Bitcoin script types

Bitcoin doesn't make a distinction between Bitcoin script coming from different sources, like scriptPubKey in transaction output or witness and sigScript in transaction input. There are many other possible script containers for Bitcoin script: redeem script, witness script, tapscript. In fact, any "script" of [bitcoin::Script] type can be used for inputs and outputs. What is a valid script for one will be a valid script for the other; the only req. is formatting of opcodes & pushes. That would mean that in principle every input script can be used as an output script, but not vice versa. But really what makes a "script" is just the fact that it's formatted correctly.

While all Scripts represent the same type semantically, there is a clear distinction at the logical level: Bitcoin script has the property to be committed into some other Bitcoin script – in a nested structures like in several layers, like redeemScript inside of sigScript used for P2SH, or tapScript within witnessScript coming from witness field for Taproot. These nested layers do distinguish on the information they contain, since some of them only commit to the hashes of the nested scripts ([bitcoin::ScriptHash], [WitnessProgram]) or public keys ([bitcoin::PubkeyHash], [bitcoin::WPubkeyHash]), while other contain the full source of the script.

The present type system represents a solution to the problem: it distinguish different logical types by introducing Script wrapper types. It defines [LockScript] as bottom layer of a script hierarchy, containing no other script commitments (in form of their hashes). It also defines types above on it: [PubkeyScript] (for whatever is there in scriptPubkey field of a TxOut), [SigScript] (for whatever comes from sigScript field of [bitcoin::TxIn]), [RedeemScript] and [TapScript]. Then, there are conversion functions, which, for instance, can analyse [PubkeyScript] and if it is a custom script or P2PK return a [LockScript] type - or otherwise fail with error. So with this type system one is always sure which logical information it does contain.

Type derivation

The following charts represent possible relations between script types:

LockScript
_________________________________
^      ^  ^    ^                ^
|      |  |    |                |
[txout.scriptPubKey] <===> PubkeyScript --?--/P2PK & custom/---+      |  |    |                |
|  |    |                |
[txin.sigScript] <===> SigScript --+--?!--/P2(W)PKH/--(#=PubkeyHash)--+  |    |                |
|                                     |    |                |
|                           (#=ScriptHash) |                |
|                                     |    |                |
+--?!--> RedeemScript --+--?!------/P2SH/  |                |
|                  |                |
/P2WSH-in-P2SH/  /#=V0_WitnessProgram_P2WSH/  |
|                  |                |
+--?!--> WitnessScript              |
^^      |                |
|| /#=V1_WitnessProgram/ |
||      |                |
[?txin.witness] <=====================================================++      +--?---> TapScript

Legend:

• [source] <===> : data source
• [?source] <===> : data source which may be absent
• --+--: algorithmic branching (alternative computation options)
• --?-->: a conversion exists, but it may fail (returns [Option] or [Result])
• --?!-->: a conversion exists, but it may fail; however one of alternative branches must always succeed
• ----->: a conversion exists which can't fail
• --/format/--: a format implied by scriptPubKey program
• --(#=type)--: the hash of the value following -> must match to the value of the <type>

Type conversion

LockScript -+-> (PubkeyScript + RedeemScript) -+-> SigScript
|                                  +-> WitnessScript
+-> PubkeyScript
|
+-> TapScript

PubkeyScript --?--> LockScript