From 6989307fb709fdf795baabc27171853f2a603441 Mon Sep 17 00:00:00 2001
From: Andreas Rossberg <rossberg@chromium.org>
Date: Thu, 20 Jul 2017 21:56:06 +0200
Subject: [PATCH] [spec/test/interpreter] Allow underscores in numbers

---
 document/core/text/values.rst |  11 +-
 interpreter/exec/int.ml       |   2 +
 interpreter/text/lexer.mll    |  14 +-
 test/core/float_literals.wast | 350 ++++++++++++++++++++++++++++++++++
 test/core/int_literals.wast   | 102 ++++++++++
 5 files changed, 469 insertions(+), 10 deletions(-)

diff --git a/document/core/text/values.rst b/document/core/text/values.rst
index 28e0c371a1..67c22d0f8a 100644
--- a/document/core/text/values.rst
+++ b/document/core/text/values.rst
@@ -27,6 +27,7 @@ Integers
 ~~~~~~~~
 
 All :ref:`integers <syntax-int>` can be written in either decimal or hexadecimal notation.
+In both cases, digits can optionally be separated by underscores.
 
 .. math::
    \begin{array}{llclll@{\qquad}l}
@@ -43,10 +44,10 @@ All :ref:`integers <syntax-int>` can be written in either decimal or hexadecimal
    \\[1ex]
    \production{decimal number} & \Tnum &::=&
      d{:}\Tdigit &\Rightarrow& d \\ &&|&
-     n{:}\Tnum~~d{:}\Tdigit &\Rightarrow& 10\cdot n + d \\
+     n{:}\Tnum~~\text{\_}^?~~d{:}\Tdigit &\Rightarrow& 10\cdot n + d \\
    \production{hexadecimal number} & \Thexnum &::=&
      h{:}\Thexdigit &\Rightarrow& h \\ &&|&
-     n{:}\Thexnum~~h{:}\Thexdigit &\Rightarrow& 16\cdot n + h \\
+     n{:}\Thexnum~~\text{\_}^?~~h{:}\Thexdigit &\Rightarrow& 16\cdot n + h \\
    \end{array}
 
 The allowed syntax for integer literals depends on size and signedness.
@@ -88,10 +89,12 @@ Floating-Point
    \begin{array}{llclll@{\qquad\qquad}l}
    \production{decimal floating-point fraction} & \Tfrac &::=&
      \epsilon &\Rightarrow& 0 \\ &&|&
-     d{:}\Tdigit~q{:}\Tfrac &\Rightarrow& (d+q)/10 \\
+     d{:}\Tdigit~~q{:}\Tfrac &\Rightarrow& (d+q)/10 \\ &&|&
+     d{:}\Tdigit~~\text{\_}~~p{:}\Tdigit~~q{:}\Tfrac &\Rightarrow& (d+(p+q)/10)/10 \\
    \production{hexadecimal floating-point fraction} & \Thexfrac &::=&
      \epsilon &\Rightarrow& 0 \\ &&|&
-     h{:}\Thexdigit~q{:}\Thexfrac &\Rightarrow& (h+q)/16 \\
+     h{:}\Thexdigit~~q{:}\Thexfrac &\Rightarrow& (h+q)/16 \\ &&|&
+     h{:}\Thexdigit~~\text{\_}~~~~p{:}\Thexdigit~~q{:}\Thexfrac &\Rightarrow& (h+(p+q)/16)/16 \\
    \production{decimal floating-point number} & \Tfloat &::=&
      p{:}\Tnum~\text{.}~q{:}\Tfrac
        &\Rightarrow& p+q \\ &&|&
diff --git a/interpreter/exec/int.ml b/interpreter/exec/int.ml
index a2574fe3ae..d81e6af4a1 100644
--- a/interpreter/exec/int.ml
+++ b/interpreter/exec/int.ml
@@ -245,12 +245,14 @@ struct
     let len = String.length s in
     let rec parse_hex i num =
       if i = len then num else
+      if s.[i] = '_' then parse_hex (i + 1) num else
       let digit = of_int (hex_digit s.[i]) in
       require (le_u num (shr_u minus_one (of_int 4)));
       parse_hex (i + 1) (logor (shift_left num 4) digit)
     in
     let rec parse_dec i num =
       if i = len then num else
+      if s.[i] = '_' then parse_dec (i + 1) num else
       let digit = of_int (dec_digit s.[i]) in
       require (lt_u num max_upper || num = max_upper && le_u digit max_lower);
       parse_dec (i + 1) (add (mul num ten) digit)
diff --git a/interpreter/text/lexer.mll b/interpreter/text/lexer.mll
index 3bb532d51c..2a41d8cfe7 100644
--- a/interpreter/text/lexer.mll
+++ b/interpreter/text/lexer.mll
@@ -91,8 +91,8 @@ let opt = Lib.Option.get
 let sign = '+' | '-'
 let digit = ['0'-'9']
 let hexdigit = ['0'-'9''a'-'f''A'-'F']
-let num = digit+
-let hexnum = hexdigit+
+let num = digit ('_'? digit)*
+let hexnum = hexdigit ('_'? hexdigit)*
 
 let letter = ['a'-'z''A'-'Z']
 let symbol =
@@ -123,11 +123,13 @@ let character =
 
 let nat = num | "0x" hexnum
 let int = sign nat
+let frac = num
+let hexfrac = hexnum
 let float =
-    sign? num '.' digit*
-  | sign? num ('.' digit*)? ('e' | 'E') sign? num
-  | sign? "0x" hexnum '.' hexdigit*
-  | sign? "0x" hexnum ('.' hexdigit*)? ('p' | 'P') sign? num
+    sign? num '.' frac?
+  | sign? num ('.' frac?)? ('e' | 'E') sign? num
+  | sign? "0x" hexnum '.' hexfrac?
+  | sign? "0x" hexnum ('.' hexfrac?)? ('p' | 'P') sign? num
   | sign? "inf"
   | sign? "nan"
   | sign? "nan:" "0x" hexnum
diff --git a/test/core/float_literals.wast b/test/core/float_literals.wast
index 1ab14f1a01..c85bfb275e 100644
--- a/test/core/float_literals.wast
+++ b/test/core/float_literals.wast
@@ -72,6 +72,28 @@
   (func (export "f64_dec.max_subnormal") (result i64) (i64.reinterpret/f64 (f64.const 2.2250738585072011e-308)))
   (func (export "f64_dec.max_finite") (result i64) (i64.reinterpret/f64 (f64.const 1.7976931348623157e+308)))
   (func (export "f64_dec.trailing_dot") (result i64) (i64.reinterpret/f64 (f64.const 1.e100)))
+
+  (func (export "f32-dec-sep1") (result f32) (f32.const 1_000_000))
+  (func (export "f32-dec-sep2") (result f32) (f32.const 1_0_0_0))
+  (func (export "f32-dec-sep3") (result f32) (f32.const 100_3.141_592))
+  (func (export "f32-dec-sep4") (result f32) (f32.const 99e+1_3))
+  (func (export "f32-dec-sep5") (result f32) (f32.const 122_000.11_3_54E0_2_3))
+  (func (export "f32-hex-sep1") (result f32) (f32.const 0xa_0f_00_99))
+  (func (export "f32-hex-sep2") (result f32) (f32.const 0x1_a_A_0_f))
+  (func (export "f32-hex-sep3") (result f32) (f32.const 0xa0_ff.f141_a59a))
+  (func (export "f32-hex-sep4") (result f32) (f32.const 0xf0P+1_3))
+  (func (export "f32-hex-sep5") (result f32) (f32.const 0x2a_f00a.1f_3_eep2_3))
+
+  (func (export "f64-dec-sep1") (result f64) (f64.const 1_000_000))
+  (func (export "f64-dec-sep2") (result f64) (f64.const 1_0_0_0))
+  (func (export "f64-dec-sep3") (result f64) (f64.const 100_3.141_592))
+  (func (export "f64-dec-sep4") (result f64) (f64.const 99e-1_23))
+  (func (export "f64-dec-sep5") (result f64) (f64.const 122_000.11_3_54e0_2_3))
+  (func (export "f64-hex-sep1") (result f64) (f64.const 0xa_f00f_0000_9999))
+  (func (export "f64-hex-sep2") (result f64) (f64.const 0x1_a_A_0_f))
+  (func (export "f64-hex-sep3") (result f64) (f64.const 0xa0_ff.f141_a59a))
+  (func (export "f64-hex-sep4") (result f64) (f64.const 0xf0P+1_3))
+  (func (export "f64-hex-sep5") (result f64) (f64.const 0x2a_f00a.1f_3_eep2_3))
 )
 
 (assert_return (invoke "f32.nan") (i32.const 0x7fc00000))
@@ -135,3 +157,331 @@
 (assert_return (invoke "f64_dec.max_subnormal") (i64.const 0xfffffffffffff))
 (assert_return (invoke "f64_dec.max_finite") (i64.const 0x7fefffffffffffff))
 (assert_return (invoke "f64_dec.trailing_dot") (i64.const 0x54b249ad2594c37d))
+
+(assert_return (invoke "f32-dec-sep1") (f32.const 1000000))
+(assert_return (invoke "f32-dec-sep2") (f32.const 1000))
+(assert_return (invoke "f32-dec-sep3") (f32.const 1003.141592))
+(assert_return (invoke "f32-dec-sep4") (f32.const 99e+13))
+(assert_return (invoke "f32-dec-sep5") (f32.const 122000.11354e23))
+(assert_return (invoke "f32-hex-sep1") (f32.const 0xa0f0099))
+(assert_return (invoke "f32-hex-sep2") (f32.const 0x1aa0f))
+(assert_return (invoke "f32-hex-sep3") (f32.const 0xa0ff.f141a59a))
+(assert_return (invoke "f32-hex-sep4") (f32.const 0xf0P+13))
+(assert_return (invoke "f32-hex-sep5") (f32.const 0x2af00a.1f3eep23))
+
+(assert_return (invoke "f64-dec-sep1") (f64.const 1000000))
+(assert_return (invoke "f64-dec-sep2") (f64.const 1000))
+(assert_return (invoke "f64-dec-sep3") (f64.const 1003.141592))
+(assert_return (invoke "f64-dec-sep4") (f64.const 99e-123))
+(assert_return (invoke "f64-dec-sep5") (f64.const 122000.11354e23))
+(assert_return (invoke "f64-hex-sep1") (f64.const 0xaf00f00009999))
+(assert_return (invoke "f64-hex-sep2") (f64.const 0x1aa0f))
+(assert_return (invoke "f64-hex-sep3") (f64.const 0xa0ff.f141a59a))
+(assert_return (invoke "f64-hex-sep4") (f64.const 0xf0P+13))
+(assert_return (invoke "f64-hex-sep5") (f64.const 0x2af00a.1f3eep23))
+
+(assert_malformed
+  (module quote "(global f32 (f32.const _100))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const +_100))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const -_100))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 99_))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 1__000))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const _1.0))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 1.0_))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 1_.0))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 1._0))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const _1e1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 1e1_))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 1_e1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 1e_1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const _1.0e1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 1.0e1_))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 1.0_e1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 1.0e_1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 1.0e+_1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 1.0e_+1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const _0x100))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 0_x100))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 0x_100))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 0x00_))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 0xff__ffff))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 0x_1.0))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 0x1.0_))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 0x1_.0))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 0x1._0))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 0x_1p1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 0x1p1_))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 0x1_p1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 0x1p_1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 0x_1.0p1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 0x1.0p1_))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 0x1.0_p1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 0x1.0p_1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 0x1.0p+_1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f32 (f32.const 0x1.0p_+1))")
+  "unknown operator"
+)
+
+(assert_malformed
+  (module quote "(global f64 (f64.const _100))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const +_100))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const -_100))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 99_))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 1__000))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const _1.0))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 1.0_))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 1_.0))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 1._0))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const _1e1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 1e1_))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 1_e1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 1e_1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const _1.0e1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 1.0e1_))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 1.0_e1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 1.0e_1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 1.0e+_1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 1.0e_+1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const _0x100))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 0_x100))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 0x_100))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 0x00_))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 0xff__ffff))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 0x_1.0))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 0x1.0_))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 0x1_.0))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 0x1._0))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 0x_1p1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 0x1p1_))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 0x1_p1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 0x1p_1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 0x_1.0p1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 0x1.0p1_))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 0x1.0_p1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 0x1.0p_1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 0x1.0p+_1))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global f64 (f64.const 0x1.0p_+1))")
+  "unknown operator"
+)
diff --git a/test/core/int_literals.wast b/test/core/int_literals.wast
index f0ec041785..73b37cc5a9 100644
--- a/test/core/int_literals.wast
+++ b/test/core/int_literals.wast
@@ -22,6 +22,16 @@
   (func (export "i64.not_octal") (result i64) (return (i64.const 010)))
   (func (export "i64.unsigned_decimal") (result i64) (return (i64.const 18446744073709551615)))
   (func (export "i64.plus_sign") (result i64) (return (i64.const +42)))
+
+  (func (export "i32-dec-sep1") (result i32) (i32.const 1_000_000))
+  (func (export "i32-dec-sep2") (result i32) (i32.const 1_0_0_0))
+  (func (export "i32-hex-sep1") (result i32) (i32.const 0xa_0f_00_99))
+  (func (export "i32-hex-sep2") (result i32) (i32.const 0x1_a_A_0_f))
+
+  (func (export "i64-dec-sep1") (result i64) (i64.const 1_000_000))
+  (func (export "i64-dec-sep2") (result i64) (i64.const 1_0_0_0))
+  (func (export "i64-hex-sep1") (result i64) (i64.const 0xa_f00f_0000_9999))
+  (func (export "i64-hex-sep2") (result i64) (i64.const 0x1_a_A_0_f))
 )
 
 (assert_return (invoke "i32.test") (i32.const 195940365))
@@ -47,3 +57,95 @@
 (assert_return (invoke "i64.not_octal") (i64.const 10))
 (assert_return (invoke "i64.unsigned_decimal") (i64.const -1))
 (assert_return (invoke "i64.plus_sign") (i64.const 42))
+
+(assert_return (invoke "i32-dec-sep1") (i32.const 1000000))
+(assert_return (invoke "i32-dec-sep2") (i32.const 1000))
+(assert_return (invoke "i32-hex-sep1") (i32.const 0xa0f0099))
+(assert_return (invoke "i32-hex-sep2") (i32.const 0x1aa0f))
+
+(assert_return (invoke "i64-dec-sep1") (i64.const 1000000))
+(assert_return (invoke "i64-dec-sep2") (i64.const 1000))
+(assert_return (invoke "i64-hex-sep1") (i64.const 0xaf00f00009999))
+(assert_return (invoke "i64-hex-sep2") (i64.const 0x1aa0f))
+
+(assert_malformed
+  (module quote "(global i32 (i32.const _100))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global i32 (i32.const +_100))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global i32 (i32.const -_100))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global i32 (i32.const 99_))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global i32 (i32.const 1__000))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global i32 (i32.const _0x100))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global i32 (i32.const 0_x100))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global i32 (i32.const 0x_100))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global i32 (i32.const 0x00_))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global i32 (i32.const 0xff__ffff))")
+  "unknown operator"
+)
+
+(assert_malformed
+  (module quote "(global i64 (i64.const _100))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global i64 (i64.const +_100))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global i64 (i64.const -_100))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global i64 (i64.const 99_))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global i64 (i64.const 1__000))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global i64 (i64.const _0x100))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global i64 (i64.const 0_x100))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global i64 (i64.const 0x_100))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global i64 (i64.const 0x00_))")
+  "unknown operator"
+)
+(assert_malformed
+  (module quote "(global i64 (i64.const 0xff__ffff))")
+  "unknown operator"
+)