hypercast-wasm 0.6.1

data/lib/hypercast.rb

@@ -0,0 +1,538 @@
+require "date"
+require_relative "hypercast/native_platform"
+require_relative "hypercast/runtime"
+
+# Allocation-lean scalar casts — booleans, numerics, exact decimals, UUIDs, temporals —
+# from one Rust core, reached through whichever of two backends this install can load:
+# the core linked straight into a Magnus native extension (shipped precompiled in the
+# platform gems), or the native libhypercast shared library called through Fiddle (the
+# universal gem bundles one per supported platform). HyperCast::BACKEND names the one that
+# loaded; the selection logic is at the bottom of this file. Every door returns a verdict: Success or
+# Fault (a closed reason plus the offending span), never an exception for bad data — the
+# only exceptions here are caller bugs (a malformed NumFormat, an undeclared option, text
+# that is not a String), never data, and the same exception on every backend.
+#
+# Consume with Ruby's own pattern matching over the two Data case types:
+#
+#   case HyperCast.i32("(1,234)", HyperCast::NumFormat::INVARIANT)
+#   in HyperCast::Success(value:) then puts "got #{value}"          # -1234
+#   in HyperCast::Fault(reason:, offset:) then puts "#{reason} at #{offset}"
+#   end
+#
+# Door names mirror the native ABI (i32, f64, timestamp, ...) so the polyglot surface reads
+# identically across bindings. Ruby-flavored fidelity: Integer is unbounded (u64 comes back
+# as the true unsigned value), Time carries full nanoseconds across the whole 0001-9999
+# window, time-of-day is an exact Integer of nanoseconds since midnight, durations come
+# back as exact Rational seconds, and the decimal door returns an exact Decimal (sign,
+# 96-bit magnitude, base-10 scale) that never rounds — no truncation anywhere, and no
+# wrapping: Ruby and the JVM are the fidelity kings of this roster.
+module HyperCast
+  # This gem's own version — kept in lockstep with hypercast.gemspec by the
+  # prepare-release workflow, so the two can never drift apart again.
+  VERSION = "0.6.1"
+
+  # The success case of a verdict: a cast value.
+  Success = Data.define(:value)
+
+  # The failure case: a closed reason Symbol (:empty, :malformed, :out_of_range) plus the
+  # offending span, in the units String#[] slices by on the text you passed: character
+  # offsets for text (in any encoding — the core reads UTF-8 and the span is mapped back),
+  # byte offsets for a binary (ASCII-8BIT) String, whose characters are its bytes. Either
+  # way `text[offset, length]` is the offending text, with no mapping on your side.
+  # Nothing is captured — slicing it out of the input is the caller's choice.
+  Fault = Data.define(:reason, :offset, :length)
+
+  # The native core's failure codes, mapped to the closed reason Symbols a Fault carries.
+  REASONS = { 1 => :empty, 2 => :malformed, 3 => :out_of_range }.freeze
+
+  # Caller-declared numeric notation for the integer, real and decimal doors — declared out
+  # loud (INVARIANT, or a literal), never defaulted, the same stance every binding takes.
+  # The currency symbol is the one field a culture table would fill in: it is declared
+  # here, never looked up, and honored only while the CURRENCY flag is set — declared with
+  # the flag off, the symbol is a :malformed Fault at the symbol, and the flag with no
+  # symbol ("" — the default) matches nothing. Equal separators, or a symbol longer than
+  # 16 UTF-8 bytes or carrying an ASCII digit or ASCII whitespace, are a caller bug
+  # (ArgumentError), never a verdict.
+  NumFormat = Data.define(:decimal_sep, :group_sep, :flags, :currency) do
+    # Validates the declared separators up front — single characters, and distinct from
+    # each other — and the currency symbol (a String of at most 16 UTF-8 bytes with no
+    # ASCII digit or ASCII whitespace, since those would collide with the digit scan and
+    # the trimming around the symbol), so a malformed format fails loudly as the caller bug
+    # it is. All three are stored transcoded to UTF-8, the encoding the core reads — a
+    # separator handed over in another encoding (a no-break space read from a Latin-1
+    # file, say) is the same code point on every backend rather than whatever its bytes
+    # happened to spell in the encoding it arrived in.
+    def initialize(decimal_sep:, group_sep:, flags:, currency: "")
+      raise ArgumentError, "separators must be single characters" unless
+        decimal_sep.is_a?(String) && decimal_sep.length == 1 &&
+        group_sep.is_a?(String) && group_sep.length == 1
+
+      decimal_sep = decimal_sep.encode(Encoding::UTF_8)
+      group_sep = group_sep.encode(Encoding::UTF_8)
+      raise ArgumentError, "decimal and group separators must differ; both are #{decimal_sep.inspect}" if
+        decimal_sep == group_sep
+      raise ArgumentError, "currency symbol must be a String; got #{currency.inspect}" unless currency.is_a?(String)
+
+      symbol = currency.encode(Encoding::UTF_8)
+      limit = self.class::CURRENCY_MAX_BYTES
+      raise ArgumentError, "currency symbol #{currency.inspect} exceeds #{limit} UTF-8 bytes" if
+        symbol.bytesize > limit
+      raise ArgumentError, "currency symbol #{currency.inspect} must not contain an ASCII digit or whitespace" if
+        symbol.match?(/[0-9\t\n\f\r ]/)
+
+      super(decimal_sep: decimal_sep, group_sep: group_sep, flags: flags, currency: symbol)
+    end
+
+    # The 32-byte little-endian form the native ABI's NumFormat struct expects: the two
+    # separators as code points, the flags, the symbol's byte length, then the symbol's
+    # UTF-8 bytes zero-padded to 16.
+    def packed
+      [decimal_sep.ord, group_sep.ord, flags, currency.bytesize].pack("L<L<L<L<") + [currency].pack("a16")
+    end
+  end
+
+  # The widest currency symbol the native ABI carries inline, in UTF-8 bytes. Assigned out
+  # here rather than inside the Data.define block above: a constant written in that block
+  # is scoped lexically, and landed on HyperCast instead of on NumFormat.
+  NumFormat::CURRENCY_MAX_BYTES = 16
+
+  # Permit the group separator between digits (sizes not validated — between digits is the rule).
+  GROUPING = 1
+  # Permit accounting parentheses as negation: (1,234) is -1234.
+  PARENTHESES = 1 << 1
+  # Permit exponent notation. Integer doors reject a negative exponent.
+  EXPONENT = 1 << 2
+  # Permit 0x/&H/0b two's-complement radix prefixes (0xFF is -1 for an i8).
+  RADIX_PREFIXES = 1 << 3
+  # Permit a trailing %, dividing by 100. Real and decimal doors only.
+  PERCENT = 1 << 4
+  # Resolve the ./, roles per input from structure instead of the declared separators
+  # (which are ignored while this flag is set). Detection, not sniffing: a repeated
+  # separator is grouping ("1.234.567,89"); with both present the rightmost is the
+  # decimal; a single separator with a non-3-digit right run is the decimal ("3,1415");
+  # with exactly 3 digits right, only a 0 integer part proves decimal ("0,785").
+  # Genuinely ambiguous input ("12.185", "1,000") is a :malformed Fault at the separator,
+  # never guessed.
+  SEPARATOR_DETECT = 1 << 5
+  # Permit the format's declared currency symbol at either edge of the numeric body —
+  # leading ("$5", "-$5", "$ -5") or trailing ("5 €", "1.234,50 kr."), once, with optional
+  # ASCII whitespace between symbol and digits; accounting parentheses wrap the symbol
+  # along with the digits ("($5)"). With no symbol declared the flag matches nothing.
+  # Integer, real and decimal doors.
+  CURRENCY = 1 << 6
+  # Every lenience on (SEPARATOR_DETECT is a separator policy, not a lenience, and is
+  # deliberately not included).
+  ALL_STYLES = GROUPING | PARENTHESES | EXPONENT | RADIX_PREFIXES | PERCENT | CURRENCY
+
+  # The invariant profile — '.' decimal, ',' grouping, every lenience on, no currency
+  # symbol declared.
+  NumFormat::INVARIANT = NumFormat.new(decimal_sep: ".", group_sep: ",", flags: ALL_STYLES)
+
+  # The detection profile — every lenience on, ./, roles resolved per input by
+  # SEPARATOR_DETECT's structural rules.
+  NumFormat::DETECT = NumFormat.new(decimal_sep: ".", group_sep: ",",
+                                    flags: ALL_STYLES | SEPARATOR_DETECT)
+
+  # The declared unit of a Unix-epoch value — no magnitude guessing, ever.
+  UNIX_PRECISIONS = { seconds: 1, milliseconds: 2, microseconds: 3, nanoseconds: 4 }.freeze
+
+  # The date system an Excel serial number is expressed in. Spreadsheets carry no marker
+  # for this — it is a workbook-level setting — so the caller states it, the same way
+  # UNIX_PRECISIONS and DATE_ORDERS are declared rather than guessed.
+  EXCEL_EPOCHS = { y1900: 1, y1904: 2 }.freeze
+
+  # The declared field order of a separated calendar date — no guessing, ever: "1/7/2026"
+  # is January 7th (:month_day_year, the en-US order) or July 1st (:day_month_year, the
+  # en-GB order) only because the caller said which.
+  DATE_ORDERS = { year_month_day: 1, month_day_year: 2, day_month_year: 3 }.freeze
+
+  # An exact decimal, the decimal door's value: an unbounded Integer magnitude (the core's
+  # 96-bit unsigned, so at most 2**96 - 1), an Integer base-10 scale (0..28 — the number of
+  # places the magnitude is shifted right), and a negative flag. The value is
+  # (negative ? -1 : 1) * magnitude * 10**-scale. The triple is canonical: exact trailing
+  # zeros in the fraction are always trimmed, so the scale is minimal ("1.10", "1.1" and
+  # "1.1000" are all magnitude 11, scale 1; "100" stays magnitude 100, scale 0), and zero
+  # is scale 0 and never negative. Nothing but a zero is ever dropped — text carrying more
+  # precision than the core holds is an :out_of_range Fault, never an approximation.
+  # Convert with to_r (exact Rational), to_s (the canonical text), or to_d (BigDecimal).
+  Decimal = Data.define(:magnitude, :scale, :negative) do
+    # True for a negative value; zero is never negative.
+    def negative?
+      negative
+    end
+
+    # The value as an exact Rational.
+    def to_r
+      value = Rational(magnitude, 10**scale)
+      negative ? -value : value
+    end
+
+    # The canonical text form — "-1234.5", "-0.025", "0": the minimal scale rendered
+    # plainly, no exponent — the same string every binding renders and the conformance
+    # corpus pins as `value`.
+    def to_s
+      digits = magnitude.to_s.rjust(scale + 1, "0")
+      text = scale.zero? ? digits : "#{digits[0, digits.length - scale]}.#{digits[-scale, scale]}"
+      negative ? "-#{text}" : text
+    end
+
+    # The value as a BigDecimal, built exactly from the canonical text. Requires the
+    # bigdecimal gem (a bundled gem since Ruby 3.4, not a dependency of this one — add it
+    # to your Gemfile under Bundler), loaded lazily on first use so a consumer who never
+    # calls this pays nothing for it.
+    def to_d
+      require "bigdecimal"
+      BigDecimal(to_s)
+    end
+  end
+
+  class << self
+    # Presents a verdict optionally: an :empty fault becomes nil (Ruby's absent),
+    # everything else flows through untouched.
+    def optional(verdict)
+      return nil if verdict in Fault(reason: :empty)
+
+      verdict
+    end
+
+    # Whether a backend actually loaded and exports the ABI this binding was built against
+    # — what a consumer with a fallback of its own checks before committing to these doors.
+    # Probed once (a native_version round trip: the cheapest call the core has), cached,
+    # and never raises: a missing shared library, an unsupported platform, or an older core
+    # without the version export all answer false. The doors themselves keep their own
+    # behavior — the first call on an unavailable backend raises its precise LoadError —
+    # this only answers the question quietly.
+    def available?
+      return @available unless @available.nil?
+
+      @available = begin
+        native_version.is_a?(String)
+      rescue LoadError, StandardError
+        false
+      end
+    end
+
+    # Casts boolean text: true/false plus the conventions untrusted sources actually send
+    # (t/f, yes/no, y/n, 1/0, on/off, enabled/disabled, active/inactive,
+    # checked/unchecked, in/out), ASCII case-insensitive.
+    def bool(text)
+      plain(:cast_bool, text, 1) { |out| out.unpack1("C") != 0 }
+    end
+
+    { i8: "c", i16: "s<", i32: "l<", i64: "q<", u8: "C", u16: "S<", u32: "L<", u64: "Q<" }
+      .each do |door, unpack|
+      sizes = { "c" => 1, "C" => 1, "s<" => 2, "S<" => 2, "l<" => 4, "L<" => 4, "q<" => 8, "Q<" => 8 }
+      size = sizes.fetch(unpack)
+      # Resolved once here, not inside the method: interpolating a Symbol per call built a
+      # String and interned it on every integer cast.
+      symbol = :"cast_#{door}"
+      # Integer doors: the target type's own range, declared grouping, accounting parens,
+      # non-negative exponent, and 0x/&H/0b two's-complement radix prefixes. Ruby Integer
+      # is unbounded, so u64 comes back as the true unsigned value.
+      define_method(door) do |text, format|
+        numeric(symbol, text, format, size) { |out| out.unpack1(unpack) }
+      end
+    end
+
+    # Casts real text to an IEEE single (widened losslessly on the way out): finite values
+    # only, declared separators and grouping, parens, exponent, and trailing percent.
+    def f32(text, format)
+      numeric(:cast_f32, text, format, 4) { |out| out.unpack1("e") }
+    end
+
+    # Casts real text to an IEEE double. Notation rules as f32.
+    def f64(text, format)
+      numeric(:cast_f64, text, format, 8) { |out| out.unpack1("E") }
+    end
+
+    # Casts decimal text to an exact Decimal — the real doors' grammar (declared separators
+    # and grouping, parens, exponent, trailing percent, declared currency), but no float is
+    # ever formed: "0.1" is one tenth, "50%" is exactly 0.5 (magnitude 5, scale 1), and
+    # "1.10" is canonical magnitude 11, scale 1 — exact trailing zeros are always trimmed.
+    # A magnitude past 2**96 - 1, or more fractional precision than 28 places can hold, is
+    # an :out_of_range Fault — the door never rounds.
+    def decimal(text, format)
+      numeric(:cast_decimal, text, format, 16) do |out|
+        lo, hi, scale, negative = out.unpack("Q<L<CCx2")
+        Decimal.new(magnitude: (hi << 64) | lo, scale: scale, negative: negative != 0)
+      end
+    end
+
+    # Casts UUID text — all five .NET Guid formats (D/N/B/P/X) plus urn:uuid:/GUID:/UUID:
+    # prefixes — to Ruby's UUID lingua franca: the lowercase hyphenated String (the same
+    # shape SecureRandom.uuid returns).
+    def uuid(text)
+      plain(:cast_uuid, text, 16) do |out|
+        out.unpack("H8H4H4H4H12").join("-")
+      end
+    end
+
+    # Casts an RFC 3339 instant — zone mandatory — to a UTC Time at full nanosecond
+    # fidelity across the whole 0001-9999 window.
+    def timestamp(text)
+      plain(:cast_timestamp, text, 16) { |out| instant(out) }
+    end
+
+    # Casts an integer Unix-epoch value under a caller-declared unit Symbol
+    # (:seconds/:milliseconds/:microseconds/:nanoseconds) to a UTC Time. An unknown unit
+    # is a caller bug (KeyError), never a verdict.
+    def unix(text, precision)
+      declared(:cast_unix, text, UNIX_PRECISIONS.fetch(precision), 16) { |out| instant(out) }
+    end
+
+    # Casts an Excel date serial under a caller-declared epoch Symbol (:y1900/:y1904) to a
+    # UTC Time. The whole part counts days from the system's own day zero and the fraction
+    # is the time of day, so "45292.75" is 2024-01-01T18:00:00Z; a cell carries no zone and
+    # none is invented.
+    #
+    # The 1900 system contains a day that never existed: serial 60 is 1900-02-29, kept
+    # deliberately because Lotus 1-2-3 wrongly treated 1900 as a leap year and Excel copied
+    # the bug for file compatibility. It is :out_of_range here — the same verdict .date gives
+    # the text "1900-02-29" — so every serial above it is shifted one day against a naive
+    # count. An unknown epoch is a caller bug (KeyError), never a verdict.
+    def excel_serial(text, epoch)
+      declared(:cast_excel_serial, text, EXCEL_EPOCHS.fetch(epoch), 16) { |out| instant(out) }
+    end
+
+    # Casts a calendar date to a Date. With no order declared: the strict ISO 8601
+    # yyyy-MM-dd form only. With a declared order Symbol (:year_month_day,
+    # :month_day_year, :day_month_year), also the separated forms — "1/7/2026" is
+    # January 7th or July 1st only because the caller said which; an unknown order is a
+    # caller bug (KeyError), never a verdict.
+    def date(text, order = nil)
+      if order.nil?
+        plain(:cast_date, text, 4) do |out|
+          year, month, day = out.unpack("S<CC")
+          Date.new(year, month, day)
+        end
+      else
+        declared(:cast_date_ordered, text, DATE_ORDERS.fetch(order), 4) do |out|
+          year, month, day = out.unpack("S<CC")
+          Date.new(year, month, day)
+        end
+      end
+    end
+
+    # Casts a zone-less civil date-time — the shape untrusted feeds actually send
+    # ("1/7/2026 3:04 PM", "2026-01-07 15:04:05") — under a declared order Symbol to a
+    # stdlib DateTime with exact Rational seconds. No zone is *read*: the text names no
+    # instant, and the parse applies no offset. Ruby has no zone-less date-time type, so
+    # the value rides a DateTime, whose offset defaults to +00:00 — that zero is a carrier
+    # artifact, not data (the same caveat PHP's UTC-labeled DateTimeImmutable carries);
+    # fusing a real zone is the caller's job, and timestamp stays the strict RFC 3339
+    # instant door. An unknown order is a caller bug (KeyError).
+    def datetime(text, order)
+      declared(:cast_datetime, text, DATE_ORDERS.fetch(order), 16) do |out|
+        year, month, day, nanos = out.unpack("S<CCx4Q<")
+        second_of_day, frac = nanos.divmod(1_000_000_000)
+        hour, rest = second_of_day.divmod(3600)
+        minute, second = rest.divmod(60)
+        DateTime.new(year, month, day, hour, minute, second + Rational(frac, 1_000_000_000))
+      end
+    end
+
+    # Casts an ISO 24-hour time-of-day to an exact Integer of nanoseconds since midnight
+    # (Ruby has no time-of-day type; the integer keeps every digit).
+    def time(text)
+      plain(:cast_time, text, 8) { |out| out.unpack1("Q<") }
+    end
+
+    # Casts a duration (ISO 8601 fixed components, invariant colon form, or protobuf JSON
+    # seconds) to exact Rational seconds — full fidelity across the core's ±10,000-year
+    # window, no wrapping and no truncation.
+    def duration(text)
+      plain(:cast_duration, text, 16) do |out|
+        seconds, nanos = out.unpack("q<l<")
+        Rational(seconds * 1_000_000_000 + nanos, 1_000_000_000)
+      end
+    end
+
+    # The version of the native core actually loaded, as "major.minor.patch" — read from
+    # the library itself, not from this gem, so a consumer can prove the core behind the
+    # doors is the one this binding was built against (and name the mismatch when it is
+    # not). The cheapest possible probe that the backend resolved at all: takes nothing,
+    # cannot fail.
+    def native_version
+      word = packed_version
+      "#{word >> 16}.#{(word >> 8) & 0xFF}.#{word & 0xFF}"
+    end
+
+    private
+
+    # Encodings whose bytes already are the UTF-8 (or byte-identical) form the core reads.
+    BYTE_COMPATIBLE = [Encoding::UTF_8, Encoding::US_ASCII, Encoding::ASCII_8BIT].freeze
+
+    # Presents the input as UTF-8 bytes: already-compatible text crosses as-is (Fiddle
+    # passes a String's own bytes for void* — no copy of them); only foreign encodings pay
+    # a transcode, and text that cannot be transcoded raises String#encode's own
+    # Encoding:: error, the same one the Magnus extension raises through the same call.
+    #
+    # Text that is not a String is a caller bug, and the same TypeError on every backend:
+    # the extension takes its argument through Ruby's implicit String conversion (to_str,
+    # or a TypeError), so this does too.
+    def utf8(text)
+      unless text.is_a?(String)
+        converted = String.try_convert(text) or
+          raise TypeError, "no implicit conversion of #{text.class} into String"
+        text = converted
+      end
+      BYTE_COMPATIBLE.include?(text.encoding) ? text : text.encode(Encoding::UTF_8)
+    end
+
+    # Fiddle spells a null pointer as nil — the core's contract for empty input.
+    def input_ptr(bytes)
+      bytes.empty? ? nil : bytes
+    end
+
+    # One 24-byte scratch allocation per thread, reused by every call: out-value at 0
+    # (16 bytes covers every door, and malloc's alignment covers the decimal's 8), fault
+    # span at 16. Two Fiddle::Pointer.malloc(..., RUBY_FREE) calls per cast — each
+    # registering a GC finalizer — measured as the dominant per-call cost by an order of
+    # magnitude. The NumFormat no longer lives here: each format owns its own packed
+    # pointer (see packed_cache), so a numeric call copies nothing into scratch before
+    # crossing. Allocated through Runtime.buffer, which loads the library first, so a
+    # missing one is reported as such before Fiddle is touched.
+    def scratch
+      Thread.current[:hypercast_scratch] ||= begin
+        base = Runtime.buffer(24)
+        [base, base + 16]
+      end
+    end
+
+    # Presents a native return code as the verdict union: 0 yields a Success, a failure
+    # code becomes a Fault over `bytes` (the UTF-8 the core read), and -1 (contract
+    # violation) is a binding bug that raises.
+    def verdict(rc, fault, bytes)
+      if rc.zero?
+        Success.new(value: yield)
+      elsif rc == -1
+        raise "hypercast: libhypercast reported a contract violation — a binding bug, please report it"
+      else
+        offset, length = characters(bytes, *fault[0, 8].unpack("L<L<"))
+        Fault.new(reason: REASONS.fetch(rc), offset: offset, length: length)
+      end
+    end
+
+    # The core's byte span in the units String#[] slices by: an identity for a binary
+    # String (its characters are its bytes) and for ASCII text (`ascii_only?` reads the
+    # cached coderange — no scan), a byte-to-character remap otherwise. Character counts
+    # survive transcoding, so a span mapped on the UTF-8 form indexes the caller's own
+    # String whatever encoding it arrived in. Failure path only: a Success never pays.
+    def characters(bytes, offset, length)
+      return [offset, length] if bytes.encoding == Encoding::ASCII_8BIT || bytes.ascii_only?
+
+      [bytes.byteslice(0, offset).length, bytes.byteslice(offset, length).length]
+    end
+
+    # The shared body of every format-free door: one native call over the scratch buffers.
+    def plain(symbol, text, out_size)
+      bytes = utf8(text)
+      out, fault = scratch
+      rc = Runtime.function(symbol).call(input_ptr(bytes), bytes.bytesize, out, fault)
+      verdict(rc, fault, bytes) { yield(out[0, out_size]) }
+    end
+
+    # The shared body of the integer/real/decimal doors: plain, plus the format's own
+    # packed pointer, passed straight through — no per-call copy of the 32 bytes into
+    # scratch.
+    def numeric(symbol, text, format, out_size)
+      bytes = utf8(text)
+      out, fault = scratch
+      rc = Runtime.function(symbol).call(input_ptr(bytes), bytes.bytesize, packed_cache[format], out, fault)
+      verdict(rc, fault, bytes) { yield(out[0, out_size]) }
+    end
+
+    # The shared body of the four doors that take a caller-declared u32 — a precision, an
+    # epoch, or a field order — already resolved from its Symbol by the door. These bodies
+    # (plain, numeric, declared, and packed_version below) are the whole Fiddle crossing.
+    def declared(symbol, text, code, out_size)
+      bytes = utf8(text)
+      out, fault = scratch
+      rc = Runtime.function(symbol).call(input_ptr(bytes), bytes.bytesize, code, out, fault)
+      verdict(rc, fault, bytes) { yield(out[0, out_size]) }
+    end
+
+    # The loaded core's version word, major << 16 | minor << 8 | patch, straight from the
+    # library's zero-argument hypercast_version export.
+    def packed_version
+      Runtime.function(:hypercast_version).call
+    end
+
+    # How many formats packed_cache remembers before it starts forgetting the oldest. Far
+    # more than a process that reuses its formats ever holds; what it bounds is the one
+    # that builds a NumFormat per request.
+    PACKED_CACHE_LIMIT = 64
+
+    # Identity-keyed memo (compare_by_identity — a pointer hash, not Data's structural
+    # #hash over three Strings and an Integer) of a native 32-byte RawNumFormat per format
+    # object, filled once from NumFormat#packed. Formats are reused constants in practice,
+    # so the common call finds its pointer in one lookup and packs nothing. The Hash holds
+    # the format, so a key can never be a recycled address; the race is benign (idempotent).
+    #
+    # Bounded, oldest entry out first (a Hash keeps insertion order): unbounded, it held
+    # every format ever cast with and its 32-byte allocation for the life of the process,
+    # which a consumer building a format per request experienced as a leak. A format
+    # evicted and then used again is simply packed again, and an evicted pointer stays
+    # alive for any call already holding it — the caller's own reference outlives the cache's.
+    def packed_cache
+      @packed_cache ||= Hash.new do |cache, format|
+        cache.shift while cache.size >= PACKED_CACHE_LIMIT
+        pointer = Runtime.buffer(32)
+        pointer[0, 32] = format.packed
+        cache[format] = pointer
+      end.compare_by_identity
+    end
+
+    # Builds a UTC Time from the core's protobuf-shaped {seconds, nanos} pair, exactly.
+    def instant(bytes)
+      seconds, nanos = bytes.unpack("q<l<")
+      Time.at(seconds, nanos, :nanosecond, in: "UTC")
+    end
+  end
+end
+
+# --- backend selection: the Magnus extension, when present, replaces the doors above in
+# place on this module (no delegation layer) — Fiddle's measured 1.6 µs per-call floor
+# drops to an ordinary extension call. The pure-Fiddle definitions stay the universal
+# zero-compile fallback — the last resort, for a Ruby or a platform no platform gem covers;
+# precompiled platform gems are how the extension ships without ever making a consumer
+# compile anything, and they carry no Fiddle library at all.
+#
+# HYPERCAST_PURE forces Fiddle. It is a testing and diagnostic switch — CI runs the whole
+# suite through it, and it is how a suspected extension bug is ruled in or out — not a
+# setting a deployment needs. It is read for presence, not value, so "0" and the empty
+# string force it too. Inside a platform gem there is no library for it to load: BACKEND
+# still reads :fiddle, HyperCast.available? answers false, and the first door call raises a
+# LoadError naming the universal gem, which is where the Fiddle backend lives.
+HyperCast::BACKEND =
+  if ENV["HYPERCAST_PURE"]
+    :fiddle
+  else
+    # Two layouts, and both have to work. A released platform gem is a "fat" gem carrying one
+    # extension per supported Ruby ABI under lib/hypercast/<minor>/ (see the Rakefile's
+    # native:gem task for why an ABI-per-file is unavoidable — Magnus has no `abi3`
+    # equivalent), and `rake native:dev` — the local dev loop — stages its own build at that
+    # same versioned path. CI's in-job staging additionally drops a single extension flat at
+    # lib/, which is also where a hand copy of `cargo ruby-ext`'s output goes (the alias only
+    # builds into rust/target/ruby/release/; nothing but that rake task copies it here).
+    # Trying the versioned path first and the flat one second means neither has to know the
+    # other exists.
+    #
+    # A miss on both is not an error: it means this Ruby/platform combination has no
+    # precompiled extension, which is precisely what the Fiddle backend is for. A platform
+    # with no shared library either still lands on Fiddle, so the first call raises its own
+    # precise error — a "not found" LoadError naming the path, or
+    # NativePlatform::UnsupportedPlatformError naming the platform. HyperCast.available? is
+    # the quiet way to ask.
+    begin
+      require "hypercast/#{RUBY_VERSION[/\d+\.\d+/]}/hypercast_native"
+      :native
+    rescue LoadError
+      begin
+        require "hypercast_native"
+        :native
+      rescue LoadError
+        :fiddle
+      end
+    end
+  end

metadata

@@ -0,0 +1,84 @@
+--- !ruby/object:Gem::Specification
+name: hypercast-wasm
+version: !ruby/object:Gem::Version
+  version: 0.6.1
+platform: ruby
+authors:
+- Brian Buvinghausen
+bindir: bin
+cert_chain: []
+date: 1980-01-02 00:00:00.000000000 Z
+dependencies:
+- !ruby/object:Gem::Dependency
+  name: rake
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '13.0'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '13.0'
+- !ruby/object:Gem::Dependency
+  name: yard
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '0.9'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '0.9'
+description: |
+  The hypercast gem's Magnus extension, prebuilt for wasm32-wasip1, for a ruby.wasm
+  interpreter built with `rbwasm build`: put this gem in that Gemfile instead of hypercast
+  and the extension is linked into the interpreter. Same API and the same Rust core as the
+  hypercast gem. Not for a host Ruby, where the hypercast gem is the one to install.
+executables: []
+extensions:
+- ext/hypercast_native/extconf.rb
+extra_rdoc_files: []
+files:
+- LICENSE
+- README.md
+- ext/hypercast_native/3.4/hypercast_native.a.gz
+- ext/hypercast_native/4.0/hypercast_native.a.gz
+- ext/hypercast_native/extconf.rb
+- lib/hypercast.rb
+- lib/hypercast/native_platform.rb
+- lib/hypercast/runtime.rb
+homepage: https://github.com/SkunkWerkx/HyperCast
+licenses:
+- MIT
+metadata:
+  source_code_uri: https://github.com/SkunkWerkx/HyperCast
+  changelog_uri: https://github.com/SkunkWerkx/HyperCast/blob/master/CHANGELOG.md
+  bug_tracker_uri: https://github.com/SkunkWerkx/HyperCast/issues
+  documentation_uri: https://github.com/SkunkWerkx/HyperCast/tree/master/ruby#ruby-in-the-browser
+  rubygems_mfa_required: 'true'
+rdoc_options: []
+require_paths:
+- lib
+required_ruby_version: !ruby/object:Gem::Requirement
+  requirements:
+  - - ">="
+    - !ruby/object:Gem::Version
+      version: '3.3'
+required_rubygems_version: !ruby/object:Gem::Requirement
+  requirements:
+  - - ">="
+    - !ruby/object:Gem::Version
+      version: '0'
+requirements: []
+rubygems_version: 4.0.20
+specification_version: 4
+summary: 'HyperCast for ruby.wasm: the Magnus extension, linked in by rbwasm build'
+test_files: []