Pure Go Full Text Search of PDF Files

This library implements full text search for PDF files.

• The public APIs are in index_search.go.

The are some command lines programs that demonstrate the library's functionality.

• examples/pdf_search_demo.go demonstrates the main APIs.
• examples/pdf_search_verify.go verifies the consistency of the in-memory and on-disk APIs.
• examples/index.go builds an index over a set of PDFs.
• examples/search.go searches the index build by examples/index.go.

Installation

git clone https://github.com/PaperCutSoftware/pdfsearch
cd pdfsearch/examples
go build pdf_search_demo.go
go build pdf_search_verify.go
go build index.go
go build search.go

examples/pdf_search_demo.go

Usage: ./pdf_search_demo -f <PDF path> <search term>

Example: ./pdf_search_demo -f PDF32000_2008.pdf cubic Bézier curve

The example will search PDF32000_2008.pdf for cubic Bézier curve.

pdf_search_demo.go shows how to use the APIs in index_search.go to

• create indexes over PDF files,
• search those indexes using full-text search, and
• mark up PDF files with the locations of the search matches on pages.

It has 3 types of index

• On-disk. These can be as large as your disk but are slower.
• In-memory with the index stored in a Go struct. Faster but limited to (virtual) memory size.
• In-memory with the index serialized to a []byte. Useful for non-Go callers such as web apps.

examples/index.go

Usage: ./index <file pattern>

Example: ./index ~/climate/**/*.pdf

The example creates an on-disk index over the PDFs in ~/climate/ and its subdirectories.

examples/search.go

Usage: ./search <search term>

Example: ./search integrated assessment model

The example searches the on-disk index created by examples/index.go for integrated assessment model.

Libraries

index_search.go uses UniDoc for PDF parsing and bleve for search.