PDF Reader

A Go library for reading PDF files, with active CJK text extraction support.

Requires Go 1.25+ (go.mod directive).

Forked from ledongthuc/pdf (upstream inactive since 2024). Original lineage: rsc/pdf.

Features

• Plain text extraction with context/cancellation support
• Styled text extraction (font name, size, position)
• Text grouped by row
• Document metadata API (/Info dict: title, author, dates, …)
• Outline (table of contents) with resolved page numbers
• CJK predefined CMap decoders:
  • Japanese Shift-JIS (90ms-RKSJ-H/V, 90pv-RKSJ-H)
  • CJK UCS-2 BE (UniGB-UCS2-H/V, UniCNS-UCS2-H/V, UniJIS-UCS2-H/V, UniKS-UCS2-H/V)
  • Simplified Chinese GBK / GB-EUC / GBKp-EUC (GBK-EUC-H/V, GB-EUC-H/V, GBKp-EUC-H/V)
  • Traditional Chinese Big5-ETen / ETenms (ETen-B5-H/V, ETenms-B5-H/V)
  • Korean UHC / KSC-EUC / UHC-HW (KSCms-UHC-H/V, KSC-EUC-H/V, KSCms-UHC-HW-H/V)

Install

go get github.com/Detective-XH/pdf

Examples

See the examples/ folder for runnable programs.

Read plain text

package main

import (
	"bytes"
	"context"
	"fmt"

	"github.com/Detective-XH/pdf"
)

func main() {
	f, r, err := pdf.Open("./sample.pdf")
	if err != nil {
		panic(err)
	}
	defer f.Close()

	var buf bytes.Buffer
	b, err := r.GetPlainText(context.Background())
	if err != nil {
		panic(err)
	}
	buf.ReadFrom(b)
	fmt.Println(buf.String())
}

Read styled text

package main

import (
	"context"
	"fmt"

	"github.com/Detective-XH/pdf"
)

func main() {
	f, r, err := pdf.Open("./sample.pdf")
	if err != nil {
		panic(err)
	}
	defer f.Close()

	sentences, err := r.GetStyledTexts(context.Background())
	if err != nil {
		panic(err)
	}
	for _, s := range sentences {
		fmt.Printf("font=%s size=%.1f x=%.1f y=%.1f text=%s\n",
			s.Font, s.FontSize, s.X, s.Y, s.S)
	}
}

Read text by row

package main

import (
	"fmt"
	"os"

	"github.com/Detective-XH/pdf"
)

func main() {
	f, r, err := pdf.Open(os.Args[1])
	if err != nil {
		panic(err)
	}
	defer f.Close()

	for i := 1; i <= r.NumPage(); i++ {
		p := r.Page(i)
		if p.V.IsNull() {
			continue
		}
		rows, _ := p.GetTextByRow()
		for _, row := range rows {
			fmt.Printf("row %d:", row.Position)
			for _, word := range row.Content {
				fmt.Printf(" %s", word.S)
			}
			fmt.Println()
		}
	}
}

Fork status

Area	Status
Upstream sync	Merged through upstream@HEAD (2024)
Shift-JIS CMaps	Added
UCS-2 BE CMaps	Added
GBK / GB-EUC / GBKp-EUC CMaps	Added
Big5-ETen / ETenms CMaps	Added
UHC / KSC-EUC / UHC-HW CMaps	Added
Metadata API (r.Info())	Added
Outline page numbers (Outline.Page)	Added
Context / cancellation	Added
Crash/CPU-spike on PDFs with inline images (upstream #57)	Fixed — readHexString EOF guard + Interpret inline-image skip
Upstream PRs incorporated	#37, #42, #45, #58, #61, #63, #64, #66

Resolved upstream issues

Issue	Title	How it was fixed	Status
#13	Load Reader from bytes instead of file path	OpenBytes(src []byte) added in read.go	Directly fixed
#16	GetTextByRow returns disordered text	sort.Sort → sort.Stable in GetTextByRow/GetTextByColumn	Directly fixed
#18	GetTextByRow X/Y always 0	Td/TD/T*/TL wired in walkTextBlocks; BT resets position; currentTL tracks leading	Directly fixed
#20	%%EOF search window too small; valid PDFs rejected	Expanded search window from 100 → 1024 bytes (with clamp for small files); added findStartxrefFallback reverse-scan for %%EOF placed further than 1024 bytes before end	Directly fixed
#21	unknown encoding UniGB-UCS2-H	Same fix as #55 — ucs2BEEncoder handles UniGB-UCS2-H	Directly fixed
#22	Handle space after header	Relaxed byte-8 check in NewReaderEncrypted to accept space/tab	Directly fixed
#27	GetTextByRow returns empty rows	Td in walkTextBlocks now updates currentX/currentY additively instead of emitting a spurious empty walker call; TD and TL wired; T* decrements Y by leading	Directly fixed
#30	crash when encountering some CJK text amongst English	dictEncoder rewrite; maxObjectDepth guard; readArray EOF fix	Directly fixed
#31	Expose page dimensions	Page.MediaBox() and Page.CropBox() added; both walk the page-tree inheritance chain; CropBox falls back to MediaBox when absent	Directly fixed
#44	Cannot read Chinese	GBK / Big5 / UniGB / UniCNS CMaps all wired in getEncoder()	Directly fixed
#48	\n added by recent version breaks old systems	Removed showText("\n") from case "BT": — BT is matrix-init, not line-break	Directly fixed
#55	GetPlainText do not support encoding "UniGB-UCS2-H"	ucs2BEEncoder wired for all 8 Uni*-UCS2-H/V CMap names	Directly fixed
#57	Crash when image is in there (malformed PNG)	case "ID": skip in ps.go Interpret(); readHexString EOF guard in lex.go	Directly fixed
#59	Streaming / range-over-func API for large PDFs	(*Reader).Pages() iter.Seq2[int, Page] and (Page).Texts() iter.Seq[Text] added; Texts merges same-style runs matching GetStyledTexts output	Directly fixed
#60	Parse PDF, some content appears garbled	Removed shared fonts map from (*Reader).GetPlainText; each page now passes nil so (*Page).GetPlainText builds a fresh per-page font map	Directly fixed
#67 / #26	Text in Form XObjects not extracted	case "Do": added to all three interpreter callbacks (interpretPlain, interpretWalk, interpret); each handler looks up the named XObject in the current /Resources/XObject dict, confirms Subtype /Form, builds a child state with fonts re-resolved from the XObject's own /Resources/Font, and recursively calls Interpret() — merging extracted text back into the parent output; depth capped at 10 to guard against cycles	Directly fixed