Fixes#683 and #691
The lexer's raw mode for reading macro args already attempted
to handle semicolons inside string literals, versus outside ones
which start comments. This change reuses the same function for
reading string literals in normal and raw modes, also handling:
- Commas in strings versus between macro args
- Character escapes
- {Interpolations} and \1-\9 args inside vs. outside strings
- Multi-line string literals
Macro args now allow escaping '\', '"', and '\\'.
A consistent model for expanding macro args and interpolations,
within macro args, string literals, and normal context:
- "{S}" should always equal the contents of S
- "\1" should always act like quoting the value of \1
In preparation for an upcoming change
Makes for nicer error messages, complaining about EOF instead of newlines
The hack had to be kept for the lexer raw mode to avoid a bug;
see the relevant code comment for more info.
Simplify the mess that was option setting (2 redundant variables !?)
Move options to a separate file
Have "modules" own their options, and OPT only access them (less redundancy)
Simplify code, respect naming conventions better
Fixes#669Closes#368Closes#624
Deprecate PRINTT, PRINTV, PRINTI, and PRINTF
Default STRFMT("%f") to 5 fractional digits like "{f:}"
Any use of string formatting will share this default
Document the existing `ROUND`, `CEIL`, and `FLOOR` functions
Also update the trig function docs for searchability
Implement `POW` and `LOG`
Addresses part of #675
Implement ** for integer exponents
** has higher precedence than -, like Python, so -3**4 == -(3**4) == 81
After discussion (starting there:
https://github.com/gbdev/rgbds/pull/594#issuecomment-706437458
), it was decided that plain nested macros should not be
allowed.
Since #590 is fixed, EQUS can be used as a workaround;
multiline strings (#589) will make that easier on the
user when implemented.
Fixes#588, supersedes and closes#594.
Additionally, closes#388.
The old "find symbol with auto scope" function is now three:
- One finds the exact name passed to it, skipping any checks
This is useful e.g. if such checks were already performed.
- One checks that the name is not scoped, and calls the first.
This is useful for names that cannot be scoped, such as checking for EQUS.
Doing this instead of the third should improve performance somehwat, since
this specific case is hit by the lexer each time an identifier is read.
- The last one checks if the name should be expanded (`.loc` → `Glob.loc`),
and that the local part is not scoped. This is essentially the old function.
Since the lexer buffer wraps, the refilling gets handled in two steps:
First, iff the buffer would wrap, the buffer is refilled until its end.
Then, if more characters are requested, that amount is refilled too.
An important detail is that `read()` may not return as many characters as
requested; for this reason, the first step checks if its `read()` was
"full", and skips the second step otherwise.
This is also where a bug lied.
After a *lot* of trying, I eventually managed to reproduce the bug on an
OpenBSD VM, and after adding a couple of `assert`s in `peekInternal`, this
is what happened, starting at line 724:
0. `lexerState->nbChars` is 0, `lexerState->index` is 19;
1. We end up with `target` = 42, and `writeIndex` = 19;
2. 42 + 19 is greater than `LEXER_BUF_SIZE` (= 42), so the `if` is entered;
3. Within the first `readChars`, **`read` only returns 16 bytes**,
advancing `writeIndex` to 35 and `target` to 26;
4. Within the second `readChars`, a `read(26)` is issued, overflowing the
buffer.
The bug should be clear now: **the check at line 750 failed to work!** Why?
Because `readChars` modifies `writeIndex`.
The fix is simply to cache the number of characters expected, and use that.
