just/src/parser.rs

840 lines
20 KiB
Rust
Raw Normal View History

2017-11-16 23:30:08 -08:00
use common::*;
use itertools;
2017-11-17 17:28:06 -08:00
use TokenKind::*;
use CompilationErrorKind::*;
2017-11-16 23:30:08 -08:00
use recipe_resolver::resolve_recipes;
use assignment_resolver::resolve_assignments;
pub struct Parser<'a> {
text: &'a str,
tokens: itertools::PutBack<vec::IntoIter<Token<'a>>>,
recipes: Map<&'a str, Recipe<'a>>,
assignments: Map<&'a str, Expression<'a>>,
assignment_tokens: Map<&'a str, Token<'a>>,
exports: Set<&'a str>,
}
impl<'a> Parser<'a> {
pub fn new(text: &'a str, tokens: Vec<Token<'a>>) -> Parser<'a> {
Parser {
text: text,
tokens: itertools::put_back(tokens),
recipes: empty(),
assignments: empty(),
assignment_tokens: empty(),
exports: empty(),
}
}
fn peek(&mut self, kind: TokenKind) -> bool {
let next = self.tokens.next().unwrap();
let result = next.kind == kind;
self.tokens.put_back(next);
result
}
fn accept(&mut self, kind: TokenKind) -> Option<Token<'a>> {
if self.peek(kind) {
self.tokens.next()
} else {
None
}
}
fn accept_any(&mut self, kinds: &[TokenKind]) -> Option<Token<'a>> {
for kind in kinds {
if self.peek(*kind) {
return self.tokens.next();
}
}
None
}
fn accepted(&mut self, kind: TokenKind) -> bool {
self.accept(kind).is_some()
}
fn expect(&mut self, kind: TokenKind) -> Option<Token<'a>> {
if self.peek(kind) {
self.tokens.next();
None
} else {
self.tokens.next()
}
}
fn expect_eol(&mut self) -> Option<Token<'a>> {
self.accepted(Comment);
if self.peek(Eol) {
self.accept(Eol);
None
} else if self.peek(Eof) {
None
} else {
self.tokens.next()
}
}
fn unexpected_token(&self, found: &Token<'a>, expected: &[TokenKind]) -> CompilationError<'a> {
2017-11-17 17:28:06 -08:00
found.error(UnexpectedToken {
2017-11-16 23:30:08 -08:00
expected: expected.to_vec(),
found: found.kind,
})
}
fn recipe(
&mut self,
2017-11-17 17:28:06 -08:00
name: &Token<'a>,
2017-11-16 23:30:08 -08:00
doc: Option<Token<'a>>,
quiet: bool,
2017-11-17 17:28:06 -08:00
) -> CompilationResult<'a, ()> {
2017-11-16 23:30:08 -08:00
if let Some(recipe) = self.recipes.get(name.lexeme) {
2017-11-17 17:28:06 -08:00
return Err(name.error(DuplicateRecipe {
2017-11-16 23:30:08 -08:00
recipe: recipe.name,
first: recipe.line_number
}));
}
let mut parsed_parameter_with_default = false;
let mut parsed_variadic_parameter = false;
let mut parameters: Vec<Parameter> = vec![];
loop {
let plus = self.accept(Plus);
let parameter = match self.accept(Name) {
Some(parameter) => parameter,
None => if let Some(plus) = plus {
return Err(self.unexpected_token(&plus, &[Name]));
} else {
break
},
};
let variadic = plus.is_some();
if parsed_variadic_parameter {
2017-11-17 17:28:06 -08:00
return Err(parameter.error(ParameterFollowsVariadicParameter {
2017-11-16 23:30:08 -08:00
parameter: parameter.lexeme,
}));
}
if parameters.iter().any(|p| p.name == parameter.lexeme) {
2017-11-17 17:28:06 -08:00
return Err(parameter.error(DuplicateParameter {
2017-11-16 23:30:08 -08:00
recipe: name.lexeme, parameter: parameter.lexeme
}));
}
let default;
if self.accepted(Equals) {
if let Some(string) = self.accept_any(&[StringToken, RawString]) {
default = Some(CookedString::new(&string)?.cooked);
} else {
let unexpected = self.tokens.next().unwrap();
return Err(self.unexpected_token(&unexpected, &[StringToken, RawString]));
}
} else {
default = None
}
if parsed_parameter_with_default && default.is_none() {
2017-11-17 17:28:06 -08:00
return Err(parameter.error(RequiredParameterFollowsDefaultParameter{
2017-11-16 23:30:08 -08:00
parameter: parameter.lexeme,
}));
}
parsed_parameter_with_default |= default.is_some();
parsed_variadic_parameter = variadic;
parameters.push(Parameter {
default: default,
name: parameter.lexeme,
token: parameter,
variadic: variadic,
});
}
if let Some(token) = self.expect(Colon) {
// if we haven't accepted any parameters, an equals
// would have been fine as part of an assignment
if parameters.is_empty() {
return Err(self.unexpected_token(&token, &[Name, Plus, Colon, Equals]));
} else {
return Err(self.unexpected_token(&token, &[Name, Plus, Colon]));
}
}
let mut dependencies = vec![];
let mut dependency_tokens = vec![];
while let Some(dependency) = self.accept(Name) {
if dependencies.contains(&dependency.lexeme) {
2017-11-17 17:28:06 -08:00
return Err(dependency.error(DuplicateDependency {
2017-11-16 23:30:08 -08:00
recipe: name.lexeme,
dependency: dependency.lexeme
}));
}
dependencies.push(dependency.lexeme);
dependency_tokens.push(dependency);
}
if let Some(token) = self.expect_eol() {
return Err(self.unexpected_token(&token, &[Name, Eol, Eof]));
}
let mut lines: Vec<Vec<Fragment>> = vec![];
let mut shebang = false;
if self.accepted(Indent) {
while !self.accepted(Dedent) {
if self.accepted(Eol) {
lines.push(vec![]);
continue;
}
if let Some(token) = self.expect(Line) {
2017-11-17 17:28:06 -08:00
return Err(token.error(Internal{
2017-11-16 23:30:08 -08:00
message: format!("Expected a line but got {}", token.kind)
}))
}
let mut fragments = vec![];
while !(self.accepted(Eol) || self.peek(Dedent)) {
if let Some(token) = self.accept(Text) {
if fragments.is_empty() {
if lines.is_empty() {
if token.lexeme.starts_with("#!") {
shebang = true;
}
} else if !shebang
&& !lines.last().and_then(|line| line.last())
.map(Fragment::continuation).unwrap_or(false)
&& (token.lexeme.starts_with(' ') || token.lexeme.starts_with('\t')) {
2017-11-17 17:28:06 -08:00
return Err(token.error(ExtraLeadingWhitespace));
2017-11-16 23:30:08 -08:00
}
}
fragments.push(Fragment::Text{text: token});
} else if let Some(token) = self.expect(InterpolationStart) {
return Err(self.unexpected_token(&token, &[Text, InterpolationStart, Eol]));
} else {
fragments.push(Fragment::Expression{
expression: self.expression(true)?
});
if let Some(token) = self.expect(InterpolationEnd) {
return Err(self.unexpected_token(&token, &[InterpolationEnd]));
}
}
}
lines.push(fragments);
}
}
self.recipes.insert(name.lexeme, Recipe {
line_number: name.line,
name: name.lexeme,
doc: doc.map(|t| t.lexeme[1..].trim()),
dependencies: dependencies,
dependency_tokens: dependency_tokens,
parameters: parameters,
private: &name.lexeme[0..1] == "_",
lines: lines,
shebang: shebang,
quiet: quiet,
});
Ok(())
}
2017-11-17 17:28:06 -08:00
fn expression(&mut self, interpolation: bool) -> CompilationResult<'a, Expression<'a>> {
2017-11-16 23:30:08 -08:00
let first = self.tokens.next().unwrap();
let lhs = match first.kind {
Name => Expression::Variable {name: first.lexeme, token: first},
Backtick => Expression::Backtick {
raw: &first.lexeme[1..first.lexeme.len()-1],
token: first
},
RawString | StringToken => {
Expression::String{cooked_string: CookedString::new(&first)?}
}
_ => return Err(self.unexpected_token(&first, &[Name, StringToken])),
};
if self.accepted(Plus) {
let rhs = self.expression(interpolation)?;
Ok(Expression::Concatination{lhs: Box::new(lhs), rhs: Box::new(rhs)})
} else if interpolation && self.peek(InterpolationEnd) {
Ok(lhs)
} else if let Some(token) = self.expect_eol() {
if interpolation {
return Err(self.unexpected_token(&token, &[Plus, Eol, InterpolationEnd]))
} else {
Err(self.unexpected_token(&token, &[Plus, Eol]))
}
} else {
Ok(lhs)
}
}
2017-11-17 17:28:06 -08:00
fn assignment(&mut self, name: Token<'a>, export: bool) -> CompilationResult<'a, ()> {
2017-11-16 23:30:08 -08:00
if self.assignments.contains_key(name.lexeme) {
2017-11-17 17:28:06 -08:00
return Err(name.error(DuplicateVariable {variable: name.lexeme}));
2017-11-16 23:30:08 -08:00
}
if export {
self.exports.insert(name.lexeme);
}
let expression = self.expression(false)?;
self.assignments.insert(name.lexeme, expression);
self.assignment_tokens.insert(name.lexeme, name);
Ok(())
}
2017-11-17 17:28:06 -08:00
pub fn justfile(mut self) -> CompilationResult<'a, Justfile<'a>> {
2017-11-16 23:30:08 -08:00
let mut doc = None;
loop {
match self.tokens.next() {
Some(token) => match token.kind {
Eof => break,
Eol => {
doc = None;
continue;
}
Comment => {
if let Some(token) = self.expect_eol() {
2017-11-17 17:28:06 -08:00
return Err(token.error(Internal {
2017-11-16 23:30:08 -08:00
message: format!("found comment followed by {}", token.kind),
}));
}
doc = Some(token);
}
At => if let Some(name) = self.accept(Name) {
2017-11-17 17:28:06 -08:00
self.recipe(&name, doc, true)?;
2017-11-16 23:30:08 -08:00
doc = None;
} else {
let unexpected = &self.tokens.next().unwrap();
return Err(self.unexpected_token(unexpected, &[Name]));
},
Name => if token.lexeme == "export" {
let next = self.tokens.next().unwrap();
if next.kind == Name && self.accepted(Equals) {
self.assignment(next, true)?;
doc = None;
} else {
self.tokens.put_back(next);
2017-11-17 17:28:06 -08:00
self.recipe(&token, doc, false)?;
2017-11-16 23:30:08 -08:00
doc = None;
}
} else if self.accepted(Equals) {
self.assignment(token, false)?;
doc = None;
} else {
2017-11-17 17:28:06 -08:00
self.recipe(&token, doc, false)?;
2017-11-16 23:30:08 -08:00
doc = None;
},
_ => return Err(self.unexpected_token(&token, &[Name, At])),
},
None => return Err(CompilationError {
text: self.text,
index: 0,
line: 0,
column: 0,
width: None,
2017-11-17 17:28:06 -08:00
kind: Internal {
2017-11-16 23:30:08 -08:00
message: "unexpected end of token stream".to_string()
}
}),
}
}
if let Some(token) = self.tokens.next() {
2017-11-17 17:28:06 -08:00
return Err(token.error(Internal {
2017-11-16 23:30:08 -08:00
message: format!("unexpected token remaining after parsing completed: {:?}", token.kind)
}))
}
resolve_recipes(&self.recipes, &self.assignments, self.text)?;
for recipe in self.recipes.values() {
for parameter in &recipe.parameters {
if self.assignments.contains_key(parameter.token.lexeme) {
2017-11-17 17:28:06 -08:00
return Err(parameter.token.error(ParameterShadowsVariable {
2017-11-16 23:30:08 -08:00
parameter: parameter.token.lexeme
}));
}
}
for dependency in &recipe.dependency_tokens {
if !self.recipes[dependency.lexeme].parameters.is_empty() {
2017-11-17 17:28:06 -08:00
return Err(dependency.error(DependencyHasParameters {
2017-11-16 23:30:08 -08:00
recipe: recipe.name,
dependency: dependency.lexeme,
}));
}
}
}
resolve_assignments(&self.assignments, &self.assignment_tokens)?;
Ok(Justfile {
recipes: self.recipes,
assignments: self.assignments,
exports: self.exports,
})
}
}
#[cfg(test)]
mod test {
use super::*;
use brev;
use testing::parse_success;
2017-11-17 23:59:55 -08:00
use tokenize;
2017-11-16 23:30:08 -08:00
2017-11-17 17:28:06 -08:00
macro_rules! summary_test {
($name:ident, $input:expr, $expected:expr $(,)*) => {
#[test]
fn $name() {
let input = $input;
let expected = $expected;
let justfile = parse_success(input);
let actual = format!("{:#}", justfile);
if actual != expected {
println!("got:\n\"{}\"\n", actual);
println!("\texpected:\n\"{}\"", expected);
assert_eq!(actual, expected);
}
}
}
2017-11-16 23:30:08 -08:00
}
2017-11-17 23:59:55 -08:00
macro_rules! error_test {
(
name: $name:ident,
input: $input:expr,
index: $index:expr,
line: $line:expr,
column: $column:expr,
width: $width:expr,
kind: $kind:expr,
) => {
#[test]
fn $name() {
let input = $input;
let expected = CompilationError {
text: input,
index: $index,
line: $line,
column: $column,
width: $width,
kind: $kind,
};
let tokens = tokenize(input).unwrap();
let parser = Parser::new(input, tokens);
if let Err(error) = parser.justfile() {
assert_eq!(error.text, expected.text);
assert_eq!(error.index, expected.index);
assert_eq!(error.line, expected.line);
assert_eq!(error.column, expected.column);
assert_eq!(error.width, expected.width);
assert_eq!(error.kind, expected.kind);
assert_eq!(error, expected);
} else {
panic!("parse succeeded but expected: {}\n{}", expected, input);
}
}
}
}
summary_test! {
parse_empty,
2017-11-17 17:28:06 -08:00
"
2017-11-16 23:30:08 -08:00
# hello
2017-11-17 17:28:06 -08:00
",
"",
}
2017-11-16 23:30:08 -08:00
2017-11-17 23:59:55 -08:00
summary_test! {
parse_string_default,
2017-11-17 17:28:06 -08:00
r#"
2017-11-16 23:30:08 -08:00
foo a="b\t":
2017-11-17 17:28:06 -08:00
"#,
r#"foo a='b\t':"#,
}
2017-11-16 23:30:08 -08:00
2017-11-17 23:59:55 -08:00
summary_test! {
parse_variadic,
2017-11-17 17:28:06 -08:00
r#"
2017-11-16 23:30:08 -08:00
foo +a:
2017-11-17 17:28:06 -08:00
"#,
r#"foo +a:"#,
}
2017-11-16 23:30:08 -08:00
2017-11-17 23:59:55 -08:00
summary_test! {
parse_variadic_string_default,
2017-11-17 17:28:06 -08:00
r#"
2017-11-16 23:30:08 -08:00
foo +a="Hello":
2017-11-17 17:28:06 -08:00
"#,
r#"foo +a='Hello':"#,
}
2017-11-16 23:30:08 -08:00
2017-11-17 23:59:55 -08:00
summary_test! {
parse_raw_string_default,
2017-11-17 17:28:06 -08:00
r#"
2017-11-16 23:30:08 -08:00
foo a='b\t':
2017-11-17 17:28:06 -08:00
"#,
r#"foo a='b\\t':"#,
}
2017-11-16 23:30:08 -08:00
2017-11-17 23:59:55 -08:00
summary_test! {
parse_export,
2017-11-17 17:28:06 -08:00
r#"
2017-11-16 23:30:08 -08:00
export a = "hello"
2017-11-17 17:28:06 -08:00
"#,
r#"export a = "hello""#,
}
2017-11-16 23:30:08 -08:00
2017-11-17 23:59:55 -08:00
summary_test! {
parse_complex,
2017-11-17 17:28:06 -08:00
"
2017-11-16 23:30:08 -08:00
x:
y:
z:
foo = \"xx\"
bar = foo
goodbye = \"y\"
hello a b c : x y z #hello
#! blah
#blarg
{{ foo + bar}}abc{{ goodbye\t + \"x\" }}xyz
1
2
3
2017-11-17 17:28:06 -08:00
",
"bar = foo
2017-11-16 23:30:08 -08:00
foo = \"xx\"
goodbye = \"y\"
hello a b c: x y z
#! blah
#blarg
{{foo + bar}}abc{{goodbye + \"x\"}}xyz
1
2
3
x:
y:
2017-11-17 17:28:06 -08:00
z:"
}
2017-11-16 23:30:08 -08:00
2017-11-17 23:59:55 -08:00
summary_test! {
parse_shebang,
2017-11-17 17:28:06 -08:00
"
2017-11-16 23:30:08 -08:00
practicum = 'hello'
install:
\t#!/bin/sh
\tif [[ -f {{practicum}} ]]; then
\t\treturn
\tfi
2017-11-17 17:28:06 -08:00
",
"practicum = \"hello\"
2017-11-16 23:30:08 -08:00
install:
#!/bin/sh
if [[ -f {{practicum}} ]]; then
\treturn
2017-11-17 17:28:06 -08:00
fi",
}
2017-11-16 23:30:08 -08:00
2017-11-17 23:59:55 -08:00
summary_test! {
parse_simple_shebang,
"a:\n #!\n print(1)",
"a:\n #!\n print(1)",
}
summary_test! {
parse_assignments,
2017-11-17 17:28:06 -08:00
r#"a = "0"
2017-11-16 23:30:08 -08:00
c = a + b + a + b
b = "1"
"#,
2017-11-17 17:28:06 -08:00
r#"a = "0"
2017-11-16 23:30:08 -08:00
b = "1"
2017-11-17 17:28:06 -08:00
c = a + b + a + b"#,
}
2017-11-16 23:30:08 -08:00
2017-11-17 23:59:55 -08:00
summary_test! {
parse_assignment_backticks,
2017-11-17 17:28:06 -08:00
"a = `echo hello`
2017-11-16 23:30:08 -08:00
c = a + b + a + b
b = `echo goodbye`",
2017-11-17 17:28:06 -08:00
"a = `echo hello`
2017-11-16 23:30:08 -08:00
b = `echo goodbye`
2017-11-17 17:28:06 -08:00
c = a + b + a + b",
}
2017-11-16 23:30:08 -08:00
2017-11-17 23:59:55 -08:00
summary_test! {
parse_interpolation_backticks,
2017-11-17 17:28:06 -08:00
r#"a:
2017-11-16 23:30:08 -08:00
echo {{ `echo hello` + "blarg" }} {{ `echo bob` }}"#,
2017-11-17 17:28:06 -08:00
r#"a:
2017-11-16 23:30:08 -08:00
echo {{`echo hello` + "blarg"}} {{`echo bob`}}"#,
2017-11-17 17:28:06 -08:00
}
2017-11-16 23:30:08 -08:00
2017-11-17 23:59:55 -08:00
summary_test! {
eof_test,
2017-11-17 17:28:06 -08:00
"x:\ny:\nz:\na b c: x y z",
"a b c: x y z\n\nx:\n\ny:\n\nz:",
}
2017-11-16 23:30:08 -08:00
2017-11-17 23:59:55 -08:00
summary_test! {
string_quote_escape,
2017-11-17 17:28:06 -08:00
r#"a = "hello\"""#,
r#"a = "hello\"""#,
}
2017-11-16 23:30:08 -08:00
2017-11-17 23:59:55 -08:00
summary_test! {
string_escapes,
2017-11-17 17:28:06 -08:00
r#"a = "\n\t\r\"\\""#,
r#"a = "\n\t\r\"\\""#,
}
2017-11-16 23:30:08 -08:00
2017-11-17 23:59:55 -08:00
summary_test! {
parameters,
2017-11-17 17:28:06 -08:00
"a b c:
{{b}} {{c}}",
"a b c:
{{b}} {{c}}",
}
2017-11-16 23:30:08 -08:00
2017-11-17 23:59:55 -08:00
error_test! {
name: missing_colon,
input: "a b c\nd e f",
index: 5,
line: 0,
column: 5,
width: Some(1),
kind: UnexpectedToken{expected: vec![Name, Plus, Colon], found: Eol},
}
error_test! {
name: missing_default_eol,
input: "hello arg=\n",
index: 10,
line: 0,
column: 10,
width: Some(1),
kind: UnexpectedToken{expected: vec![StringToken, RawString], found: Eol},
}
error_test! {
name: missing_default_eof,
input: "hello arg=",
index: 10,
line: 0,
column: 10,
width: Some(0),
kind: UnexpectedToken{expected: vec![StringToken, RawString], found: Eof},
}
error_test! {
name: missing_default_colon,
input: "hello arg=:",
index: 10,
line: 0,
column: 10,
width: Some(1),
kind: UnexpectedToken{expected: vec![StringToken, RawString], found: Colon},
}
error_test! {
name: missing_default_backtick,
input: "hello arg=`hello`",
index: 10,
line: 0,
column: 10,
width: Some(7),
kind: UnexpectedToken{expected: vec![StringToken, RawString], found: Backtick},
}
error_test! {
name: parameter_after_variadic,
input: "foo +a bbb:",
index: 7,
line: 0,
column: 7,
width: Some(3),
kind: ParameterFollowsVariadicParameter{parameter: "bbb"},
}
error_test! {
name: required_after_default,
input: "hello arg='foo' bar:",
index: 16,
line: 0,
column: 16,
width: Some(3),
kind: RequiredParameterFollowsDefaultParameter{parameter: "bar"},
}
error_test! {
name: missing_eol,
input: "a b c: z =",
index: 9,
line: 0,
column: 9,
width: Some(1),
kind: UnexpectedToken{expected: vec![Name, Eol, Eof], found: Equals},
}
error_test! {
name: duplicate_parameter,
input: "a b b:",
index: 4,
line: 0,
column: 4,
width: Some(1),
kind: DuplicateParameter{recipe: "a", parameter: "b"},
}
error_test! {
name: parameter_shadows_varible,
input: "foo = \"h\"\na foo:",
index: 12,
line: 1,
column: 2,
width: Some(3),
kind: ParameterShadowsVariable{parameter: "foo"},
}
error_test! {
name: dependency_has_parameters,
input: "foo arg:\nb: foo",
index: 12,
line: 1,
column: 3,
width: Some(3),
kind: DependencyHasParameters{recipe: "b", dependency: "foo"},
}
error_test! {
name: duplicate_dependency,
input: "a b c: b c z z",
index: 13,
line: 0,
column: 13,
width: Some(1),
kind: DuplicateDependency{recipe: "a", dependency: "z"},
}
error_test! {
name: duplicate_recipe,
input: "a:\nb:\na:",
index: 6,
line: 2,
column: 0,
width: Some(1),
kind: DuplicateRecipe{recipe: "a", first: 0},
}
error_test! {
name: duplicate_variable,
input: "a = \"0\"\na = \"0\"",
index: 8,
line: 1,
column: 0,
width: Some(1),
kind: DuplicateVariable{variable: "a"},
}
error_test! {
name: extra_whitespace,
input: "a:\n blah\n blarg",
index: 10,
line: 2,
column: 1,
width: Some(6),
kind: ExtraLeadingWhitespace,
}
error_test! {
name: interpolation_outside_of_recipe,
input: "{{",
index: 0,
line: 0,
column: 0,
width: Some(2),
kind: UnexpectedToken{expected: vec![Name, At], found: InterpolationStart},
}
error_test! {
name: unclosed_interpolation_delimiter,
input: "a:\n echo {{ foo",
index: 15,
line: 1,
column: 12,
width: Some(0),
kind: UnexpectedToken{expected: vec![Plus, Eol, InterpolationEnd], found: Dedent},
}
error_test! {
name: plus_following_parameter,
input: "a b c+:",
index: 5,
line: 0,
column: 5,
width: Some(1),
kind: UnexpectedToken{expected: vec![Name], found: Plus},
2017-11-17 17:28:06 -08:00
}
2017-11-16 23:30:08 -08:00
2017-11-17 17:28:06 -08:00
#[test]
fn readme_test() {
let mut justfiles = vec![];
let mut current = None;
2017-11-16 23:30:08 -08:00
2017-11-17 17:28:06 -08:00
for line in brev::slurp("README.asc").lines() {
if let Some(mut justfile) = current {
if line == "```" {
justfiles.push(justfile);
current = None;
} else {
justfile += line;
justfile += "\n";
current = Some(justfile);
}
} else if line == "```make" {
current = Some(String::new());
2017-11-16 23:30:08 -08:00
}
}
2017-11-17 17:28:06 -08:00
for justfile in justfiles {
parse_success(&justfile);
}
2017-11-16 23:30:08 -08:00
}
}