$ npm install ursa-optional
--
This Node module provides a fairly complete set of wrappers for the RSA public/private key crypto functionality of OpenSSL.
It is being passivly fixed for nodeJS 8, 10 and 12 (always the current LTS versions). If you find it doesn't work for you, please file a bug (see below). Most likely you'll find a suting PR here. Just drop it in your issue, and I'll cherry-pick the changes into a new release.
If this package does not work on Windows, then see the additional installation requirements
NOTE: This package will never fail to install, but might throw an error if the compiled dependency is missing on runtime. This is a feature, not a bug, to allow the usage of a pureJS fallback library, such as node-forge
This package was transfered from Medium and NodePrime to quartzjer to JoshKaufman on 8-2017
After that, in 2018 this package was forked to a new package named ursa-optional
by mkg20001, with the added benefit of now being optional, to allow the usage of a fallback pureJS library
Pull requests are welcomed to help maintain it
// openssl genrsa -out certs/server/my-server.key.pem 2048
// openssl rsa -in certs/server/my-server.key.pem -pubout -out certs/client/my-server.pub
'use strict';
var fs = require('fs')
, ursa = require('ursa')
, crt
, key
, msg
;
key = ursa.createPrivateKey(fs.readFileSync('./certs/server/my-server.key.pem'));
crt = ursa.createPublicKey(fs.readFileSync('./certs/client/my-server.pub'));
console.log('Encrypt with Public');
msg = crt.encrypt("Everything is going to be 200 OK", 'utf8', 'base64');
console.log('encrypted', msg, '\n');
console.log('Decrypt with Private');
msg = key.decrypt(msg, 'base64', 'utf8');
console.log('decrypted', msg, '\n');
console.log('############################################');
console.log('Reverse Public -> Private, Private -> Public');
console.log('############################################\n');
console.log('Encrypt with Private (called public)');
msg = key.privateEncrypt("Everything is going to be 200 OK", 'utf8', 'base64');
console.log('encrypted', msg, '\n');
console.log('Decrypt with Public (called private)');
msg = crt.publicDecrypt(msg, 'base64', 'utf8');
console.log('decrypted', msg, '\n');
npm install ursa
Or grab the source and
npm install
npm test
Or
node ./test/test.js
On Windows, you'll need to install some dependencies first:
C:\OpenSSL-Win32
or C:\OpenSSL-Win64
)Error: The specified module could not be found.
, copy libeay32.dll
from the OpenSSL bin directory to this module's bin directory, or to Windows\System32. If you don't have libeay32.dll
, you may have to install OpenSSL v1.0.2, not v1.1.x.npm install -g node-gyp
)
node-gyp
installation or use can have
issues on Windows. The node-gyp
README file has detailed instructions
if you have difficulties. This post
is also a good reference.This library aims to be convenient to use, allowing one to pass in and get back regular string objects. However, it is also meant to be reasonably easy to use efficiently, allowing one to pass in and get back Buffer objects. Using Buffers is always the more efficient option.
All methods that can deal with strings take one or more arguments indicating
the encoding to use when interpreting an argument or generating a result.
These are limited to the usual encoding names that are valid for use with
Buffers: base64
binary
hex
and utf8
. If an encoding is left undefined
and the argument is a string, then the encoding is always assumed to be
utf8
. If an argument is a Buffer, then the encoding (if defined at all)
is ignored. An undefined output encoding is always interpreted as a request
for a Buffer result.
The library knows how to read and output PEM format files for both public and private keys, and it can generate new private keys (aka keypairs).
The usual public-encryption / private-decryption operations by default
use padding mode RSA_PKCS1_OAEP_PADDING
, which is the recommended
mode for all new applications (as of this writing). Note that this mode
builds-in a random element into every encryption operation, making it
unnecessary to waste time or effort adding randomness in at a higher layer.
This default may be overridden to use the older mode RSA_PKCS1_PADDING
if needed.
The less well-understood private-encryption / public-decryption operations
(used for building signature mechanisms) by default use padding
mode RSA_PKCS1_PADDING
. This doesn't build in any randomness (but that's
not usually a problem for applications that use these operations). This
default may be overridden to use RSA_NO_PADDING
if needed.
See the doc comments and tests for the excruciating details, but here's a quick rundown of the available top-level exports and instance methods:
Create and return a private key (aka a keypair) read in from the given PEM-format file. If defined, the given password is used to decrypt the PEM file.
The encoding, if specified, applies to both other arguments.
See "Public Key Methods" below for more details.
Create and return a private key from the given components.
Create and return a public key from the given components.
Convenient shorthand for assert(ursa.isKey(obj))
.
Convenient shorthand for assert(ursa.isPrivateKey(obj))
.
Convenient shorthand for assert(ursa.isPublicKey(obj))
.
Coerce the given key value into a key object (either public or private), returning it. If given a private key object, this just returns it as-is. If given a string or Buffer, it tries to parse it as PEM. Anything else will result in an error.
Coerce the given key value into a private key object, returning it. If given a private key object, this just returns it as-is. If given a string or Buffer, it tries to parse it as PEM. Anything else will result in an error.
Coerce the given key value into a public key object, returning it. If given a private key object, this just returns it as-is. If given a string or Buffer, it tries to parse it as PEM. Anything else will result in an error.
Create and return a public key read in from the given PEM-format file. See "Public Key Methods" below for more details.
Create and return a signer which can sign a hash generated with the named
algorithm (such as "sha256"
or "md5"
). See "Signer Methods" below
for more details.
This function is similar to crypto.createSign()
, except this function
takes a hash algorithm name (e.g., "sha256"
) and not a crypto+hash name
combination (e.g., "RSA-SHA256"
).
Create and return a verifier which can verify a hash generated with the
named algorithm (such as "sha256"
or "md5"
). See "Verifier Methods" below
for more details.
This function is similar to crypto.createVerify()
, except this function
takes a hash algorithm name (e.g., "sha256"
) and not a crypto+hash name
combination (e.g., "RSA-SHA256"
).
This returns true
if and only if both arguments are key objects of
the same type (public or private) and their contents match.
Create and return a freshly-generated private key (aka a keypair). The first argument indicates the number of bits in the modulus (1024 or more is generally considered secure). The second argument indicates the exponent value, which must be odd (65537 is the typical value; 3 and 17 are also common). Both arguments are optional and default to 2048 and 65537 (respectively).
This method will throw if modulusBits
is less than 512
(because
it's pretty crazy to want a key with that few bits) or if exponent
is even (because RSA only works for odd exponents).
Using the command-line openssl
tool, this operation is
equivalent to:
openssl genrsa -out key-name.pem <modulusBits>
for exponent 65537, or for exponent 3 with the additional option
-3
. (That tool doesn't support other exponents.)
Return true
if the given object is a key object (public or private) that
was created by this module. Return false
if not.
Return true
if the given object is a private key object that
was created by this module. Return false
if not.
Return true
if the given object is a public key object that
was created by this module. Return false
if not.
Note that, even though all the public key operations work on private keys, this function only returns true if the given object is a public key, per se.
This returns true
if and only if both arguments are key objects of
some sort (either can be public or private, and they don't have to
be the same) and their public aspects match each other.
This returns publicKey
from ssh-rsa public key-string. First argument
must be a string like ssh-rsa AAAAB3Nz.... user@localhost
or Buffer of pubKey bits.
Return the SSH-style public key fingerprint of the given SSH-format
public key (which was, perhaps, the result of a call to
toPublicSsh()
on a key object).
This is no more and no less than an MD5 hash of the given SSH-format public key. This function doesn't actually check to see if the given key is valid (garbage in, garbage out).
Using the command-line ssh-keygen
tool, this operation is
equivalent to:
ssh-keygen -l -f key-name.sshpub
This operation is also equivalent to this:
cat key-name.sshpub | awk '{print $2}' | base64 --decode | md5
These are all the methods available on public keys. These methods are also available on private keys (since private keys have all the underlying data necessary to perform the public-side operations).
This performs the "public encrypt" operation on the given buffer. The result is always a byte sequence that is the same size as the key associated with the instance. (For example, if the key is 2048 bits, then the result of this operation will be 2048 bits, aka 256 bytes.)
The input buffer is limited to be no larger than the key size minus 41 bytes.
If no padding mode is specified, the default, and recommended, mode
is ursa.RSA_PKCS1_OAEP_PADDING
. The mode
ursa.RSA_PKCS1_PADDING
is also supported.
Get the public exponent as an unsigned big-endian byte sequence.
Get the public modulus as an unsigned big-endian byte sequence.
This is a friendly wrapper for verifying signatures. The given buffer
is hashed using the named algorithm, and the result is verified
against the given signature. This returns true
if the hash and
signature match and the signature was produced by the appropriate
private key. This returns false
if the signature is a valid signature
(structurally) but doesn't match. This throws an exception in other
cases.
The encoding, if specified, applies to both buffer-like arguments. The algorithm must always be a string.
If use_pss_padding
is truthy then RSASSA-PSS
padding is used when verifying the signature. salt_len
, if specified, is
the length of the PSS salt (in bytes) or one of the following:
RSA_PKCS1_SALT_LEN_HLEN
(the same as the hash length, default).RSA_PKCS1_SALT_LEN_MAX
(maximum permissable value).This performs the "public decrypt" operation on the given buffer. The result is always a byte sequence that is no more than the size of the key associated with the instance. (For example, if the key is 2048 bits, then the result of this operation will be no more than 2048 bits, aka 256 bytes.)
If no padding mode is specified, the default, and recommended, mode
is ursa.RSA_PKCS1_PADDING
. The mode ursa.RSA_NO_PADDING
is also supported.
This converts the public key data into a PEM-format file.
This converts the public key data into an SSH-format file. This is the
file format one finds in SSH's authorized_keys
and known_hosts
files.
When used in such files, the contents are base64-encoded and prefixed with
the label ssh-rsa
. Depending on context, the line a key appears on may
also have a host name prefix (in known_hosts
) or comment suffix
(in authorized_keys
).
Using the command-line ssh-keygen
tool, this operation is equivalent to:
ssh-keygen -y -f key-name.pem > key-name.sshpub
Return the SSH-style public key fingerprint of this key. See
ursa.sshFingerprint()
, above, for more details.
This performs an RSA public-verify on the given hash buffer, which
should be the result of performing the hash operation named by
the algorithm (such as "sha256"
or "md5"
) on some data. The
signature buffer is checked to see if it contains a private-signed
statement of the algorithm and hash. The method returns true
if
the signature and hash match, or false
if the signature and hash
don't match but the signature is at least a valid signature of
some sort. In any other situation, this throws an exception.
The encoding, if specified, applies to both buffer-like arguments. The algorithm must always be a string.
This method is the underlying one used as part of the implementation
of the higher-level and much friendlier ursa.createVerifier()
and
hashAndVerify()
.
This is an internal method that is used in the implementation of
ursa.isKey()
ursa.isPrivateKey()
ursa.isPublicKey()
and
associated assertion functions. When called externally, it will
always return undefined
.
These are the methods available on private keys, above and beyond what is available for public keys.
This performs the "private decrypt" operation on the given buffer. The result is always a byte sequence that is no more than the size of the key associated with the instance. (For example, if the key is 2048 bits, then the result of this operation will be no more than 2048 bits, aka 256 bytes.)
If no padding mode is specified, the default, and recommended, mode
is ursa.RSA_PKCS1_OAEP_PADDING
. The mode
ursa.RSA_PKCS1_PADDING
is also supported.
Get the private exponent as an unsigned big-endian byte sequence. The returned exponent is not encrypted in any way, so this method should be used with caution.
This is a friendly wrapper for producing signatures. The given buffer is hashed using the named algorithm, and the result is signed using the private key held by this instance. The return value of this method is the signature.
If use_pss_padding
is truthy then RSASSA-PSS
padding is used when generating the signature. The salt_len
, if specified, is
the length of the PSS salt (in bytes) or one of the following:
RSA_PKCS1_SALT_LEN_HLEN
(the same as the hash length, default).RSA_PKCS1_SALT_LEN_RECOVER
(assume RSA_PKCS1_SALT_LEN_MAX
was used when the padding was added).This performs the "private encrypt" operation on the given buffer. The result is always a byte sequence that is the same size as the key associated with the instance. (For example, if the key is 2048 bits, then the result of this operation will be 2048 bits, aka 256 bytes.)
The input buffer is limited to be no larger than the key size minus 12 bytes.
If no padding mode is specified, the default, and recommended, mode
is ursa.RSA_PKCS1_PADDING
. The mode ursa.RSA_NO_PADDING
is also supported.
This performs an RSA private-sign on the given hash buffer, which
should be the result of performing the hash operation named by
the algorithm (such as "sha256"
or "md5"
) on some data. The
result of this operation may later be passed to verify()
on the
corresponding public key.
This method is the underlying one used as part of the implementation
of the higher-level and much friendlier ursa.createSigner()
and
hashAndSign()
.
This converts the private key data into a PEM-format file. The result is not encrypted, so it behooves the user of this method to take care with the result if the key is sensitive from a security standpoint, which is often the case with such things. (YMMV of course.)
These are the methods available on signer objects, which are returned
by ursa.createSigner()
. These are similar to the objects returned
from crypto.createSign()
.
Update the hash in-progress with the given data.
Get the final hash of the data, and sign it using the private key. The return value is the signature, suitable for later verification.
These are the methods available on verifier objects, which are returned
by ursa.createVerifier()
. These are similar to the objects returned
from crypto.createVerify()
.
Update the hash in-progress with the given data.
Get the final hash of the data, and verify that the given signature both matches it and was produced by the private key corresponding to the given public key.
This returns true
if the signature and hash match appropriately,
or false
if the signature appears to be generally valid (e.g.
structurally) yet doesn't match. This throws an exception in all
other cases.
Allowed padding modes for public/private encryption/decryption:
ursa.RSA_PKCS1_PADDING
ursa.RSA_NO_PADDING
ursa.RSA_PKCS1_OAEP_PADDING
Questions, comments, bug reports, and pull requests are all welcome. Submit them at the project on GitHub.
Bug reports that include steps-to-reproduce (including code) are the best. Even better, make them in the form of pull requests that update the test suite. Thanks!
Current (2015+) maintenance by Jeremie Miller.
Previous (2014) maintenance sponsored by NodePrime.
Original Author (2012): Dan Bornstein (personal website), supported by The Obvious Corporation (now Medium).
With contribution from:
With thanks to:
Updates Copyright 2014 NodePrime, Inc.. Original Copyright 2012 The Obvious Corporation.
Licensed under the Apache License, Version 2.0.
See the top-level file [LICENSE.txt](LICENSE.txt)
and
(http://www.apache.org/licenses/LICENSE-2.0).
© 2010 - cnpmjs.org x YWFE | Home | YWFE