This repository was archived by the owner on Jul 7, 2020. It is now read-only.
panic on some PDFs + suspect memory leak #9
Description
I have the following Go program that uses this library:
package main
import (
"fmt"
"os"
"strconv"
"rsc.io/pdf"
)
func main() {
if len(os.Args) < 2 || os.Args[1] == "-h" || os.Args[1] == "--help" {
fmt.Println("usage: pdfpage file.pdf [pnum]")
os.Exit(1)
}
reader, err := pdf.Open(os.Args[1])
if err != nil {
fmt.Println(err)
os.Exit(2)
}
if len(os.Args) == 3 {
var pnum int
var err error
if pnum, err = strconv.Atoi(os.Args[2]); err != nil {
pnum = 1
}
fmt.Printf("PAGE %d\n", pnum)
printPage(reader, pnum)
} else {
for pnum := 1; pnum <= reader.NumPage(); pnum++ {
fmt.Printf("PAGE %d\n", pnum)
printPage(reader, pnum)
fmt.Println("")
}
}
}
func printPage(reader *pdf.Reader, pnum int) {
page := reader.Page(pnum)
if page.V.IsNull() {
fmt.Printf("failed to read page %d\n", pnum)
os.Exit(3)
}
for _, chunk := range page.Content().Text {
fmt.Printf("x=%06.2f y=%06.2f w=%06.2f %q %s %.1fpt\n",
chunk.X, chunk.Y, chunk.W, chunk.S, chunk.Font,
chunk.FontSize)
}
}This builds and runs fine and for many PDFs gives the expected output (although it is rather slow).
However I have a few PDFs which produce a panic:
PAGE 1
panic: malformed PDF: reading at offset 0: stream not present
goroutine 1 [running]:
rsc.io/pdf.(*buffer).errorf(0xc4200d3948, 0x507f70, 0x27, 0xc4200d36d0, 0x2, 0x2)
/home/mark/app/go/src/rsc.io/pdf/lex.go:82 +0x74
rsc.io/pdf.(*buffer).reload(0xc4200d3948, 0x8)
/home/mark/app/go/src/rsc.io/pdf/lex.go:95 +0x193
rsc.io/pdf.(*buffer).readByte(0xc4200d3948, 0x599da0)
/home/mark/app/go/src/rsc.io/pdf/lex.go:71 +0x69
rsc.io/pdf.(*buffer).readToken(0xc4200d3948, 0xc42000aca0, 0x1000)
/home/mark/app/go/src/rsc.io/pdf/lex.go:135 +0x4a
rsc.io/pdf.Interpret(0xc42006e060, 0x37, 0x4d78a0, 0xc42000ab60, 0xc4200d3b08)
/home/mark/app/go/src/rsc.io/pdf/ps.go:64 +0x1c6
rsc.io/pdf.Page.Content(0xc42006e060, 0x37, 0x4db2e0, 0xc420014810, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0)
/home/mark/app/go/src/rsc.io/pdf/page.go:613 +0x326
main.printPage(0xc42006e060, 0x1)
/home/mark/app/go/src/pdfpage2/main.go:47 +0xa8
main.main()
/home/mark/app/go/src/pdfpage2/main.go:35 +0x25d
I also have a 647 page PDF for which the program outputs the first 22 pages, then outputs PAGE 23 and then just sits there eating memory and using ~25% CPU. That particular page has some Japanese characters but I don't know if they are Unicode text or paths.