Skip to content

Latest commit

 

History

History
702 lines (607 loc) · 33.5 KB

language_spec.md

File metadata and controls

702 lines (607 loc) · 33.5 KB

Tapescript

Overview

This scripting language was inspired by Bitcoin script. The main use envisioned is as a method for embedding ACL into decentralized data structures, e.g. hash DAGs. The core mechanisms under the hood are the following:

  1. Tape: the Tape takes the script bytes and advances a pointer as the bytes are read. It also keeps a dict of flags which control how certain ops execute and a dict of contracts mapping contract_id to dict contracts used for checking transfers between compatible protocols.
  2. Stack: the Stack provides a memory structure similar to the stack used in Forth and Bitcoin script. Most ops interact with the stack, and most values are stored in the stack. The stack can contain only bytes items; each item is limited in size; and the number of items is also limited.
  3. Cache: the dict cache is where special values are held, e.g. a timestamp or the parts of a message to be used for checking signatures. Additionally, it is possible to move items from stack to cache or vice versa with the limitation that only bytes cache keys can be used for these operations while interpreter values are stored with str cache keys and thus cannot be written by scripts.

Syntax

The syntax of tapescript is simple and unforgiving, and it takes two forms: the bytes fed into the interpreter and the human-readable source code fed into the compiler which produces bytecode for the VM. The syntax for bytecode is just the op code followed by its arguments. The syntax for human-readable code is outlined below.

All symbols must be separated by whitespace, but which whitespace is used does not matter. (Making source code easy on the eyes is still recommended, as it is in any language; see the Style section below for detailed recommendations.) Any op name, value, or bracket/parenthesis is a symbol.

Values

All values in human-readable tapescript are prefixed by an encoding char:

  • d: any value starting with d is a decimal int (e.g. d123)
  • f: any value starting with f is a decimal float (e.g. f-3.21)
  • s: any value starting with s is is a string (e.g. s"hello world")
  • x: any value starting with x is hexadecimal bytes (e.g. x00ff)

Note that string escapes are not currently supported. It is especially important to note that an escaped quotation mark will not be interpreted by the compiler as escaped and will instead terminate the string and cause a syntax error.

Comments

Everything between two hashtags or double quotes is disregarded as a comment.

Calling ops

To call an op, write the op name followed by any argument(s). For example, OP_PUSH s"hello world" will convert the utf-8 string "hello world" into bytes and push it onto the stack, utilizing OP_PUSH1. Note that each op has an alias equal to the name of the op without the OP_ prefix.

Subtapes

Some features are implemented using subtapes by reading a 2 byte uint size argument from the tape, then reading that many bytes from the tape as the subtape definition, then executing the subtape. Subtape definitions have a maximum size of 64KiB (2^16-1 bytes). The following features use subtapes:

  • OP_IF
  • OP_IF_ELSE
  • OP_DEF
  • OP_TRY_EXCEPT
  • OP_EVAL
  • OP_MERKLEVAL
  • OP_TAPROOT
  • OP_LOOP

Note that OP_EVAL does not require any arguments because it reads the top item from the stack as the subtape definition rather than parsing one out from the tape. OP_MERKLEVAL and OP_TAPROOT call OP_EVAL if the supplied script validates as part of the cryptographic commitment.

Conditional programming

Tapescript includes three conditional operators: OP_IF, OP_IF_ELSE, and OP_MERKLEVAL. OP_IF_ELSE is used under the hood for the human-readable syntax of OP_IF ( hoisted_statements ) { if_body } ELSE { else_body }.

OP_IF/OP_IF_ELSE

Unlike in c- type languages, the condition is pulled from the stack by OP_IF/ OP_IF_ELSE, so the condition is not provided as an argument to the operators.

OP_IF {
    OP_PUSH s"true value found on the stack"
} ELSE {
    OP_PUSH s"false value found on the stack"
}

However, conditional statement hoisting is available, so any statements in parentheses after OP_IF will be hoisted, e.g. IF ( <statements> ) { <body> } will be compiled as <statements> IF { <body> }.

If OP_TRUE was called, or non-null bytes were put on the stack in some other way, before this script ran, the first branch will execute. If OP_FALSE was called, or if null bytes were put on the stack, before this script ran, the second branch will execute.

The bodies for both clauses must be between parentheses, or the final clause must be terminated by END_IF, e.g. OP_IF OP_DUP ELSE OP_POP0 END_IF.

OP_MERKLEVAL

Tapescript includes a streamlined mechanism for branching scripts that hide the unexecuted branches behind cryptographic commitments. The syntax for a locking script using this mechanism is simply OP_MERKLEVAL <root sha256 hash>. This op reads 32 bytes from the tape as the root digest; calls OP_DUP then OP_SHA256 twice; moves stack item at index 2 to the top and calls OP_SHA256 once; calls OP_XOR; calls OP_SHA256; pushes root hash onto the stack; calls OP_EQUAL_VERIFY; then finally calls OP_EVAL.

The unlocking script must provide the code of the branch to execute, the hash of the unexecuted branch, and anything needed to execute the branch (in reverse order). This generalizes to any number of levels of branches, but there can be only two branches per level. These form a Merkle-tree like script structure.

See "Example 5: merklized script" in the script_examples.md file for a thorough example of how this works and how it compares to using OP_IF_ELSE for conditional execution and cryptographic script commitments.

Tools are provided that generate the locking and unlocking scripts for use with OP_MERKLEVAL. See the "#### Merklized Scripts" section of the readme for more details.

OP_TAPROOT

Tapescript includes an implementation of the taproot mechanism whereby a script commitment and a public key are combined into a single root commitment which allows for two execution branches: checking a signature against the root commitment (which is a valid public key) and executing the committed script after proving that the committed script and public key combine to form the root. The locking script is OP_TAPROOT <root commitment>.

The key-spend unlocking script takes the following form: PUSH <sig>, where the sig is a signature created with the private key corresponding to the committed public key tweaked by adding an ed25519 scalar derived from sha256(pubkey + sha256(script)). This signature then validates against the root commitment, itself a tweaked public key.

The script-spend unlocking script takes the following form: PUSH <script> PUSH <pubkey>. When the locking script runs, OP_TAPROOT will verify that the supplied script and pubkey combine to form the root commitment, then it will execute the script. Any additional conditions encoded in the script must be fulfilled prior; in practice the committed script will be another locking script, and the unlocking script will be a combination of the unlocking script for the script and then the unlocking proof for OP_TAPROOT.

By using an OP_MERKLEVAL locking script as the committed script, OP_TAPROOT provides an equivalent script experience as the Taproot+MAST upgrade to Bitcoin.

Exception handling

Some ops, such as OP_VERIFY and OP_CHECK_EPOCH_VERIFY, will raise exceptions under certain conditions. If these ops are called within an OP_TRY block, the exception will be caught, serialized, and put into the cache under the key x45, then the EXCEPT block will be executed. Example:

OP_TRY {
    OP_VERIFY
    OP_PUSH x20
} EXCEPT {
    OP_READ_CACHE x45
    OP_PUSH s"ScriptExecutionError|OP_VERIFY check failed"
    OP_EQUAL
}

The above results in a 20 hexadecimal value if it executed without error, true if it raised a ScriptExecutionError with the given error message, and false is some other error was raised.

This feature can be combined with soft forks for conditional logic.

OP_TRY {
    OP_PUSH s"this is a new feature"
    OP_SOME_SOFT_FORK_RAISES_ERROR d1
    OP_PUSH s"old nodes will always execute this"
} EXCEPT {
    OP_PUSH s"old nodes will not execute this"
}

Defining and calling functions

A function can be defined using OP_DEF. Up to 256 functions can be defined, and all statements between the opening and closing curly braces ({ and }) or before a terminal END_DEF will be executed when the function is called using OP_CALL. Each definition is referenced by an integer 0-255. Note that the definition number passed as an argument for OP_CALL must be a value like any other, while the integer used with OP_DEF can be a plain int or a value.

Also note that the compiler currently does not support defining functions within other functions. The interpreter can probably handle it, but that would be undocumented behavior. If I get feedback that indicates this would be a useful feature to support, I will revisit the topic.

OP_DEF 0 {
    OP_DUP
    OP_SHA256
}

OP_DEF 1
    OP_SHAKE256 d20
END_DEF

OP_PUSH x0123
OP_CALL d0
OP_CALL d1

This will define two functions, put the 2 bytes x0123 onto the stack, then call the two functions in sequence. The result will be a stack with x0123 and shake_256(sha256(b'\x01\x23').digest()).digest(20).

Loops

As of 0.4.0, tapescript supports loops. Loop logic is similar to conditional logic: using OP_LOOP { clause }, it is possible to run the clause in a loop as long as the top value of the stack evaluates to True. For example, to count down from 10 to 0:

push d10
loop {
    push d1
    swap2
    subtract_ints d2
}

If the top value evaluates to False, the loop code will not run. Additionally, if the loop runs for more than the callstack_limit, a ScriptExecutionError will be raised to prevent locking up the runtime with an infinite loop.

Macros and Variables

As of v0.3.0, the compiler supports macros and variables of a sort.

Macros

Macros are code templates that take in arguments and return the source code with the template values replaced by the macro arguments.

To define a macro, use the following syntax:

!= name [ arg1 arg2 etc ] {
    OP_SOMETHING arg1
    OP_PUSH arg2
    OP_SOMETHING_ELSE
    # etc #
}

To invoke a macro, use the following:

!name [ arg1 arg2 etc ]

Only positional arguments are supported, and the values supplied to macro calls must be valid. Each macro invocation will be followed by a separate compilation of just the resulting code, and that bytecode will be inserted if successful.

Variables

Variables are simply syntactic sugar for using the cache as a set of registers. The syntax is simple: @= name [ values ] or @= name int to set and @name to copy the values onto the stack. The first setting syntax pushes values onto the stack and then puts them from the stack into the cache. The second setting syntax simply pulls values from the stack. Variables cannot be used for on-tape arguments to ops, e.g. OP_MERKLEVAL @root will not work.

Comptime

Version 0.6.0 added two comptime features: inline compilation and inline execution. The inline compilation feature syntax is ~ { ops } and replaces the code section with a hexadecimal value symbol equal to the compiled ops. The inline execution feature syntax is ~! { ops }; it compiles and executes the ops, then pops the top item of the Stack, and replaces the code section with that item as a hexadecimal value symbol -- if the Stack was empty, it outputs nothing.

Because this is implemented as a preprocessor step, it is recursive. Example:

push ~! {
    push ~! { push d1 random mod_int d2 }
    push ~! { push d1 random mod_int d2 }
    if ( equal ) {
        push ~ { true }
    } else {
        push ~ { false }
    }
}

This will replace ~ { true } with the byte code for OP_TRUE as a hexadecimal value, then it will replace ~ { false } with the byte code for OP_FALSE, then it will replace ~! { push d1 random } with a random byte as a hex symbol, then it will compile and execute the whole OP_IF_ELSE, then it will replace the whole outer comptime exec block with the result. It is a contrived example, but it works and can be tested by copying and pasting into the REPL (expect the stack to contain 0x00 half of the time and 0x01 the other half of the time).

A more pragmatic example is to generate cryptographic commitments. Example from tests/vectors/correspondent_locking_script.src:

# locking script #
OP_DUP
OP_SHAKE256 d20
OP_PUSH ~! {
    push ~ {
        # committed script #
        OP_IF {
            OP_PUSH x09f5067410b240ac3aa3143016f2285f32fd6eb86ee0efe34248a25bb57bb937
            OP_CHECK_SIG x00
        } ELSE {
            OP_PUSH x1481cd547c77799b4551f1e2947a9ad350bafe972ba55c827ef78279a096343f
            OP_PUSH xcdf907630128847e63dc0b6156b331b29f56cf899e5689b61da3747382d1a80a
            OP_SWAP d1 d2
            OP_CHECK_SIG_VERIFY x00
            OP_CHECK_SIG x00
        }
    }
    shake256 d20
}
OP_EQUAL_VERIFY
OP_EVAL

This will create a short (27 byte) locking script that is unlocked by satisfying the committed script and then pushing the committed script to the stack. The locking script will duplicate the top item on the stack, hash it, push the hash of the committed script, verify the top two items are equal, then execute the committed script.

Style

While style of human-readable source scripts is not enforced, the following are encouraged:

  • Each statement invoking an op should be on its own line except when a few such statements are logically connected and result in one or zero values, e.g. two PUSH ops followed by ADD_INTS d2 or an op followed by VERIFY.
  • The bodies of functions and conditional clauses should be indented. I recommend 4 spaces per indentation level.
  • The opening bracket of a function should be at the end of the line starting OP_DEF, and the closing bracket of the function should be on its own line following the final statement of the function body. If END_DEF is used instead of brackets, then it should be on its own deindented line.
  • The opening parenthesis should be at the end of the line starting OP_IF or ELSE, and the closing parenthesis should be on a new line following the final statement of the conditional clause body.
  • ELSE should be on the same line as the closing parenthesis of the previous conditional clause, i.e. } ELSE { should be its own line.
  • If END_IF is used instead of a closing parenthesis, it should be on its own deindented line following the final statement of the conditional clause.
  • The type prefix of a value should be lowercase. If not, at least be consistent.
  • The opening bracket of a try...except block should be at the end of the line starting OP_TRY, and the closing bracket be on its own line following the statements in the block. If an EXCEPT block is specified, it should be on the same line as the closing bracket of the previous block. If END_TRY is used instead of a closing bracket, then it should be on its own line.
  • Brackets and parenthesis are recommended instead END_DEF/END_IF/END_TRY. Choose a single convention and be consistent.

Ops

Below is a list of ops, the arguments for each, and a brief explanation of what each does. See docs.md for more in-depth details about each op.

List of ops

In the op call syntax below, prefixed values within brackets are items on the stack, in the order in which they must be pushed onto the stack.

  • OP_FALSE - puts x00 onto stack
  • OP_TRUE - puts xFF onto stack
  • OP_PUSH val - puts val onto stack; uses one of OP_PUSH0, OP_PUSH1, or OP_PUSH2, depending on the size of the val
  • OP_PUSH0 val - puts val onto stack; val must be exactly 1 byte
  • OP_PUSH1 size val - puts val onto stack; val byte length must be <256; size must be the length of the val
  • OP_PUSH2 size val - puts val onto stack; val byte length must be <65536; size must be the length of the val
  • OP_GET_MESSAGE sigflags - constructs the message from the sigfields and puts it onto the stack; runs signature extension plugins beforehand
  • OP_POP0 - takes the top item from the stack and puts in the cache at b'P'
  • OP_POP1 count - takes the top count items from the stack and puts them in the cache at b'P'
  • OP_SIZE - counts the number of items on the stack and puts it onto the stack
  • OP_WRITE_CACHE size cache_key count - takes count number of items from the stack and stores them at cache_key; size must be the length of cache_key
  • OP_READ_CACHE size cache_key - takes values from the cache at cache_key and puts them onto the stack; size must be the length of cache_key
  • OP_READ_CACHE_SIZE size cache_key - counts the number of items in the cache at cache_key; size must be the length of cache_key
  • [cache_key] OP_READ_CACHE_STACK - takes an item from the stack as a cache_key, reads the items in the cache at that location, and puts them onto the stack
  • [cache_key] OP_READ_CACHE_STACK_SIZE - takes an item from the stack as a cache_key, counts the number of items in the cache at that location, and puts the count onto the stack
  • [... int1] OP_ADD_INTS count - takes count number of ints from the stack, adds them together, and puts the sum onto the stack
  • [... int1] OP_SUBTRACT_INTS count - takes count number of ints from the stack, subtracts count-1 of them from the first one, and puts the difference onto the stack
  • [int1 ...] OP_MULT_INTS count - takes count number of ints from the stack, multiplies them together, and puts the product onto the stack
  • [int1] OP_DIV_INT size divisor - takes an int from the stack, divides it by the divisor, and puts the quotient onto the stack; size must be the byte length of the divisor
  • [int1 int2] OP_DIV_INTS - takes two ints from the stack, divides the first by the second, and puts the quotient onto the stack
  • [int1] OP_MOD_INT size divisor - takes an int from the stack, divides it by the divisor, and puts the remainder onto the stack; size must be the byte length of the divisor
  • [int2 int1] OP_MOD_INTS - takes two ints from the stack, divides int1 by int2, and puts the remainder onto the stack
  • [... float1] OP_ADD_FLOATS count - takes count number of floats from the stack, adds them together, and puts the sum onto the stack
  • [... float1] OP_SUBTRACT_FLOATS count - takes count number of floats from
  • the stack, subtracts count-1 of them from the first one, and puts the difference onto the stack
  • [float1] OP_DIV_FLOAT divisor - takes a float from the stack, divides it by divisor, and puts the quotient onto the stack; divisor must be a 4-byte float
  • [float1 float2] OP_DIV_FLOATS - takes 2 floats from the stack, divides float1 by float2, and puts the quotient onto the stack
  • OP_MOD_FLOAT divisor - takes a float from the stack, divides it by divisor, and puts the remainder onto the stack
  • OP_MOD_FLOATS - takes 2 floats from the stack, divides the second by the first, and puts the remainder onto the stack
  • [... point1] OP_ADD_POINTS count - takes count ed25519 points from the stack, adds them together, and puts the resulting ed25519 point onto the stack
  • [item] OP_COPY count - copies the top value on the stack count times
  • [item] OP_DUP - duplicates the top stack value
  • [item] OP_SHA256 - replaces the top value of the stack with its sha256 digest
  • [item] OP_SHAKE256 size - replaces the top value of the stack with its size length shake_256 digest
  • [bool] OP_VERIFY - takes a value from the stack and raises an error if it does not evaluate to True
  • [item1 item2] OP_EQUAL - takes 2 values from the stack and puts True onto the stack if they are the same or False if they are not
  • [item1 item2] OP_EQUAL_VERIFY - calls OP_EQUAL and then OP_VERIFY
  • [sig vkey] OP_CHECK_SIG allowed_flags - takes a VerifyKey and signature from the stack, builds a message from the cache values sigfield[1-8] depending upon the sigflags allowed by allowed_flags and appended to the signature, checks if the signature is valid for the VerifyKey and message, and puts True onto the stack if the signature validated or False if it did not
  • [sig vkey] OP_CHECK_SIG_VERIFY allowed_flags - calls OP_CHECK_SIG allowed_flags then OP_VERIFY
  • [constraint] OP_CHECK_TIMESTAMP - takes an unsigned int from the stack as a constraint, takes a timestamp from the cache at "timestamp", compares the timestamp to the constraint, and puts False onto the stack if the timestamp is less than the constraint or if the "ts_threshold" flag was set and exceeded by the difference between the timestamp and the current time (i.e. if the timestamp is more than ts_threshold into the future) and puts True onto the stack otherwise
  • [constraint] OP_CHECK_TIMESTAMP_VERIFY - calls OP_CHECK_TIMESTAMP then OP_VERIFY
  • [constraint] OP_CHECK_EPOCH - takes an unsigned int from the stack as a constraint, subtracts the current time from the constraint, and puts False onto the stack if the "epoch_threshold" flag is met or exceeded by the difference and puts True onto the stack otherwise
  • [constraint] OP_CHECK_EPOCH_VERIFY - calls OP_CHECK_EPOCH then OP_VERIFY
  • OP_DEF handle size def_body - defines a function; see section above
  • OP_CALL handle - calls a function; see section above
  • [bool] OP_IF length clause - runs conditional code; see section above
  • [bool] OP_IF_ELSE length1 clause1 length2 clause2 - runs conditional code; see section above
  • [script] OP_EVAL - takes a value from the stack and runs it as a script
  • [item] OP_NOT - takes a value from the stack and puts the inverse boolean value onto the stack
  • [int] OP_RANDOM - pulls an int from the stack and puts a random byte string that long onto the stack
  • OP_RETURN - ends the script; since functions and conditional clauses are run as subtapes, OP_RETURN ends only the local execution and returns to the outer context
  • OP_SET_FLAG number - sets the tape flag number to the default value
  • OP_UNSET_FLAG number - unsets the tape flag number
  • OP_DEPTH - puts the size of the stack onto the stack as signed int
  • [...] OP_SWAP idx1 idx2 - swaps the items at the given indices on the stack
  • [item1 item2] OP_SWAP2 - swaps the order of the top two items on the stack
  • [...] OP_REVERSE count - reverses the order of the top count items on the stack
  • [item2 item1] OP_CONCAT - takes two values from the stack, concatenates item2 + item1, and puts the result onto the stack
  • [item] OP_SPLIT idx - takes a value from the stack, splits it at the given idx, and puts the two resulting byte strings onto the stack
  • [str1 str2] OP_CONCAT_STR - takes 2 utf-8 strings from the stack, concatenates str2 + str1, and puts the result onto the stack
  • [str] OP_SPLIT_STR idx - takes a utf-8 string from the stack, splits at idx, and puts the 2 resulting strings onto the stack
  • [...] OP_CHECK_TRANSFER - checks proofs of a transfer; see section below
  • [commitment script] OP_MERKLEVAL hash - enforces cryptographic commitment to branching script; see section above
  • OP_TRY_EXCEPT size1 try_body size2 except_body - executes the first block; if an exception is raised, it is serialized into a string and put on the stack, then the second block is executed
  • [v2 v1] OP_LESS - pulls 2 values v1 and v2 from stack; puts (v1<v2) onto stack
  • [v2 v1] OP_LESS_OR_EQUAL - pulls 2 values v1 and v2 from stack; puts (v1<=v2) onto stack
  • OP_GET_VALUE key - puts the read-only cache value(s) at the str key onto the stack
  • [f2 f1] OP_FLOAT_LESS - takes floats f1 and f2 from the stack and puts (f1<f2) onto the stack
  • [f2 f1] OP_FLOAT_LESS_OR_EQUAL - takes floats f1 and f2 from the stack and puts (f1<=f2) onto the stack
  • [int] OP_INT_TO_FLOAT - takes int from stack and puts it back as a float
  • [f32] OP_FLOAT_TO_INT - takes float from stack and puts it back as an int
  • [bool] OP_LOOP length clause - runs the clause in a loop as long as the top value on the stack is not null; errors if the callstack limit is exceeded
  • [...] OP_CHECK_MULTISIG allowed_flags m n - takes n vkeys and m signatures from stack; puts true onto the stack if each of the signatures is valid for one of the vkeys and if each vkey is used only once; otherwise, puts false onto the stack
  • [...] OP_CHECK_MULTISIG_VERIFY allowed_flags m n - calls OP_CHECK_MULTISIG then OP_VERIFY
  • [seed] OP_SIGN flags - takes a signing key seed from the stack, signs a message constructed from sigfields not blanked by the flags, and puts that signature onto the stack.
  • [message seed] OP_SIGN_STACK - takes a signing key seed and message from the stack, signs the message, and puts the signature onto the stack.
  • [sig msg vkey] OP_CHECK_SIG_STACK - takes a verify key, message, and signature from the stack; puts True onto the stack if the signature was valid for the vkey and message, otherwise puts False onto the stack.
  • [seed] OP_DERIVE_SCALAR - takes a seed from stack; derives an ed25519 private key scalar from it; puts the scalar onto the stack and into cache[b'x'] if tape.flags[1].
  • [scalar] OP_CLAMP_SCALAR is_key - reads byte from tape as bool is_key; pulls a value from the stack; clamps the value as an ed25519 private key if is_key else as normal scalar; puts clamped scalar onto the stack.
  • [...] OP_ADD_SCALARS count - takes count values from stack; uses ed25519 scalar addition to sum them; put the sum onto the stack.
  • [... minuend] OP_SUBTRACT_SCALARS count - takes count values from stack; uses ed25519 scalar subtraction to subtract count-1 values from the first value; put the difference onto the stack.
  • [scalar] OP_DERIVE_POINT - takes a value from the stack as a scalar; generates an ed25519 curve point from it; puts the point onto the stack and into cache[b'X'] if tape.flags[2].
  • [... minuend] OP_SUBTRACT_POINTS count - takes count values from the stack as ed25519 points; subtracts count-1 of them from the first using ed25519 inverse group operator; puts difference onto the stack.
  • [seed m T] OP_MAKE_ADAPTER_SIG_PUBLIC - takes tweak point T, message m, and prvkey seed from stack; derives key scalar x from seed and nonce r from seed and m; derives nonce point R from r; generates signature adapter sa; puts R and sa onto stack; sets cache[b'r'] to r if tape.flags[3]; sets cache[b'R'] to R if tape.flages[4]; sets cache[b'T'] to T if tape.flags[6]; sets cache[b'sa'] if tape.flags[9].
  • [m t seed] OP_MAKE_ADAPTER_SIG_PRIVATE - takes prvkey seed, tweak scalar t, and message m from the stack; derives prvkey scalar x from seed; derives pubkey X from x; derives private nonce r from seed and m; derives public nonce point R from r; derives public tweak point T from t; creates signature adapter sa; puts T, R, and sa onto stack; sets cache keys b't' to t if tape.flags[5], b'T' to T if tapeflags[6], b'R' to R if tape.flags[4], and b'sa' to sa if tape.flags[8] (can be used in code with @t, @T, @R, and @sa). Values seed and t should be 32 bytes each. Values T, R, and sa are all public 32 byte values and necessary for verification; t is used to decrypt the signature.
  • [sa R m T X] OP_CHECK_ADAPTER_SIG - takes public key X, tweak point T, message m, nonce point R, and signature adapter sa from the stack; puts True onto the stack if the signature adapter is valid and False otherwise.
  • [sa R t] OP_DECRYPT_ADAPTER_SIG - takes tweak scalar t, nonce point R, and signature adapter sa from stack; calculates nonce RT; decrypts signature s from sa; puts s onto stack; puts RT onto the stack; sets cache keys b's' to s if tape.flags[9] and b'RT' to RT if tape.flags[7] (can be used in code with @s and @RT).
  • [... argcount contract_id] OP_INVOKE - takes an item from the stack as a contract ID; takes a uint from the stack as count; takes count items from the stack as arguments; tries to invoke the contract's abi method, passing it the arguments; puts any return values onto the stack. Raises ScriptExecutionError if the contract is missing. Raises TypeError if the return value type is not bytes or NoneType. If allowed by tape.flags[0], will put any return values into cache at key b'IR'.
  • [item2 item1] OP_XOR - takes two items from the stack; bitwise XORs them together; puts result onto the stack. Can be used in boolean logic as boolean values are just bytes.
  • [item2 item1] OP_OR - takes two items from the stack; bitwise ORs them together; puts result onto the stack. Can be used in boolean logic as boolean values are just bytes.
  • [item2 item1] OP_AND - takes two items from the stack; bitwise ANDs them together; puts result onto the stack. Can be used in boolean logic as boolean values are just bytes.
  • [...] OP_CHECK_TEMPLATE sigflags - pulls a template from the stack for every sigfield indicated in the sigflags and validates the associated sigfield against the template by running the "check_template" plugins or, if there are none, by doing an equality comparison; if all template checks pass, puts 0xff onto the stack, otherwise puts 0x00 onto the stack; runs the signature extension plugins beforehand if tape.flags[10] is set, which is default behavior.
  • [...] OP_CHECK_TEMPLATE_VERIFY sigflags - runs OP_CHECK_TEMPLATE sigflags then OP_VERIFY
  • [...] OP_TAPROOT root - if the top item in the stack is a public key, verify the supplied script (2nd item from stack top) and the public key combine into the root using sha256 and ed25519, then execute the supplied script if they do or remove the script from the stack and put 0x00 onto the stack if they do not; else verify the top item is a signature that validates against the root as the public key, and put 0xFF onto stack if it is or 0x00 onto the stack otherwise
  • NOP count - removes count values from the stack; dummy ops useful for soft fork updates

OP_CHECK_TRANSFER

Pulls an item from the stack, interpreting as an unsigned int count; takes an item from the stack as a contract_id; takes an item from the stack as an amount; takes an item from the stack as a serialized constraint; takes an item from the stack as a destination (address, locking script hash, etc); takes the count number of items from the stack as sources; takes the count number of items from the stack as txn_proofs; verifies that the aggregate of the transfers to the destination from the sources equals or exceeds the amount; verifies that the transfers were valid using the proofs and the contract code; verifies the constraint was followed for each txn proof; and puts True onto the stack if successful and False otherwise. Sources and proofs must be in corresponding order on the stack.

For this to work, the contract must be loaded into the tape's contracts dict at the bytes contract_id dict key. This can be done by passing a contracts dict into run_script or run_auth_script. If the contract should be loaded for all script executions, instead it can be added with add_contract(contract_id, contract). The contract must be an instance of a class implementing the CanCheckTransfer interface with following functions:

  • verify_txn_proof(txn_proof: bytes) -> bool
  • verify_transfer(txn_proof: bytes, source: bytes, destination: bytes) -> bool
  • verify_txn_constraint(txn_proof: bytes, constraint: bytes) -> bool
  • calc_txn_aggregates(txn_proofs: list[bytes], scope: bytes = None) -> dict[bytes, int]

The contract should be the source of the values put onto the stack and passed to the contract functions by OP_CHECK_TRANSFER count or at least sharing an interface with the source of those values.

The first three functions will be called on each transaction proof, and a False returned for any of them will result in False placed onto the stack. Then, calc_txn_aggregates will be called and supplied the list of txn proofs, and the result for the destination will be taken out of the result of that function call; if it is equal to or greater than the amount and all proofs were valid, it puts True onto the stack, else it puts False onto the stack.

OP_INVOKE

Takes an item from the stack as contract_id; takes a uint from the stack as argcount; takes argcount items from the stack as arguments; tries to invoke the contract's abi method, passing it the arguments; puts any return values onto the stack. Raises ScriptExecutionError if the contract is missing or does not implement the CanBeInvoked interface. Raises TypeError if the return value type is not bytes or NoneType. If allowed by tape.flag[0], will put any return values into cache at key b'IR'.

Example:

# file somecontract.py
from tapescript import int_to_bytes

class SomeContract:
    def abi(self, args: list[bytes]) -> list[bytes]:
        if not len(args):
            return [b'\x00']
        avg_size = sum([len(a) for a in args]) / len(args)
        return [int_to_bytes(int(avg_size))]

# file bootstrap.py
from hashlib import shake_256
from inspect import getsource
import somecontract

def boot():
    contract_id = shake_256(bytes(getsource(somecontract), 'utf-8')).digest(20)
    add_contract(contract_id, somecontract.SomeContract)

# file test.py
from bootstrap import boot
from tapescript import compile_script, run_script, bytes_to_int

boot()

script = '''
push xfeedbeef
push s"yellow submarine"
push d2
push x66b58394825b07bc65e504697654d7dd43640f26
invoke
'''

_, stack, cache = run_script(compile_script(script))
assert stack.qsize() == 1
assert bytes_to_int(stack.get()) == 10