Add support for E PluralOperand (#407)

* Add support for E PluralOperand * Add the FromStr * Fix ecma402 trait * Apply reviewers feedback * Add roundtrip parse/serialize test * Fix ECMA402 test for plurals * Move `e` to not be an operand * Move to serialize C as `c` and switch tests to use it. * Re-add AST node for E
unicode-org · Jan 9, 2021 · 2a9adbc · 2a9adbc
1 parent d0ee68b
commit 2a9adbc
Show file tree

Hide file tree

Showing 12 changed files with 123 additions and 39 deletions.
diff --git a/components/ecma402/src/pluralrules.rs b/components/ecma402/src/pluralrules.rs
@@ -214,6 +214,7 @@ pub(crate) mod internal {
                     w: 1,
                     f: 50,
                     t: 5,
+                    c: 0,
                 },
             }];
             for test in tests {

diff --git a/components/plurals/src/operands.rs b/components/plurals/src/operands.rs
@@ -25,6 +25,7 @@ use std::str::FromStr;
 ///    w: 0,
 ///    f: 0,
 ///    t: 0,
+///    c: 0,
 /// }, PluralOperands::from(2_usize))
 /// ```
 ///
@@ -39,6 +40,7 @@ use std::str::FromStr;
 ///    w: 3,
 ///    f: 567,
 ///    t: 567,
+///    c: 0,
 /// }), "-1234.567".parse())
 /// ```
 ///
@@ -53,6 +55,7 @@ use std::str::FromStr;
 ///    w: 2,
 ///    f: 45,
 ///    t: 45,
+///    c: 0,
 /// }), "123.45".parse())
 /// ```
 #[derive(Debug, Clone, Copy, PartialEq)]
@@ -67,6 +70,8 @@ pub struct PluralOperands {
     pub f: u64,
     /// Visible fraction digits without trailing zeros
     pub t: u64,
+    /// Exponent of the power of 10 used in compact decimal formatting
+    pub c: usize,
 }
 
 impl PluralOperands {
@@ -99,6 +104,15 @@ impl From<IOError> for OperandsError {
     }
 }
 
+fn get_exponent(input: &str) -> Result<(&str, usize), OperandsError> {
+    if let Some(e_idx) = input.find('e') {
+        let e = usize::from_str(&input[e_idx + 1..])?;
+        Ok((&input[..e_idx], e))
+    } else {
+        Ok((input, 0))
+    }
+}
+
 impl FromStr for PluralOperands {
     type Err = OperandsError;
 
@@ -115,9 +129,10 @@ impl FromStr for PluralOperands {
             num_fraction_digits,
             fraction_digits0,
             fraction_digits,
+            exponent,
         ) = if let Some(sep_idx) = abs_str.find('.') {
             let int_str = &abs_str[..sep_idx];
-            let dec_str = &abs_str[(sep_idx + 1)..];
+            let (dec_str, exponent) = get_exponent(&abs_str[(sep_idx + 1)..])?;
 
             let integer_digits = u64::from_str(int_str)?;
 
@@ -140,10 +155,12 @@ impl FromStr for PluralOperands {
                 num_fraction_digits,
                 fraction_digits0,
                 fraction_digits,
+                exponent,
             )
         } else {
+            let (abs_str, exponent) = get_exponent(abs_str)?;
             let integer_digits = u64::from_str(abs_str)?;
-            (integer_digits, 0, 0, 0, 0)
+            (integer_digits, 0, 0, 0, 0, exponent)
         };
 
         Ok(Self {
@@ -152,6 +169,7 @@ impl FromStr for PluralOperands {
             w: num_fraction_digits,
             f: fraction_digits0,
             t: fraction_digits,
+            c: exponent,
         })
     }
 }
@@ -166,6 +184,7 @@ macro_rules! impl_integer_type {
                     w: 0,
                     f: 0,
                     t: 0,
+                    c: 0,
                 }
             }
         }
@@ -187,6 +206,7 @@ macro_rules! impl_signed_integer_type {
                     w: 0,
                     f: 0,
                     t: 0,
+                    c: 0,
                 })
             }
         }
@@ -232,6 +252,7 @@ impl From<&FixedDecimal> for PluralOperands {
             w,
             f,
             t,
+            c: 0,
         }
     }
 }
diff --git a/components/plurals/src/rules/ast.rs b/components/plurals/src/rules/ast.rs
@@ -264,6 +264,11 @@ pub enum Operand {
     F,
     /// Visible fraction digits without trailing zeros
     T,
+    /// Compact decimal exponent value:
+    ///   exponent of the power of 10 used in compact decimal formatting
+    C,
+    /// Currently, synonym for ‘c’. however, may be redefined in the future
+    E,
 }
 
 /// An incomplete AST representation of a plural rule. Comprises a vector of `RangeListItems`.

diff --git a/components/plurals/src/rules/lexer.rs b/components/plurals/src/rules/lexer.rs
@@ -21,6 +21,7 @@ pub enum Token {
     Integer,
     Ellipsis,
     Tilde,
+    E,
 }
 
 #[derive(Debug)]
@@ -165,6 +166,8 @@ impl<'l> Lexer<'l> {
                         Token::Ellipsis
                     }
                     b'~' => Token::Tilde,
+                    b'e' => Token::E,
+                    b'c' => Token::Operand(ast::Operand::C),
                     b => return Err(LexerError::UnknownToken(*b)),
                 };
                 return Ok(Some(token));

diff --git a/components/plurals/src/rules/mod.rs b/components/plurals/src/rules/mod.rs
@@ -35,7 +35,8 @@
 //!     v: 0,
 //!     w: 0,
 //!     f: 0,
-//!     t: 0
+//!     t: 0,
+//!     c: 0,
 //! };
 //! ```
 //!

diff --git a/components/plurals/src/rules/parser.rs b/components/plurals/src/rules/parser.rs
@@ -174,6 +174,7 @@ impl<'p> Parser<'p> {
 
     fn get_expression(&mut self) -> Result<Option<ast::Expression>, ParserError> {
         let operand = match self.lexer.peek() {
+            Some(Token::E) => ast::Operand::E,
             Some(Token::Operand(op)) => *op,
             Some(Token::At) | None => return Ok(None),
             _ => return Err(ParserError::ExpectedOperand),
@@ -302,6 +303,20 @@ impl<'p> Parser<'p> {
                 self.lexer.next();
             }
         }
+
+        if self.take_if(Token::E) {
+            s.push('e');
+            match self.lexer.peek() {
+                Some(Token::Zero) => s.push('0'),
+                Some(Token::Number(v)) => {
+                    s.push_str(&v.to_string());
+                }
+                _ => {
+                    return Err(ParserError::ExpectedValue);
+                }
+            }
+            self.lexer.next();
+        }
         if s.is_empty() {
             Err(ParserError::ExpectedValue)
         } else {

diff --git a/components/plurals/src/rules/resolver.rs b/components/plurals/src/rules/resolver.rs
@@ -63,6 +63,7 @@ fn calculate_expression(expression: &ast::Expression, operands: &PluralOperands)
         ast::Operand::V => operands.v as u64,
         ast::Operand::W => operands.w as u64,
         ast::Operand::T => operands.t,
+        ast::Operand::C | ast::Operand::E => operands.c as u64,
     };
     if let Some(modulus) = &expression.modulus {
         value.checked_rem_euclid(modulus.0)

diff --git a/components/plurals/src/rules/serializer.rs b/components/plurals/src/rules/serializer.rs
@@ -105,6 +105,8 @@ fn serialize_operand(operand: ast::Operand, w: &mut impl fmt::Write) -> fmt::Res
         ast::Operand::W => w.write_char('w'),
         ast::Operand::F => w.write_char('f'),
         ast::Operand::T => w.write_char('t'),
+        ast::Operand::C => w.write_char('c'),
+        ast::Operand::E => w.write_char('e'),
     }
 }
 
@@ -115,7 +117,7 @@ fn serialize_rangelist(rl: &ast::RangeList, w: &mut impl fmt::Write) -> fmt::Res
         if first {
             first = false;
         } else {
-            w.write_str(",")?;
+            w.write_str(", ")?;
         }
         serialize_rangelistitem(rli, w)?
     }
@@ -144,9 +146,6 @@ pub fn serialize_samples(samples: &ast::Samples, w: &mut impl fmt::Write) -> fmt
     if let Some(sample_list) = &samples.integer {
         w.write_str(" @integer ")?;
         serialize_sample_list(sample_list, w)?;
-    } else {
-        // Quirk of the current serializer
-        w.write_str("  ")?;
     }
     if let Some(sample_list) = &samples.decimal {
         w.write_str(" @decimal ")?;

diff --git a/components/plurals/tests/fixtures/mod.rs b/components/plurals/tests/fixtures/mod.rs
@@ -47,14 +47,15 @@ impl From<&FixedDecimalInput> for FixedDecimal {
 #[derive(Debug, Clone, Deserialize)]
 #[serde(untagged)]
 pub enum PluralOperandsInput {
-    List((f64, u64, usize, usize, u64, u64)),
+    List((f64, u64, usize, usize, u64, u64, usize)),
     Struct {
         n: Option<f64>,
         i: Option<u64>,
         v: Option<usize>,
         w: Option<usize>,
         f: Option<u64>,
         t: Option<u64>,
+        c: Option<usize>,
     },
     String(String),
     Number(isize),
@@ -69,13 +70,23 @@ impl From<PluralOperandsInput> for PluralOperands {
                 w: operands.3,
                 f: operands.4,
                 t: operands.5,
+                c: operands.6,
             },
-            PluralOperandsInput::Struct { n, i, v, w, f, t } => Self {
+            PluralOperandsInput::Struct {
+                n,
+                i,
+                v,
+                w,
+                f,
+                t,
+                c,
+            } => Self {
                 i: i.unwrap_or_else(|| n.unwrap_or(0_f64) as u64),
                 v: v.unwrap_or(0),
                 w: w.unwrap_or(0),
                 f: f.unwrap_or(0),
                 t: t.unwrap_or(0),
+                c: c.unwrap_or(0),
             },
             PluralOperandsInput::String(num) => num
                 .parse()