I need to take a pdf byte stream returned from a call to backend service and pass it to a service expecting multipart form data. I've attempted many iterations but have not been successful. Currently I am using this code:

Hi Theresa,
For sure not providing the payloads with the multipart boundaries will result in MIME multipart message is not complete if the Content-Type header indicates a multipart/form-data and you simply pass in your buffer which would simply be your pdf file.  As for the error File validation failed: File type is unknown, I don't believe application/binary is a valid Content-Type MIME value.  Have you tried application/pdf for the Content-Type header of your pdf file?
Best Regards,
Steve Linn

I need to take a pdf byte stream returned from a call to backend service and pass it to a service expecting multipart form data. I've attempted many iterations but have not been successful. Currently I am using this code:

Thanks Steve, 
I'm still getting {"Message":"Unexpected end of MIME multipart stream. MIME multipart message is not complete."}.

This is my current code:

------------------------------
Theresa Furlong

Original Message:
Sent: Fri November 01, 2024 02:27 PM
From: Steve Linn
Subject: PDF stream to multi-part form data in gateway script

Hi Theresa,
For sure not providing the payloads with the multipart boundaries will result in MIME multipart message is not complete if the Content-Type header indicates a multipart/form-data and you simply pass in your buffer which would simply be your pdf file.  As for the error File validation failed: File type is unknown, I don't believe application/binary is a valid Content-Type MIME value.  Have you tried application/pdf for the Content-Type header of your pdf file?
Best Regards,
Steve Linn

------------------------------
Steve Linn
Senior Consulting I/T Specialist
IBM

Original Message:
Sent: Fri November 01, 2024 07:37 AM
From: Theresa Furlong
Subject: PDF stream to multi-part form data in gateway script

I need to take a pdf byte stream returned from a call to backend service and pass it to a service expecting multipart form data. I've attempted many iterations but have not been successful. Currently I am using this code:

------------------------------
Theresa Furlong
------------------------------

Hi Teresa, 
A couple of points. 

First, trace data.  I'm not sure how you're getting this data, but my assumption about the base64 decoded data is just the way  the trace is presenting the data.  A .pdf is going to contain binary data which is not going to create a well formed JSON property in the trace object, therefore it must be getting converted to base64 for that reason. 

Second, I took your code and pasted it into a new API and posted a .pdf file to it.  What I received as a response was missing the ending boundary and also a lot of my pdf file.  Some of the .pdf file that I did receive appeared to have translated/garbled characters as well which supported my theory that your issue was due to the transformation of your buffer to a string which by default does a utf-8 encoding.  Some of those characters you can see get garbled, and I'm guess that if within a string a hex 00 is encountered, that's taken as an end of string and your response is being truncated.  The reason your buffer gets an implicit toString() function is that your code is doing string concatenation using your buffer, which will toString() your buffer.

I've posted this in other multipart threads in this forum, but your answer is going to be to deal with buffers only to prevent any implicit toString() functions.  That would mean doing your string concatenations where they make sense, but creating a buffer from that string and using your buffer with your pdf data asis.  If you place each buffer in an array, you can do a Buffer.concat to produce one final buffer from the multiple parts.  My testing this appears to work but please test and verify.  Also note that I did remove your final cflf after your final boundary as I didn't think that was necessary.

context.message.body.readAsBuffer(function (error, buffer) {    if (error) {        context.reject('InvalidData', 'Invalid Data format');    }    var boundary = "--------------------------123456789123456789123456";    //Line break    var crlf = '\r\n';    //Create the multipart message payload using a buffer array    var bufferArray = [];    bufferArray.push(Buffer.from('--' + boundary + crlf + 'Content-Type: application/pdf' +                                 crlf + 'Content-Disposition: form-data; name="test"; fileName = "test.pdf"' + crlf + crlf));    bufferArray.push(buffer);    bufferArray.push(Buffer.from(crlf + '--' + boundary + '--'));    var contentType = 'multipart/form-data; boundary=' + '--------------------------123456789123456789123456';    context.message.body.write(Buffer.concat(bufferArray));    context.message.header.set('Content-Type', contentType);});

Best Regards,

Steve Linn

Thanks Steve, 
I'm still getting {"Message":"Unexpected end of MIME multipart stream. MIME multipart message is not complete."}.

This is my current code:

Hi Theresa,
For sure not providing the payloads with the multipart boundaries will result in MIME multipart message is not complete if the Content-Type header indicates a multipart/form-data and you simply pass in your buffer which would simply be your pdf file.  As for the error File validation failed: File type is unknown, I don't believe application/binary is a valid Content-Type MIME value.  Have you tried application/pdf for the Content-Type header of your pdf file?
Best Regards,
Steve Linn

I need to take a pdf byte stream returned from a call to backend service and pass it to a service expecting multipart form data. I've attempted many iterations but have not been successful. Currently I am using this code:

Steve, 
Thank you so much. That makes complete sense, and the call does work now. I really appreciate the assistance. 

Hi Teresa, 
A couple of points. 

First, trace data.  I'm not sure how you're getting this data, but my assumption about the base64 decoded data is just the way  the trace is presenting the data.  A .pdf is going to contain binary data which is not going to create a well formed JSON property in the trace object, therefore it must be getting converted to base64 for that reason. 

Second, I took your code and pasted it into a new API and posted a .pdf file to it.  What I received as a response was missing the ending boundary and also a lot of my pdf file.  Some of the .pdf file that I did receive appeared to have translated/garbled characters as well which supported my theory that your issue was due to the transformation of your buffer to a string which by default does a utf-8 encoding.  Some of those characters you can see get garbled, and I'm guess that if within a string a hex 00 is encountered, that's taken as an end of string and your response is being truncated.  The reason your buffer gets an implicit toString() function is that your code is doing string concatenation using your buffer, which will toString() your buffer.

I've posted this in other multipart threads in this forum, but your answer is going to be to deal with buffers only to prevent any implicit toString() functions.  That would mean doing your string concatenations where they make sense, but creating a buffer from that string and using your buffer with your pdf data asis.  If you place each buffer in an array, you can do a Buffer.concat to produce one final buffer from the multiple parts.  My testing this appears to work but please test and verify.  Also note that I did remove your final cflf after your final boundary as I didn't think that was necessary.

context.message.body.readAsBuffer(function (error, buffer) {    if (error) {        context.reject('InvalidData', 'Invalid Data format');    }    var boundary = "--------------------------123456789123456789123456";    //Line break    var crlf = '\r\n';    //Create the multipart message payload using a buffer array    var bufferArray = [];    bufferArray.push(Buffer.from('--' + boundary + crlf + 'Content-Type: application/pdf' +                                 crlf + 'Content-Disposition: form-data; name="test"; fileName = "test.pdf"' + crlf + crlf));    bufferArray.push(buffer);    bufferArray.push(Buffer.from(crlf + '--' + boundary + '--'));    var contentType = 'multipart/form-data; boundary=' + '--------------------------123456789123456789123456';    context.message.body.write(Buffer.concat(bufferArray));    context.message.header.set('Content-Type', contentType);});

Best Regards,

Steve Linn

Thanks Steve, 
I'm still getting {"Message":"Unexpected end of MIME multipart stream. MIME multipart message is not complete."}.

This is my current code:

Hi Theresa,
For sure not providing the payloads with the multipart boundaries will result in MIME multipart message is not complete if the Content-Type header indicates a multipart/form-data and you simply pass in your buffer which would simply be your pdf file.  As for the error File validation failed: File type is unknown, I don't believe application/binary is a valid Content-Type MIME value.  Have you tried application/pdf for the Content-Type header of your pdf file?
Best Regards,
Steve Linn

I need to take a pdf byte stream returned from a call to backend service and pass it to a service expecting multipart form data. I've attempted many iterations but have not been successful. Currently I am using this code: