Failed to run pdf2json: ENOENT no such file or directory

[styfle]

Steps to reproduce

git clone https://github.com/friedhelmensch/node_now_pdf2json_issue
cd node_now_pdf2json_issue
ncc build index.js
node dist

Error

Error: ENOENT: no such file or directory, open '/Users/styfle/Desktop/foobar/node_now_pdf2json_issue/dist/baseshared/util.js'
    at Object.openSync (fs.js:438:3)
    at Object.readFileSync (fs.js:343:35)
    at module.exports.179._pdfjsFiles.forEach (/Users/styfle/Desktop/foobar/node_now_pdf2json_issue/dist/index.js:1756:66)
    at Array.forEach (<anonymous>)
    at Object.179 (/Users/styfle/Desktop/foobar/node_now_pdf2json_issue/dist/index.js:1756:13)
    at __webpack_require__ (/Users/styfle/Desktop/foobar/node_now_pdf2json_issue/dist/index.js:22:30)
    at Object.932 (/Users/styfle/Desktop/foobar/node_now_pdf2json_issue/dist/index.js:26354:10)
    at __webpack_require__ (/Users/styfle/Desktop/foobar/node_now_pdf2json_issue/dist/index.js:22:30)
    at Object.318 (/Users/styfle/Desktop/foobar/node_now_pdf2json_issue/dist/index.js:7517:19)
    at __webpack_require__ (/Users/styfle/Desktop/foobar/node_now_pdf2json_issue/dist/index.js:22:30)

[guybedford]

Here's the code path in play here:

const _pdfjsFiles = [
    'shared/util.js',
    'shared/colorspace.js',
    'shared/pattern.js',
    'shared/function.js',
    'shared/annotation.js',

    'core/core.js',
    'core/obj.js',
    'core/charsets.js',
    'core/crypto.js',
    'core/evaluator.js',
    'core/fonts.js',
    'core/font_renderer.js',
    'core/glyphlist.js',
    'core/image.js',
    'core/metrics.js',
    'core/parser.js',
    'core/stream.js',
    'core/worker.js',
    'core/jpx.js',
    'core/jbig2.js',
    'core/bidi.js',
    'core/jpg.js',
    'core/chunked_stream.js',
    'core/pdf_manager.js',
    'core/cmap.js',
    'core/cidmaps.js',

    'display/canvas.js',
    'display/font_loader.js',
    'display/metadata.js',
    'display/api.js'
];

_pdfjsFiles.forEach( (fieldName, idx, arr) => _fileContent += fs.readFileSync(_basePath + fieldName, 'utf8') );

eval(_fileContent)

The main issue with this from an analysis point of view is that it's a new type of analysis to backtrack from a readFileSync statement to try to piece together the whole expression.

Furthermore it is followed by an eval so this seems like it may be better handled by a special casing again.

[styfle]

@guybedford

Would changing the source code solve the problem?

- _pdfjsFiles.forEach( (fieldName, idx, arr) => _fileContent += fs.readFileSync(_basePath + fieldName, 'utf8') );
+ _pdfjsFiles.forEach( (fieldName, idx, arr) => _fileContent += fs.readFileSync(__dirname + '/../base/' + fieldName, 'utf8') );

Or is it simply the fact that they went up a directory (../), the cause of the problem?

[guybedford]

The problem is more that we don't know to emit the assets in the list of _pdfjsFiles, or how to relocate their references. This is because of the level of abstraction between _basePath and _pdfjsFiles in the readFileSync expression.

[guybedford]

@styfle if we had something like your example, then the wildcard emission proposal would handle this though, yes, even without the file list analysis.

[styfle]

@guybedford If "wildcard emission proposal" is referring to #297 then I think we should prioritize #297 and then we could probably submit a PR to pdf2json with my patch above.

[guybedford]

Actually we know exactly what _basePath is in our analysis, so this should work with just #297 implemented.

[styfle]

Fixed in #378

[JinhooBong]

I'm a bit new to github issues so it's a bit confusing to follow where the solution can be found. But I'm struggling with this issue with pdf2json causing failures on my nextjs app hosted on vercel.
I am hitting this error: "Error: ENOENT: no such file or directory, open '/var/task/node_modules/pdf2json/base/shared/util.js'"

[PaulChase]

I'm a bit new to github issues so it's a bit confusing to follow where the solution can be found. But I'm struggling with this issue with pdf2json causing failures on my nextjs app hosted on vercel. I am hitting this error: "Error: ENOENT: no such file or directory, open '/var/task/node_modules/pdf2json/base/shared/util.js'"

I am also facing the same issue with my nextjs app hosted on netlify

[JinhooBong]

Hey @PaulChase, I followed this thread modesty/pdf2json#330 and was able to resolve my issue. Updating the package it was pulling from to a fix from a fellow dev seemed to fix it for me. Hopefully this helps!