Change implementation strategy of cashassembly, add scripts, comments support and tests

source/cash-assembly/defaults.ts

@@ -1,61 +1,41 @@
 /**
- * Detects whether a string is a pure CashAssembly expression.
- *
- * CashAssembly expressions look like `$(<variable>)` or `$(<a> <b>)`. This pattern checks
- * that the entire string is one such expression and nothing else. It will not match if
- * there is other text surrounding the expression.
- *
- * For example:
- *   `$(<fee>)`          matches (a full expression)
- *   `OP_DUP $(<fee>)`   does not match (extra text before it)
- *   `$()`               does not match (empty expression)
- */
-export const CASHASSEMBLY_EXPRESSION_PATTERN = /^\$\([^)]+\)$/;
-
-/**
- * Finds all CashAssembly evaluations embedded in a larger string.
- *
- * An evaluation looks like `$(...)`, for example `$(<fee>)` or `$(<a> <b>)`. This pattern
- * locates every occurrence in the input and returns them all (global flag `g`).
- * Empty evaluations `$()` are intentionally excluded because they reference no variables.
- *
- * For example, scanning `"OP_DUP <$(<pubkeyHash>)> OP_HASH160 $(<fee>)"` would return
- * `['$(<pubkeyHash>)', '$(<fee>)']`.
- */
-export const CASHASSEMBLY_EVALUATION_PATTERN = /\$\([^)]+\)/g;
-
-/**
- * Extracts variable names from angle-bracket references inside a CashAssembly evaluation.
- *
- * Inside an evaluation like `$(<pubkeyHash> <fee>)`, variables are referenced as `<name>`.
- * This pattern captures the name between the brackets. The global flag `g` allows iterating
- * over every variable reference in a single evaluation string.
- *
- * For example, running this against `$(<pubkeyHash> <fee>)` would return the variable names
- * `["pubkeyHash", "fee"]`.
- */
-export const CASHASSEMBLY_VARIABLE_PATTERN = /<([^>]+)>/g;
-
-/**
- * Identifies CashAssembly literal tokens that appear inside angle-bracket push statements.
- *
- * Inside an evaluation, not everything between `<` and `>` is a variable name. Literals are
- * also valid push contents: numeric literals (e.g. `<0>`, `<32>`), hex literals (`<0x02>`),
- * binary literals (`<0b1010>`), and string literals (`<"minting">`, `<'hello'>`). This pattern
- * matches any captured token that starts with a digit or a quote character.
- */
-export const CASHASSEMBLY_LITERAL_TOKEN_PATTERN = /^[0-9"']/;
-
-/**
- * Matches a single dot variable method reference inside an angle-bracket identifier.
+ * Matches a single dot variable method reference inside an angle bracket identifier.
  *
  * Used to detect primitive method references such as `expiry.toIso8601`.
  *
- * For example:
- *   `expiry.toIso8601`                         matches (base `expiry`, method `toIso8601`)
- *   `requestedSatoshis`                        does not match (no method)
- *   `key.schnorr_signature.all_outputs`        does not match (more than one dot)
- *   `key.public_key`                           matches the pattern shape but it is only resolved
- *                                              when `type` maps to a primitive
+ * For example `expiry.toIso8601` matches with base expiry and method toIso8601.
+ * `requestedSatoshis` does not match because it has no method.
+ * `key.schnorr_signature.all_outputs` does not match because it has more than one dot.
  */
 export const CASHASSEMBLY_VARIABLE_METHOD_REFERENCE_PATTERN = /^([^.]+)\.([^.]+)$/;
+
+/**
+ * Character count of the `$(` evaluation opener, which is two characters long.
+ *
+ * Adding this length moves the read position to the first character after that sequence so scanning
+ * starts on the evaluation contents rather than on the opening parenthesis.
+ */
+export const CASHASSEMBLY_EVALUATION_START_LENGTH = 2;
+
+/**
+ * Character count of the `//` and `/*` comment openers, which are both two characters long.
+ *
+ * Adding this length moves the read position to the first character after that sequence so the '*'
+ * of a block comment opener cannot be reused as the '*' of its closer.
+ */
+export const CASHASSEMBLY_COMMENT_START_LENGTH = 2;
+
+/**
+ * Character count of the block comment closer, which is two characters long.
+ *
+ * Adding this length moves the read position to the first character after that sequence so the
+ * closing slash cannot pair with the next character and open another comment.
+ */
+export const CASHASSEMBLY_BLOCK_COMMENT_END_LENGTH = 2;
+
+/**
+ * `$()` with nothing between the parentheses.
+ *
+ * Callers use this to detect an evaluation that has no contents to compile.
+ */
+export const EMPTY_CASHASSEMBLY_EVALUATION = '$()';