问题
I need to inject a png into a PDF using HummusJS.
In the production version of my API, I'll be receiving a post request containing a base64 that I'll convert to binary reader stream. (In the example below, I'm using a test read from a local file.)
Here's my test input stream (which is defined in a "create" object factory):
hummusReadStreamObject (filePath) {
let fileData = fs.readFileSync(filePath).toString('hex');
let result = []
for (var i = 0; i < fileData.length; i+=2) {
result.push('0x'+fileData[i]+''+fileData[i+1])
}
return new hummus.PDFRStreamForBuffer(result)
},
I generate a PNG (I have confirmed, the PNG is written and valid) and then perform modification during the stream's "finally" event
png.on('finish', function () {
create.modifiedPDFResponseStream({
pngFileName,
readStream: create.hummusReadStreamObject('/foo/file/path'),
res
})
})
the modifiedPDFResponseStream
now pulls in the png and is supposed to append it to the file:
modifiedPDFResponseStream ({ pngFileName, readStream, res }) {
const writeStream = new hummus.PDFStreamForResponse(res)
const pdfWriter = hummus.createWriterToModify(
readStream,
writeStream,
{ log: `path/to/error.txt` })
debugger
const pageModifier = new hummus.PDFPageModifier(pdfWriter,0);
if (pngFileName) {
const ctx = pageModifier.startContext().getContext()
ctx.drawImage(2, 2, `./path/to/${pngFileName}`)
pageModifier.endContext().writePage()
pdfWriter.end()
}
}
I sense that I'm close to a solution, the error logs do not report any problems, but i receive the following exception when debugging via Chrome:
events.js:183 Uncaught Error: write after end
at write_ (_http_outgoing.js:622:15)
at ServerResponse.write (_http_outgoing.js:617:10)
at PDFStreamForResponse.write
is the issue something to do with the fact that the stream is being populated during the .png's .on('finish', ...)
event? If so, is there a synchronous approach I could take that would mitigate this problem?
回答1:
inspired by the solution from @JAM in "wait for all streams to finish - stream a directory of files"
I solved this issue by having the .pngWriteStream
method return a promise and having the writestream resolve on finish:
return new Promise((resolve, reject) => {
try {
writeStream.on('finish', resolve)
} catch (e) {
$.logger.error(e.message)
reject()
}
so instead of:
const png = create.pngWriteStream({ ... })
png.on('finish', function () {
create.modifiedPDFResponseStream({
pngFileName,
readStream: create.hummusReadStreamObject('/foo/file/path'),
res
})
})
I await the resolution of the promise
const png = await create.pngWriteStream({
fileName: pngFileName,
value: '123456789' })
and the img file is available for hummus.js to write to the pdf!
create.modifiedPDFResponseStream({
pngFileName,
readStream: create.hummusReadStreamObject('/foo/file/path'),
writeStream: new hummus.PDFStreamForResponse(res)
})
来源:https://stackoverflow.com/questions/51731191/how-to-modify-a-stream-from-request-to-a-new-response-stream-in-hummusjs