Nine operations on one dropped file: merge several PDFs, extract pages ("1-3,5,8-" grammar, order as written), split to single pages, rotate by ±90/180, text watermark (stamp or tiled, any size/opacity/angle/colour), lossless compress, view or strip metadata, PDF → PNG/JPG at 1.5–3× scale, and images → PDF.
Structure work is pdf-lib, rendering is pdf.js with the worker and standard fonts served from this site. Every byte stays in the page. One honest gap: encryption is not offered — pdf-lib cannot write encrypted PDFs, and a password-protected input is reported clearly rather than failing obscurely.
Frequently asked questions
How many pages can it handle?
Structure operations (rotate, watermark, metadata, compress) have no page limit — a 2,000-page file loads in ~0.6s and rotates fully in ~0.6s. Merge is bound by total input size (a few hundred MB is safe). Split caps at 500 pages because every page becomes an in-memory blob. PDF→image renders page by page, so memory stays flat.
Is the compression lossless?
Yes — it re-packs objects and cross-references the way Word’s "minimum size" save does. Images are never re-encoded: that would be lossy. Files already saved tightly may gain a few bytes of zip overhead.
Why no encryption?
The only mature browser library for PDF structure (pdf-lib) has never implemented writing encrypted files — a long-standing upstream gap. Pretending otherwise would mean shipping something that silently produces broken files.
中文说明
拖入文件即得的九种操作：合并多个 PDF、按页码提取（"1-3,5,8-" 语法，按书写顺序）、拆分为单页、±90/180 旋转、文字水印（居中或平铺，字号/透明度/角度/颜色可调）、无损压缩、查看或清除元数据、PDF 转 PNG/JPG（1.5–3 倍分辨率）、图片合成 PDF。
结构操作走 pdf-lib，渲染走 pdf.js（worker 与标准字体由本站托管）。所有字节不出页面。一个诚实的缺口：不提供加密——pdf-lib 无法写出加密 PDF，遇到带密码的输入会明确提示而非晦涩报错。
常见问题
最多支持多少页？
结构操作（旋转/水印/元数据/压缩）页数无硬限——2000 页文件载入约 0.6 秒、全页旋转保存约 0.6 秒。合并受总输入体积限制（几百 MB 内安全）。拆分上限 500 页（每页生成一个内存中的 blob）。PDF 转图片逐页渲染，内存恒定。
压缩是无损的吗？
是——重打包对象与交叉引用表，与 Word"最小体积"另存同思路。图片绝不重编码：那是有损的。本已压得很紧的文件可能多几个字节的重打包开销。
为什么不支持加密？
浏览器端唯一成熟的 PDF 结构库（pdf-lib）从未实现写出加密文件——长期的上游缺口。假装支持只会产出悄悄损坏的文件。