Hi!
>>>>> "peter" == peter <peter@stripped> writes:
peter> ChangeSet
peter> 1.1243 02/05/10 12:12:10 peter@stripped +10 -0
peter> Add SHA1 and SHA functions which both corresponds to SHA1 digest message
peter> generation.
<cut>
peter> --- 1.45/sql/item_strfunc.cc Thu Apr 18 10:53:55 2002
peter> +++ 1.46/sql/item_strfunc.cc Fri May 10 12:12:00 2002
<cut>
peter> +
peter> +
peter> +
peter> +/* Functions for SHA1 Function object. */
peter> +
peter> +
peter> +
When it comes to function comments, I have tried to have 2 empty lines
before a function comment and one empty line afterwords.
peter> +String *Item_func_sha::val_str(String *str)
peter> +{
peter> + String * sptr= args[0]->val_str(str);
peter> + if (sptr) /* If we got value different from NULL */
peter> + {
peter> + SHA1Context context; /* Context used to generate SHA1 hash */
peter> + uint8_t digest[SHA1HashSize]; /* Temporary buffer to store 160bit digest
> */
peter> + int err; /* error code for SHA library functions */
peter> +
peter> + null_value=0;
peter> + SHA1Reset(&context); /* We do not have to check for error here */
peter> + SHA1Input(&context,(const unsigned char *) sptr->ptr(),
> sptr->length());
peter> + /* No need to check error as the only case would be too long message */
For which function call can we in theory get an error. SHA1Input() or
SHA1Result() ?
peter> + err=SHA1Result(&context,digest);
<cut>
peter> /*
peter> ** Concatinate args with the following premissess
The multi-line comment syntax we should use are as follows:
/*
Comment
*/
Not
/*
** Comment
*/
(Yes, I know we have to correct a lot of old code regarding this...)
peter> +++ include/sha1.h 02/05/10 12:12:00
<cut>
peter> #ifndef _SHA1_H_
peter> #define _SHA1_H_
peter> #include <stdint.h>
peter> /*
peter> * If you do not have the ISO standard stdint.h header file, then you
peter> * must typdef the following:
peter> * name meaning
peter> * uint32_t unsigned 32 bit integer
peter> * uint8_t unsigned 8 bit integer (i.e., unsigned char)
peter> * int_least16_t integer of >= 16 bits
peter> *
peter> */
Instead of the above typedefs, you should use the following one that
are always defined by my_global.h:
uint32
int16
uint8
This will remove the dependenc of stdint.h, which may not exists on
all systems.
peter> #ifndef _SHA_enum_
peter> #define _SHA_enum_
Why do you need the above define ?
If you are concerned it's defined by some standard include file, you
should rename the enum to something else.
peter> enum
peter> {
peter> shaSuccess = 0,
peter> shaNull, /* Null pointer parameter */
peter> shaInputTooLong, /* input data too long */
peter> shaStateError /* called Input after Result */
peter> };
peter> #endif
peter> #define SHA1HashSize 20
Instead of doing an unamed enum, you should use a named one:
enum sha_result_codes
{
shaSuccess = 0,
shaNull, /* Null pointer parameter */
shaInputTooLong, /* input data too long */
shaStateError /* called Input after Result */
};
peter> --- New file ---
peter> +++ mysys/sha1.c 02/05/10 12:12:01
peter> * SHA-1 is designed to work with messages less than 2^64 bits
peter> * long. Although SHA-1 allows a message digest to be generated
peter> * for messages of any number of bits less than 2^64, this
peter> * implementation only works with messages with a length that is
peter> * a multiple of the size of an 8-bit character.
What does the above mean in practice?
That this version only works on 8 bit machines ?
peter> int SHA1Result( SHA1Context *context,
peter> uint8_t Message_Digest[SHA1HashSize])
peter> {
peter> int i;
peter> if (!context || !Message_Digest)
peter> {
peter> return shaNull;
peter> }
I would like to have the above surrounded by an #ifndef DBUG_OFF test.
(mysys should be able to normally assume that the caller knows what to do)
<cut>
peter> int SHA1Input( SHA1Context *context,
peter> const uint8_t *message_array,
peter> unsigned length)
peter> {
peter> if (!length)
peter> {
peter> return shaSuccess;
peter> }
peter> if (!context || !message_array)
peter> {
peter> return shaNull;
peter> }
The above test is not needed (or should be ifdef-ed)
peter> if (context->Computed)
peter> {
context-> Corrupted = shaStateError;
peter> return shaStateError;
peter> }
peter> if (context->Corrupted)
peter> {
peter> return context->Corrupted;
peter> }
peter> while(length-- && !context->Corrupted)
peter> {
context-> Message_Block[context->Message_Block_Index++] =
peter> (*message_array & 0xFF);
peter> context-> Length_Low += 8;
peter> if (context->Length_Low == 0)
peter> {
peter> context-> Length_High++;
peter> if (context->Length_High == 0)
peter> {
peter> /* Message is too long */
peter> context-> Corrupted = 1;
peter> }
peter> }
Why not add a longlong Length strucutre argument instead of
having 'Length_Low' and 'Length_high' ?
<cut>
Otherwise ok; Please fix the above and then push this into the 4.0
tree!
Regards,
Monty
