boluopai.paper-retrieval
Find relevant Chinese/English papers, books and chapters, reconcile citations
and editions, and verify the actual acquired text. The exact Skill identifier is
paper-retrieval. Copy this complete directory into your host's Skill directory
and invoke its SKILL.md through that host, or use the helper directly.
Examples: “Find the author's papers since 2020 and separate bibliography coverage from obtained full texts”; “Obtain the available author manuscript for this DOI”; “Check whether this PDF is the complete second edition or a preview.”
Companion methods and execution dependencies
| Dependency | When and why | Obtain or supply |
|---|---|---|
| research | Companion method for evidence selection, current investigation and conflicting sources; use with the scholarly workflow | Download the complete public research package beside this package. It is a Skill, not a search engine. |
| browser-relay | Conditional on controlled/authenticated pages; owns exact targets, session reuse, readback, placement and cleanup | Download the complete sibling browser-relay package and bind the host’s real authorized browser/download API. The Skill does not install a browser carrier or copy authentication. |
| network | Conditional on evidenced connectivity faults; distinguish transport from access/capture failures | Download the network companion when using that method, configure your own targets, and use actual diagnosis capabilities within authorization. It is conditional, not permission to alter routing. |
| Search/HTTP/browser tools, Python, Poppler, optional curl | Actual execution capabilities, not companion Skills | Supply the tools and install the helper dependencies below. Downloading a Skill does not install them or establish library access. |
The local method's division of responsibility is retained: research selects and assesses evidence; the browser owner controls the exact authorized surface; paper-retrieval reconciles the scholarly object and acquired bytes. The inspector can run independently; the complete collection method composes these roles.
The host needs an available search/HTTP/browser tool for discovery and acquisition.
The local helper requires Python 3.10+ and Poppler (pdfinfo, pdftotext on PATH).
Install Poppler through your platform's package manager, for example
brew install poppler on macOS or sudo apt install poppler-utils on Debian/Ubuntu.
The public direct-download example also uses curl.
python3 scripts/inspect_pdf.py --help
python3 scripts/inspect_pdf.py /path/to/candidate.pdf
python3 -m unittest discover -s tests -v
The helper prints JSON with byte size, SHA-256, page count and endpoint text; it rejects non-PDF responses and parsing errors. It does not certify identity, missing middle pages, edition equivalence or completeness. Inspect the printed work against its source record, and render image-only pages. There is no universal retrieval promise, credential store, authenticated capture adapter, CAJ converter or restriction-evasion mechanism. Existing library access must be established through the recipient's own authorized interface. Record incomplete items and verified identifiers in the conversation as well as the collection manifest.
See source and verification guidance. This original portable adaptation derives its scholarly workflow from the maintainer's local Skill, removing host paths, personal records and private browser/library machinery. Original package material is MIT licensed. Poppler and retrieved publications retain their own rights; no third-party code or publication is bundled.
中文说明
本 Skill 用于检索相关的中英文论文、学术书籍与章节，核对题录和版本，并验证实际取得的文本。准确标识为 paper-retrieval。将整个目录复制到宿主的 Skill 目录，通过宿主调用 SKILL.md，也可直接运行脚本。
示例：“收集这位作者 2020 年以来的论文，区分题录覆盖与已取得全文”； “寻找这个 DOI 可获取的作者稿”；“检查这个 PDF 是完整第二版还是预览”。
| 配套方法或依赖 | 用途与取得方式 |
|---|---|
| research | 证据选择、当前调查与冲突来源判断的配套 Skill；将完整的公开 research 包下载到本包旁边。它不是搜索引擎。 |
浏览器／会话负责人（本地源名为 browser-relay） |
仅在受控或登录页面需要：负责准确目标、授权会话复用、结果确认、窗口放置与清理。由宿主提供相应方法和真实浏览器／下载 API；本包不附私人适配器或公开 browser-relay 包。 |
网络诊断负责人（本地源名为 network） |
仅在有证据的连接故障需要：使用接收方已有的授权诊断能力，区分传输与访问／捕获问题。无需强制安装 Mac 网络包，也不由此授权更改路由。 |
| 搜索／HTTP／浏览器工具、Python、Poppler、可选 curl | 是实际执行能力，需自行提供或按下文安装；下载 Skill 不等于安装工具或取得图书馆权限。 |
本地方法的职责关系仍保留：research 选择和判断证据，浏览器负责人管理准确的授权界面，paper-retrieval 核对学术对象与取得的文件。检查脚本可单独运行；完整收藏方法组合这些职责。
检索和下载需要宿主已有的搜索、HTTP 或浏览器工具。本地脚本需要 Python
3.10+ 和 PATH 中的 Poppler pdfinfo、pdftotext。macOS 可用
brew install poppler，Debian/Ubuntu 可用 sudo apt install poppler-utils；公开直链下载示例还需 curl。上方命令用于帮助、文件检查和本地测试。
脚本输出字节数、SHA-256、页数及首尾页文字，拒绝非 PDF 与解析失败；不会自动确认作品身份、内部缺页、版本等价或完整性。应核对正文与来源题录，无文字页需渲染检查。本包不保证任意全文可得，不包含凭据库、登录捕获适配器、 CAJ 转换器或限制规避机制。图书馆权限仍由使用者的授权界面确认。未完成作品及已验证标识应同时出现在对话交付和收藏清单中。
来源与验证指南保留方法说明。这是从维护者本地 Skill 独立改写的可移植版本，移除了私人记录、主机路径与专用浏览器和图书馆机制。原创内容采用 MIT 许可；Poppler 和取得的出版物保留各自权利，本包不附第三方代码或出版物。