nushell/crates/nu-protocol/src/value.rs

308 lines
8.3 KiB
Rust
Raw Normal View History

pub mod column_path;
mod convert;
mod debug;
pub mod dict;
pub mod evaluate;
pub mod primitive;
Add Range and start Signature support This commit contains two improvements: - Support for a Range syntax (and a corresponding Range value) - Work towards a signature syntax Implementing the Range syntax resulted in cleaning up how operators in the core syntax works. There are now two kinds of infix operators - tight operators (`.` and `..`) - loose operators Tight operators may not be interspersed (`$it.left..$it.right` is a syntax error). Loose operators require whitespace on both sides of the operator, and can be arbitrarily interspersed. Precedence is left to right in the core syntax. Note that delimited syntax (like `( ... )` or `[ ... ]`) is a single token node in the core syntax. A single token node can be parsed from beginning to end in a context-free manner. The rule for `.` is `<token node>.<member>`. The rule for `..` is `<token node>..<token node>`. Loose operators all have the same syntactic rule: `<token node><space><loose op><space><token node>`. The second aspect of this pull request is the beginning of support for a signature syntax. Before implementing signatures, a necessary prerequisite is for the core syntax to support multi-line programs. That work establishes a few things: - `;` and newlines are handled in the core grammar, and both count as "separators" - line comments begin with `#` and continue until the end of the line In this commit, multi-token productions in the core grammar can use separators interchangably with spaces. However, I think we will ultimately want a different rule preventing separators from occurring before an infix operator, so that the end of a line is always unambiguous. This would avoid gratuitous differences between modules and repl usage. We already effectively have this rule, because otherwise `x<newline> | y` would be a single pipeline, but of course that wouldn't work.
2019-12-04 22:14:52 +01:00
pub mod range;
mod serde_bigdecimal;
mod serde_bigint;
use crate::type_name::{ShellTypeName, SpannedTypeName};
use crate::value::dict::Dictionary;
use crate::value::evaluate::Evaluate;
use crate::value::primitive::Primitive;
Add Range and start Signature support This commit contains two improvements: - Support for a Range syntax (and a corresponding Range value) - Work towards a signature syntax Implementing the Range syntax resulted in cleaning up how operators in the core syntax works. There are now two kinds of infix operators - tight operators (`.` and `..`) - loose operators Tight operators may not be interspersed (`$it.left..$it.right` is a syntax error). Loose operators require whitespace on both sides of the operator, and can be arbitrarily interspersed. Precedence is left to right in the core syntax. Note that delimited syntax (like `( ... )` or `[ ... ]`) is a single token node in the core syntax. A single token node can be parsed from beginning to end in a context-free manner. The rule for `.` is `<token node>.<member>`. The rule for `..` is `<token node>..<token node>`. Loose operators all have the same syntactic rule: `<token node><space><loose op><space><token node>`. The second aspect of this pull request is the beginning of support for a signature syntax. Before implementing signatures, a necessary prerequisite is for the core syntax to support multi-line programs. That work establishes a few things: - `;` and newlines are handled in the core grammar, and both count as "separators" - line comments begin with `#` and continue until the end of the line In this commit, multi-token productions in the core grammar can use separators interchangably with spaces. However, I think we will ultimately want a different rule preventing separators from occurring before an infix operator, so that the end of a line is always unambiguous. This would avoid gratuitous differences between modules and repl usage. We already effectively have this rule, because otherwise `x<newline> | y` would be a single pipeline, but of course that wouldn't work.
2019-12-04 22:14:52 +01:00
use crate::value::range::{Range, RangeInclusion};
use crate::{ColumnPath, PathMember};
use bigdecimal::BigDecimal;
use indexmap::IndexMap;
use nu_errors::ShellError;
Add Range and start Signature support This commit contains two improvements: - Support for a Range syntax (and a corresponding Range value) - Work towards a signature syntax Implementing the Range syntax resulted in cleaning up how operators in the core syntax works. There are now two kinds of infix operators - tight operators (`.` and `..`) - loose operators Tight operators may not be interspersed (`$it.left..$it.right` is a syntax error). Loose operators require whitespace on both sides of the operator, and can be arbitrarily interspersed. Precedence is left to right in the core syntax. Note that delimited syntax (like `( ... )` or `[ ... ]`) is a single token node in the core syntax. A single token node can be parsed from beginning to end in a context-free manner. The rule for `.` is `<token node>.<member>`. The rule for `..` is `<token node>..<token node>`. Loose operators all have the same syntactic rule: `<token node><space><loose op><space><token node>`. The second aspect of this pull request is the beginning of support for a signature syntax. Before implementing signatures, a necessary prerequisite is for the core syntax to support multi-line programs. That work establishes a few things: - `;` and newlines are handled in the core grammar, and both count as "separators" - line comments begin with `#` and continue until the end of the line In this commit, multi-token productions in the core grammar can use separators interchangably with spaces. However, I think we will ultimately want a different rule preventing separators from occurring before an infix operator, so that the end of a line is always unambiguous. This would avoid gratuitous differences between modules and repl usage. We already effectively have this rule, because otherwise `x<newline> | y` would be a single pipeline, but of course that wouldn't work.
2019-12-04 22:14:52 +01:00
use nu_source::{AnchorLocation, HasSpan, Span, Spanned, Tag};
use num_bigint::BigInt;
use serde::{Deserialize, Serialize};
use std::path::PathBuf;
use std::time::SystemTime;
#[derive(Debug, Clone, PartialEq, PartialOrd, Eq, Ord, Hash, Serialize, Deserialize)]
pub enum UntaggedValue {
Primitive(Primitive),
Row(Dictionary),
Table(Vec<Value>),
// Errors are a type of value too
Error(ShellError),
Block(Evaluate),
}
impl UntaggedValue {
pub fn retag(self, tag: impl Into<Tag>) -> Value {
Value {
value: self,
tag: tag.into(),
}
}
pub fn data_descriptors(&self) -> Vec<String> {
match self {
UntaggedValue::Primitive(_) => vec![],
UntaggedValue::Row(columns) => columns
.entries
.keys()
.into_iter()
.map(|x| x.to_string())
.collect(),
UntaggedValue::Block(_) => vec![],
UntaggedValue::Table(_) => vec![],
UntaggedValue::Error(_) => vec![],
}
}
pub fn into_value(self, tag: impl Into<Tag>) -> Value {
Value {
value: self,
tag: tag.into(),
}
}
pub fn into_untagged_value(self) -> Value {
Value {
value: self,
tag: Tag::unknown(),
}
}
pub fn is_true(&self) -> bool {
match self {
UntaggedValue::Primitive(Primitive::Boolean(true)) => true,
_ => false,
}
}
pub fn is_some(&self) -> bool {
!self.is_none()
}
pub fn is_none(&self) -> bool {
match self {
UntaggedValue::Primitive(Primitive::Nothing) => true,
_ => false,
}
}
pub fn is_error(&self) -> bool {
match self {
UntaggedValue::Error(_err) => true,
_ => false,
}
}
pub fn expect_error(&self) -> ShellError {
match self {
UntaggedValue::Error(err) => err.clone(),
_ => panic!("Don't call expect_error without first calling is_error"),
}
}
pub fn expect_string(&self) -> &str {
match self {
UntaggedValue::Primitive(Primitive::String(string)) => &string[..],
_ => panic!("expect_string assumes that the value must be a string"),
}
}
#[allow(unused)]
pub fn row(entries: IndexMap<String, Value>) -> UntaggedValue {
UntaggedValue::Row(entries.into())
}
pub fn table(list: &Vec<Value>) -> UntaggedValue {
UntaggedValue::Table(list.to_vec())
}
pub fn string(s: impl Into<String>) -> UntaggedValue {
UntaggedValue::Primitive(Primitive::String(s.into()))
}
pub fn line(s: impl Into<String>) -> UntaggedValue {
UntaggedValue::Primitive(Primitive::Line(s.into()))
}
pub fn column_path(s: Vec<impl Into<PathMember>>) -> UntaggedValue {
UntaggedValue::Primitive(Primitive::ColumnPath(ColumnPath::new(
s.into_iter().map(|p| p.into()).collect(),
)))
}
pub fn int(i: impl Into<BigInt>) -> UntaggedValue {
UntaggedValue::Primitive(Primitive::Int(i.into()))
}
pub fn pattern(s: impl Into<String>) -> UntaggedValue {
UntaggedValue::Primitive(Primitive::String(s.into()))
}
pub fn path(s: impl Into<PathBuf>) -> UntaggedValue {
UntaggedValue::Primitive(Primitive::Path(s.into()))
}
pub fn bytes(s: impl Into<u64>) -> UntaggedValue {
UntaggedValue::Primitive(Primitive::Bytes(s.into()))
}
pub fn decimal(s: impl Into<BigDecimal>) -> UntaggedValue {
UntaggedValue::Primitive(Primitive::Decimal(s.into()))
}
pub fn binary(binary: Vec<u8>) -> UntaggedValue {
UntaggedValue::Primitive(Primitive::Binary(binary))
}
Add Range and start Signature support This commit contains two improvements: - Support for a Range syntax (and a corresponding Range value) - Work towards a signature syntax Implementing the Range syntax resulted in cleaning up how operators in the core syntax works. There are now two kinds of infix operators - tight operators (`.` and `..`) - loose operators Tight operators may not be interspersed (`$it.left..$it.right` is a syntax error). Loose operators require whitespace on both sides of the operator, and can be arbitrarily interspersed. Precedence is left to right in the core syntax. Note that delimited syntax (like `( ... )` or `[ ... ]`) is a single token node in the core syntax. A single token node can be parsed from beginning to end in a context-free manner. The rule for `.` is `<token node>.<member>`. The rule for `..` is `<token node>..<token node>`. Loose operators all have the same syntactic rule: `<token node><space><loose op><space><token node>`. The second aspect of this pull request is the beginning of support for a signature syntax. Before implementing signatures, a necessary prerequisite is for the core syntax to support multi-line programs. That work establishes a few things: - `;` and newlines are handled in the core grammar, and both count as "separators" - line comments begin with `#` and continue until the end of the line In this commit, multi-token productions in the core grammar can use separators interchangably with spaces. However, I think we will ultimately want a different rule preventing separators from occurring before an infix operator, so that the end of a line is always unambiguous. This would avoid gratuitous differences between modules and repl usage. We already effectively have this rule, because otherwise `x<newline> | y` would be a single pipeline, but of course that wouldn't work.
2019-12-04 22:14:52 +01:00
pub fn range(
left: (Spanned<Primitive>, RangeInclusion),
right: (Spanned<Primitive>, RangeInclusion),
) -> UntaggedValue {
UntaggedValue::Primitive(Primitive::Range(Box::new(Range::new(left, right))))
}
pub fn boolean(s: impl Into<bool>) -> UntaggedValue {
UntaggedValue::Primitive(Primitive::Boolean(s.into()))
}
pub fn duration(secs: u64) -> UntaggedValue {
UntaggedValue::Primitive(Primitive::Duration(secs))
}
pub fn system_date(s: SystemTime) -> UntaggedValue {
UntaggedValue::Primitive(Primitive::Date(s.into()))
}
pub fn nothing() -> UntaggedValue {
UntaggedValue::Primitive(Primitive::Nothing)
}
}
#[derive(Debug, Clone, PartialOrd, PartialEq, Ord, Eq, Hash, Serialize, Deserialize)]
pub struct Value {
pub value: UntaggedValue,
pub tag: Tag,
}
impl std::ops::Deref for Value {
type Target = UntaggedValue;
fn deref(&self) -> &Self::Target {
&self.value
}
}
impl Value {
pub fn anchor(&self) -> Option<AnchorLocation> {
self.tag.anchor()
}
pub fn anchor_name(&self) -> Option<String> {
self.tag.anchor_name()
}
pub fn tag(&self) -> Tag {
self.tag.clone()
}
pub fn as_string(&self) -> Result<String, ShellError> {
match &self.value {
UntaggedValue::Primitive(Primitive::String(string)) => Ok(string.clone()),
UntaggedValue::Primitive(Primitive::Line(line)) => Ok(line.clone() + "\n"),
_ => Err(ShellError::type_error("string", self.spanned_type_name())),
}
}
2019-12-02 17:14:05 +01:00
pub fn as_forgiving_string(&self) -> Result<&str, ShellError> {
match &self.value {
UntaggedValue::Primitive(Primitive::String(string)) => Ok(&string[..]),
_ => Err(ShellError::type_error("string", self.spanned_type_name())),
}
}
pub fn as_path(&self) -> Result<PathBuf, ShellError> {
match &self.value {
UntaggedValue::Primitive(Primitive::Path(path)) => Ok(path.clone()),
UntaggedValue::Primitive(Primitive::String(path_str)) => {
Ok(PathBuf::from(&path_str).clone())
}
_ => Err(ShellError::type_error("Path", self.spanned_type_name())),
}
}
Add Range and start Signature support This commit contains two improvements: - Support for a Range syntax (and a corresponding Range value) - Work towards a signature syntax Implementing the Range syntax resulted in cleaning up how operators in the core syntax works. There are now two kinds of infix operators - tight operators (`.` and `..`) - loose operators Tight operators may not be interspersed (`$it.left..$it.right` is a syntax error). Loose operators require whitespace on both sides of the operator, and can be arbitrarily interspersed. Precedence is left to right in the core syntax. Note that delimited syntax (like `( ... )` or `[ ... ]`) is a single token node in the core syntax. A single token node can be parsed from beginning to end in a context-free manner. The rule for `.` is `<token node>.<member>`. The rule for `..` is `<token node>..<token node>`. Loose operators all have the same syntactic rule: `<token node><space><loose op><space><token node>`. The second aspect of this pull request is the beginning of support for a signature syntax. Before implementing signatures, a necessary prerequisite is for the core syntax to support multi-line programs. That work establishes a few things: - `;` and newlines are handled in the core grammar, and both count as "separators" - line comments begin with `#` and continue until the end of the line In this commit, multi-token productions in the core grammar can use separators interchangably with spaces. However, I think we will ultimately want a different rule preventing separators from occurring before an infix operator, so that the end of a line is always unambiguous. This would avoid gratuitous differences between modules and repl usage. We already effectively have this rule, because otherwise `x<newline> | y` would be a single pipeline, but of course that wouldn't work.
2019-12-04 22:14:52 +01:00
pub fn as_primitive(&self) -> Result<Primitive, ShellError> {
match &self.value {
UntaggedValue::Primitive(primitive) => Ok(primitive.clone()),
_ => Err(ShellError::type_error(
"Primitive",
self.spanned_type_name(),
)),
}
}
pub fn as_u64(&self) -> Result<u64, ShellError> {
match &self.value {
UntaggedValue::Primitive(primitive) => primitive.as_u64(self.tag.span),
_ => Err(ShellError::type_error("integer", self.spanned_type_name())),
}
}
}
impl Into<UntaggedValue> for &str {
fn into(self) -> UntaggedValue {
UntaggedValue::Primitive(Primitive::String(self.to_string()))
}
}
impl Into<UntaggedValue> for Value {
fn into(self) -> UntaggedValue {
self.value
}
}
impl<'a> Into<&'a UntaggedValue> for &'a Value {
fn into(self) -> &'a UntaggedValue {
&self.value
}
}
impl HasSpan for Value {
fn span(&self) -> Span {
self.tag.span
}
}
impl ShellTypeName for Value {
fn type_name(&self) -> &'static str {
ShellTypeName::type_name(&self.value)
}
}
impl ShellTypeName for UntaggedValue {
fn type_name(&self) -> &'static str {
match &self {
UntaggedValue::Primitive(p) => p.type_name(),
UntaggedValue::Row(_) => "row",
UntaggedValue::Table(_) => "table",
UntaggedValue::Error(_) => "error",
UntaggedValue::Block(_) => "block",
}
}
}
impl From<Primitive> for UntaggedValue {
fn from(input: Primitive) -> UntaggedValue {
UntaggedValue::Primitive(input)
}
}
impl From<String> for UntaggedValue {
fn from(input: String) -> UntaggedValue {
UntaggedValue::Primitive(Primitive::String(input))
}
}