使用rga在pdf中搜索关键字

我们一般会用 grep/ag/rg 在文本文件中搜索关键字. 但是某一天我想在一堆 pdf 文件中搜索关键字了怎么办才好呢?在网上搜了一下还真找到一个工具:rga(ripgrep-all)

ripgrep-all 对 ripgrep 进行了封装，借助了其他一些程序(比如pandoc)实现了对PDF，DOCX，ODT，EPUB，SQLite数据库，MKV/MP4内嵌字幕，zip/gz压缩包等文件的正则表达式搜索

更夸张的是，ripgrep-all 可以借助 tesseract 使用OCR技术识别图片和PDF内图片中的文字，不过由于OCR本身很慢因此默认是没有使用该能力的。可以通过 --rga-adapters=+pdfpages,tesseract 来启用该能力。

比如，我要在一堆pdf书籍中搜索包含 Philosophy 的内容，

rga -i Philosophy ~/ebooks/*.pdf |head

/home/lujun9972/ebooks/Automate it.pdf:Page 20: philosophy, and provides you with the knowledge required to develope your own APIs. It /home/lujun9972/ebooks/Automate it.pdf:Page 52: asynchronous philosophy. The following code represents an asynchronous server /home/lujun9972/ebooks/Automate it.pdf:Page 53: But what is asynchronous about this? In fact, tornado works on the philosophy of a single- /home/lujun9972/ebooks/Automate it.pdf:Page 147: works on the philosophy of TwiML, also known as the Twilio markup language. /home/lujun9972/ebooks/Automate it.pdf:Page 155: and-forward philosophy. Today, e-mail messages are stored asynchronously on the servers /home/lujun9972/ebooks/Automate it.pdf:Page 264: The python-telegram-bot module works on standard event-driven philosophy. A bot /home/lujun9972/ebooks/Beginning RPG Maker MV, 2nd Edition.pdf:Page 78: book, as my design philosophy here is to learn the engine). Creating a game with the engine just happens /home/lujun9972/ebooks/Beginning RPG Maker MV, 2nd Edition.pdf:Page 78: entries ourselves, of course! It’s a great opportunity for me to delve into basic RPG game design philosophy, /home/lujun9972/ebooks/Blogging For Dummies, 6th Edition.pdf:Page 19: that forgives mistakes unless you try to hide them. In keeping with the philosophy /home/lujun9972/ebooks/Build Your Own PaaS with Docker.pdf:Page 107: ease of use. Their philosophy gives a hint why:

而且 ripgrep-all 会对搜索过得内容进行缓存，因此你会发现第一次搜索时会比较慢，但是第二次搜索时会明显快很多。但你可以通过 --rga-no-cache 选项来禁止缓存

ripgrep-all借助的外部工具和支持的文件类型包括:

ffmpeg

从mkv,mp4,avi视频文件中抽取元数据和字幕

pandoc

将 epbu,odt,docx,fb2,ipynb 转换成markdown文本文件进行搜索

pdftotext

从pdf中抽取纯文本进行搜索

各种解压缩程序

读取zip,tgz,tbz2,gz,bz2,xz,zst,tar压缩包中的内容进行搜索

sqlite

将sqlite数据库(db,db3,sqlite,sqlite3)转换成纯文本格式进行搜索

tesseract

通过OCR技术将jpg，png图片中的内容抽取出来

pdfpages

将pdf转换成png文件，需要跟 tesseract 连用将图片中的内容抽取出来。