barefoot_js 0.39.8 → 0.40.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 727248cd6dcc7a533846854d5a8d4675ac946390d5750df49b03e25dad8cd3e5
-  data.tar.gz: 508a38a3a2d56d99af75875be59b6989103ab3adf3fe6e42bae3e040cd69d608
+  metadata.gz: 5683e188c57d0ed7b6101db57828d8bfb0a6f8ffdbb6939052e5132fbd099fd9
+  data.tar.gz: 546f0e92374f3947eae8965b2771a3ba6928526eddaa8b9f1e3ced6883deed1f
 SHA512:
-  metadata.gz: e91debd9fe87d53e154cd53a7be218efd5c3dec07e64f3b57cb3c4cd0d2406f1a5846f84bde2d9cb49f73a5236c6979649e89f50879f2511e31dda9191ed060b
-  data.tar.gz: 5b617131eebba68e976d362993fba1f85e8cadd86fff1258ab029d44a4bdbdf64013aa3b8d739229deff1497c411af7205e016b05ff1bfbf877cf85b2ab27e22
+  metadata.gz: 99c0e7ceaf4e6e92c1a0b9c462a43983b91de28555e6032c11e0b904156088e7ba08177043f3a3658140a18de3d1cae365746c01193d87b5c83e5db6cbf21a1a
+  data.tar.gz: dc2aa7213b1e63646a13f0c9340377731ab62baf7731ffcbed0bb517a78ff79d866f08a3a7c985990f666f01532a79237be42c346791d7bff213fa1e83f3a5c6

data/lib/barefoot_js.rb

@@ -1034,18 +1034,28 @@ module BarefootJS
     end
 
     # `Array.prototype.indexOf(x)` / `.lastIndexOf(x)` -- value-equality
-    # search. Non-array receivers return -1.
+    # search. A String receiver is `String.prototype.indexOf`, counted in
+    # UTF-16 code units like JS (#3613). Other receivers return -1.
     def index_of(recv, elem)
+      return utf16_index_of(recv, elem, false) if recv.is_a?(String)
+
       array_index_of(recv, elem, false)
     end
 
     def last_index_of(recv, elem)
+      return utf16_index_of(recv, elem, true) if recv.is_a?(String)
+
       array_index_of(recv, elem, true)
     end
 
     # `Array.prototype.at(i)` -- negative indices count from the end;
     # out-of-bounds -> nil (renders as '' via `h`, matching JS `undefined`).
+    # A String receiver is indexed in UTF-16 code units (#3613). Half of a
+    # surrogate pair, a lone surrogate in JS, comes back as U+FFFD: a
+    # UTF-8 Ruby String can't hold one, so the halves compare equal and
+    # don't rejoin (limitation `string-at-lone-surrogate`).
     def at(recv, i)
+      return utf16_at(recv, i) if recv.is_a?(String)
       return nil unless recv.is_a?(Array)
       return nil if i.nil?
 
@@ -1237,9 +1247,10 @@ module BarefootJS
     def starts_with(recv, prefix, position = nil)
       s = recv.nil? ? '' : string(recv)
       p = prefix.nil? ? '' : string(prefix)
+      # `position` counts UTF-16 code units, like JS (#3613).
       unless position.nil?
-        n = clamp_index(position.to_i, s.length)
-        s = s[n..] || ''
+        units = utf16_units(s)
+        return utf16_has_at?(units, utf16_units(p), clamp_index(to_integer_or_infinity(position), units.length))
       end
       s.start_with?(p)
     end
@@ -1248,9 +1259,11 @@ module BarefootJS
     def ends_with(recv, suffix, end_position = nil)
       s = recv.nil? ? '' : string(recv)
       x = suffix.nil? ? '' : string(suffix)
+      # `endPosition` counts UTF-16 code units, like JS (#3613).
       unless end_position.nil?
-        e = clamp_index(end_position.to_i, s.length)
-        s = s[0...e]
+        units = utf16_units(s)
+        needle = utf16_units(x)
+        return utf16_has_at?(units, needle, clamp_index(to_integer_or_infinity(end_position), units.length) - needle.length)
       end
       s.end_with?(x)
     end
@@ -1706,6 +1719,55 @@ module BarefootJS
       -1
     end
 
+    # `s` as JS sees it: an Array of UTF-16 code units. JS string positions
+    # (`indexOf`, `at`, `startsWith`'s position, ...) count these, so an
+    # astral character such as '😀' is two units (#3613).
+    def utf16_units(s)
+      s.to_s.encode('UTF-16LE', invalid: :replace, undef: :replace).unpack('v*')
+    end
+
+    MAX_SAFE_INDEX = 2**53
+
+    # JS `ToIntegerOrInfinity` for a string position or index (#3613): NaN
+    # -> 0, truncation toward zero, and +/-Infinity clamped to +/-2**53 so
+    # the caller's range check treats it as out of range instead of
+    # `Float#to_i` raising FloatDomainError.
+    def to_integer_or_infinity(value)
+      n = number(value)
+      return 0 if n.respond_to?(:nan?) && n.nan?
+      return MAX_SAFE_INDEX if n >= MAX_SAFE_INDEX
+      return -MAX_SAFE_INDEX if n <= -MAX_SAFE_INDEX
+
+      n.to_i
+    end
+
+    def utf16_has_at?(hay, needle, at)
+      return false if at.negative? || at + needle.length > hay.length
+
+      hay[at, needle.length] == needle
+    end
+
+    def utf16_index_of(recv, elem, reverse)
+      hay = utf16_units(recv)
+      needle = utf16_units(string(elem))
+      last = hay.length - needle.length
+      indices = reverse ? last.downto(0) : 0.upto(last)
+      indices.each { |i| return i if utf16_has_at?(hay, needle, i) }
+      -1
+    end
+
+    def utf16_at(recv, i)
+      return nil if i.nil?
+
+      units = utf16_units(recv)
+      idx = to_integer_or_infinity(i)
+      idx += units.length if idx.negative?
+      return nil if idx.negative? || idx >= units.length
+
+      u = units[idx]
+      (0xD800..0xDFFF).cover?(u) ? "\uFFFD" : [u].pack('U')
+    end
+
     def clamp_index(n, len)
       n = 0 if n.negative?
       n = len if n > len

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: barefoot_js
 version: !ruby/object:Gem::Version
-  version: 0.39.8
+  version: 0.40.0
 platform: ruby
 authors:
 - kobaken