Add functions for DER parsing a BIT STRING.

net/der/parse_values.h

Index: net/der/parse_values.h
diff --git a/net/der/parse_values.h b/net/der/parse_values.h
index fbe1d88b80b694e384f63379812c9bc536fc02e6..724ecf7648253812c2c98bb0bdf26bae479764b5 100644
--- a/net/der/parse_values.h
+++ b/net/der/parse_values.h
@@ -30,6 +30,19 @@ NET_EXPORT bool ParseBoolRelaxed(const Input& in, bool* out) WARN_UNUSED_RESULT;
 // uint64_t, is negative, or if there is an error reading the integer.
 NET_EXPORT bool ParseUint64(const Input& in, uint64_t* out) WARN_UNUSED_RESULT;
 
+// Reads a DER-encoded ASN.1 BIT STRING value from |in| and puts the resulting
+// octet string into |bytes|, and the number of unused bits into |unused_bits|.
+// Returns true on success, otherwise returns false and does not modify the
+// out-parameters.
+//
+// On success it can be assumed that:
+//   * |*unused_bits| < 8
+//   * The final byte of |bytes| has its |*unused_bits| least significant bits
+//     set to 0.
+NET_EXPORT bool ParseBitString(const Input& in,
+                               Input* bytes,
+                               uint8_t* unused_bits) WARN_UNUSED_RESULT;

[nharper, 2015/07/21 23:23:47]

Should unused_bits be a size_t* instead? I realize that it's on the wire as a
uint8_t, but a size_t makes more sense semantically. (This would apply to the
methods in parser.h as well.)

[eroman, 2015/07/22 17:05:23]

I would argue for a uint8_t on the grounds that it is sufficient to represent
the needed range.

A size_t in my mind implies a length/offset in memory, which is not the case
here. It may also be overkill to use a (potentially) 64-bit number to represent
a total of 8 total values :)

+
 struct GeneralizedTime {
   uint16_t year;
   uint8_t month;