feat: add raw byte access methods and \x escape to String class
Add `byteAt(index)`, `codePointAt(index)`, and `bytes` sequence to String, enabling direct byte-level manipulation of UTF-8 encoded strings. Introduce `\x` hex escape sequence in string literals for specifying raw byte values. Refactor Unicode escape parsing into a generic `readHexEscape` function supporting variable digit counts. Implement `wrenUtf8Decode` utility for decoding UTF-8 sequences from byte buffers. Add comprehensive tests for `byteAt` including boundary conditions and error cases.
This commit is contained in:
@@ -0,0 +1,38 @@
|
||||
// Bytes: 11111
|
||||
// 012345678901234
|
||||
// Chars: sø mé ஃ thî ng
|
||||
var s = "søméஃthîng"
|
||||
|
||||
IO.print(s.byteAt(0)) // expect: 115
|
||||
IO.print(s.byteAt(1)) // expect: 195
|
||||
IO.print(s.byteAt(2)) // expect: 184
|
||||
IO.print(s.byteAt(3)) // expect: 109
|
||||
IO.print(s.byteAt(4)) // expect: 195
|
||||
IO.print(s.byteAt(5)) // expect: 169
|
||||
IO.print(s.byteAt(6)) // expect: 224
|
||||
IO.print(s.byteAt(7)) // expect: 174
|
||||
IO.print(s.byteAt(8)) // expect: 131
|
||||
IO.print(s.byteAt(9)) // expect: 116
|
||||
IO.print(s.byteAt(10)) // expect: 104
|
||||
IO.print(s.byteAt(11)) // expect: 195
|
||||
IO.print(s.byteAt(12)) // expect: 174
|
||||
IO.print(s.byteAt(13)) // expect: 110
|
||||
IO.print(s.byteAt(14)) // expect: 103
|
||||
|
||||
IO.print(s.byteAt(-15)) // expect: 115
|
||||
IO.print(s.byteAt(-14)) // expect: 195
|
||||
IO.print(s.byteAt(-13)) // expect: 184
|
||||
IO.print(s.byteAt(-12)) // expect: 109
|
||||
IO.print(s.byteAt(-11)) // expect: 195
|
||||
IO.print(s.byteAt(-10)) // expect: 169
|
||||
IO.print(s.byteAt(-9)) // expect: 224
|
||||
IO.print(s.byteAt(-8)) // expect: 174
|
||||
IO.print(s.byteAt(-7)) // expect: 131
|
||||
IO.print(s.byteAt(-6)) // expect: 116
|
||||
IO.print(s.byteAt(-5)) // expect: 104
|
||||
IO.print(s.byteAt(-4)) // expect: 195
|
||||
IO.print(s.byteAt(-3)) // expect: 174
|
||||
IO.print(s.byteAt(-2)) // expect: 110
|
||||
IO.print(s.byteAt(-1)) // expect: 103
|
||||
|
||||
IO.print("\0".byteAt(0)) // expect: 0
|
||||
@@ -0,0 +1 @@
|
||||
IO.print("string".byteAt(12.34)) // expect runtime error: Index must be an integer.
|
||||
@@ -0,0 +1 @@
|
||||
IO.print("string".byteAt("not num")) // expect runtime error: Index must be a number.
|
||||
@@ -0,0 +1 @@
|
||||
IO.print("string".byteAt(6)) // expect runtime error: Index out of bounds.
|
||||
@@ -0,0 +1 @@
|
||||
IO.print("string".byteAt(-7)) // expect runtime error: Index out of bounds.
|
||||
@@ -0,0 +1,6 @@
|
||||
// Bytes: 11111
|
||||
// 012345678901234
|
||||
// Chars: sø mé ஃ thî ng
|
||||
var s = "søméஃthîng"
|
||||
|
||||
IO.print(s.bytes is StringByteSequence) // expect: true
|
||||
@@ -0,0 +1,38 @@
|
||||
// Bytes: 11111
|
||||
// 012345678901234
|
||||
// Chars: sø mé ஃ thî ng
|
||||
var s = "søméஃthîng"
|
||||
|
||||
IO.print(s.codePointAt(0)) // expect: 115
|
||||
IO.print(s.codePointAt(1)) // expect: 248
|
||||
IO.print(s.codePointAt(2)) // expect: -1
|
||||
IO.print(s.codePointAt(3)) // expect: 109
|
||||
IO.print(s.codePointAt(4)) // expect: 233
|
||||
IO.print(s.codePointAt(5)) // expect: -1
|
||||
IO.print(s.codePointAt(6)) // expect: 2947
|
||||
IO.print(s.codePointAt(7)) // expect: -1
|
||||
IO.print(s.codePointAt(8)) // expect: -1
|
||||
IO.print(s.codePointAt(9)) // expect: 116
|
||||
IO.print(s.codePointAt(10)) // expect: 104
|
||||
IO.print(s.codePointAt(11)) // expect: 238
|
||||
IO.print(s.codePointAt(12)) // expect: -1
|
||||
IO.print(s.codePointAt(13)) // expect: 110
|
||||
IO.print(s.codePointAt(14)) // expect: 103
|
||||
|
||||
IO.print(s.codePointAt(-15)) // expect: 115
|
||||
IO.print(s.codePointAt(-14)) // expect: 248
|
||||
IO.print(s.codePointAt(-13)) // expect: -1
|
||||
IO.print(s.codePointAt(-12)) // expect: 109
|
||||
IO.print(s.codePointAt(-11)) // expect: 233
|
||||
IO.print(s.codePointAt(-10)) // expect: -1
|
||||
IO.print(s.codePointAt(-9)) // expect: 2947
|
||||
IO.print(s.codePointAt(-8)) // expect: -1
|
||||
IO.print(s.codePointAt(-7)) // expect: -1
|
||||
IO.print(s.codePointAt(-6)) // expect: 116
|
||||
IO.print(s.codePointAt(-5)) // expect: 104
|
||||
IO.print(s.codePointAt(-4)) // expect: 238
|
||||
IO.print(s.codePointAt(-3)) // expect: -1
|
||||
IO.print(s.codePointAt(-2)) // expect: 110
|
||||
IO.print(s.codePointAt(-1)) // expect: 103
|
||||
|
||||
IO.print("\0".codePointAt(0)) // expect: 0
|
||||
@@ -0,0 +1,3 @@
|
||||
// The first two bytes of a three-octet sequence.
|
||||
var s = "\xe0\xae"
|
||||
IO.print(s.codePointAt(0)) // expect: -1
|
||||
@@ -0,0 +1 @@
|
||||
IO.print("string".codePointAt(12.34)) // expect runtime error: Index must be an integer.
|
||||
@@ -0,0 +1 @@
|
||||
IO.print("string".codePointAt("not num")) // expect runtime error: Index must be a number.
|
||||
@@ -0,0 +1 @@
|
||||
IO.print("string".codePointAt(6)) // expect runtime error: Index out of bounds.
|
||||
@@ -0,0 +1 @@
|
||||
IO.print("string".codePointAt(-7)) // expect runtime error: Index out of bounds.
|
||||
@@ -0,0 +1,21 @@
|
||||
// Bytes:
|
||||
// 012345678
|
||||
// Chars: sø mé ஃ
|
||||
var bytes = "søméஃ".bytes
|
||||
|
||||
IO.print(bytes.iterate(null)) // expect: 0
|
||||
IO.print("".bytes.iterate(null)) // expect: false
|
||||
|
||||
IO.print(bytes.iterate(0)) // expect: 1
|
||||
IO.print(bytes.iterate(1)) // expect: 2
|
||||
IO.print(bytes.iterate(2)) // expect: 3
|
||||
IO.print(bytes.iterate(3)) // expect: 4
|
||||
IO.print(bytes.iterate(4)) // expect: 5
|
||||
IO.print(bytes.iterate(5)) // expect: 6
|
||||
IO.print(bytes.iterate(6)) // expect: 7
|
||||
IO.print(bytes.iterate(7)) // expect: 8
|
||||
IO.print(bytes.iterate(8)) // expect: false
|
||||
|
||||
// Out of bounds.
|
||||
IO.print(bytes.iterate(123)) // expect: false
|
||||
IO.print(bytes.iterate(-1)) // expect: false
|
||||
@@ -0,0 +1 @@
|
||||
"str".bytes.iterate(12.34) // expect runtime error: Iterator must be an integer.
|
||||
@@ -0,0 +1 @@
|
||||
"str".bytes.iterate("not num") // expect runtime error: Iterator must be a number.
|
||||
@@ -0,0 +1,24 @@
|
||||
// Bytes:
|
||||
// 012345678
|
||||
// Chars: sø mé ஃ
|
||||
var bytes = "søméஃ".bytes
|
||||
|
||||
IO.print(bytes.iteratorValue(0)) // expect: 115
|
||||
IO.print(bytes.iteratorValue(1)) // expect: 195
|
||||
IO.print(bytes.iteratorValue(2)) // expect: 184
|
||||
IO.print(bytes.iteratorValue(3)) // expect: 109
|
||||
IO.print(bytes.iteratorValue(4)) // expect: 195
|
||||
IO.print(bytes.iteratorValue(5)) // expect: 169
|
||||
IO.print(bytes.iteratorValue(6)) // expect: 224
|
||||
IO.print(bytes.iteratorValue(7)) // expect: 174
|
||||
IO.print(bytes.iteratorValue(8)) // expect: 131
|
||||
|
||||
IO.print(bytes.iteratorValue(-9)) // expect: 115
|
||||
IO.print(bytes.iteratorValue(-8)) // expect: 195
|
||||
IO.print(bytes.iteratorValue(-7)) // expect: 184
|
||||
IO.print(bytes.iteratorValue(-6)) // expect: 109
|
||||
IO.print(bytes.iteratorValue(-5)) // expect: 195
|
||||
IO.print(bytes.iteratorValue(-4)) // expect: 169
|
||||
IO.print(bytes.iteratorValue(-3)) // expect: 224
|
||||
IO.print(bytes.iteratorValue(-2)) // expect: 174
|
||||
IO.print(bytes.iteratorValue(-1)) // expect: 131
|
||||
@@ -0,0 +1 @@
|
||||
"abcd".bytes.iteratorValue(12.34) // expect runtime error: Index must be an integer.
|
||||
@@ -0,0 +1 @@
|
||||
"abcd".bytes.iteratorValue("not num") // expect runtime error: Index must be a number.
|
||||
@@ -0,0 +1 @@
|
||||
"abcd".bytes.iteratorValue(4) // expect runtime error: Index out of bounds.
|
||||
@@ -0,0 +1 @@
|
||||
"abcd".bytes.iteratorValue(-5) // expect runtime error: Index out of bounds.
|
||||
@@ -0,0 +1,24 @@
|
||||
// Bytes:
|
||||
// 012345678
|
||||
// Chars: sø mé ஃ
|
||||
var bytes = "søméஃ".bytes
|
||||
|
||||
IO.print(bytes[0]) // expect: 115
|
||||
IO.print(bytes[1]) // expect: 195
|
||||
IO.print(bytes[2]) // expect: 184
|
||||
IO.print(bytes[3]) // expect: 109
|
||||
IO.print(bytes[4]) // expect: 195
|
||||
IO.print(bytes[5]) // expect: 169
|
||||
IO.print(bytes[6]) // expect: 224
|
||||
IO.print(bytes[7]) // expect: 174
|
||||
IO.print(bytes[8]) // expect: 131
|
||||
|
||||
IO.print(bytes[-9]) // expect: 115
|
||||
IO.print(bytes[-8]) // expect: 195
|
||||
IO.print(bytes[-7]) // expect: 184
|
||||
IO.print(bytes[-6]) // expect: 109
|
||||
IO.print(bytes[-5]) // expect: 195
|
||||
IO.print(bytes[-4]) // expect: 169
|
||||
IO.print(bytes[-3]) // expect: 224
|
||||
IO.print(bytes[-2]) // expect: 174
|
||||
IO.print(bytes[-1]) // expect: 131
|
||||
@@ -0,0 +1 @@
|
||||
"abcd".bytes[12.34] // expect runtime error: Index must be an integer.
|
||||
@@ -0,0 +1 @@
|
||||
"abcd".bytes["not num"] // expect runtime error: Index must be a number.
|
||||
@@ -0,0 +1 @@
|
||||
"abcd".bytes[4] // expect runtime error: Index out of bounds.
|
||||
@@ -0,0 +1 @@
|
||||
"abcd".bytes[-5] // expect runtime error: Index out of bounds.
|
||||
Reference in New Issue
Block a user