这篇文章能帮你做什么
下面给出一组仅面向 PDF 工具类别的「在线工具生成型 AI 提示词（Prompts）」。每条 Prompt 都要求 AI 产出一个可运行/可部署的在线工具项目：包含完整源码、清晰文件结构、运行命令、部署说明，以及可执行的测试用例或 QA checklist。
在线 PDF 书签/目录生成 + 层级校验与修复工具
把没有书签的 PDF 生成 Outline/Bookmarks；对已有书签做层级、跳转、重复与空标题校验，并输出修复建议与可回滚的补丁。
英文 Prompt:
You are to generate a complete, runnable web app that creates and validates PDF bookmarks (outline) and table-of-contents mapping.
Requirements:
- Tech stack: Next.js (Type ) + a minimal Node runtime compatible with serverless and Docker.
- Features:
1) Upload PDF (drag-drop) and show basic info (page count, file size, data).
2) Extract existing outline/bookmarks and render as a tree view.
3) Validate outline: empty s, duplicated siblings, invalid destinations, out-of-range page refs, non-monotonic page ordering, overly deep nesting.
4) Generate outline from user-provided TOC text (paste), with rules: detect headings, infer levels, map to page numbers, and allow manual adjustments.
5) Export results: JSON report + downloadable patched PDF + a diff summary.
6) Safety: never store files permanently; process in memory or ephemeral storage.
Deliverables (must include all):
- Full source code with a clear file tree.
- Exact commands to run locally (install, dev, build, start).
- Dockerfile + docker-compose for production.
- Deployment notes for Vercel and a generic VPS.
- At least 8 tests (unit/integration) OR a detailed QA checklist (>= 12 items) covering edge cases.
Implementation notes:
- Use a well-known PDF library and explain trade-offs.
- Include input size limits, error handling, and progress indicators.
- Provide sample PDFs generation for testing (no image generation; use text-only PDF generation).
中文释义: 生成一个可直接上线的在线 PDF 书签工具：既能从目录文本生成书签，也能对现有书签做强校验并输出报告/修复版 PDF，适合出版/归档/资料整理流程。
在线 PDF Outline 提取 + 可视化树编辑 + JSON/CSV 导出工具
把 PDF 书签提取成结构化数据，支持在网页里调整层级与标题，导出 JSON/CSV 供后续自动化流程使用。
英文 Prompt:
Build a runnable web tool: "PDF Outline Extractor & Editor".
Stack:
- Frontend: React + Vite + Type .
- Backend: Node.js (Express) API.
Core flows:
1) Upload PDF -> backend extracts outline/bookmarks and returns a normalized tree model (id, , level, dest page, dest coords).
2) UI shows tree with inline edit, drag-drop reorder, promote/demote level, and search.
3) Validate tree rules: cycles, invalid levels, missing destinations, duplicate s within same parent.
4) Export: JSON and CSV.
5) Optional: apply edited outline back to a new PDF and allow download.
Deliverables:
- Complete project code + file tree.
- Run commands.
- Docker setup.
- API docs (OpenAPI or a concise markdown).
- Testing: at least 6 automated tests plus a QA checklist.
Constraints:
- No permanent storage.
- Clear error messages for encrypted PDFs, malformed PDFs, and huge files.
中文释义: 适合把 PDF 书签结构转成可机器处理的数据，并在浏览器里快速修正层级与标题，再导出供脚本/工作流消费。
在线 PDF 跳转链接检查 + 断链/循环引用报告工具
扫描 PDF 内部链接（目录、注释链接、URL），检查目标页是否存在、是否形成循环跳转，并生成可下载的审计报告。
英文 Prompt:
Create a web app that audits s inside a PDF.
Stack: Nuxt 3 (Type ) + Node API.
Features:
- Upload PDF.
- Extract all annotations and outline destinations.
- Validate: target pages exist, destination coordinates sane, detect repeated chains that form loops, count external URLs and flag suspicious schemes.
- Produce an HTML report view + downloadable JSON.
- Provide a "fix suggestions" section (e.g., remove broken s, normalize destinations).
Deliverables:
- Full code, run/build/deploy steps.
- Dockerfile.
- QA checklist >= 12 items with sample cases.
- Include a small to generate synthetic PDFs with annotations for testing (text-only).
中文释义: 做资料发布/课程讲义/产品说明书时常见的目录跳转问题可以一次扫出来，并输出审计报告，避免用户点目录无反应。
在线 PDF 元数据清理 + 规范化（ /Author/Producer）工具
批量清理敏感元数据，统一标题/作者/主题字段，并输出处理前后对比与可回滚记录。
英文 Prompt:
Generate a complete web tool: "PDF data Cleaner".
Stack: SvelteKit + Type .
Functions:
- Upload one or multiple PDFs.
- Show extracted data fields.
- Provide presets: "Remove all", "Keep only", "Enterprise standard".
- Apply changes and allow downloading cleaned PDFs individually or as a zip.
- Provide an audit log and a before/after diff view.
Deliverables:
- Full source code, file tree.
- Commands to run.
- Docker deploy + reverse proxy notes.
- Tests or QA checklist (>= 10) including multi-file batch and very large PDF handling.
中文释义: 适合对外发布前做隐私与一致性处理：例如去掉 Producer/Creator 暴露的工具链信息，统一 Author/ 规范。
在线 PDF 文本检索 + 关键字定位 + 章节命中统计工具
对 PDF 做全文索引，支持关键字高亮定位与页码跳转，输出命中次数、命中页分布和章节统计。
英文 Prompt:
Build a web app that searches text inside PDFs and generates a search analytics report.
Stack: Fastify (Node) + React.
Features:
- Upload PDF.
- Extract text per page (handle multi-column layout best-effort).
- Search keywords (multi-keyword AND/OR, regex mode optional).
- Show results with page preview (render PDF page) and highlight occurrences.
- Export a report: keyword -> counts, pages, snippets.
Deliverables:
- Complete runnable project.
- Docker.
- Deployment notes.
- Automated tests for text extraction and search edge cases, or a QA checklist >= 12.
中文释义: 适合做手册/合同/规范文档的快速检索与审计：把关键条款出现在哪些页、出现多少次，一眼看清。
在线 PDF 去敏（遮盖）与替换预览工具
对身份证号/手机号/邮箱等敏感字段做定位与遮盖，生成不可恢复的 Redaction 版本，并提供替换前后对照预览。
英文 Prompt:
Create a web- d PDF redaction tool.
Stack: Next.js + Type .
Requirements:
- Upload PDF.
- Detect sensitive patterns (configurable regex presets for phone/email/id).
- Let user confirm/redraw redaction rectangles on a page canvas.
- Produce a redacted PDF where content is truly removed/covered (not just an overlay).
- Provide a safety note explaining limitations.
Deliverables:
- Full code + file tree.
- Local run/build commands.
- Docker.
- QA checklist >= 12 items, including verifying redaction is not recoverable by copy/paste or text extraction.
中文释义: 适用于对外共享文件前的合规处理：先自动识别，再人工确认，最后输出真正去敏的 PDF，而不是表层遮挡。
在线 PDF 合并/拆分（按书签/按页码范围）+ 规则预览回滚工具
按书签或页码范围批量拆分，或把多份 PDF 合并；支持先预览规则结果，再一键执行并保留回滚记录。
英文 Prompt:
Build an online PDF splitter/merger with rule preview.
Stack: Remix + Node.
Features:
- Upload multiple PDFs.
- Split modes: by page ranges; by bookmarks top-level; by file size target.
- Preview: show how many output files, their page ranges, and names.
- Execute: produce outputs and allow download as zip.
- Include a rollback log (client-side) and a shareable config JSON.
Deliverables:
- Full project code + file tree.
- Commands.
- Docker.
- Tests or QA checklist (>= 10).
中文释义: 适合整理资料包：先按目录拆分成章节 PDF，或者把分散的附件合并成一个发布版，并且在执行前就能看到结果预览。
在线 PDF 字体嵌入/子集化审计 + 报告导出工具
检查 PDF 使用了哪些字体、是否嵌入/是否子集化，输出可读报告与风险提示（例如跨平台显示风险）。
英文 Prompt:
Create a web tool that audits fonts in PDFs.
Stack: Astro (frontend) + Node API.
Capabilities:
- Upload PDF.
- Extract font resources per page if possible; list font names, embedding flags, subset usage, and warnings.
- Export report as JSON and a nicely formatted HTML.
- Provide suggestions: how to re-export with embedded fonts.
Deliverables:
- Full runnable code.
- Docker.
- QA checklist >= 10 items.
- Mention limitations depending on PDF library capabilities.
中文释义: 很多 PDF 在不同设备显示异常往往是字体未嵌入导致；这个工具能在发布前做一次字体资源体检并输出报告。
使用建议（同主题补充）
如果你要把这些 Prompt 交给 AI 执行，建议每次只选一条 Prompt 生成一个独立项目，然后根据真实样例 PDF 做回归测试。对于批量处理类工具，优先补齐：进度条、错误重试、并发限制、以及对超大文件的降级策略。
