DEV Community: QiuLiang

艺术二维码发布前的工程验收：透明背景、导出文件与静态链接

QiuLiang — Mon, 14 Sep 2026 07:11:27 +0000

本文与 Halo 产品有关联，由人工智能结合当前实现辅助研究与撰写，不是第三方独立测评，也不承诺未经实测的扫码成功率。

预览成功不等于最终文件成功

二维码生成器的链路是输入内容、创建模块矩阵、输出矢量图、栅格化为图片、进入排版软件，最后发布或印刷。每一步都可能改变真实识别条件。

Halo 的图片与打印文档复用矢量渲染结果，但图片可以包含透明背景，打印文档使用不透明背景。单独打开文件能扫描，不代表叠放到照片上还能扫描：透明区域露出照片，局部对比度随之变化。

用对照版本定位失败阶段

用同一最终网址生成深色方形模块、浅色背景、无标志的基础版本。
分别测试预览、下载文件和最终排版文件，记录哪一步开始失败。
基础版本正常时，逐项增加渐变、艺术图案和标志，每次只改变一项。
透明版本在真实物料底色上复测；必要时加覆盖码与留白的浅色底板。
按实际比例打印样品，用两种手机测试，不用放大的屏幕预览替代实物。

二维码开发方说明普通二维码需要四个模块宽的边缘留白，不应被裁切线、文字或海报纹理侵入。原始说明提供尺寸计算示例，但不是所有艺术样式的扫码保证。

识别失败与网页失败要分开

手机没有识别出内容时，检查图像、留白、尺寸和对比度。手机识别出正确网址但网页打不开时，检查目标网址和网络链路。混淆两类问题，容易不断调整颜色，却没有修复失效链接。

建议把编码目标、最终文件名、物料版本、测试设备和打开结果记录为验收记录。这是建议流程，没有虚构某一批设备的测试结果。

静态链接与追踪短链的工程边界

静态二维码直接保存最终内容。网页内容更新但网址不变时，可以维护原页面；网址本身改变时，原码不会自动改变。追踪短链依赖中间服务持续可用，不能凭接口今天返回成功就宣称长期可靠。

Halo 当前可选追踪使用内存与本地文件，不承诺长期持久性。重要名片和活动可关闭追踪，直接编码自己控制的最终网址。浏览器设计历史也是本地保存，不是跨设备云端作品库。

继续阅读与实践

透明二维码排版与打印检查清单说明如何在真实海报中应用这些步骤。

也可以打开 Halo 二维码生成器制作自己的对照版本。是否适合你的物料，仍以最终文件与实物测试为准。

A Practical Client-Side Workflow for Splitting PDF Pages

QiuLiang — Tue, 08 Sep 2026 07:08:37 +0000

PDF utilities are a good fit for client-side processing. A user selects a local file, the browser reads the required pages, and the result is generated without a document-processing round trip to an application server.

That approach has three practical benefits:

private documents do not need to be uploaded;
processing begins immediately after file selection;
hosting remains simple because the public interface can be served as static assets.

The user experience that matters

Implementation details are important, but users mostly care about a predictable flow:

Select a PDF.
See its page count.
Enter a clear range such as 1-3, 7, 10-12.
Receive useful validation when the range is invalid.
Preview what will be exported.
Download a correctly named result.

Large files should not make the interface appear frozen. Progress feedback, disabled controls during processing, and actionable errors make a major difference.

Privacy language should be specific

Avoid vague promises such as “100% secure.” Explain what actually happens: whether processing occurs locally, whether the file is transmitted, and what telemetry is collected. Users can then make an informed decision.

Test cases worth keeping

a one-page document;
encrypted or malformed PDFs;
ranges beyond the final page;
duplicate and unordered page selections;
filenames containing non-Latin characters;
a large scanned document on a mobile device.

I use this checklist for the browser-based PDF page splitter.

The accompanying step-by-step PDF splitting guide covers common desktop and browser workflows.

I rebuilt my PDF workflow so a file never has to leave my laptop

QiuLiang — Wed, 12 Aug 2026 13:47:26 +0000

For years my "PDF workflow" meant: open a free website, upload the file,
pray it
2
wasn't a bank statement, download the result. I never liked it. So I built
the thing
I wanted.
3
4
5
wezzik.com is 16 PDF tools that all run in the browser. Merge, split,
6
7
rotate,
compress to an exact size, convert to Word/Excel/Image, OCR, even an AI
summary.
The part that matters: nothing is uploaded. Open DevTools while you
process a 50MB
file and the Network tab stays empty.
8
9
10
11
It's static HTML and JS, pdf-lib and pdf.js doing the heavy lifting,
Tesseract for
to, which
OCR, hosted on Vercel. No backend means there's nothing to send your file
is the only privacy guarantee that actually holds.
12
13
14
15
If you process PDFs with any regularity, give it a try. And if there's a
tool missing
that fits the local-only model, tell me, I'll build .

A PDF toolkit that runs 100% in the browser, no upload, no signup

QiuLiang — Wed, 12 Aug 2026 13:40:49 +0000