Comments
This commit is contained in:
parent
8127bcfe89
commit
0eeb21c885
45
src/lexer.cc
45
src/lexer.cc
@ -7,9 +7,52 @@ constexpr std::string_view Valid_Operator_Chars =
|
||||
"<>=!" // comparisons
|
||||
;
|
||||
|
||||
void Lexer::skip_whitespace_and_comments()
|
||||
{
|
||||
for (;;) {
|
||||
bool done_something = false;
|
||||
|
||||
while (consume_if(unicode::is_space)) {
|
||||
done_something = true;
|
||||
}
|
||||
|
||||
// #! line comments
|
||||
if (consume_if('#', '!')) {
|
||||
done_something = true;
|
||||
while (peek() && peek() != '\n') {
|
||||
consume();
|
||||
}
|
||||
}
|
||||
|
||||
// -- line and multiline coments
|
||||
if (consume_if('-', '-')) {
|
||||
done_something = true;
|
||||
if (consume_if('-')) {
|
||||
// multiline
|
||||
unsigned count = 0;
|
||||
while (count < 3) if (consume_if('-')) {
|
||||
++count;
|
||||
} else {
|
||||
consume();
|
||||
count = 0;
|
||||
}
|
||||
while (consume_if('-')) {}
|
||||
} else {
|
||||
// single line
|
||||
while (peek() && peek() != '\n') {
|
||||
consume();
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
if (not done_something)
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
auto Lexer::next_token() -> Result<Token>
|
||||
{
|
||||
while (consume_if(unicode::is_space)) {}
|
||||
skip_whitespace_and_comments();
|
||||
start();
|
||||
|
||||
if (peek() == 0) {
|
||||
|
@ -196,6 +196,9 @@ struct Lexer
|
||||
|
||||
auto next_token() -> Result<Token>;
|
||||
|
||||
// Utility function for next_token()
|
||||
void skip_whitespace_and_comments();
|
||||
|
||||
// Finds next rune in source
|
||||
auto peek() const -> u32;
|
||||
|
||||
@ -218,6 +221,18 @@ struct Lexer
|
||||
return condition && (consume(), true);
|
||||
}
|
||||
|
||||
inline auto consume_if(auto first, auto second) -> bool
|
||||
{
|
||||
if (consume_if(first)) {
|
||||
if (consume_if(second)) {
|
||||
return true;
|
||||
} else {
|
||||
rewind();
|
||||
}
|
||||
}
|
||||
return false;
|
||||
}
|
||||
|
||||
// Goes back last rune
|
||||
void rewind();
|
||||
|
||||
|
@ -51,12 +51,32 @@ static void expect_token_type_and_location(
|
||||
expect(eq(result->location, location), sl) << "tokenized source is at different place then expected";
|
||||
}
|
||||
|
||||
static void expect_empty_file(
|
||||
std::string_view source,
|
||||
reflection::source_location const& sl = reflection::source_location::current())
|
||||
{
|
||||
Lexer lexer{source};
|
||||
auto result = lexer.next_token();
|
||||
expect(!result.has_value(), sl) << "could not produce any tokens from empty file";
|
||||
if (not result.has_value()) {
|
||||
expect(result.error() == errors::End_Of_File, sl) << "could not produce any tokens from empty file";
|
||||
}
|
||||
}
|
||||
|
||||
suite lexer_test = [] {
|
||||
"Empty file"_test = [] {
|
||||
Lexer lexer{""};
|
||||
auto result = lexer.next_token();
|
||||
expect(!result.has_value() >> fatal) << "could not produce any tokens from empty file";
|
||||
expect(result.error() == errors::End_Of_File) << "could not produce any tokens from empty file";
|
||||
expect_empty_file("");
|
||||
};
|
||||
|
||||
"Comments"_test = [] {
|
||||
expect_empty_file("#!/bin/sh");
|
||||
expect_empty_file("-- line comment");
|
||||
expect_token_type_and_value(Token::Type::Numeric, "--- block comment --- 0", "0");
|
||||
expect_empty_file(R"musique(
|
||||
--- hello
|
||||
multiline comment
|
||||
---
|
||||
)musique");
|
||||
};
|
||||
|
||||
"Simple token types"_test = [] {
|
||||
|
Loading…
Reference in New Issue
Block a user