roc/compiler/parse/src/string_literal.rs
2021-02-26 15:09:52 +01:00

326 lines
13 KiB
Rust

use crate::ast::{Attempting, EscapedChar, StrLiteral, StrSegment};
use crate::expr;
use crate::parser::Progress::*;
use crate::parser::{
allocated, ascii_char, ascii_hex_digits, loc, parse_utf8, unexpected, unexpected_eof,
ParseResult, Parser, State, SyntaxError,
};
use bumpalo::collections::vec::Vec;
use bumpalo::Bump;
pub fn parse<'a>() -> impl Parser<'a, StrLiteral<'a>, SyntaxError<'a>> {
use StrLiteral::*;
move |arena: &'a Bump, mut state: State<'a>| {
let mut bytes = state.bytes.iter();
// String literals must start with a quote.
// If this doesn't, it must not be a string literal!
match bytes.next() {
Some(&byte) => {
if byte != b'"' {
return Err(unexpected(0, Attempting::StrLiteral, state));
}
}
None => {
return Err(unexpected_eof(arena, state, 0));
}
}
// Advance past the opening quotation mark.
state = state.advance_without_indenting(1)?;
// At the parsing stage we keep the entire raw string, because the formatter
// needs the raw string. (For example, so it can "remember" whether you
// wrote \u{...} or the actual unicode character itself.)
//
// Since we're keeping the entire raw string, all we need to track is
// how many characters we've parsed. So far, that's 1 (the opening `"`).
let mut segment_parsed_bytes = 0;
let mut segments = Vec::new_in(arena);
macro_rules! escaped_char {
($ch:expr) => {
// Record the escaped char.
segments.push(StrSegment::EscapedChar($ch));
// Advance past the segment we just added
state = state.advance_without_indenting(segment_parsed_bytes)?;
// Reset the segment
segment_parsed_bytes = 0;
};
}
macro_rules! end_segment {
($transform:expr) => {
// Don't push anything if the string would be empty.
if segment_parsed_bytes > 1 {
// This function is always called after we just parsed
// something which signalled that we should end the
// current segment - so use segment_parsed_bytes - 1 here,
// to exclude that char we just parsed.
let string_bytes = &state.bytes[0..(segment_parsed_bytes - 1)];
match parse_utf8(string_bytes) {
Ok(string) => {
state = state.advance_without_indenting(string.len())?;
segments.push($transform(string));
}
Err(reason) => {
return state.fail(arena, MadeProgress, reason);
}
}
}
// Depending on where this macro is used, in some
// places this is unused.
#[allow(unused_assignments)]
{
// This function is always called after we just parsed
// something which signalled that we should end the
// current segment.
segment_parsed_bytes = 1;
}
};
}
while let Some(&byte) = bytes.next() {
// This is for the byte we just grabbed from the iterator.
segment_parsed_bytes += 1;
match byte {
b'"' => {
// This is the end of the string!
if segment_parsed_bytes == 1 && segments.is_empty() {
match bytes.next() {
Some(b'"') => {
// If the very first three chars were all `"`,
// then this literal begins with `"""`
// and is a block string.
return parse_block_string(arena, state, &mut bytes);
}
_ => {
// Advance 1 for the close quote
return Ok((
MadeProgress,
PlainLine(""),
state.advance_without_indenting(1)?,
));
}
}
} else {
end_segment!(StrSegment::Plaintext);
let expr = if segments.len() == 1 {
// We had exactly one segment, so this is a candidate
// to be StrLiteral::Plaintext
match segments.pop().unwrap() {
StrSegment::Plaintext(string) => StrLiteral::PlainLine(string),
other => {
let vec = bumpalo::vec![in arena; other];
StrLiteral::Line(vec.into_bump_slice())
}
}
} else {
Line(segments.into_bump_slice())
};
// Advance the state 1 to account for the closing `"`
return Ok((MadeProgress, expr, state.advance_without_indenting(1)?));
};
}
b'\n' => {
// This is a single-line string, which cannot have newlines!
// Treat this as an unclosed string literal, and consume
// all remaining chars. This will mask all other errors, but
// it should make it easiest to debug; the file will be a giant
// error starting from where the open quote appeared.
return Err(unexpected(
state.bytes.len() - 1,
Attempting::StrLiteral,
state,
));
}
b'\\' => {
// We're about to begin an escaped segment of some sort!
//
// Record the current segment so we can begin a new one.
// End it right before the `\` char we just parsed.
end_segment!(StrSegment::Plaintext);
// This is for the byte we're about to parse.
segment_parsed_bytes += 1;
// This is the start of a new escape. Look at the next byte
// to figure out what type of escape it is.
match bytes.next() {
Some(b'(') => {
// Advance past the `\(` before using the expr parser
state = state.advance_without_indenting(2)?;
let original_byte_count = state.bytes.len();
// This is an interpolated variable.
// Parse an arbitrary expression, then give a
// canonicalization error if that expression variant
// is not allowed inside a string interpolation.
let (_progress, loc_expr, new_state) =
skip_second!(loc(allocated(expr::expr(0))), ascii_char(b')'))
.parse(arena, state)?;
// Advance the iterator past the expr we just parsed.
for _ in 0..(original_byte_count - new_state.bytes.len()) {
bytes.next();
}
segments.push(StrSegment::Interpolated(loc_expr));
// Reset the segment
segment_parsed_bytes = 0;
state = new_state;
}
Some(b'u') => {
// Advance past the `\u` before using the expr parser
state = state.advance_without_indenting(2)?;
let original_byte_count = state.bytes.len();
// Parse the hex digits, surrounded by parens, then
// give a canonicalization error if the digits form
// an invalid unicode code point.
let (_progress, loc_digits, new_state) = between!(
ascii_char(b'('),
loc(ascii_hex_digits()),
ascii_char(b')')
)
.parse(arena, state)?;
// Advance the iterator past the expr we just parsed.
for _ in 0..(original_byte_count - new_state.bytes.len()) {
bytes.next();
}
segments.push(StrSegment::Unicode(loc_digits));
// Reset the segment
segment_parsed_bytes = 0;
state = new_state;
}
Some(b'\\') => {
escaped_char!(EscapedChar::Backslash);
}
Some(b'"') => {
escaped_char!(EscapedChar::Quote);
}
Some(b'r') => {
escaped_char!(EscapedChar::CarriageReturn);
}
Some(b't') => {
escaped_char!(EscapedChar::Tab);
}
Some(b'n') => {
escaped_char!(EscapedChar::Newline);
}
_ => {
// Invalid escape! A backslash must be followed
// by either an open paren or else one of the
// escapable characters (\n, \t, \", \\, etc)
return Err(unexpected(
state.bytes.len() - 1,
Attempting::StrLiteral,
state,
));
}
}
}
_ => {
// All other characters need no special handling.
}
}
}
// We ran out of characters before finding a closed quote
Err(unexpected_eof(arena, state.clone(), state.bytes.len()))
}
}
fn parse_block_string<'a, I>(
arena: &'a Bump,
state: State<'a>,
bytes: &mut I,
) -> ParseResult<'a, StrLiteral<'a>, SyntaxError<'a>>
where
I: Iterator<Item = &'a u8>,
{
// So far we have consumed the `"""` and that's it.
let mut parsed_chars = 3;
let mut prev_byte = b'"';
let mut quotes_seen = 0;
// start at 3 to omit the opening `"`.
let mut line_start = 3;
let mut lines: Vec<'a, &'a str> = Vec::new_in(arena);
for byte in bytes {
parsed_chars += 1;
// Potentially end the string (unless this is an escaped `"`!)
match byte {
b'"' if prev_byte != b'\\' => {
if quotes_seen == 2 {
// three consecutive qoutes, end string
// Subtract 3 from parsed_chars so we omit the closing `"`.
let line_bytes = &state.bytes[line_start..(parsed_chars - 3)];
return match parse_utf8(line_bytes) {
Ok(line) => {
// state = state.advance_without_indenting(parsed_chars)?;
// lines.push(line);
// Ok((StrLiteral::Block(lines.into_bump_slice()), state))
Err((
MadeProgress,
SyntaxError::NotYetImplemented(format!(
"TODO parse this line in a block string: {:?}",
line
)),
state,
))
}
Err(reason) => state.fail(arena, MadeProgress, reason),
};
}
quotes_seen += 1;
}
b'\n' => {
// note this includes the newline
let line_bytes = &state.bytes[line_start..parsed_chars];
match parse_utf8(line_bytes) {
Ok(line) => {
lines.push(line);
quotes_seen = 0;
line_start = parsed_chars;
}
Err(reason) => {
return state.fail(arena, MadeProgress, reason);
}
}
}
_ => {
quotes_seen = 0;
}
}
prev_byte = *byte;
}
// We ran out of characters before finding 3 closing quotes
Err(unexpected_eof(arena, state, parsed_chars))
}