Pdf-lib: Parsing never ends

[mlecoq]

Hi, I try to parse a pdf but it never ends

const index = PDFObjectIndex.create();
const pdfParser = new PDFParser();
const parsedPdf = pdfParser.parse(data, index);

The pdf is this one : https://mle-sandbox.s3.amazonaws.com/bug7.pdf

For the moment I don't see where it's stuck

[mlecoq]

In fact it will ends but it takes much time. I have added some logs and most of time is spent in parsing numbers

[mlecoq]

@Hopding you told me you're working on performance improvements for pdf-lib. Do you think these improvements could improve parsing time for this file ?

[Hopding]

Hello @mlecoq! I am indeed working on several performance improvements. The changes improve parsing times significantly for most PDF files. So I suspect it will help with the one your having trouble with.

I am unable to access the file via the link you shared. I get the following error when I tap on the link:

AccessDeniedAccess DeniedACFADBEE3B14786DV89MhuHrho0pR1NJX65rnSvf0V+ActBeUUE9CzLrXGlEqI9fYlPFxZ9q6refXwNxuHIDlGb70s8=

If you’re able to fix this, I can try it out with the new parser and let you know what I find.

Also, all the code for the rewrite is located here: https://github.com/Hopding/pdf-lib/tree/Rewrite. Feel free to try it out for yourself, if you want. I’m nearly finished now. I expect to cut a beta release no later than this weekend.

[mlecoq]

Ooops forgot to make the file public. You should have access now. I will try with the rewrite !

It takes around 8 minutes to be parsed with pdf-lib 0.6.3

[mlecoq]

I have changed my code a bit:

const start = new Date();

const pdf = fs.readFileSync(`${__dirname}/pdf/bug7.pdf`);

const pdfParser = new PDFParser(pdf);

pdfParser.parseDocument().then(parsedPdf => {
    console.log(parsedPdf);
    const end = new Date() - start;
    console.log('Execution time: %dms', end);
});

Parsing is done within 413ms against 8min currently (impressive)

[philipjmurphy]

Out of curiosity, was the performance improvement due to the code change or update to pdf-lib or both?

[Hopding]

@philipjmurphy The parsing speed improvement is due to the extensive rewrite I’ve been working on in the Rewrite branch. The code sample that @mlecoq shared is using the new PDFParser from that branch.

I’m intending to release this rewrite as v1.0.0. As I mentioned, I plan to cut a beta release no later than this weekend. If you’re interested, I’d be happy if you’d help test it out!

[philipjmurphy]

Wow, those results are amazing. Sure, let me know if you'd like any particular test or I can use some of my own business PDFs with it.

[mlecoq]

fixed in 1.0.0-beta.1