这是一个作为PVLDB第16卷出版流程一部分的资源合集。它主要提供检查PDF/A合规性的工具，并包含各种实现PDF/A合规的资源和操作指南。

注：这是一份实时文档，请随时关注更新内容，也欢迎随时参与贡献。

开始使用检查PDF/A合规性的工具

摆好工具

检查脚本需要 Python 3 带 PyPDF2，poppler（针对 pdffonts）和 VeraPDF（需要 Java）。在 macOS（用 Homebrew）、Debian/Ubuntu 或 Fedora 上，一个命令就能安装所有内容：

它要求输入密码安装系统包，且可以安全重启。运行 ./setup.sh -h 以获取选项。

如需手动安装工具，请遵循_tools/INSTALL_TOOLS.md。

检查所有设备是否就位：

简约例子

把你的论文复制到 _testdir（默认输入文件夹，git 忽略了），然后两次检查：

通过字体和PDF/A检查的纸张打印：

使用 --dir 检查另一个文件夹，使用 -h 查看所有选项及其默认值。[check_fonts_pdfa.py 还会将所需修正列表写入 corrections.txt。

How to ensure PDF/A compliance

This is a collection of resources started as part of the PVLDB Vol 16 Publication Process. It primarily hosts tool to check PDF/A compliance and also include various resources and how-tos regarding achieving PDF/A compliance.

Note: This is a live document, so please check back for updated content and feel free to contribute as well.

Getting started with the tools to check PDF/A compliance

Set up the tools

The checking scripts need Python 3 with PyPDF2, poppler (for pdffonts) and veraPDF (which needs Java). On macOS (with Homebrew), Debian/Ubuntu or Fedora, one command installs everything:

./setup.sh

It asks for your password to install system packages and is safe to re-run. Run ./setup.sh -h for options.

To install the tools by hand instead, follow _tools/INSTALL_TOOLS.md.

Check that everything is in place:

python3 -c "import PyPDF2; print(PyPDF2.__version__)"
pdffonts -v
_tools/veraPDF/verapdf --version

Minimal example

Copy your paper into _testdir (the default input folder, ignored by git) and run both checks:

mkdir -p _testdir
cp /path/to/p1234-smith.pdf _testdir/
./check_fonts_pdfa.py     # PDF/A compliance, embedded fonts, no Type 3 fonts
./check_format.py         # common PVLDB formatting issues

A paper that passes the font and PDF/A check prints:

	Check file _testdir/p1234-smith.pdf
... OK: PDF/A compliant, all fonts are embedded and there are no Type 3 fonts.

Use --dir to check a different folder, and -h to see all options and their defaults. check_fonts_pdfa.py also writes the list of needed corrections to corrections.txt.

If a check fails, PDFA-how-to.md explains how to fix it. If your paper declares PDF/A-1b and fails, try declaring PDF/A-2b with \hypersetup{pdfapart=2,pdfaconformance=b}. PVLDB accepts PDF/A-2, which allows things PDF/A-1 forbids, such as transparency in figures. Then re-run the check.

Proposed workflow: check your figures too

Font problems often come from a single figure. To find it, check the figures one by one:

cp /path/to/paper/figures/*.pdf _testdir/
./check_fonts_pdfa.py

Look for contains Type 3 fonts or has N missing fonts. For figures, ignore not a valid PDF/A and don’t run check_format.py.

How to prepare your PDF

Click here to see a set of advice on how to prepare your PDF so that it is PDF/A compliant.

AI Agents

AI coding agents can use the getting-started skill in .claude/skills/pvldb-pdfa-getting-started/SKILL.md, which also covers interpreting and fixing failures.