”按钮，进入世界地图自由探索。</p>
<p>3. 与NPC对话时仔细查看选项，可触发事件或领取任务。</p>
<p>4. 打开“武学”界面，可为角色装备和升级已学功法。</p>
<p>5. 进入“天机阁”查看榜单信息，并领取对应的排名奖励。</p>
<p>6. 在“活动”界面选择比武大会或比武招亲，即可参与挑战。</p>
Need "操作指南" section's points may use quotes, allowed. Good.
Review paragraph: Need avoid numbering and use "一段连续短评". We'll write one p. Need maybe "炙热江湖" appears. It may be okay.
Let's craft a review around 150-180 chars. Let's count exactly to ensure ≤200.
Candidate:
"炙热江湖以自由的路线、鲜活的NPC和密集的榜单活动，构成了一个很有生气的武侠世界。免广告体验让人更容易沉浸其中，随机事件与人物成长则让每次闯荡都不重样。虽然画面不算极致，但玩法深度和江湖氛围足够打动玩家。若你向往快意恩仇、自定命运的江湖，这款游戏值得一试。"
Let's count characters including punctuation. Need count accurately. I'll count by units.
"炙热江湖以自由的路线、鲜活的NPC和密集的榜单活动，构成了一个很有生气的武侠世界。免广告体验让人更容易沉浸其中，随机事件与人物成长则让每次闯荡都不重样。虽然画面不算极致，但玩法深度和江湖氛围足够打动玩家。若你向往快意恩仇、自定命运的江湖，这款游戏值得一试。"
Let's count with Chinese punctuation. I'll do careful enumeration.
Take string segments:
A: "炙热江湖以自由的路线、鲜活的NPC和密集的榜单活动，"
Count "炙热江湖"4, "以"1=5, "自由"2=7, "的"1=8, "路线"2=10, "、"=11, "鲜活"2=13, "的"1=14, "NPC"3=17, "和"1=18, "密集"2=20, "的"1=21, "榜单"2=23, "活动"2=25, "，"=26. A=26.
B: "构成了一个很有生气的武侠世界。"
"构成"2=28, "了"1=29, "一个"2=31, "很有"2=33, "生气"2=35, "的"1=36, "武侠"2=38, "世界"2=40, "。"=41. B=15 (A+B 41).
C: "免广告体验让人更容易沉浸其中，"
"免"1=42, "广告"2=44, "体验"2=46, "让人"2=48, "更"1=49, "容易"2=51, "沉浸"2=53, "其中"2=55, "，"=56. C=15.
D: "随机事件与人物成长则让每次闯荡都不重样。"
"随机"2=58, "事件"2=60, "与"1=61, "人物"2=63, "成长"2=65, "则"1=66, "让"1=67, "每次"2=69, "闯荡"2=71, "都"1=72, "不重样"3=75? "不"1, "重"1? Actually "重样"2, so "不重样"=3, sum 75, "。"=76. D count 20? Let's recalc: from 56 to 76 =20. Good.
E: "虽然画面不算极致，但玩法深度和江湖氛围足够打动玩家。"
"虽然"2=78, "画面"2=80, "不算"2=82, "极致"2=84, "，"=85, "但"1=86, "玩法"2=88, "深度"2=90, "和"1=91, "江湖"2=93, "氛围"2=95, "足够"2=97, "打动"2=99, "玩家"2=101, "。"=102. E count 26? From 76 to 102 =26. Good.
F: "若你向往快意恩仇、自定命运的江湖，这款游戏值得一试。"
"若"1=103, "你"1=104, "向往"2=106, "快意"2=108, "恩仇"2=110, "、"=111, "自定"2=113, "命运"2=115, "的"1=116, "江湖"2=118, "，"=119, "这款"2=121, "游戏"2=123, "值得"2=125, "一试"2=127, "。"=128. Total 128. Good under 200.
Need ensure "NPC" letters counted as 3 but total still okay.
Need maybe final paragraph "全文不超过200字" means the entire final one, yes.
Potential issue: The first paragraph "《炙热江湖》免广告版" uses "免广告版" which includes "广告" but app name in title is "炙热江湖" — fine. Need maybe "《炙热江湖》免广告版" is okay. It is a game "免广告版"; but "应用名称" must be "炙热江湖" naturally in first paragraph. It appears.
Need check if "应用截图由系统在生成后按小节顺序自动插入" means we shouldn't mention screenshots. Fine.
Need ensure no `section` tags, no HTML comments. We'll output only p/h3.
Need consider "每个小节先 <h3>（标题须与下列 {H3title} 完全一致）" — our titles are exact.
Need consider "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**" — yes.
Need consider "每条单独成行" — in HTML, each <p> is a block. Good.
Need maybe "全篇统一为同一种序号格式" means no "1." vs "1、". We use "1. " consistently.
Need maybe "数字+英文句点+空格" — after space, text. We'll include a space after "1.".
Need ensure no trailing punctuation? Fine.
Let's think about possible issue with "游戏介绍 → 亮点 → 优势 → 魅力 → 玩法 → 操作指南 → 测评" from "小节顺序" and "标题须与下列 {H3title} 完全一致" — In the detailed tasks, for "游戏介绍" title says "<h3>炙热江湖游戏介绍</h3>（与 {H3title}/{h3} 一致）". Good.
Need maybe "游戏介绍" should be "炙热江湖游戏介绍", not "游戏介绍" alone. Yes.
Let's examine whether "从词池随机抽 3～6 个非测评小节" might imply we should output only 3-6 non-eval sections, not all six. But because the prompt's "各小节任务" includes all six, and "全文共 4～7 个 <h3>" includes 7 possible, outputting all six is within. However "随机抽 3～6" is specifically "已从词池随机抽 3～6 个非测评小节" — if generated by a random selector, it should have selected some sections; but we don't know which. The "各小节任务" might have been constructed to include only selected ones. But here it includes all six. Could be because the "小节顺序" section lists all possible but "各小节任务" includes all? Let's re-read user's text: "【各小节任务与已代入模板】 --- 小节关键词「游戏介绍」：标题必须是 ... --- 小节关键词「亮点」：... --- ..." This is a list of tasks for all categories in the pool, not necessarily selected? 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" but then "【各小节任务与已代入模板】" indeed includes every possible section. Maybe the model is expected to output all sections in the "小节顺序" because the random selection has already happened indirectly? Hmm.
Let's 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>）：每个小节先 <h3>（标题须与下列 {H3title} 完全一致），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**（本次要求：用序号分条：每条单独成行，行首为「数字+英文句点+空格」（示例 1. 2. 3.），全篇保持同一格式）；**每节要点条数以该小节下方【条数】为准**（在 3～6 条之间且各节独立）。"
This parenthetical appears in "整体结构" before "【小节顺序】". It says output only according to "下列「小节顺序」" — but "下列" refers to the "小节顺序" list that follows, which includes all seven. It says "已从词池随机抽 3～6 个非测评小节" but then "若含「介绍」「简介」类关键词会固定排... 其余小节再按其下列顺序". That suggests the "小节顺序" list is the order to sort selected sections, not necessarily all. But then "全文共 4～7 个 <h3>" confirms variable count.
In our specific prompt, there is no indication which sections were selected beyond that "各小节任务" includes all. Could be because the selection logic in the actual pipeline selected all six? But "随机抽 3～6 个" can't select all six? It can: if random number is 6, it selects all six non-eval sections. So yes, possible selected all six. The phrase "并优先保留采集中有原文者" — if no original, maybe random draw from the pool could be all six. Since "各小节任务" includes all six with no original, maybe the random draw indeed selected all six? But "随机抽 3~6" is a fixed instruction; if all six included, it's a valid max. Also "全文共 4～7 个 h3" with eval gives 7. So outputting all is safe.
Need maybe "各小节条数" in 3-6: Our 游戏介绍 has 5, 操作指南 has 6, all within 3-6. Good.
Need ensure "首段 <p>" is not too long? It's fine.
Need maybe "禁止输出 <img>" — no img. Good.
Need maybe "不要输出 <html><body> 等外壳" — no.
Need "只输出一段可直接写入数据库的 HTML 正文片段" — we will output exactly fragment, no extra. Good.
Let's think about the "测评" section: It says "标题必须是 <h3>炙热江湖测评</h3>，随后**仅 1 个** <p>。" We will output h3 and p. Good.
Need maybe "不要使用 1、2、3 或分条序号" in eval. We won't.
Need maybe "全文不超过 200 字" for eval. Our eval 128 Chinese characters. Good.
Potential issue: "应用名称（须在首段 <p> 正文中自然出现）：炙热江湖" — first paragraph should contain "炙热江湖" not only "《炙热江湖》免广告版" — yes.
Need maybe "标题：炙热江湖" in first paragraph? No, that's from template. We incorporate.
Need perhaps "首段模板 {head}/{text} 均已代入为首段原文" — We should "改写应用介绍" not necessarily same as "首段". We have a rewrite.
Let's consider if using "《炙热江湖》" with book title marks counts as "应用名称" yes.
Need maybe "原文首段：炙热江湖免广告版是一款..." It says "应用名称：炙热江湖" not "免广告版". So we should not say "《炙热江湖》免广告版" too much? It's okay because the original is "炙热江湖免广告版", but app name is "炙热江湖". It might be seen as "炙热江湖免广告版" is the app name? The requirement says "应用名称（须在首段 <p> 正文中自然出现）：炙热江湖", so at least "炙热江湖". Good.
Need perhaps "免广告版" should be "免广告版" because original has it. Good.
Need "语气参考" maybe not necessary.
Let's maybe improve section content to avoid repetition and maintain "app store copy" style. We can use "你将" etc. But numbered points should be concise. Need "每条一句、每条约50字内" — our points are within.
Let's verify each point length:
Game intro:
1. 炙热江湖是一款以自由武侠世界为背景的角色扮演游戏。 (22)
2. 玩家可选择独行江湖、创立门派或投身朝廷，玩法自由。 (23)
3. 游戏中的NPC会随时间成长，互动可触发各种随机事件。 (24)
4. 天机阁定期发布天地榜、龙凤榜等榜单，内容十分丰富。 (27)
5. 比武大会与比武招亲等活动让江湖生活更加多彩。 (21)
Good.
Highlights:
1. 免广告设计让玩家可以无干扰地沉浸于江湖历练。 (21)
2. NPC成长与随机事件令每次探索都充满意外惊喜。 (21)
3. 多元身份和榜单比武带来极强的长期可玩性。 (19)
Good.
Advantages:
1. 自由路线与多线身份让玩家拥有远超同类的选择空间。(23)
2. 天机阁榜单和江湖活动持续带来清晰目标与丰富奖励。(24)
3. 无广告打扰，搭配丰富NPC交互，营造沉浸式体验。(23)
Good.
Charm:
1. 每个人都能按心意走出独特江湖路，代入感极强。(21)
2. NPC会成长、会回应，让这个虚拟世界更有人情味。(21)
3. 从初出茅庐到名动江湖，成长的成就感令人沉浸。(22)
Good.
Gameplay:
1. 自由探索地图，通过对话和选项触发随机事件与任务。(23)
2. 修炼武学、结交侠客、发展门派，逐步提升自身实力。(24)
3. 参加比武大会和比武招亲，争夺丰厚奖励与江湖声望。(24)
4. 借助天地榜、龙凤榜等排名系统，不断挑战更高目标。(24)
Good.
Operation:
1. 进入游戏后先创建角色，并选择想要的身份背景。(20)
2. 在主界面点击“江湖”按钮，进入世界地图自由探索。(23)
3. 与NPC对话时仔细查看选项，可触发事件或领取任务。(24)
4. 打开“武学”界面，可为角色装备和升级已学功法。(24)
5. 进入“天机阁”查看榜单信息，并领取对应的排名奖励。( 
