NodeJS, How to avoid sending read-only headers in AWS Lambda#Edge? - node.js

Question

I have set up Lambda on Cloudfront to resize images dynamically on the fly using width and height parameters on images , hence I added the code below
(I didn't write this code, and I can't remember where I got it from)
const querystring = require('querystring');
const http = require('http');
const https = require('https');
const fs = require('fs');
const child = require('child_process');
// headers that cloudfront does not allow in the http response
const blacklistedHeaders = [
/^connection$/i,
/^content-length$/i,
/^expect$/i,
/^keep-alive$/i,
/^proxy-authenticate$/i,
/^proxy-authorization$/i,
/^proxy-connection$/i,
/^trailer$/i,
/^upgrade$/i,
/^x-accel-buffering$/i,
/^x-accel-charset$/i,
/^x-accel-limit-rate$/i,
/^x-accel-redirect$/i,
/^X-Amz-Cf-.*/i,
/^X-Amzn-.*/i,
/^X-Cache.*/i,
/^X-Edge-.*/i,
/^X-Forwarded-Proto.*/i,
/^X-Real-IP$/i
];
exports.handler = (event, context, callback) => {
console.log(JSON.stringify(event, null, 2));
const request = event.Records[0].cf.request;
const origin = request.origin.custom;
const protocol = origin.protocol;
const tmpPath = '/tmp/sourceImage';
const targetPath = '/tmp/targetImage';
const getFile = origin.protocol === 'https' ?
https.get :
http.get;
const options = querystring.parse(request.querystring);
const maxSize = 5000;
const width = Math.min(options.width || maxSize, maxSize);
const height = Math.min(options.height || maxSize, maxSize);
// make sure input values are numbers
if (Number.isNaN(width) || Number.isNaN(height)) {
console.log('Invalid input');
context.succeed({
status: '400',
statusDescription: 'Invalid input'
});
return;
}
// dowload the file from the origin server
getFile(`${origin.protocol}://${origin.domainName}${origin.path}${request.uri}`, (res) => {
const statusCode = res.statusCode;
console.log(res.headers);
// grab headers from the origin request and reformat them
// to match the lambda#edge return format
const originHeaders = Object.keys(res.headers)
// some headers we get back from the origin
// must be filtered out because they are blacklisted by cloudfront
.filter((header) => blacklistedHeaders.every((blheader) => !blheader.test(header)))
.reduce((acc, header) => {
acc[header.toLowerCase()] = [
{
key: header,
value: res.headers[header]
}
];
return acc;
}, {})
if (statusCode === 200) {
const writeStream = fs.createWriteStream(tmpPath);
res
.on('error', (e) => {
context.succeed({
status: '500',
statusDescription: 'Error downloading the image'
});
})
.pipe(writeStream)
writeStream
.on('finish', () => {
console.log('image downloaded');
try {
// invoke ImageMagick to resize the image
const stdout = child.execSync(
`convert ${tmpPath} -resize ${width}x${height}\\> -quality 80 ${targetPath}`
);
} catch(e) {
console.log('ImageMagick error');
console.log(e.stderr.toString());
context.succeed({
status: '500',
statusDescription: 'Error resizing image'
});
return;
}
const image = fs.readFileSync(targetPath).toString('base64');
context.succeed({
bodyEncoding: 'base64',
body: image,
headers: originHeaders,
status: '200',
statusDescription: 'OK'
});
})
.on('error', (e) => {
console.log(e);
context.succeed({
status: '500',
statusDescription: 'Error writing the image to a file'
});
})
} else {
// grap the status code from the origin request
// and return to the viewer
console.log('statusCode: ', statusCode);
context.succeed({
status: statusCode.toString(),
headers: originHeaders
});
}
})
};
I am getting a 502 error on some images, not all of them
The Lambda function result failed validation: The function tried to add, delete, or change a read-only header.