uint128_t stream_md5( stream_t* stream )
{
int64_t cur, ic, offset, lastc, num;
md5_t* md5;
uint128_t ret = {0};
unsigned char buf[1025];
bool ignore_lf = false;
FOUNDATION_ASSERT( stream );
if( stream->vtable->md5 )
return stream->vtable->md5( stream );
if( !stream || stream_is_sequential( stream ) || !( stream->mode & STREAM_IN ) )
return ret;
FOUNDATION_ASSERT( stream->vtable->read );
cur = stream_tell( stream );
stream_seek( stream, 0, STREAM_SEEK_BEGIN );
md5 = md5_allocate();
while( !stream_eos( stream ) )
{
num = (int64_t)stream->vtable->read( stream, buf, 1024 );
if( !num )
continue;
if( stream->mode & STREAM_BINARY )
md5_digest_raw( md5, buf, (size_t)num );
else
{
//If last buffer ended with CR, ignore a leading LF
offset = 0;
lastc = 0;
if( ignore_lf && ( buf[0] == '\n' ) )
offset = 1;
ignore_lf = false;
//Treat all line endings (LF, CR, CR+LF) as Unix style LF. If file has mixed line endings (for example, one line ending in a single CR and next ending in a single LF), it will not work!
for( ic = 0; ic < num; ++ic )
{
bool was_cr = ( buf[ic] == '\r' );
bool was_lf = ( buf[ic] == '\n' );
if( was_cr || was_lf )
{
if( was_cr && ( ic >= 1023 ) )
ignore_lf = true;
buf[ic] = '\n';
md5_digest_raw( md5, buf + lastc, (size_t)( ic - lastc + 1 ) ); //Include the LF
if( was_cr && ( buf[ic+1] == '\n' ) ) //Check for CR+LF
++ic;
lastc = ic + 1;
}
}
if( lastc < num )
md5_digest_raw( md5, buf + lastc, (size_t)( num - lastc ) );
}
}
stream_seek( stream, cur, STREAM_SEEK_BEGIN );
md5_finalize( md5 );
ret = md5_get_digest_raw( md5 );
md5_deallocate( md5 );
return ret;
}
uint128_t
stream_md5(stream_t* stream) {
size_t cur, ic, lastc, num, limit;
md5_t md5;
uint128_t ret = uint128_null();
unsigned char buf[1025];
bool ignore_lf = false;
if (stream->vtable->md5)
return stream->vtable->md5(stream);
if (stream_is_sequential(stream) || !(stream->mode & STREAM_IN))
return ret;
cur = stream_tell(stream);
stream_seek(stream, 0, STREAM_SEEK_BEGIN);
md5_initialize(&md5);
limit = sizeof(buf)-1;
buf[limit] = 0;
while (!stream_eos(stream)) {
num = stream->vtable->read(stream, buf, limit);
if (!num)
continue;
if (stream->mode & STREAM_BINARY)
md5_digest(&md5, buf, (size_t)num);
else {
//If last buffer ended with CR, ignore a leading LF
lastc = 0;
if (ignore_lf && (buf[0] == '\n'))
lastc = 1;
ignore_lf = false;
//Digest one line at a time
//Treat all line endings (LF, CR, CR+LF) as Unix style LF. If file has mixed line endings
//(for example, one line ending in a single CR and next is empty and ending in a single LF),
//it will not work!
/*lint -e{850} */
for (ic = lastc; ic < num && ic < limit; ++ic) {
bool was_cr = (buf[ic] == '\r');
bool was_lf = (buf[ic] == '\n');
if (was_cr || was_lf) {
if (was_cr && (ic == limit-1))
ignore_lf = true; //Make next buffer ignore leading LF as it is part of CR+LF
buf[ic] = '\n';
md5_digest(&md5, buf + lastc, (size_t)((ic - lastc) + 1)); //Include the LF
if (was_cr && (buf[ic + 1] == '\n')) //Check for CR+LF
++ic;
lastc = ic + 1;
}
}
if (lastc < num)
md5_digest(&md5, buf + lastc, (size_t)(num - lastc));
}
}
stream_seek(stream, (ssize_t)cur, STREAM_SEEK_BEGIN);
md5_digest_finalize(&md5);
ret = md5_get_digest_raw(&md5);
md5_finalize(&md5);
return ret;
}
