Add common interface for cryptographic hash functions to lib/crypto.dart.

Add common interface for cryptographic hash functions to lib/crypto.dart.

Implement sha1 and sha256.

R=sgjesse@google.com,iposva@google.com
BUG=
TEST=

Committed: https://code.google.com/p/dart/source/detail?r=6941

[Søren Gjesse, 2012-04-25 13:02:40]

LGTM, this looks nice and clean!

Maybe we should consider moving more functions to the sha1_utils.dart or maybe
new file crypto_utils.dart (e.g _rotr and _rotl) as they might be useful for
other crypto algorithms. Of cause it does not realy matter which file they are
in.

https://chromiumcodereview.appspot.com/10134055/diff/1/lib/crypto/sha1.dart
File lib/crypto/sha1.dart (right):

https://chromiumcodereview.appspot.com/10134055/diff/1/lib/crypto/sha1.dart#n...
lib/crypto/sha1.dart:7: // Construct a SHA256 hasher object.
SHA256 -> SHA1

[Ben Laurie (Google), 2012-04-25 13:17:13]

https://chromiumcodereview.appspot.com/10134055/diff/1/lib/crypto/crypto.dart
File lib/crypto/crypto.dart (right):

https://chromiumcodereview.appspot.com/10134055/diff/1/lib/crypto/crypto.dart...
lib/crypto/crypto.dart:17: * used again.
"should not"? You mean "will throw an exception"? Or "will bite you unexpectedly
at some future date"? :-)

In any case, why? Why not allow digest to be called as often as people want?

Also, what about clear()?

https://chromiumcodereview.appspot.com/10134055/diff/1/tests/lib/src/crypto/s...
File tests/lib/src/crypto/sha1Test.dart (right):

https://chromiumcodereview.appspot.com/10134055/diff/1/tests/lib/src/crypto/s...
tests/lib/src/crypto/sha1Test.dart:25: final expected_values = const [
Where did these values come from? What about standard test vectors?

[Mads Ager (google), 2012-04-25 13:39:34]

https://chromiumcodereview.appspot.com/10134055/diff/1/lib/crypto/crypto.dart
File lib/crypto/crypto.dart (right):

https://chromiumcodereview.appspot.com/10134055/diff/1/lib/crypto/crypto.dart...
lib/crypto/crypto.dart:17: * used again.
On 2012/04/25 13:17:13, Ben Laurie (Google) wrote:
> "should not"? You mean "will throw an exception"? Or "will bite you
unexpectedly
> at some future date"? :-)

Good point. I should throw an exception at least on calling update after digest
has been called. digest adds the padding to the data and therefore adding more
data makes no sense. I think throwing an exception in both methods if digest has
been called is clearest.

> 
> In any case, why? Why not allow digest to be called as often as people want?
> 
> Also, what about clear()?

Create a new hasher if you want a hasher for new data.

https://chromiumcodereview.appspot.com/10134055/diff/1/lib/crypto/sha1.dart
File lib/crypto/sha1.dart (right):

https://chromiumcodereview.appspot.com/10134055/diff/1/lib/crypto/sha1.dart#n...
lib/crypto/sha1.dart:7: // Construct a SHA256 hasher object.
On 2012/04/25 13:02:40, Søren Gjesse wrote:
> SHA256 -> SHA1

Done.

https://chromiumcodereview.appspot.com/10134055/diff/1/tests/lib/src/crypto/s...
File tests/lib/src/crypto/sha1Test.dart (right):

https://chromiumcodereview.appspot.com/10134055/diff/1/tests/lib/src/crypto/s...
tests/lib/src/crypto/sha1Test.dart:25: final expected_values = const [
On 2012/04/25 13:17:13, Ben Laurie (Google) wrote:
> Where did these values come from? What about standard test vectors?

These values came from another library. Adding standard tests could be a good
addition to these smoke tests.

[Ivan Posva, 2012-04-25 13:44:21]

https://chromiumcodereview.appspot.com/10134055/diff/1/tests/lib/src/crypto/s...
File tests/lib/src/crypto/sha1Test.dart (right):

https://chromiumcodereview.appspot.com/10134055/diff/1/tests/lib/src/crypto/s...
tests/lib/src/crypto/sha1Test.dart:25: final expected_values = const [
On 2012/04/25 13:17:13, Ben Laurie (Google) wrote:
> Where did these values come from? What about standard test vectors?

These values were generated using a different SHA-1 implementation. If standard
test vectors exist, then adding them would be a great addition.

[Ben Laurie (Google), 2012-04-25 13:55:35]

On 2012/04/25 13:39:34, Mads Ager wrote:
> https://chromiumcodereview.appspot.com/10134055/diff/1/lib/crypto/crypto.dart
> File lib/crypto/crypto.dart (right):
> 
>
https://chromiumcodereview.appspot.com/10134055/diff/1/lib/crypto/crypto.dart...
> lib/crypto/crypto.dart:17: * used again.
> On 2012/04/25 13:17:13, Ben Laurie (Google) wrote:
> > "should not"? You mean "will throw an exception"? Or "will bite you
> unexpectedly
> > at some future date"? :-)
> 
> Good point. I should throw an exception at least on calling update after
digest
> has been called. digest adds the padding to the data and therefore adding more
> data makes no sense. I think throwing an exception in both methods if digest
has
> been called is clearest.
> 
> > 
> > In any case, why? Why not allow digest to be called as often as people want?
> > 
> > Also, what about clear()?
> 
> Create a new hasher if you want a hasher for new data.
> 
> https://chromiumcodereview.appspot.com/10134055/diff/1/lib/crypto/sha1.dart
> File lib/crypto/sha1.dart (right):
> 
>
https://chromiumcodereview.appspot.com/10134055/diff/1/lib/crypto/sha1.dart#n...
> lib/crypto/sha1.dart:7: // Construct a SHA256 hasher object.
> On 2012/04/25 13:02:40, Søren Gjesse wrote:
> > SHA256 -> SHA1
> 
> Done.
> 
>
https://chromiumcodereview.appspot.com/10134055/diff/1/tests/lib/src/crypto/s...
> File tests/lib/src/crypto/sha1Test.dart (right):
> 
>
https://chromiumcodereview.appspot.com/10134055/diff/1/tests/lib/src/crypto/s...
> tests/lib/src/crypto/sha1Test.dart:25: final expected_values = const [
> On 2012/04/25 13:17:13, Ben Laurie (Google) wrote:
> > Where did these values come from? What about standard test vectors?
> 
> These values came from another library. Adding standard tests could be a good
> addition to these smoke tests.

http://csrc.nist.gov/groups/STM/cavp/documents/shs/shabittestvectors.zip
http://csrc.nist.gov/groups/STM/cavp/documents/shs/shabytetestvectors.zip

(from http://csrc.nist.gov/groups/STM/cavp/index.html#03).

[Ben Laurie (Google), 2012-04-25 13:55:43]

https://chromiumcodereview.appspot.com/10134055/diff/1/lib/crypto/crypto.dart
File lib/crypto/crypto.dart (right):

https://chromiumcodereview.appspot.com/10134055/diff/1/lib/crypto/crypto.dart...
lib/crypto/crypto.dart:17: * used again.
On 2012/04/25 13:39:34, Mads Ager wrote:
> Create a new hasher if you want a hasher for new data.

In, e.g., HMAC I need two instances of the same hash. So, here I am, in
HMAC(Hash hash). How do I create the second one?

[Mads Ager (google), 2012-04-25 14:22:31]

On 2012/04/25 13:55:43, Ben Laurie (Google) wrote:
> https://chromiumcodereview.appspot.com/10134055/diff/1/lib/crypto/crypto.dart
> File lib/crypto/crypto.dart (right):
> 
>
https://chromiumcodereview.appspot.com/10134055/diff/1/lib/crypto/crypto.dart...
> lib/crypto/crypto.dart:17: * used again.
> On 2012/04/25 13:39:34, Mads Ager wrote:
> > Create a new hasher if you want a hasher for new data.
> 
> In, e.g., HMAC I need two instances of the same hash. So, here I am, in
> HMAC(Hash hash). How do I create the second one?

Fair question. Adding reset if/when we need it at a later point will luckily be
easy. :)

[Ben Laurie (Google), 2012-04-25 14:27:26]

On 2012/04/25 14:22:31, Mads Ager wrote:
> On 2012/04/25 13:55:43, Ben Laurie (Google) wrote:
> >
https://chromiumcodereview.appspot.com/10134055/diff/1/lib/crypto/crypto.dart
> > File lib/crypto/crypto.dart (right):
> > 
> >
>
https://chromiumcodereview.appspot.com/10134055/diff/1/lib/crypto/crypto.dart...
> 
> Fair question. Adding reset if/when we need it at a later point will luckily
be
> easy. :)

clone() is also likely to be useful for the same reason.