Open wunderkind2k1 opened 2 years ago
Thanks for reporting this @wunderkind2k1, I'll take a look!
Let me know if I can be of any help. Sven
Thanks Sven, I plan on looking at this on Monday. You're more than welcome to have a crack at fixing it yourself and opening a PR if you wish!
Can confirm this is also an issue using Windows 10 with the included binaries:
pdf_1.3_NHS_Constitution.tif
is the result when passing a filename to write to, and can be openedfoo.tif
is the result when writing to stdout and then writing to file using fs.writeFileSync()
, and Windows Photo Viewer throws an error when openedAs you can see, both files are the same size, so no issue with them being cut short. Not really sure on how to fix this. 😞
Ok. Thank you for verifying it. I think its kind of complicated to track down... maybe related to the stream events of the child process? I will try to find some time and have a deeper look into it.
hello folks ! I have a question about the path of poppler-utils. how can I add the path of poppler-utils to Poppler object when the application is running in server ? and also can you please provide me the commands of installing poppler data and utils in Alpine os ? thanks. @Fdawgs
hello folks ! I have a question about the path of poppler-utils. how can I add the path of poppler-utils to Poppler object when the application is running in server ? and also can you please provide me the commands of installing poppler data and utils in Alpine os ? thanks. @Fdawgs
Hi Mustafa,
would you please be so kind open and open an other issue for such a question the next time? It does not belong to this issue.
Nevertheless:
"how can I add the path of poppler-utils to Poppler object when the application is running in server ?" -> you put it in the constructor of the Poppler 'object'. like so:
const poppler = new Poppler(<INSERT YOUR PATH TO POPPLER_UTILS>);
"and also can you please provide me the commands of installing poppler data and utils in Alpine os ?" -> this should do the trick:
apk add poppler-utils
Hope this helps.
Hi @wunderkind2k1 sorry for using this issue for non-relevant question.
anyway, Thanks for your helps.
Hi. A little update on this: I had the chance to dig deeper into this during the holidays.
I changed src/index.js a little to get an error:
After running this testcode:
const fs = require("fs");
const path = require("path");
const { Poppler } = require("./index");
const findCairoBinaryPath = require("./findCairoBinaryPath");
const file = fs.readFileSync(
path.join(__dirname, "../test_files/issue_430.pdf")
);
(async () => {
try {
const currentPathToPOPPLER = findCairoBinaryPath();
const poppler = new Poppler(currentPathToPOPPLER);
const res = await poppler.pdfToCairo(file, undefined, {
tiffFile: true,
tiffCompression: "jpeg",
//pngFile: true,
singleFile: true,
firstPageToConvert: 1,
lastPageToConvert: 1,
});
if (res instanceof Error) {
console.log(`Error:\n${JSON.stringify(res)}`);
return;
}
console.log(res.length);
fs.writeFileSync("foo.tif", res, { encoding: "binary" });
} catch (e) {
console.log(`Error: ${e.toString()}`);
}
})();
This is the resulting error:
Error: Error: TIFFAppendToStrip: Maximum TIFF file size exceeded.
TiffWriter: Error writing tiff row 16
// repetitions removed
TIFFAppendToStrip: Maximum TIFF file size exceeded.
TiffWriter: Error writing tiff row 1648
JPEGLib: Application transferred too few scanlines.
To me this looks like there is a problem with the underlying libs like libtiff then. But I haven't been able to find the reason yet, especially as I am wondering why a call to
cat test_files/issue_430.pdf | pdftocairo -tiff -tiffcompression jpeg -singlefile -f 1 -l 1 - - > bar.tif
produces a nice bar.tif
file then.
Thanks for continuing to look at this @wunderkind2k1!
@wunderkind2k1 Did you find a solution to this?
I have just added streaming support to poppler.pdfToCairo in a fork. I tried rendering to TIFF and got the same problem. Just wanted to chime in and inform that this does not seem to be a buffer problem, since my stream version behaves the same.
The file size of my TIFF files are 8 bytes. Rendering to JPG files is no problem with file sizes about 3-9 MB
I can confirm that the exact same pdftocario call works fine manually. The resulting TIFF file is 58 MB in my case. I.e. the last argument to my manually called pdftocairo is "-", which will direct the output to stdout. Piping this into a file works great.
The only difference between my manual call and poppler.pdfToCairo is the spawn command.
stream.end
is called from somewhere.This post might relate to this: https://github.com/nodejs/node/issues/12921 But I didn't find anything useful.
I didn't find a solution to this until now. I ended up doing the tiff conversion with libvips... which has been a solution but a unnecessarily complex one.
Are you using libvips for converting from pdf->tiff? Or are you using pdftocario for png export and libvips for png->tiff?
If the latter is true, you might want to use my stream version, since this makes it easy to pipe the result from poppler.pdfToCairo straigth into Sharp (Sharp uses libvips).
I extracted a png with poppler and used sharp to convert it to a tiff which than has been fed into an ocr... But I am not working on that project anymore. Thanks for sharing your lib - I might use it if needed <3
This issue may be related: https://gitlab.freedesktop.org/poppler/poppler/-/issues/1334.
TL;DR: The tiff library needs random access to the output file.
That sounds like the explanation! Thank you for sharing!
Prerequisites
[X] I have written a descriptive issue title
[X] I have searched existing issues to ensure it has not already been reported
[X] I agree to follow the Code of Conduct that this project adheres to
API/app/plugin version
5.1.6
Node.js version
v16.13.2
Operating system
macOS
Operating system version (i.e. 20.04, 11.3, 10)
12.4
Description
I am trying to fetch a single page from a pdf without writing it down in a separate image file. This works beautifully for jpg's and png's as documented in https://github.com/Fdawgs/node-poppler/blob/master/README.md#popplerpdftocairo. For tiff files it's a different story though: It works only if I give an output file as a second parameter in the
pdfToCairo
function, but not when I use 'undefined'. The result ist way to small - like it is only the header of the tiff file or something.I checked wether my poppler version (22.05.0) is working correctly on the comand line. It does. pdftocairo -tiff -f 1 -l 1 -singlefile example.pdf - > example.tiff works perfectly on the shell.
As far as I can see - node-poppler does send the correct params to the spawned child process. But the result is a very short string - sth like this:
From debugging index.js in node-poppler I can see that
is only called once for the whole childprocess. This could be the problem.
Steps to Reproduce
This code should be enough to see that foo.tif is not a valid tiff file. You can use any or the provided pdf file: example.pdf
Additional information: Though I wrote this code on OSX I also tried it on a docker container with alpine linux expecting the behaviour to be an OSX glitch. But I could also reproduce the problem on linux successfully.
Expected Behaviour
The expected behaviour should be equal for all possible output formats - meaning when using an 'undefined' outputfile and the -singleFile Option the resulting string should contain valid image data.