hopp: Add SetDeadline methods to Conn and Trans

Still need: dynamic decoding, static decoding, static encoding

.gitignore

@@ -0,0 +1 @@
+/generate/test

cmd/hopp-generate/main.go

@@ -4,6 +4,7 @@ import "os"
 import "fmt"
 import "strings"
 import "path/filepath"
+import "git.tebibyte.media/sashakoshka/goparse"
 import "git.tebibyte.media/sashakoshka/hopp/generate"
 
 func main() {
@@ -18,7 +19,7 @@ func main() {
 	input, err := os.Open(source)
 	handleErr(1, err)
 	defer input.Close()
-	protocol, err := generate.ParseReader(input)
+	protocol, err := generate.ParseReader(source, input)
 	handleErr(1, err)
 
 	absDestination, err := filepath.Abs(destination)
@@ -30,14 +31,18 @@ func main() {
 
 	output, err := os.Create(destination)
 	handleErr(1, err)
-	err = protocol.Generate(output, packageName)
+	generator := generate.Generator {
+		Output:      output,
+		PackageName: packageName,
+	}
+	_, err = generator.Generate(protocol)
 	handleErr(1, err)
 	fmt.Fprintf(os.Stderr, "%s: OK\n", name)
 }
 
 func handleErr(code int, err error) {
 	if err != nil {
-		fmt.Fprintf(os.Stderr, "%s: %v\n", os.Args[0], err)
+		fmt.Fprintf(os.Stderr, "%s: %v\n", os.Args[0], parse.Format(err))
 		os.Exit(code)
 	}
 }

codec.go

@@ -0,0 +1,47 @@
+package hopp
+
+import "fmt"
+
+type anyInt16 interface { ~uint16 | ~int16 }
+type anyInt64 interface { ~uint64 | ~int64 }
+
+// decodeI16 decodes a 16 bit integer from the given data.
+func decodeI16[T anyInt16](data []byte) (T, error) {
+	if len(data) != 2 { return 0, fmt.Errorf("decoding int16: %w", ErrWrongBufferLength) }
+	return T(data[0]) << 8 | T(data[1]), nil
+}
+
+// encodeI16 encodes a 16 bit integer into the given buffer.
+func encodeI16[T anyInt16](buffer []byte, value T) error {
+	if len(buffer) != 2 { return fmt.Errorf("encoding int16: %w", ErrWrongBufferLength) }
+	buffer[0] = byte(value >> 8)
+	buffer[1] = byte(value)
+	return nil
+}
+
+// decodeI64 decodes a 64 bit integer from the given data.
+func decodeI64[T anyInt64](data []byte) (T, error) {
+	if len(data) != 8 { return 0, fmt.Errorf("decoding int64: %w", ErrWrongBufferLength) }
+	return T(data[0]) << 56 |
+		T(data[1]) << 48 |
+		T(data[2]) << 40 |
+		T(data[3]) << 32 |
+		T(data[4]) << 24 |
+		T(data[5]) << 16 |
+		T(data[6]) <<  8 |
+		T(data[7]), nil
+}
+
+// encodeI64 encodes a 64 bit integer into the given buffer.
+func encodeI64[T anyInt64](buffer []byte, value T) error {
+	if len(buffer) != 8 { return fmt.Errorf("encoding int64: %w", ErrWrongBufferLength) }
+	buffer[0] = byte(value >> 56)
+	buffer[1] = byte(value >> 48)
+	buffer[2] = byte(value >> 40)
+	buffer[3] = byte(value >> 32)
+	buffer[4] = byte(value >> 24)
+	buffer[5] = byte(value >> 16)
+	buffer[6] = byte(value >> 8)
+	buffer[7] = byte(value)
+	return nil
+}

codec/decode.go

@@ -1,108 +0,0 @@
-package codec
-
-import "io"
-
-// Decodable is any type that can decode itself from a decoder.
-type Decodable interface {
-	// Decode reads data from decoder, replacing the data of the object. It
-	// returns the amount of bytes written, and an error if the write
-	// stopped early.
-	Decode(decoder *Decoder) (n int, err error)
-}
-
-// Decoder wraps an [io.Reader] and decodes data from it.
-type Decoder struct {
-	io.Reader
-}
-
-// ReadFull calls [io.ReadFull] on the reader.
-func (this *Decoder) ReadFull(buffer []byte) (n int, err error) {
-	return io.ReadFull(this, buffer)
-}
-
-// ReadByte decodes a single byte from the input reader.
-func (this *Decoder) ReadByte() (value byte, n int, err error) {
-	uncasted, n, err := this.ReadUint8()
-	return byte(uncasted), n, err
-}
-
-// ReadInt8 decodes an 8-bit signed integer from the input reader.
-func (this *Decoder) ReadInt8() (value int8, n int, err error) {
-	uncasted, n, err := this.ReadUint8()
-	return int8(uncasted), n, err
-}
-
-// ReadUint8 decodes an 8-bit unsigned integer from the input reader.
-func (this *Decoder) ReadUint8() (value uint8, n int, err error) {
-	buffer := [1]byte { }
-	n, err = this.ReadFull(buffer[:])
-	return uint8(buffer[0]), n, err
-}
-
-// ReadInt16 decodes an 16-bit signed integer from the input reader.
-func (this *Decoder) ReadInt16() (value int16, n int, err error) {
-	uncasted, n, err := this.ReadUint16()
-	return int16(uncasted), n, err
-}
-
-// ReadUint16 decodes an 16-bit unsigned integer from the input reader.
-func (this *Decoder) ReadUint16() (value uint16, n int, err error) {
-	buffer := [2]byte { }
-	n, err = this.ReadFull(buffer[:])
-	return uint16(buffer[0]) << 8 |
-		uint16(buffer[1]), n, err
-}
-
-// ReadInt32 decodes an 32-bit signed integer from the input reader.
-func (this *Decoder) ReadInt32() (value int32, n int, err error) {
-	uncasted, n, err := this.ReadUint32()
-	return int32(uncasted), n, err
-}
-
-// ReadUint32 decodes an 32-bit unsigned integer from the input reader.
-func (this *Decoder) ReadUint32() (value uint32, n int, err error) {
-	buffer := [4]byte { }
-	n, err = this.ReadFull(buffer[:])
-	return uint32(buffer[0]) << 24 |
-		uint32(buffer[1]) << 16 |
-		uint32(buffer[2]) << 8 |
-		uint32(buffer[3]), n, err
-}
-
-// ReadInt64 decodes an 64-bit signed integer from the input reader.
-func (this *Decoder) ReadInt64() (value int64, n int, err error) {
-	uncasted, n, err := this.ReadUint64()
-	return int64(uncasted), n, err
-}
-
-// ReadUint64 decodes an 64-bit unsigned integer from the input reader.
-func (this *Decoder) ReadUint64() (value uint64, n int, err error) {
-	buffer := [8]byte { }
-	n, err = this.ReadFull(buffer[:])
-	return uint64(buffer[0]) << 56 |
-		uint64(buffer[1]) << 48 |
-		uint64(buffer[2]) << 48 |
-		uint64(buffer[3]) << 32 |
-		uint64(buffer[4]) << 24 |
-		uint64(buffer[5]) << 16 |
-		uint64(buffer[6]) << 8 |
-		uint64(buffer[7]), n, err
-}
-
-// ReadGBEU decodes a growing unsigned integer of up to 64 bits from the input
-// reader.
-func (this *Decoder) ReadGBEU() (value uint64, n int, err error) {
-	var fullValue uint64
-	for {
-		chunk, nn, err := this.ReadByte()
-		if err != nil { return 0, n, err }
-		n += nn
-		
-		fullValue *= 0x80
-		fullValue += uint64(chunk & 0x7F)
-		ccb := chunk >> 7
-		if ccb == 0 {
-			return fullValue, n, nil
-		}
-	}
-}

codec/encode.go

@@ -1,102 +0,0 @@
-package codec
-
-import "io"
-
-// Encodable is any type that can write itself to an encoder.
-type Encodable interface {
-	// Encode sends data to encoder. It returns the amount of bytes written,
-	// and an error if the write stopped early.
-	Encode(encoder *Encoder) (n int, err error)
-}
-	
-// Encoder wraps an [io.Writer] and encodes data to it.
-type Encoder struct {
-	io.Writer
-}
-
-// WriteByte encodes a single byte to the output writer.
-func (this *Encoder) WriteByte(value byte) (n int, err error) {
-	return this.WriteByte(uint8(value))
-}
-
-// WriteInt8 encodes an 8-bit signed integer to the output writer.
-func (this *Encoder) WriteInt8(value int8) (n int, err error) {
-	return this.WriteUint8(uint8(value))
-}
-
-// WriteUint8 encodes an 8-bit unsigned integer to the output writer.
-func (this *Encoder) WriteUint8(value uint8) (n int, err error) {
-	return this.Write([]byte { byte(value) })
-}
-
-// WriteInt16 encodes an 16-bit signed integer to the output writer.
-func (this *Encoder) WriteInt16(value int16) (n int, err error) {
-	return this.WriteUint16(uint16(value))
-}
-
-// WriteUint16 encodes an 16-bit unsigned integer to the output writer.
-func (this *Encoder) WriteUint16(value uint16) (n int, err error) {
-	return this.Write([]byte {
-		byte(value >> 8),
-		byte(value),
-	})
-}
-
-// WriteInt32 encodes an 32-bit signed integer to the output writer.
-func (this *Encoder) WriteInt32(value int32) (n int, err error) {
-	return this.WriteUint32(uint32(value))
-}
-
-// WriteUint32 encodes an 32-bit unsigned integer to the output writer.
-func (this *Encoder) WriteUint32(value uint32) (n int, err error) {
-	return this.Write([]byte {
-		byte(value >> 24),
-		byte(value >> 16),
-		byte(value >> 8),
-		byte(value),
-	})
-}
-
-// WriteInt64 encodes an 64-bit signed integer to the output writer.
-func (this *Encoder) WriteInt64(value int64) (n int, err error) {
-	return this.WriteUint64(uint64(value))
-}
-
-// WriteUint64 encodes an 64-bit unsigned integer to the output writer.
-func (this *Encoder) WriteUint64(value uint64) (n int, err error) {
-	return this.Write([]byte {
-		byte(value >> 56),
-		byte(value >> 48),
-		byte(value >> 40),
-		byte(value >> 32),
-		byte(value >> 24),
-		byte(value >> 16),
-		byte(value >> 8),
-		byte(value),
-	})
-}
-
-// EncodeGBEU encodes a growing unsigned integer of up to 64 bits to the output
-// writer.
-func (this *Encoder) EncodeGBEU(value uint64) (n int, err error) {
-	// increase if go somehow gets support for over 64 bit integers. we
-	// could also make an expanding int type in goutil to use here, or maybe
-	// there is one in the stdlib. keep this int64 version as well though
-	// because its ergonomic.
-	buffer := [16]byte { }
-	
-	window := (GBEUSize(value) - 1) * 7
-	index := 0
-	for window >= 0 {
-		chunk := uint8(value >> window) & 0x7F
-		if window > 0 {
-			chunk |= 0x80
-		}
-		buffer[index] = chunk
-		
-		index += 1
-		window -= 7
-	}
-	
-	return this.Write(buffer[:])
-}

codec/measure.go

@@ -1,11 +0,0 @@
-package codec
-
-// GBEUSize returns the size (in octets) of a GBEU integer.
-func GBEUSize(value uint64) int {
-	length := 0
-	for {
-		value >>= 7
-		length ++
-		if value == 0 { return length }
-	}
-}

connection.go

@@ -2,7 +2,7 @@ package hopp
 
 import "io"
 import "net"
-// import "time"
+import "time"
 
 const defaultSizeLimit int64 = 1024 * 1024 // 1 megabyte
 
@@ -23,8 +23,13 @@ type Conn interface {
 	// be called in a loop to avoid the connection locking up.
 	AcceptTrans() (Trans, error)
 
+	// SetDeadline operates is [net.Conn.SetDeadline] but for OpenTrans
+	// and AcceptTrans calls.
+	SetDeadline(t time.Time) error	
 	// SetSizeLimit sets a limit (in bytes) for how large messages can be.
-	// By default, this limit is 1 megabyte.
+	// By default, this limit is 1 megabyte. Note that this is only
+	// enforced when sending and receiving byte slices, and it does not
+	// apply to [Trans.SendWriter] or [Trans.ReceiveReader].
 	SetSizeLimit(limit int64)
 }
 
@@ -40,8 +45,6 @@ type Trans interface {
 	// unique within the connection. This method is safe for concurrent use.
 	ID() int64
 		
-	// TODO: add methods for setting send and receive deadlines
-		
 	// Send sends a message. This method is not safe for concurrent use.
 	Send(method uint16, data []byte) error
 	// SendWriter sends data written to an [io.Writer]. The writer must be
@@ -57,4 +60,12 @@ type Trans interface {
 	// previously opened through this function will be discarded. This
 	// method is not safe for concurrent use, and neither is its result.
 	ReceiveReader() (method uint16, data io.Reader, err error)
+
+	// See the documentation for [net.Conn.SetDeadline].
+	SetDeadline(time.Time) error
+	// TODO
+	// // See the documentation for [net.Conn.SetReadDeadline].
+	// SetReadDeadline(t time.Time) error
+	// // See the documentation for [net.Conn.SetWriteDeadline].
+	// SetWriteDeadline(t time.Time) error
 }

design/branched-generated-encoder.md

@@ -0,0 +1,128 @@
+# Branched Generated Decoder
+
+Pasted here because Tebitea is down
+
+## The problem
+
+TAPE is designed so that the decoder can gloss over data it does not understand.
+Technically the protocol allows for this, but I completely forgot to implement
+this in the generated decoder, oops. This would be trivial if TAPE messages were
+still flat tables, but they aren't, because those aren't useful enough. So,
+let's analyze the problem.
+
+## When it happens
+
+There are two reasons something might not match up with the expected data:
+
+The first and most obvious is unrecognized keys. If the key is not in the set of
+recognized keys for a KTV, it should leave the corresponding struct field blank.
+Once #6 has been implemented, throw an error if the data was not optional.
+
+The second is wrong types. If we are expecting KTV and get SBA, we should leave
+the data as empty. The aforementioned concern about #6 also applies here. We
+don't need to worry about special cases at the structure root, because it would
+be technically possible to make the structure root an option, so it really is
+just a normal value. Until #6, we will leave that blank too.
+
+## Preliminary ideas
+
+The first is going to be pretty simple. All we need to do is have a skimmer
+function that skims over TAPE data very, and then call that on the KTV value
+each time we run into a mystery key. It should only return an error if the
+structure of the data is malformed in such a way that it cannot continue to the
+next one. This should be stored in the tape package alongside the dynamic
+decoding functions, because they will essentially function the same way and
+could probably share lots of code.
+
+The second is a bit more complicated because of the existence of KTV and OTA
+because they are aggregate types. Go types work a bit differently, as if you
+have an array of an array of an array of ints, that information is represented
+in one place, whereas TAPE doesn't really do that. All of that information is
+sort of buried within the data structure, so we don't know what we will be
+decoding before we actually do it. Whenever we encounter a type we don't expect,
+we would need to abort decoding of the entire data structure, and then skim over
+whatever detritus is left, which would literally be in a half-decoded state. The
+fact that the code is generated flat and thus cannot use return or defer
+statements contributes to the complexity of this problem. We need to go up, but
+we can't. There is no up, only forward.
+
+Of course, the dynamic decoder does not have this problem in the first place
+because it doesn't expect anything, and constructs the destination to fit
+whatever it sees in the TAPE structure as it is decoding it. KTVs are completely
+dynamic because they are implemented as maps, so the only time it needs to
+completely comprehend a type is with OTAs. There is a function called typeOf
+that gets the type of the current tag and returns it as a reflect.Type, which
+necessitates recursion and peeking at OTAs and their elements.
+
+We could try to do the same thing in the generated decoder, comparing the
+determined type against the expected type to try to figure out whether we should
+decode an array or a table, etc. This is immediately problematic as it requires
+memory to be allocated, both for the peek buffer and the resulting tree of type
+information. If we end up with some crazy way to keep track of the types, that's
+only one half of the allocation problem and we would still be spending extra
+cycles going over all of that twice.
+
+## Performance constraints
+
+The generated decoder is supposed to blaze through data, and it can't do that if
+it does all the singing and dancing that the dynamic decoder does. It's time for
+some performance constraints:
+
+- No allocations, except as required to build the destination for the data
+- No redundant work
+- So, no freaking peeking
+- It should take well under 500 lines of generated code to decode one message of
+reasonable size (i.e. be careful not to bloat the binary)
+
+I'm not really going to do my usual thing here of making a slow version and
+speeding it up over time based on evidence and experimentation because these
+constraints inform the design so much it would be impossible to continue without
+them. I am 99% confident that these constraints will allow for an acceptable
+baseline of performance (for generated code) and we can still profile and
+micro-optimize later. This is good enough for me.
+Heavy solution
+
+There is a solution that might work very well which involves completely redoing
+the generated decoding code. We could create a function for every source type to
+destination type mapping that exists in protocol, and then compose them all
+together. The decoding methods for each message or type would be wrappers around
+the correct function for their root TAPE -> Go type mapping. The main benefit of
+this is it would make this problem a lot more manageable because the interface
+points between the data would be represented by function boundaries. This would
+allow the use of return and defer statements, and would allow more code sharing,
+producing a smaller binary. Go would probably inline these where needed.
+
+Would this work? Probably. More investigation is required to make sure. I want
+to stop re-writing things I don't need to. On the other hand, it is just the
+decoder.
+
+## Light solution
+
+TODO: find a solution that satisfies the performance constraints, keeps the same
+identical interface, and works off the same code. I am convinced this is doable,
+and it might even allow us to extract more data from an unexpected structure.
+However, continuing this way might introduce unmanageable complexity. It is
+already a little unmanageable and I am just one pony (kind of).
+
+## Implementation
+
+Heavy solution is going to work here, applied to only the points of
+`Generator.generateDecodeValue` where it decodes an aggregate data structure.
+That way, only minimal amounts of code need to be redone.
+
+Whenever a branch needs to happen, a call shall be generated, a deferred
+implementation request shall be added to a special FIFO queue within the
+generator. After generating data structures and their root decoding functions,
+the generator shall pick away at this queue until no requests remain. The
+generator shall accept new items during this process, so that recursion is
+possible. This is all to ensure it is only ever writing one function at a time
+
+The functions shall take a pointer to a type that accepts any type like (~) the
+destination's base type. We should also probably just call
+`Generator.generateDecodeValue` directly on user defined types this way, keeping
+their public `Decode` methods just for convenience.
+
+The tape package shall contain a skimming function that takes a decoder and a
+tag, and recursively consumes the decoder given the context of the tag. This
+shall be utilized by the decoder functions to skip over values if their tags
+or keys do not match up with what is expected.

design/pdl-compiler.md

@@ -56,20 +56,27 @@ static section.
 
 For each defined type, the compiler shall generate a Go type with the same name
 as written in its definition. The Go type shall be encodable, and shall have
-`Encode` and `Decode` methods as described below.
+`EncodeValue`, `DecodeValue`, and `Tag` methods as described below.
 
 ## Encoding and Decoding Methods
 
-Each encodable type shall be given an `Encode` method and a `Decode` method,
-which will take in a `codec.Encoder` and a `codec.Decoder` respectively. Both
+Each message shall be given an `Encode` method and a `Decode` method,
+which shall take in a `codec.Encoder` and a `codec.Decoder` respectively. Both
 return an `(int, error)` pair describing the amount of bytes written and an
-error if the write stopped early. `Encode` will encode the data within the
-message to the given encoder, and `Decode` will decode data from the given
+error if the write stopped early. `Encode` shall encode the data within the
+message to the given encoder, and `Decode` shall decode data from the given
 decoder and place it in the type's value. The methods shall not retain or close
 any encoders or decoders they are given. Both methods shall have pointer
-receivers. In effect, these methods will satisfy `codec.Encodable` and
+receivers. In effect, these methods shall satisfy `codec.Encodable` and
 `codec.Decodable`.
 
+Each defined type shall be given an `EncodeValue` method and a `DecodeValue`
+method, which shall both take in a `tape.Tag`, then a `codec.Encoder` and a
+`codec.Decoder` respectively. These methods shall encode and decode the value
+according to the CN given by the tag. The TN shall be ignored. The message shall
+also have a method `Tag` that takes no arguments and returns the preferred tag
+of the type including the TN and CN.
+
 ## Connection
 
 The compiler shall generate a `Conn` struct which embeds a `hopp.Conn`, which

design/pdl-language.md

@@ -7,12 +7,12 @@ PDL allows defining a protocol using HOPP and TAPE.
 | Syntax     | TN      | CN | Description
 | ---------- | ------- | -: | -----------
 | I5         | SI      |    |   
-| I8         | LI      |  0 | 
-| I16        | LI      |  1 | 
-| I32        | LI      |  3 | 
-| I64        | LI      |  7 | 
-| I128[^2]   | LI      | 15 | 
-| I256[^2]   | LI      | 31 | 
+| I8         | LSI     |  0 | 
+| I16        | LSI     |  1 | 
+| I32        | LSI     |  3 | 
+| I64        | LSI     |  7 | 
+| I128[^2]   | LSI     | 15 | 
+| I256[^2]   | LSI     | 31 | 
 | U5         | SI      |    | 
 | U8         | LI      |  0 | 
 | U16        | LI      |  1 | 

error.go

@@ -9,6 +9,7 @@ type Error string; const (
 	ErrIntegerOverflow    Error = "integer overflow"
 	ErrMessageMalformed   Error = "message is malformed"
 	ErrTablePairMissing   Error = "required table pair is missing"
+	ErrWrongBufferLength  Error = "wrong buffer length"
 )
 
 // Error implements the error interface.

generate/generate_test.go

@@ -0,0 +1,381 @@
+package generate
+
+// import "fmt"
+import "testing"
+
+// TODO: once everything has been ironed out, test that the public API of the
+// generator is equal to something specific
+
+var exampleProtocol = defaultProtocol()
+
+func init() {
+	exampleProtocol.Messages[0x0000] = Message {
+		Name: "Connect",
+		Type: TypeTableDefined {
+			Fields: map[uint16] Field {
+				0x0000: Field { Name: "Name",      Type: TypeString { } },
+				0x0001: Field { Name: "Password",  Type: TypeString { } },
+			},
+		},
+	}
+	exampleProtocol.Messages[0x0001] = Message {
+		Name: "UserList",
+		Type: TypeTableDefined {
+			Fields: map[uint16] Field {
+				0x0000: Field { Name: "Users",  Type: TypeArray { Element: TypeNamed { Name: "User" } } },
+			},
+		},
+	}
+	exampleProtocol.Messages[0x0002] = Message {
+		Name: "Pulse",
+		Type: TypeTableDefined {
+			Fields: map[uint16] Field {
+				0x0000: Field { Name: "Index",  Type: TypeInt { Bits: 5 } },
+				0x0001: Field { Name: "Offset", Type: TypeInt { Bits: 16, Signed: true }},
+				0x0002: Field { Name: "X",      Type: TypeFloat { Bits: 16 }},
+				0x0003: Field { Name: "Y",      Type: TypeFloat { Bits: 32 }},
+				0x0004: Field { Name: "Z",      Type: TypeFloat { Bits: 64 }},
+			},
+		},
+	}
+	exampleProtocol.Messages[0x0003] = Message {
+		Name: "NestedArray",
+		Type: TypeArray { Element: TypeArray { Element: TypeInt { Bits: 8 } } },
+	}
+	exampleProtocol.Messages[0x0004] = Message {
+		Name: "Integers",
+		Type: TypeTableDefined {
+			Fields: map[uint16] Field {
+				0x0000: Field { Name: "U5",  Type: TypeInt { Bits: 5  } },
+				0x0001: Field { Name: "U8",  Type: TypeInt { Bits: 8  } },
+				0x0002: Field { Name: "U16", Type: TypeInt { Bits: 16 } },
+				0x0003: Field { Name: "U32", Type: TypeInt { Bits: 32 } },
+				0x0004: Field { Name: "U64", Type: TypeInt { Bits: 64 } },
+				0x0006: Field { Name: "I8",  Type: TypeInt { Bits: 8,  Signed: true } },
+				0x0007: Field { Name: "I16", Type: TypeInt { Bits: 16, Signed: true } },
+				0x0008: Field { Name: "I32", Type: TypeInt { Bits: 32, Signed: true } },
+				0x0009: Field { Name: "I64", Type: TypeInt { Bits: 64, Signed: true } },
+				0x000B: Field { Name: "NI8", Type: TypeInt { Bits: 8,  Signed: true } },
+				0x000C: Field { Name: "NI16",Type: TypeInt { Bits: 16, Signed: true } },
+				0x000D: Field { Name: "NI32",Type: TypeInt { Bits: 32, Signed: true } },
+				0x000E: Field { Name: "NI64",Type: TypeInt { Bits: 64, Signed: true } },
+			},
+		},
+	}
+	exampleProtocol.Types["User"] = TypeTableDefined {
+		Fields: map[uint16] Field {
+			0x0000: Field { Name: "Name",      Type: TypeString { } },
+			0x0001: Field { Name: "Bio",       Type: TypeString { } },
+			0x0002: Field { Name: "Followers", Type: TypeInt { Bits: 32 } },
+		},
+	}
+}
+
+func TestGenerateRunEncodeDecode(test *testing.T) {
+	testGenerateRun(test, &exampleProtocol, "encode-decode", `
+		// imports
+	`, `
+		log.Println("MessageConnect")
+		messageConnect := MessageConnect {
+			Name: "rarity",
+			Password: "gems",
+		}
+		testEncodeDecode(
+			&messageConnect,
+			tu.S(0xE1, 0x02).AddVar(
+				[]byte { 0x00, 0x00, 0x86, 'r', 'a', 'r', 'i', 't', 'y' },
+				[]byte { 0x00, 0x01, 0x84, 'g', 'e', 'm', 's' },
+				))
+		log.Println("MessageUserList")
+		messageUserList := MessageUserList {
+			Users: []User {
+				User {
+					Name: "rarity",
+					Bio: "asdjads",
+					Followers: 0x324,
+				},
+				User {
+					Name: "deez nuts",
+					Bio: "logy",
+					Followers: 0x8000,
+				},
+				User {
+					Name: "creekflow",
+					Bio: "im creekflow",
+					Followers: 0x3894,
+				},
+			},
+		}
+		testEncodeDecode(
+			&messageUserList,
+			tu.S(0xE1, 0x01, 0x00, 0x00,
+				0xC1, 0x03, 0xE1,
+					).Add(0x03).AddVar(
+						[]byte { 0x00, 0x00, 0x86, 'r', 'a', 'r', 'i', 't', 'y' },
+						[]byte { 0x00, 0x01, 0x87, 'a', 's', 'd', 'j', 'a', 'd', 's' },
+						[]byte { 0x00, 0x02, 0x23, 0x00, 0x00, 0x03, 0x24 },
+					).Add(0x03).AddVar(
+						[]byte { 0x00, 0x00, 0x89, 'd', 'e', 'e', 'z', ' ', 'n', 'u', 't', 's' },
+						[]byte { 0x00, 0x01, 0x84, 'l', 'o', 'g', 'y' },
+						[]byte { 0x00, 0x02, 0x23, 0x00, 0x00, 0x80, 0x00 },
+					).Add(0x03).AddVar(
+						[]byte { 0x00, 0x00, 0x89, 'c', 'r', 'e', 'e', 'k', 'f', 'l', 'o', 'w' },
+						[]byte { 0x00, 0x01, 0x8C, 'i', 'm', ' ', 'c', 'r', 'e', 'e', 'k', 'f',
+							'l', 'o', 'w' },
+						[]byte { 0x00, 0x02, 0x23, 0x00, 0x00, 0x38, 0x94 },
+				))
+		log.Println("MessagePulse")
+		messagePulse := MessagePulse {
+			Index: 9,
+			Offset: -0x3521,
+			X: 45.375,
+			Y: 294.1,
+			Z: 384729384.234892034,
+		}
+		testEncodeDecode(
+			&messagePulse,
+			tu.S(0xE1, 0x05).AddVar(
+				[]byte { 0x00, 0x00, 0x09 },
+				[]byte { 0x00, 0x01, 0x41, 0xCA, 0xDF },
+				[]byte { 0x00, 0x02, 0x61, 0x51, 0xAC },
+				[]byte { 0x00, 0x03, 0x63, 0x43, 0x93, 0x0C, 0xCD },
+				[]byte { 0x00, 0x04, 0x67, 0x41, 0xB6, 0xEE, 0x81, 0x28, 0x3C, 0x21, 0xE2 },
+				))
+		log.Println("MessageNestedArray")
+		uint8s := func(n int) []uint8 {
+			array := make([]uint8, n)
+			for index := range array {
+				array[index] = uint8(index + 1) | 0xF0
+			}
+			return array
+		}
+		messageNestedArray := MessageNestedArray {
+			uint8s(6),
+			uint8s(35),
+		}
+		testEncodeDecode(
+			&messageNestedArray,
+			tu.S(0xC1, 0x02, 0xC1,
+				0x06, 0x20, 0xF1, 0xF2, 0xF3, 0xF4, 0xF5, 0xF6,
+				35,   0x20, 0xF1, 0xF2, 0xF3, 0xF4, 0xF5, 0xF6,
+				            0xF7, 0xF8, 0xF9, 0xFA, 0xFB, 0xFC,
+				            0xFD, 0xFE, 0xFF, 0xF0, 0xF1, 0xF2,
+				            0xF3, 0xF4, 0xF5, 0xF6, 0xF7, 0xF8,
+				            0xF9, 0xFA, 0xFB, 0xFC, 0xFD, 0xFE,
+				            0xFF, 0xF0, 0xF1, 0xF2, 0xF3))
+		log.Println("MessageIntegers")
+		messageIntegers := MessageIntegers {
+			U5:   0x13,
+			U8:   0xC9,
+			U16:  0x34C9,
+			U32:  0x10E134C9,
+			U64:  0x639109BC10E134C9,
+			I8:   0x35,
+			I16:  0x34C9,
+			I32:  0x10E134C9,
+			I64:  0x639109BC10E134C9,
+			NI8:  -0x35,
+			NI16: -0x34C9,
+			NI32: -0x10E134C9,
+			NI64: -0x639109BC10E134C9,
+		}
+		testEncodeDecode(
+			&messageIntegers,
+			tu.S(0xE1, 13).AddVar(
+				[]byte { 0x00, 0x00, 0x13 },
+				[]byte { 0x00, 0x01, 0x20, 0xC9 },
+				[]byte { 0x00, 0x02, 0x21, 0x34, 0xC9 },
+				[]byte { 0x00, 0x03, 0x23, 0x10, 0xE1, 0x34, 0xC9 },
+				[]byte { 0x00, 0x04, 0x27, 0x63, 0x91, 0x09, 0xBC, 0x10, 0xE1, 0x34, 0xC9 },
+				[]byte { 0x00, 0x06, 0x40, 0x35 },
+				[]byte { 0x00, 0x07, 0x41, 0x34, 0xC9 },
+				[]byte { 0x00, 0x08, 0x43, 0x10, 0xE1, 0x34, 0xC9 },
+				[]byte { 0x00, 0x09, 0x47, 0x63, 0x91, 0x09, 0xBC, 0x10, 0xE1, 0x34, 0xC9 },
+				[]byte { 0x00, 0x0B, 0x40, 0xCB },
+				[]byte { 0x00, 0x0C, 0x41, 0xCB, 0x37 },
+				[]byte { 0x00, 0x0D, 0x43, 0xEF, 0x1E, 0xCB, 0x37 },
+				[]byte { 0x00, 0x0E, 0x47, 0x9C, 0x6E, 0xF6, 0x43, 0xEF, 0x1E, 0xCB, 0x37 },
+				))
+	`)
+}
+
+func TestGenerateRunDecodeWrongType(test *testing.T) {
+	protocol := defaultProtocol()
+	protocol.Messages[0x0000] = Message {
+		Name: "Uint5",
+		Type: TypeInt { Bits: 5 },
+	}
+	protocol.Messages[0x0001] = Message {
+		Name: "Uint8",
+		Type: TypeInt { Bits: 8 },
+	}
+	protocol.Messages[0x0002] = Message {
+		Name: "Uint16",
+		Type: TypeInt { Bits: 16 },
+	}
+	protocol.Messages[0x0003] = Message {
+		Name: "Uint32",
+		Type: TypeInt { Bits: 32 },
+	}
+	protocol.Messages[0x0004] = Message {
+		Name: "Uint64",
+		Type: TypeInt { Bits: 64 },
+	}
+	protocol.Messages[0x0005] = Message {
+		Name: "Int8",
+		Type: TypeInt { Bits: 8 },
+	}
+	protocol.Messages[0x0006] = Message {
+		Name: "Int16",
+		Type: TypeInt { Bits: 16 },
+	}
+	protocol.Messages[0x0007] = Message {
+		Name: "Int32",
+		Type: TypeInt { Bits: 32 },
+	}
+	protocol.Messages[0x0008] = Message {
+		Name: "Int64",
+		Type: TypeInt { Bits: 64 },
+	}
+	protocol.Messages[0x0009] = Message {
+		Name: "String",
+		Type: TypeString { },
+	}
+	protocol.Messages[0x000A] = Message {
+		Name: "Buffer",
+		Type: TypeBuffer { },
+	}
+	protocol.Messages[0x000B] = Message {
+		Name: "StringArray",
+		Type: TypeArray { Element: TypeString { } },
+	}
+	protocol.Messages[0x000C] = Message {
+		Name: "Table",
+		Type: TypeTable { },
+	}
+	protocol.Messages[0x000D] = Message {
+		Name: "TableDefined",
+		Type: TypeTableDefined {
+			Fields: map[uint16] Field {
+				0x0000: Field { Name: "Name",      Type: TypeString { } },
+				0x0001: Field { Name: "Password",  Type: TypeString { } },
+			},
+		},
+	}
+	
+	testGenerateRun(test, &protocol, "decode-wrong-type", `
+		// imports
+	`, `
+		datas := [][]byte {
+			/* int8     */ []byte { byte(tape.LSI.WithCN(0)), 0x45 },
+			/* int16    */ []byte { byte(tape.LSI.WithCN(1)), 0x45, 0x67 },
+			/* int32    */ []byte { byte(tape.LSI.WithCN(3)), 0x45, 0x67, 0x89, 0xAB },
+			/* int64    */ []byte { byte(tape.LSI.WithCN(7)), 0x45, 0x67, 0x89, 0xAB, 0xCD, 0xEF, 0x01, 0x23 },
+			/* uint5    */ []byte { byte(tape.SI.WithCN(12)) },
+			/* uint8    */ []byte { byte(tape.LI.WithCN(0)), 0x45 },
+			/* uint16   */ []byte { byte(tape.LI.WithCN(1)), 0x45, 0x67 },
+			/* uint32   */ []byte { byte(tape.LI.WithCN(3)), 0x45, 0x67, 0x89, 0xAB },
+			/* uint64   */ []byte { byte(tape.LI.WithCN(7)), 0x45, 0x67, 0x89, 0xAB, 0xCD, 0xEF, 0x01, 0x23 },
+			/* string   */ []byte { byte(tape.SBA.WithCN(7)), 'p', 'u', 'p', 'e', 'v', 'e', 'r' },
+			/* []byte   */ []byte { byte(tape.SBA.WithCN(5)), 'b', 'l', 'a', 'r', 'g' },
+			/* []string */ []byte {
+				byte(tape.OTA.WithCN(0)), 2, byte(tape.LBA.WithCN(0)),
+				0x08, 0x45, 0x67, 0x89, 0xAB, 0xCD, 0xEF, 0x01, 0x23,
+				0x05, 0x11, 0x11, 0x11, 0x11, 0x11,
+			},
+			/* map[uint16] any */ []byte {
+				byte(tape.KTV.WithCN(0)), 2,
+				0x02, 0x23, byte(tape.LSI.WithCN(1)), 0x45, 0x67,
+				0x02, 0x23, byte(tape.LI.WithCN(3)),  0x45, 0x67, 0x89, 0xAB,
+			},
+		}
+
+		
+		for index, data := range datas {
+			log.Printf("data %2d %v [%s]", index, tape.Tag(data[0]), tu.HexBytes(data[1:]))
+			// integers should only assign to other integers
+			if index > 8 {
+				cas := func(destination Message) {
+					n, err := destination.Decode(tape.NewDecoder(bytes.NewBuffer(data)))
+					if err != nil { log.Fatalf("error: %v | n: %d", err, n) }
+					reflectValue := reflect.ValueOf(destination).Elem()
+					if reflectValue.CanInt() {
+						if reflectValue.Int() != 0 {
+							log.Fatalf(
+								"destination not zero: %v",
+								reflectValue.Elem().Interface())
+						}
+					} else {
+						if reflectValue.Uint() != 0 {
+							log.Fatalf(
+								"destination not zero: %v",
+								reflectValue.Elem().Interface())
+						}
+					}
+					if n != len(data) {
+						log.Fatalf("n not equal: %d != %d", n, len(data))
+					}
+				}
+				log.Println("- MessageInt8")
+				{ var dest MessageInt8;   cas(&dest) }
+				log.Println("- MessageInt16")
+				{ var dest MessageInt16;  cas(&dest) }
+				log.Println("- MessageInt32")
+				{ var dest MessageInt32;  cas(&dest) }
+				log.Println("- MessageInt64")
+				{ var dest MessageInt64;  cas(&dest) }
+				log.Println("- MessageUint8")
+				{ var dest MessageUint8;  cas(&dest) }
+				log.Println("- MessageUint16")
+				{ var dest MessageUint16; cas(&dest) }
+				log.Println("- MessageUint32")
+				{ var dest MessageUint32; cas(&dest) }
+				log.Println("- MessageUint64")
+				{ var dest MessageUint64; cas(&dest) }
+			}
+			arrayCase := func(destination Message) {
+				n, err := destination.Decode(tape.NewDecoder(bytes.NewBuffer(data)),)
+				if err != nil { log.Fatalf("error: %v | n: %d", err, n) }
+				reflectDestination := reflect.ValueOf(destination)
+				reflectValue := reflectDestination.Elem()
+				if reflectValue.Len() != 0 {
+					log.Fatalf("len(destination) not zero: %v", reflectValue.Interface())
+				}
+				if n != len(data) {
+					log.Fatalf("n not equal: %d != %d", n, len(data))
+				}
+			}
+			anyCase := func(destination Message) {
+				n, err := destination.Decode(tape.NewDecoder(bytes.NewBuffer(data)),)
+				if err != nil { log.Fatalf("error: %v | n: %d", err, n) }
+				reflectDestination := reflect.ValueOf(destination)
+				reflectValue := reflectDestination.Elem()
+				if reflectValue == reflect.Zero(reflectValue.Type()) {
+					log.Fatalf("len(destination) not zero: %v", reflectValue.Interface())
+				}
+				if n != len(data) {
+					log.Fatalf("n not equal: %d != %d", n, len(data))
+				}
+			}
+			// SBA/LBA types should only assign to other SBA/LBA types
+			if index != 9 && index != 10 {
+				log.Println("- MessageString")
+				{ var dest MessageString; arrayCase(&dest) }
+				log.Println("- MessageBuffer")
+				{ var dest MessageBuffer; arrayCase(&dest) }
+			}
+			// arrays should only assign to other arrays
+			if index != 11 {
+				log.Println("- MessageStringArray")
+				{ var dest MessageStringArray; arrayCase(&dest) }
+			}
+			// tables should only assign to other tables
+			if index != 12 {
+				log.Println("- MessageTable")
+				{ var dest = make(MessageTable); arrayCase(&dest) }
+				log.Println("- MessageTableDefined")
+				{ var dest MessageTableDefined; anyCase(&dest) }
+			}
+		}
+	`)
+}

generate/misc_test.go

@@ -0,0 +1,131 @@
+package generate
+
+import "os"
+import "fmt"
+import "os/exec"
+import "testing"
+import "path/filepath"
+
+func testGenerateRun(test *testing.T, protocol *Protocol, title, imports, testCase string) {
+	// reset data directory
+	dir := filepath.Join("test", title)
+	err := os.RemoveAll(dir)
+	if err != nil { test.Fatal(err) }
+	err = os.MkdirAll(dir, 0750)
+	if err != nil { test.Fatal(err) }
+
+	// open files
+	sourceFile, err := os.Create(filepath.Join(dir, "protocol.go"))
+	if err != nil { test.Fatal(err) }
+	defer sourceFile.Close()
+	mainFile, err := os.Create(filepath.Join(dir, "main.go"))
+	if err != nil { test.Fatal(err) }
+	defer mainFile.Close()
+
+	// generate protocol
+	generator := Generator {
+		Output: sourceFile,
+		PackageName: "main",
+	}
+	_, err = generator.Generate(protocol)
+	if err != nil { test.Fatal(err) }
+
+	// build static source files
+	imports = `
+		import "log"
+		import "bytes"
+		import "reflect"
+		import "git.tebibyte.media/sashakoshka/hopp/tape"
+		import tu "git.tebibyte.media/sashakoshka/hopp/internal/testutil"
+		` + imports
+	setup := `log.Println("*** BEGIN TEST CASE OUTPUT ***")`
+	teardown := `log.Println("--- END TEST CASE OUTPUT ---")`
+	static := `
+		func testEncode(message Message, correct tu.Snake) {
+			buffer := bytes.Buffer { }
+			encoder := tape.NewEncoder(&buffer)
+			n, err := message.Encode(encoder)
+			if err != nil { log.Fatalf("at %d: %v\n", n, err) }
+			encoder.Flush()
+			got := buffer.Bytes()
+			log.Printf("got:     [%s]", tu.HexBytes(got))
+			log.Println("correct:", correct)
+			if n != len(got) {
+				log.Fatalf("n incorrect: %d != %d\n", n, len(got))
+			}
+			if ok, n := correct.Check(got); !ok {
+				log.Fatalln("not equal at", n)
+			}
+		}
+
+		func testDecode(correct Message, data any) {
+			var flat []byte
+			switch data := data.(type) {
+			case []byte:   flat = data
+			case tu.Snake: flat = data.Flatten()
+			}
+			message := reflect.New(reflect.ValueOf(correct).Elem().Type()).Interface().(Message)
+			log.Println("before: ", message)
+			decoder := tape.NewDecoder(bytes.NewBuffer(flat))
+			n, err := message.Decode(decoder)
+			if err != nil { log.Fatalf("at %d: %v\n", n, err) }
+			log.Println("got:    ", message)
+			log.Println("correct:", correct)
+			if n != len(flat) {
+				log.Fatalf("n incorrect: %d != %d\n", n, len(flat))
+			}
+			if !reflect.DeepEqual(message, correct) {
+				log.Fatalln("not equal")
+			}
+		}
+
+		// TODO: possibly combine the two above functions into this one,
+		// also take a data parameter here (snake)
+		func testEncodeDecode(message Message, data tu.Snake) {buffer := bytes.Buffer { }
+			log.Println("encoding:")
+			encoder := tape.NewEncoder(&buffer)
+			n, err := message.Encode(encoder)
+			if err != nil { log.Fatalf("at %d: %v\n", n, err) }
+			encoder.Flush()
+			got := buffer.Bytes()
+			log.Printf("got:     [%s]", tu.HexBytes(got))
+			log.Println("correct:", data)
+			if n != len(got) {
+				log.Fatalf("n incorrect: %d != %d\n", n, len(got))
+			}
+			if ok, n := data.Check(got); !ok {
+				log.Fatalln("not equal at", n)
+			}
+		
+			log.Println("decoding:")
+			destination := reflect.New(reflect.ValueOf(message).Elem().Type()).Interface().(Message)
+			flat := data.Flatten()
+			log.Println("before: ", destination)
+			decoder := tape.NewDecoder(bytes.NewBuffer(flat))
+			n, err = destination.Decode(decoder)
+			if err != nil { log.Fatalf("at %d: %v\n", n, err) }
+			log.Println("got:    ", destination)
+			log.Println("correct:", message)
+			if n != len(flat) {
+				log.Fatalf("n incorrect: %d != %d\n", n, len(flat))
+			}
+			if !reflect.DeepEqual(destination, message) {
+				log.Fatalln("not equal")
+			}
+			
+		}
+		`
+	fmt.Fprintf(
+		mainFile, "package main\n%s\nfunc main() {\n%s\n%s\n%s\n}\n%s",
+		imports, setup, testCase, teardown, static)
+
+	// build and run test
+	command := exec.Command("go", "run", "./" + filepath.Join("generate", dir))
+	workingDirAbs, err := filepath.Abs("..")
+	if err != nil { test.Fatal(err) }
+	command.Dir = workingDirAbs
+	command.Env = os.Environ()
+	output, err := command.CombinedOutput()
+	test.Logf("output of %v:\n%s", command, output)
+	if err != nil { test.Fatal(err) }
+}

generate/parse.go

@@ -21,33 +21,12 @@ func Parse(lx parse.Lexer) (*Protocol, error) {
 func defaultProtocol() Protocol {
 	return Protocol {
 		Messages: make(map[uint16] Message),
-		Types: map[string] Type {
-			"U8":     TypeInt { Bits: 8   },
-			"U16":    TypeInt { Bits: 16  },
-			"U32":    TypeInt { Bits: 32  },
-			"U64":    TypeInt { Bits: 64  },
-			"U128":   TypeInt { Bits: 128 },
-			"U256":   TypeInt { Bits: 256 },
-			"I8":     TypeInt { Bits: 8,   Signed: true },
-			"I16":    TypeInt { Bits: 16,  Signed: true },
-			"I32":    TypeInt { Bits: 32,  Signed: true },
-			"I64":    TypeInt { Bits: 64,  Signed: true },
-			"I128":   TypeInt { Bits: 128, Signed: true },
-			"I256":   TypeInt { Bits: 256, Signed: true },
-			"F16":    TypeFloat { Bits: 16  },
-			"F32":    TypeFloat { Bits: 32  },
-			"F64":    TypeFloat { Bits: 64  },
-			"F128":   TypeFloat { Bits: 128 },
-			"F256":   TypeFloat { Bits: 256 },
-			"String": TypeString { },
-			"Buffer": TypeBuffer { },
-			"Table":  TypeTable { },
-		},
+		Types: map[string] Type { },
 	}
 }
 
-func ParseReader(reader io.Reader) (*Protocol, error) {
-	lx, err := Lex("test.pdl", reader)
+func ParseReader(fileName string, reader io.Reader) (*Protocol, error) {
+	lx, err := Lex(fileName, reader)
 	if err != nil { return nil, err }
 	return Parse(lx)
 }
@@ -116,6 +95,28 @@ func (this *parser) parseType() (Type, error) {
 
 	switch this.Kind() {
 	case TokenIdent:
+		switch this.Value() {
+		case "U8":     return TypeInt { Bits: 8   }, this.Next()
+		case "U16":    return TypeInt { Bits: 16  }, this.Next()
+		case "U32":    return TypeInt { Bits: 32  }, this.Next()
+		case "U64":    return TypeInt { Bits: 64  }, this.Next()
+		case "U128":   return TypeInt { Bits: 128 }, this.Next()
+		case "U256":   return TypeInt { Bits: 256 }, this.Next()
+		case "I8":     return TypeInt { Bits: 8,   Signed: true }, this.Next()
+		case "I16":    return TypeInt { Bits: 16,  Signed: true }, this.Next()
+		case "I32":    return TypeInt { Bits: 32,  Signed: true }, this.Next()
+		case "I64":    return TypeInt { Bits: 64,  Signed: true }, this.Next()
+		case "I128":   return TypeInt { Bits: 128, Signed: true }, this.Next()
+		case "I256":   return TypeInt { Bits: 256, Signed: true }, this.Next()
+		case "F16":    return TypeFloat { Bits: 16  }, this.Next()
+		case "F32":    return TypeFloat { Bits: 32  }, this.Next()
+		case "F64":    return TypeFloat { Bits: 64  }, this.Next()
+		case "F128":   return TypeFloat { Bits: 128 }, this.Next()
+		case "F256":   return TypeFloat { Bits: 256 }, this.Next()
+		case "String": return TypeString { }, this.Next()
+		case "Buffer": return TypeBuffer { }, this.Next()
+		case "Table":  return TypeTable { }, this.Next()
+		}
 		return this.parseTypeNamed()
 	case TokenLBracket:
 		return this.parseTypeArray()

generate/parse_test.go

@@ -3,7 +3,6 @@ package generate
 import "fmt"
 import "strings"
 import "testing"
-// import "reflect"
 import "git.tebibyte.media/sashakoshka/goparse"
 
 func TestParse(test *testing.T) {
@@ -12,8 +11,8 @@ func TestParse(test *testing.T) {
 		Name: "Connect",
 		Type: TypeTableDefined {
 			Fields: map[uint16] Field {
-				0x0000: Field { Name: "Name",      Type: TypeNamed { Name: "String" } },
-				0x0001: Field { Name: "Password",  Type: TypeNamed { Name: "String" } },
+				0x0000: Field { Name: "Name",      Type: TypeString { } },
+				0x0001: Field { Name: "Password",  Type: TypeString { } },
 			},
 		},
 	}
@@ -27,14 +26,14 @@ func TestParse(test *testing.T) {
 	}
 	correct.Types["User"] = TypeTableDefined {
 		Fields: map[uint16] Field {
-			0x0000: Field { Name: "Name",      Type: TypeNamed { Name: "String" } },
-			0x0001: Field { Name: "Bio",       Type: TypeNamed { Name: "String" } },
-			0x0002: Field { Name: "Followers", Type: TypeNamed { Name: "U32" } },
+			0x0000: Field { Name: "Name",      Type: TypeString { } },
+			0x0001: Field { Name: "Bio",       Type: TypeString { } },
+			0x0002: Field { Name: "Followers", Type: TypeInt { Bits: 32 } },
 		},
 	}
 	test.Log("CORRECT:", &correct)
 	
-	got, err := ParseReader(strings.NewReader(`
+	got, err := ParseReader("test.pdl", strings.NewReader(`
 	M0000 Connect {
 		0000 Name String,
 		0001 Password String,

generate/protocol.go

@@ -1,5 +1,10 @@
 package generate
 
+import "fmt"
+import "maps"
+import "slices"
+import "crypto/md5"
+
 type Protocol struct {
 	Messages map[uint16] Message
 	Types    map[string] Type
@@ -11,7 +16,7 @@ type Message struct {
 }
 
 type Type interface {
-	
+	fmt.Stringer
 }
 
 type TypeInt struct {
@@ -19,29 +24,84 @@ type TypeInt struct {
 	Signed bool
 }
 
+func (typ TypeInt) String() string {
+	output := ""
+	if typ.Signed {
+		output += "I"
+	} else {
+		output += "U"
+	}
+	output += fmt.Sprint(typ.Bits)
+	return output
+}
+
 type TypeFloat struct {
 	Bits int
 }
 
+func (typ TypeFloat) String() string {
+	return fmt.Sprintf("F%d", typ.Bits)
+}
+
 type TypeString struct { }
 
+func (TypeString) String() string {
+	return "String"
+}
+
 type TypeBuffer struct { }
 
+func (TypeBuffer) String() string {
+	return "Buffer"
+}
+
 type TypeArray struct {
 	Element Type
 }
 
+func (typ TypeArray) String() string {
+	return fmt.Sprintf("[]%v", typ.Element)
+}
+
 type TypeTable struct { }
 
+func (TypeTable) String() string {
+	return "Table"
+}
+
 type TypeTableDefined struct {
 	Fields map[uint16] Field
 }
 
+func (typ TypeTableDefined) String() string {
+	output := "{"
+	for _, key := range slices.Sorted(maps.Keys(typ.Fields)) {
+		output += fmt.Sprintf("%04X %v", key, typ.Fields[key])
+	}
+	output += "}"
+	return output
+}
+
 type Field struct {
 	Name string
 	Type Type
 }
 
+func (field Field) String() string {
+	return fmt.Sprintf("%s %v", field.Name, field.Type)
+}
+
 type TypeNamed struct {
 	Name string
 }
+
+func (typ TypeNamed) String() string {
+	return typ.Name
+}
+
+func HashType(typ Type) [16]byte {
+	// TODO: if we ever want to make the compiler more efficient, this would
+	// be a good place to start, complex string concatenation in a hot path
+	// (sorta)
+	return md5.Sum([]byte(typ.String()))
+}

internal/testutil/testutil.go

@@ -0,0 +1,173 @@
+package testutil
+
+import "fmt"
+import "slices"
+import "strings"
+import "reflect"
+
+// Snake lets you compare blocks of data where the ordering of certain parts may
+// be swapped every which way. It is designed for comparing the encoding of
+// maps where the ordering of individual elements is inconsistent.
+//
+// The snake is divided into sectors, which hold a number of variations. For a
+// sector to be satisfied by some data, some ordering of it must match the data
+// exactly. for the snake to be satisfied by some data, its sectors must match
+// the data in order, but the internal ordering of each sector doesn't matter.
+type Snake []    []     []byte
+//         snake sector variation
+
+// S returns a new snake.
+func S(data ...byte) Snake {
+	return (Snake { }).Add(data...)
+}
+
+// AddVar returns a new snake with the given sector added on to it. Successive
+// calls of this method can be chained together to create a big ass snake.
+func (sn Snake) AddVar(sector ...[]byte) Snake {
+	slice := make(Snake, len(sn) + 1)
+	copy(slice, sn)
+	slice[len(slice) - 1] = sector
+	return slice
+}
+
+// Add is like AddVar, but adds a sector with only one variation, which means it
+// does not vary, hence why the method is called that.
+func (sn Snake) Add(data ...byte) Snake {
+	return sn.AddVar(data)
+}
+
+// Check determines if the data satisfies the snake.
+func (sn Snake) Check(data []byte) (ok bool, n int) {
+	left := data
+	variations := map[int] []byte { }
+	for _, sector := range sn {
+		clear(variations)
+		for key, variation := range sector {
+			variations[key] = variation
+		}
+		for len(variations) > 0 {
+			found := false
+			for key, variation := range variations {
+				if len(left) < len(variation) { continue }
+				if !slices.Equal(left[:len(variation)], variation) { continue }
+				n += len(variation)
+				left = data[n:]
+				delete(variations, key)
+				found = true
+			}
+			if !found { return false, n }
+		}
+	}
+	if n < len(data) {
+		return false, n
+	}
+	return true, n
+}
+
+// Flatten returns the snake flattened to a byte array. The result of this
+// function always satisfies the snake.
+func (sn Snake) Flatten() []byte {
+	flat := []byte { }
+	for _, sector := range sn {
+		for _, variation := range sector {
+			flat = append(flat, variation...)
+		}
+	}
+	return flat
+}
+
+func (sn Snake) String() string {
+	if len(sn) == 0 || len(sn[0]) == 0 || len(sn[0][0]) == 0 {
+		return "EMPTY"
+	}
+
+	out := strings.Builder { }
+	for index, sector := range sn {
+		if index > 0 { out.WriteString(" : ") }
+		out.WriteRune('[')
+		for index, variation := range sector {
+			if index > 0 { out.WriteString(" / ") }
+			for _, byt := range variation {
+				fmt.Fprintf(&out, "%02x", byt)
+			}
+		}
+		out.WriteRune(']')
+	}
+	return out.String()
+}
+
+// HexBytes formats bytes into a hexadecimal string.
+func HexBytes(data []byte) string {
+	if len(data) == 0 { return "EMPTY" }
+	out := strings.Builder { }
+	for _, byt := range data {
+		fmt.Fprintf(&out, "%02x", byt)
+	}
+	return out.String()
+}
+
+// Describe returns a string representing the type and data of the given value.
+func Describe(value any) string {
+	desc := describer { }
+	desc.describe(reflect.ValueOf(value))
+	return desc.String()
+}
+
+type describer struct {
+	strings.Builder
+	indent int
+}
+
+func (this *describer) describe(value reflect.Value) {
+	value = reflect.ValueOf(value.Interface())
+	switch value.Kind() {
+	case reflect.Array, reflect.Slice:
+		this.printf("[\n")
+		this.indent += 1
+		for index := 0; index < value.Len(); index ++ {
+			this.iprintf("")
+			this.describe(value.Index(index))
+			this.iprintf("\n")
+		}
+		this.indent -= 1
+		this.iprintf("]")
+	case reflect.Struct:
+		this.printf("struct {\n")
+		this.indent += 1
+		typ := value.Type()
+		for index := range typ.NumField() {
+			indexBuffer := [1]int { index }
+			this.iprintf("%s: ", typ.Field(index).Name)
+			this.describe(value.FieldByIndex(indexBuffer[:]))
+			this.iprintf("\n")
+		}
+		this.indent -= 1
+		this.iprintf("}\n")
+	case reflect.Map:
+		this.printf("map {\n")
+		this.indent += 1
+		iter := value.MapRange()
+		for iter.Next() {
+			this.iprintf("")
+			this.describe(iter.Key())
+			this.printf(": ")
+			this.describe(iter.Value())
+			this.iprintf("\n")
+		}
+		this.indent -= 1
+		this.iprintf("}\n")
+	case reflect.Pointer:
+		this.printf("& ")
+		this.describe(value.Elem())
+	default:
+		this.printf("<%v %v>", value.Type(), value.Interface())
+	}
+}
+
+func (this *describer) printf(format string, v ...any) {
+	fmt.Fprintf(this, format, v...)
+}
+
+func (this *describer) iprintf(format string, v ...any) {
+	fmt.Fprintf(this, strings.Repeat("\t", this.indent) + format, v...)
+}

internal/testutil/testutil_test.go

@@ -0,0 +1,66 @@
+package testutil
+
+import "testing"
+
+func TestSnakeA(test *testing.T) {
+	snake := S(1, 6).AddVar(
+		[]byte { 1 },
+		[]byte { 2 },
+		[]byte { 3 },
+		[]byte { 4 },
+		[]byte { 5 },
+	).Add(9)
+
+	test.Log(snake)
+
+	ok, n := snake.Check([]byte { 1, 6, 1, 2, 3, 4, 5, 9 })
+	if !ok { test.Fatal("false negative:", n) }
+	ok, n = snake.Check([]byte { 1, 6, 5, 4, 3, 2, 1, 9 })
+	if !ok { test.Fatal("false negative:", n) }
+	ok, n = snake.Check([]byte { 1, 6, 3, 1, 4, 2, 5, 9 })
+	if !ok { test.Fatal("false negative:", n) }
+	
+	ok, n = snake.Check([]byte { 1, 6, 9 })
+	if ok { test.Fatal("false positive:", n) }
+	ok, n = snake.Check([]byte { 1, 6, 1, 2, 3, 4, 5, 6, 9 })
+	if ok { test.Fatal("false positive:", n) }
+	ok, n = snake.Check([]byte { 1, 6, 0, 2, 3, 4, 5, 6, 9 })
+	if ok { test.Fatal("false positive:", n) }
+	ok, n = snake.Check([]byte { 1, 6, 7, 1, 4, 2, 5, 9 })
+	if ok { test.Fatal("false positive:", n) }
+	ok, n = snake.Check([]byte { 1, 6, 7, 3, 1, 4, 2, 5, 9 })
+	if ok { test.Fatal("false positive:", n) }
+	ok, n = snake.Check([]byte { 1, 6, 7, 3, 1, 4, 2, 5, 9 })
+	if ok { test.Fatal("false positive:", n) }
+	ok, n = snake.Check([]byte { 1, 6, 1, 2, 3, 4, 5, 9, 10})
+	if ok { test.Fatal("false positive:", n) }
+}
+
+func TestSnakeB(test *testing.T) {
+	snake := S(1, 6).AddVar(
+		[]byte { 1 },
+		[]byte { 2 },
+	).Add(9).AddVar(
+		[]byte { 3, 2 },
+		[]byte { 0 },
+		[]byte { 1, 1, 2, 3 },
+	)
+
+	test.Log(snake)
+
+	ok, n := snake.Check([]byte { 1, 6, 1, 2, 9, 3, 2, 0, 1, 1, 2, 3})
+	if !ok { test.Fatal("false negative:", n) }
+	ok, n = snake.Check([]byte { 1, 6, 2, 1, 9, 0, 1, 1, 2, 3, 3, 2})
+	if !ok { test.Fatal("false negative:", n) }
+	
+	ok, n = snake.Check([]byte { 1, 6, 9 })
+	if ok { test.Fatal("false positive:", n) }
+	ok, n = snake.Check([]byte { 1, 6, 1, 2, 9 })
+	if ok { test.Fatal("false positive:", n) }
+	ok, n = snake.Check([]byte { 1, 6, 9, 3, 2, 0, 1, 1, 2, 3})
+	if ok { test.Fatal("false positive:", n) }
+	ok, n = snake.Check([]byte { 1, 6, 2, 9, 0, 1, 1, 2, 3, 3, 2})
+	if ok { test.Fatal("false positive:", n) }
+	ok, n = snake.Check([]byte { 1, 6, 1, 2, 9, 3, 2, 1, 1, 2, 3})
+	if ok { test.Fatal("false positive:", n) }
+}

message.go

@@ -1,52 +0,0 @@
-package hopp
-
-import "fmt"
-import "encoding"
-import "git.tebibyte.media/sashakoshka/hopp/tape"
-
-// Message is any object that can be sent or received over a HOPP connection.
-type Message interface {
-	// Method returns the method number of the message. This must be unique
-	// within the protocol, and should not change between calls.
-	Method() uint16
-	encoding.BinaryMarshaler
-	encoding.BinaryUnmarshaler
-}
-
-var _ Message = new(MessageData)
-
-// MessageData represents a message that organizes its data into table pairs. It
-// can be used to alter a protocol at runtime, transmit data with arbitrary
-// keys, etc. Bear in mind that is less performant than generating code because
-// it has to make extra memory allocations and such.
-type MessageData struct {
-	// Methd holds the method number. This should only be set once.
-	Methd uint16
-	// Pairs maps tags to values.
-	Pairs map[uint16] []byte
-}
-
-// Method returns the message's method field.
-func (this *MessageData) Method() uint16 {
-	return this.Methd
-}
-
-// MarshalBinary implements the [encoding.BinaryMarshaler] interface. The
-// message is encoded using TAPE (Table Pair Encoding).
-func (this *MessageData) MarshalBinary() ([]byte, error) {
-	buffer, err := tape.EncodePairs(this.Pairs)
-	if err != nil { return nil, fmt.Errorf("marshaling MessageData: %w", err) }
-	return buffer, nil
-}
-
-// UnmarshalBinary implements the [encoding.BinaryUnmarshaler] interface. The
-// message is decoded using TAPE (Table Pair Encoding).
-func (this *MessageData) UnmarshalBinary(buffer []byte) error {
-	this.Pairs = make(map[uint16] []byte)
-	pairs, err := tape.DecodePairs(buffer)
-	if err != nil { return fmt.Errorf("unmarshaling MessageData: %w", err) }
-	for key, value := range pairs {
-		this.Pairs[key] = value
-	}
-	return nil
-}

metadapta.go

@@ -1,11 +1,12 @@
 package hopp
 
 import "io"
+import "os"
 import "fmt"
 import "net"
 import "sync"
+import "time"
 import "sync/atomic"
-import "git.tebibyte.media/sashakoshka/hopp/tape"
 import "git.tebibyte.media/sashakoshka/go-util/sync"
 
 // TODO investigate why 30 never reaches the server, causing it to wait for ever
@@ -109,6 +110,10 @@ func (this *a) AcceptTrans() (Trans, error) {
 	}
 }
 
+func (this *a) SetDeadline(t time.Time) error {
+	return this.underlying.SetDeadline(t)
+}
+
 func (this *a) SetSizeLimit(limit int64) {
 	this.sizeLimit = limit
 }
@@ -213,6 +218,10 @@ type transA struct {
 	currentWriter io.Closer
 	writeBuffer   []byte
 	closed        atomic.Bool
+	closeErr      error
+	
+	deadline      *time.Timer
+	deadlineLock  sync.Mutex
 }
 
 func (this *transA) Close() error {
@@ -222,6 +231,11 @@ func (this *transA) Close() error {
 	return err
 }
 
+func (this *transA) closeWithError(err error) error {
+	this.closeErr = err
+	return this.Close()
+}
+
 func (this *transA) closeDontUnlist() (err error) {
 	// MUST be goroutine safe
 	this.incoming.Close()
@@ -270,9 +284,9 @@ func (this *transA) Receive() (method uint16, data []byte, err error) {
 }
 
 func (this *transA) ReceiveReader() (uint16, io.Reader, error) {
-	// if the transaction has been closed, return an io.EOF
-	if this.closed.Load() {
-		return 0, nil, io.EOF
+	// if the transaction has been closed, return an appropriate error.
+	if err := this.errIfClosed(); err != nil {
+		return 0, nil, err
 	}
 
 	// drain previous reader if necessary
@@ -290,6 +304,54 @@ func (this *transA) ReceiveReader() (uint16, io.Reader, error) {
 	return method, reader, nil
 }
 
+func (this *transA) SetDeadline(t time.Time) error {
+	this.deadlineLock.Lock()
+	defer this.deadlineLock.Unlock()
+
+	if t == (time.Time { }) {
+		if this.deadline != nil {
+			this.deadline.Stop()
+		}
+		return nil
+	}
+	
+	until := time.Until(t)
+	if this.deadline == nil {
+		this.deadline.Reset(until)
+		return nil
+	}
+	this.deadline = time.AfterFunc(until, func () {
+		this.closeWithError(os.ErrDeadlineExceeded)
+	})
+	return nil
+}
+
+// TODO
+// func (this *transA) SetReadDeadline(t time.Time) error {
+// 	// TODO
+// }
+// 
+// func (this *transA) SetWriteDeadline(t time.Time) error {
+// 	// TODO
+// }
+
+func (this *transA) errIfClosed() error {
+	if !this.closed.Load() {
+		return nil
+	}
+	return this.bestErr()
+}
+
+func (this *transA) bestErr() error {
+	if this.parent.err != nil {
+		return this.parent.err
+	}
+	if this.closeErr != nil {
+		return this.closeErr
+	}
+	return io.EOF
+}
+
 type readerA struct {
 	parent   *transA
 	leftover []byte
@@ -320,11 +382,7 @@ func (this *readerA) pull() (uint16, error) {
 	// close and return error on failure
 	this.eof = true
 	this.parent.Close()
-	if this.parent.parent.err == nil {
-		return 0, fmt.Errorf("could not receive message: %w", io.EOF)
-	} else {
-		return 0, this.parent.parent.err
-	}
+	return 0, fmt.Errorf("could not receive message: %w", this.parent.bestErr())
 }
 
 func (this *readerA) Read(buffer []byte) (int, error) {
@@ -406,9 +464,9 @@ func encodeMessageA(
 		return ErrPayloadTooLarge
 	}
 	buffer := make([]byte, 18 + len(data))
-	tape.EncodeI64(buffer[:8], trans)
-	tape.EncodeI16(buffer[8:10], method)
-	tape.EncodeI64(buffer[10:18], uint64(len(data)))
+	encodeI64(buffer[:8], trans)
+	encodeI16(buffer[8:10], method)
+	encodeI64(buffer[10:18], uint64(len(data)))
 	copy(buffer[18:], data)
 	_, err := writer.Write(buffer)
 	return err
@@ -427,11 +485,11 @@ func decodeMessageA(
 	headerBuffer := [18]byte { }
 	_, err = io.ReadFull(reader, headerBuffer[:])
 	if err != nil { return 0, 0, false, nil, err }
-	transID, err = tape.DecodeI64[int64](headerBuffer[:8])
+	transID, err = decodeI64[int64](headerBuffer[:8])
 	if err != nil { return 0, 0, false, nil, err }
-	method, err = tape.DecodeI16[uint16](headerBuffer[8:10])
+	method, err = decodeI16[uint16](headerBuffer[8:10])
 	if err != nil { return 0, 0, false, nil, err }
-	size, err := tape.DecodeI64[uint64](headerBuffer[10:18])
+	size, err := decodeI64[uint64](headerBuffer[10:18])
 	if err != nil { return 0, 0, false, nil, err }
 	chunked, size = splitCCBSize(size)
 	if size > uint64(sizeLimit) {

metadaptb.go

@@ -2,10 +2,10 @@ package hopp
 
 import "io"
 import "net"
+import "time"
 import "bytes"
 import "errors"
 import "context"
-import "git.tebibyte.media/sashakoshka/hopp/tape"
 
 // B implements METADAPT-B over a multiplexed stream-oriented transport such as
 // QUIC.
@@ -51,6 +51,10 @@ func (this *b) SetSizeLimit(limit int64) {
 	this.sizeLimit = limit
 }
 
+func (this *b) SetDeadline(t time.Time) error {
+	return this.underlying.SetDeadline(t)
+}
+
 func (this *b) newTrans(underlying Stream) *transB {
 	return &transB {
 		sizeLimit: this.sizeLimit,
@@ -125,6 +129,10 @@ func (this *transB) receiveReader() (uint16, int64, io.Reader, error) {
 	return method, size, data, nil
 }
 
+func (this *transB) SetDeadline(t time.Time) error {
+	return this.underlying.SetDeadline(t)
+}
+
 type writerB struct {
 	parent *transB
 	buffer bytes.Buffer
@@ -150,12 +158,16 @@ type MultiConn interface {
 	AcceptStream(context.Context) (Stream, error)
 	// OpenStream opens a new stream.
 	OpenStream() (Stream, error)
+	// See the documentation for [net.Conn.SetDeadline].
+	SetDeadline(time.Time) error
 }
 
 // Stream represents a single stream returned by a [MultiConn].
 type Stream interface {
 	// See documentation for [net.Conn].
 	io.ReadWriteCloser
+	// See the documentation for [net.Conn.SetDeadline].
+	SetDeadline(time.Time) error
 	// ID returns the stream ID
 	ID() int64
 }
@@ -165,8 +177,8 @@ func encodeMessageB(writer io.Writer, sizeLimit int64, method uint16, data []byt
 		return ErrPayloadTooLarge
 	}
 	buffer := make([]byte, 10 + len(data))
-	tape.EncodeI16(buffer[:2], method)
-	tape.EncodeI64(buffer[2:10], uint64(len(data)))
+	encodeI16(buffer[:2], method)
+	encodeI64(buffer[2:10], uint64(len(data)))
 	copy(buffer[10:], data)
 	_, err := writer.Write(buffer)
 	return err
@@ -187,9 +199,9 @@ func decodeMessageB(
 		if errors.Is(err, io.EOF) { return 0, 0, nil, io.ErrUnexpectedEOF }
 		return 0, 0, nil, err
 	}
-	method, err = tape.DecodeI16[uint16](headerBuffer[:2])
+	method, err = decodeI16[uint16](headerBuffer[:2])
 	if err != nil { return 0, 0, nil, err }
-	length, err := tape.DecodeI64[uint64](headerBuffer[2:10])
+	length, err := decodeI64[uint64](headerBuffer[2:10])
 	if err != nil { return 0, 0, nil, err }
 	if length > uint64(sizeLimit) {
 		return 0, 0, nil, ErrPayloadTooLarge

tape/decode.go

@@ -0,0 +1,192 @@
+package tape
+
+import "io"
+import "math"
+import "bufio"
+
+// Decodable is any type that can decode itself from a decoder.
+type Decodable interface {
+	// Decode reads data from decoder, replacing the data of the object. It
+	// returns the amount of bytes written, and an error if the write
+	// stopped early.
+	Decode(decoder *Decoder) (n int, err error)
+}
+
+// Decoder decodes data from an [io.Reader].
+type Decoder struct {
+	bufio.Reader
+}
+
+// NewDecoder creates a new decoder that reads from reader.
+func NewDecoder(reader io.Reader) *Decoder {
+	decoder := &Decoder { }
+	decoder.Reader.Reset(reader)
+	return decoder
+}
+
+// ReadFull calls [io.ReadFull] on the reader.
+func (this *Decoder) ReadFull(buffer []byte) (n int, err error) {
+	return io.ReadFull(this, buffer)
+}
+
+// ReadInt8 decodes an 8-bit signed integer from the input reader.
+func (this *Decoder) ReadInt8() (value int8, n int, err error) {
+	uncasted, n, err := this.ReadUint8()
+	return int8(uncasted), n, err
+}
+
+// ReadUint8 decodes an 8-bit unsigned integer from the input reader.
+func (this *Decoder) ReadUint8() (value uint8, n int, err error) {
+	buffer := [1]byte { }
+	n, err = this.ReadFull(buffer[:])
+	return uint8(buffer[0]), n, err
+}
+
+// ReadInt16 decodes an 16-bit signed integer from the input reader.
+func (this *Decoder) ReadInt16() (value int16, n int, err error) {
+	uncasted, n, err := this.ReadUint16()
+	return int16(uncasted), n, err
+}
+
+// ReadUint16 decodes an 16-bit unsigned integer from the input reader.
+func (this *Decoder) ReadUint16() (value uint16, n int, err error) {
+	buffer := [2]byte { }
+	n, err = this.ReadFull(buffer[:])
+	return uint16(buffer[0]) << 8 |
+		uint16(buffer[1]), n, err
+}
+
+// ReadInt32 decodes an 32-bit signed integer from the input reader.
+func (this *Decoder) ReadInt32() (value int32, n int, err error) {
+	uncasted, n, err := this.ReadUint32()
+	return int32(uncasted), n, err
+}
+
+// ReadUint32 decodes an 32-bit unsigned integer from the input reader.
+func (this *Decoder) ReadUint32() (value uint32, n int, err error) {
+	buffer := [4]byte { }
+	n, err = this.ReadFull(buffer[:])
+	return uint32(buffer[0]) << 24 |
+		uint32(buffer[1]) << 16 |
+		uint32(buffer[2]) << 8 |
+		uint32(buffer[3]), n, err
+}
+
+// ReadInt64 decodes an 64-bit signed integer from the input reader.
+func (this *Decoder) ReadInt64() (value int64, n int, err error) {
+	uncasted, n, err := this.ReadUint64()
+	return int64(uncasted), n, err
+}
+
+// ReadUint64 decodes an 64-bit unsigned integer from the input reader.
+func (this *Decoder) ReadUint64() (value uint64, n int, err error) {
+	buffer := [8]byte { }
+	n, err = this.ReadFull(buffer[:])
+	return uint64(buffer[0]) << 56 |
+		uint64(buffer[1]) << 48 |
+		uint64(buffer[2]) << 40 |
+		uint64(buffer[3]) << 32 |
+		uint64(buffer[4]) << 24 |
+		uint64(buffer[5]) << 16 |
+		uint64(buffer[6]) << 8 |
+		uint64(buffer[7]), n, err
+}
+
+// ReadIntN decodes an N-byte signed integer from the input reader.
+func (this *Decoder) ReadIntN(bytes int) (value int64, n int, err error) {
+	uncasted, n, err := this.ReadUintN(bytes)
+	return int64(uncasted), n, err
+}
+
+// ReadUintN decodes an N-byte unsigned integer from the input reader.
+func (this *Decoder) ReadUintN(bytes int) (value uint64, n int, err error) {
+	// TODO: don't make multiple read calls (without allocating)
+	buffer := [1]byte { }
+	for bytesLeft := bytes; bytesLeft > 0; bytesLeft -- {
+		nn, err := this.ReadFull(buffer[:])
+		n += nn; if err != nil { return 0, n, err }
+		value |= uint64(buffer[0]) << ((bytesLeft - 1) * 8)
+	}
+	// *read* integers too big, but don't return them.
+	if bytes > 8 { value = 0 }
+	return value, n, nil
+}
+
+// ReadFloat16 decodes a 16-bit floating point value from the input reader.
+func (this *Decoder) ReadFloat16() (value float32, n int, err error) {
+	bits, nn, err := this.ReadUint16()
+	n += nn; if err != nil { return 0, n, err }
+	return math.Float32frombits(f16bitsToF32bits(bits)), n, nil
+}
+
+// ReadFloat32 decldes a 32-bit floating point value from the input reader.
+func (this *Decoder) ReadFloat32() (value float32, n int, err error) {
+	bits, nn, err := this.ReadUint32()
+	n += nn; if err != nil { return 0, n, err }
+	return math.Float32frombits(bits), n, nil
+}
+
+// ReadFloat64 decldes a 64-bit floating point value from the input reader.
+func (this *Decoder) ReadFloat64() (value float64, n int, err error) {
+	bits, nn, err := this.ReadUint64()
+	n += nn; if err != nil { return 0, n, err }
+	return math.Float64frombits(bits), n, nil
+}
+
+// ReadTag decodes a [Tag] from the input reader.
+func (this *Decoder) ReadTag() (value Tag, n int, err error) {
+	uncasted, nn, err := this.ReadUint8()
+	n += nn; if err != nil { return 0, n, err }
+	return Tag(uncasted), n, nil
+}
+
+// f16bitsToF32bits returns uint32 (float32 bits) converted from specified uint16.
+// Taken from https://github.com/x448/float16/blob/v0.8.4/float16
+//
+// MIT License
+// 
+// Copyright (c) 2019 Montgomery Edwards⁴⁴⁸ and Faye Amacker
+// 
+// Permission is hereby granted, free of charge, to any person obtaining a copy
+// of this software and associated documentation files (the "Software"), to deal
+// in the Software without restriction, including without limitation the rights
+// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+// copies of the Software, and to permit persons to whom the Software is
+// furnished to do so, subject to the following conditions:
+// 
+// The above copyright notice and this permission notice shall be included in all
+// copies or substantial portions of the Software.
+func f16bitsToF32bits(in uint16) uint32 {
+	// All 65536 conversions with this were confirmed to be correct
+	// by Montgomery Edwards⁴⁴⁸ (github.com/x448).
+	
+	sign := uint32(in&0x8000) << 16 // sign for 32-bit
+	exp := uint32(in&0x7c00) >> 10  // exponenent for 16-bit
+	coef := uint32(in&0x03ff) << 13 // significand for 32-bit
+
+	if exp == 0x1f {
+		if coef == 0 {
+			// infinity
+			return sign | 0x7f800000 | coef
+		}
+		// NaN
+		return sign | 0x7fc00000 | coef
+	}
+
+	if exp == 0 {
+		if coef == 0 {
+			// zero
+			return sign
+		}
+
+		// normalize subnormal numbers
+		exp++
+		for coef&0x7f800000 == 0 {
+			coef <<= 1
+			exp--
+		}
+		coef &= 0x007fffff
+	}
+
+	return sign | ((exp + (0x7f - 0xf)) << 23) | coef
+}

tape/dynamic.go

@@ -0,0 +1,505 @@
+package tape
+
+// dont smoke reflection, kids!!!!!!!!!
+// totally reflectric, reflectrified, etc. this is probably souper slow but
+// certainly no slower than the built in json encoder i'd imagine.
+// TODO: add support for struct tags: `tape:"0000"`, tape:"0001"` so they can get
+// transformed into tables with a defined schema
+
+// TODO: test all of these smaller functions individually
+
+import "fmt"
+import "reflect"
+
+var dummyMap map[uint16] any
+var dummyBuffer []byte
+
+type errCantAssign string
+func (err errCantAssign) Error() string {
+	return string(err)
+}
+func errCantAssignf(format string, v ...any) errCantAssign {
+	return errCantAssign(fmt.Sprintf(format, v...))
+}
+
+// EncodeAny encodes an "any" value. Returns an error if the underlying type is
+// unsupported. Supported types are:
+//
+//   - int
+//   - int<N>
+//   - uint
+//   - uint<N>
+//   - string
+//   - []<supported type>
+//   - map[uint16]<supported type>
+func EncodeAny(encoder *Encoder, value any, tag Tag) (n int, err error) {
+	// primitives
+	reflectValue := reflect.ValueOf(value)
+	switch reflectValue.Kind() {
+	case reflect.Int:    return encoder.WriteInt32(int32(reflectValue.Int()))
+	case reflect.Uint:   return encoder.WriteUint32(uint32(reflectValue.Uint()))
+	case reflect.Int8:   return encoder.WriteInt8(int8(reflectValue.Int()))
+	case reflect.Uint8:  return encoder.WriteUint8(uint8(reflectValue.Uint()))
+	case reflect.Int16:  return encoder.WriteInt16(int16(reflectValue.Int()))
+	case reflect.Uint16: return encoder.WriteUint16(uint16(reflectValue.Uint()))
+	case reflect.Int32:  return encoder.WriteInt32(int32(reflectValue.Int()))
+	case reflect.Uint32: return encoder.WriteUint32(uint32(reflectValue.Uint()))
+	case reflect.Int64:  return encoder.WriteInt64(int64(reflectValue.Int()))
+	case reflect.Uint64: return encoder.WriteUint64(uint64(reflectValue.Uint()))
+	case reflect.String:
+		if reflectValue.Len() > MaxStructureLength {
+			return 0, ErrTooLong
+		}
+		return EncodeAny(encoder, []byte(reflectValue.String()), tag)
+	}
+	if reflectValue.CanConvert(reflect.TypeOf(dummyBuffer)) {
+		if reflectValue.Len() > MaxStructureLength {
+			return 0, ErrTooLong
+		}
+		if tag.Is(LBA) {
+			nn, err := encoder.WriteUintN(uint64(reflectValue.Len()), tag.CN() + 1)
+			n += nn; if err != nil { return n, err }
+		}
+		nn, err := encoder.Write(reflectValue.Bytes())
+		n += nn; if err != nil { return n, err }
+		return n, nil
+	}
+
+	// aggregates
+	reflectType := reflect.TypeOf(value)
+	switch reflectType.Kind() {
+	case reflect.Slice:
+		return encodeAnySlice(encoder, value, tag)
+	// case reflect.Array:
+		// TODO: we can encode arrays. but can we decode into them?
+		// that's the fucken question. maybe we just do the first
+		// return encodeAnySlice(encoder, reflect.ValueOf(value).Slice(0, reflectType.Len()).Interface(), tag)
+	case reflect.Map:
+		if reflectValue.Len() > MaxStructureLength {
+			return 0, ErrTooLong
+		}
+		if reflectType.Key() == reflect.TypeOf(uint16(0)) {
+			return encodeAnyMap(encoder, value, tag)
+		}
+		return n, fmt.Errorf("cannot encode map key %T, key must be uint16", value)
+	}
+	return n, fmt.Errorf("cannot encode type %T", value)
+}
+
+// DecodeAny decodes data and places it into destination, which must be a
+// pointer to a supported type. See [EncodeAny] for a list of supported types.
+func DecodeAny(decoder *Decoder, destination any, tag Tag) (n int, err error) {
+	reflectDestination := reflect.ValueOf(destination)
+	if reflectDestination.Kind() != reflect.Pointer {
+		return n, fmt.Errorf("expected pointer destination, not %v", destination)
+	}
+	return decodeAny(decoder, reflectDestination.Elem(), tag)
+}
+
+// unknownSlicePlaceholder is inserted by skeletonValue and informs the program
+// that the destination for the slice needs to be generated based on the item
+// tag in the OTA.
+type unknownSlicePlaceholder struct { }
+var unknownSlicePlaceholderType = reflect.TypeOf(unknownSlicePlaceholder { })
+
+// decodeAny is internal to [DecodeAny]. It takes in an addressable
+// [reflect.Value] as the destination. If the decoded value cannot fit in the
+// destination, it skims over the payload, leaves the destination empty, and
+// returns without an error.
+func decodeAny(decoder *Decoder, destination reflect.Value, tag Tag) (n int, err error) {
+	n, err = decodeAnyOrError(decoder, destination, tag)
+	if _, ok := err.(errCantAssign); ok {
+		if n > 0 { panic(fmt.Sprintf("decodeAnyOrError decoded more than it should: %d", n)) }
+		nn, err := Skim(decoder, tag)
+		n += nn; if err != nil { return n, err }
+		return n, nil
+	}
+	return n, err
+}
+
+// decodeAnyOrError is internal to [decodeAny]. It takes in an addressable
+// [reflect.Value] as the destination. If the decoded value cannot fit in the
+// destination, it decodes nothing and returns an error of type errCantAssign,
+// except for the case of a mismatched OTA element tag, wherein it will skim
+// over the rest of the payload, leave the destination empty, and return without
+// an error.
+func decodeAnyOrError(decoder *Decoder, destination reflect.Value, tag Tag) (n int, err error) {
+	err = canSet(destination.Type(), tag)
+	if err != nil { return n, err }
+	
+	switch tag.WithoutCN() {
+	case SI:
+		// SI: (none)
+		setInt(destination, uint64(tag.CN()))
+	case LI:
+		// LI: <value: IntN>
+		nn, err := decodeAndSetUint(decoder, destination, tag.CN() + 1)
+		n += nn; if err != nil { return n, err }
+	case LSI:
+		// LSI: <value: IntN>
+		nn, err := decodeAndSetInt(decoder, destination, tag.CN() + 1)
+		n += nn; if err != nil { return n, err }
+	case FP:
+		// FP: <value: FloatN>
+		nn, err := decodeAndSetFloat(decoder, destination, tag.CN() + 1)
+		n += nn; if err != nil { return n, err }
+	case SBA:
+		// SBA: <data: U8>*
+		length := tag.CN()
+		if length > MaxStructureLength {
+			return 0, ErrTooLong
+		}
+		buffer := make([]byte, length)
+		nn, err := decoder.Read(buffer)
+		n += nn; if err != nil { return n, err }
+		setByteArray(destination, buffer)
+	case LBA:
+		// LBA: <length: UN> <data: U8>*
+		length, nn, err := decoder.ReadUintN(tag.CN() + 1)
+		n += nn; if err != nil { return n, err }
+		if length > uint64(MaxStructureLength) {
+			return 0, ErrTooLong
+		}
+		buffer := make([]byte, length)
+		nn, err = decoder.Read(buffer)
+		n += nn; if err != nil { return n, err }
+		setByteArray(destination, buffer)
+	case OTA:
+		// OTA: <length: UN> <elementTag: tape.Tag> <values>*
+		length, nn, err := decoder.ReadUintN(tag.CN() + 1)
+		n += nn; if err != nil { return n, err }
+		if length > uint64(MaxStructureLength) {
+			return 0, ErrTooLong
+		}
+		lengthCast, err := Uint64ToIntSafe(length)
+		if err != nil { return n, err }
+		oneTag, nn, err := decoder.ReadTag()
+		n += nn; if err != nil { return n, err }
+		if destination.Cap() < lengthCast {
+			destination.Grow(lengthCast - destination.Cap())
+		}
+		// skip the rest of the array if the one tag doesn't
+		// match up with the destination
+		err = canSet(destination.Type().Elem(), oneTag)
+		if _, ok := err.(errCantAssign); ok {
+			for _ = range length {
+				nn, err := Skim(decoder, oneTag)
+				n += nn; if err != nil { return n, err }
+			}
+			break
+		}
+		if err != nil { return n, err }
+		destination.SetLen(lengthCast)
+		for index := range length {
+			nn, err := decodeAny(decoder, destination.Index(int(index)), oneTag)
+			n += nn
+			if _, ok := err.(errCantAssign); ok {
+				continue
+			} else if err != nil {
+				return n, err
+			}
+		}
+	case KTV:
+		// KTV: <length: UN> (<key: U16> <tag: Tag> <value>)*
+		length, nn, err := decoder.ReadUintN(tag.CN() + 1)
+		n += nn; if err != nil { return n, err }
+		if length > uint64(MaxStructureLength) {
+			return 0, ErrTooLong
+		}
+		destination.Clear()
+		for _ = range length {
+			key, nn, err := decoder.ReadUint16()
+			n += nn; if err != nil { return n, err }
+			itemTag, nn, err := decoder.ReadTag()
+			n += nn; if err != nil { return n, err }
+			value, err := skeletonValue(decoder, itemTag)
+			if err != nil { return n, err }
+			nn, err = decodeAny(decoder, value.Elem(), itemTag)
+			n += nn; if err != nil { return n, err }
+			destination.SetMapIndex(reflect.ValueOf(key), value.Elem())
+		}
+	default:
+		return n, fmt.Errorf("unknown TN %d", tag.TN())
+	}
+	return n, nil
+}
+
+// TagAny returns the correct tag for an "any" value. Returns an error if the
+// underlying type is unsupported. See [EncodeAny] for a list of supported
+// types.
+func TagAny(value any) (Tag, error) {
+	return tagAny(reflect.ValueOf(value))
+}
+
+func tagAny(reflectValue reflect.Value) (Tag, error) {
+	// primitives
+	switch reflectValue.Kind() {
+	case reflect.Int:    return LSI.WithCN(3), nil
+	case reflect.Int8:   return LSI.WithCN(0), nil
+	case reflect.Int16:  return LSI.WithCN(1), nil
+	case reflect.Int32:  return LSI.WithCN(3), nil
+	case reflect.Int64:  return LSI.WithCN(7), nil
+	case reflect.Uint:   return LI.WithCN(3), nil
+	case reflect.Uint8:  return LI.WithCN(0), nil
+	case reflect.Uint16: return LI.WithCN(1), nil
+	case reflect.Uint32: return LI.WithCN(3), nil
+	case reflect.Uint64: return LI.WithCN(7), nil
+	case reflect.String: return bufferLenTag(reflectValue.Len()), nil
+	}
+	if reflectValue.CanConvert(reflect.TypeOf(dummyBuffer)) {
+		return bufferLenTag(reflectValue.Len()), nil
+	}
+	
+	// aggregates
+	reflectType := reflectValue.Type()
+	switch reflectType.Kind() {
+	case reflect.Slice: return OTA.WithCN(IntBytes(uint64(reflectValue.Len())) - 1), nil
+	case reflect.Array: return OTA.WithCN(reflectType.Len()), nil
+	case reflect.Map:
+		if reflectType.Key() == reflect.TypeOf(uint16(0)) {
+			return KTV.WithCN(IntBytes(uint64(reflectValue.Len())) - 1), nil
+		}
+		return 0, fmt.Errorf("cannot encode map key %v, key must be uint16", reflectType.Key())
+	}
+	return 0, fmt.Errorf("cannot get tag of type %v", reflectType)
+}
+
+func encodeAnySlice(encoder *Encoder, value any, tag Tag) (n int, err error) {
+	// OTA: <length: UN> <elementTag: tape.Tag> <values>*
+	reflectValue := reflect.ValueOf(value)
+	nn, err := encoder.WriteUintN(uint64(reflectValue.Len()), tag.CN() + 1)
+	n += nn; if err != nil { return n, err }
+	reflectType := reflect.TypeOf(value)
+	oneTag, err := tagAny(reflect.Zero(reflectType.Elem()))
+	if err != nil { return n, err }
+	for index := 0; index < reflectValue.Len(); index += 1 {
+		itemTag, err := tagAny(reflectValue.Index(index))
+		if err != nil { return n, err }
+		if itemTag.CN() > oneTag.CN() { oneTag = itemTag }
+	}
+	if oneTag.Is(SBA) { oneTag += 1 << 5 }
+	nn, err = encoder.WriteUint8(uint8(oneTag))
+	n += nn; if err != nil { return n, err }
+	for index := 0; index < reflectValue.Len(); index += 1 {
+		item := reflectValue.Index(index).Interface()
+		nn, err = EncodeAny(encoder, item, oneTag)
+		n += nn; if err != nil { return n, err }
+	}
+	return n, err
+}
+
+func encodeAnyMap(encoder *Encoder, value any, tag Tag) (n int, err error) {
+	// KTV: <length: UN> (<key: U16> <tag: Tag> <value>)*
+	reflectValue := reflect.ValueOf(value)
+	nn, err := encoder.WriteUintN(uint64(reflectValue.Len()), tag.CN() + 1)
+	n += nn; if err != nil { return n, err }
+	iter := reflectValue.MapRange()
+	for iter.Next() {
+		reflectValue := iter.Value().Elem()
+		key          := iter.Key().Interface().(uint16)
+		value        := reflectValue.Interface()
+		nn, err = encoder.WriteUint16(key)
+		n += nn; if err != nil { return n, err }
+		itemTag, err := tagAny(reflectValue)
+		if err != nil { return n, err }
+		nn, err = encoder.WriteUint8(uint8(itemTag))
+		n += nn; if err != nil { return n, err }
+		nn, err = EncodeAny(encoder, value, itemTag)
+		n += nn; if err != nil { return n, err }
+	}
+	return n, nil
+}
+
+func canSet(destination reflect.Type, tag Tag) error {
+	switch tag.WithoutCN() {
+	case SI, LI, LSI:
+		switch destination.Kind() {
+		case
+			reflect.Int, reflect.Int8, reflect.Int16, reflect.Int32, reflect.Int64,
+			reflect.Uint, reflect.Uint8, reflect.Uint16, reflect.Uint32, reflect.Uint64:
+		default:
+			return errCantAssignf("cannot assign integer to %v", destination)
+		}
+	case FP:
+		switch destination.Kind() {
+		case reflect.Float32, reflect.Float64:
+		default:
+			return errCantAssignf("cannot assign float to %v", destination)
+		}
+	case SBA, LBA:
+		if destination.Kind() != reflect.Slice {
+			return errCantAssignf("cannot assign byte array to %v", destination)
+		}
+		if destination.Elem() != reflect.TypeOf(byte(0)) {
+			return errCantAssignf("cannot convert %v to *[]byte", destination)
+		}
+	case OTA:
+		if destination.Kind() != reflect.Slice {
+			return errCantAssignf("cannot assign array to %v", destination)
+		}
+	case KTV:
+		if destination != reflect.TypeOf(dummyMap) {
+			return errCantAssignf("cannot assign table to %v", destination)
+		}
+	default:
+		return fmt.Errorf("unknown TN %d", tag.TN())
+	}
+	return nil
+}
+
+// setInt expects a settable destination.
+func setInt[T int64 | uint64](destination reflect.Value, value T) {
+	switch {
+	case destination.CanInt():
+		destination.Set(reflect.ValueOf(int64(value)).Convert(destination.Type()))
+	case destination.CanUint():
+		destination.Set(reflect.ValueOf(value).Convert(destination.Type()))
+	default:
+		panic("setInt called on an unsupported type")
+	}
+}
+
+// setFloat expects a settable destination.
+func setFloat(destination reflect.Value, value float64) {
+	destination.Set(reflect.ValueOf(value).Convert(destination.Type()))
+}
+
+// setByteArrayexpects a settable destination.
+func setByteArray(destination reflect.Value, value []byte) {
+	destination.Set(reflect.ValueOf(value))
+}
+
+// decodeAndSetInt expects a settable destination.
+func decodeAndSetInt(decoder *Decoder, destination reflect.Value, bytes int) (n int, err error) {
+	value, nn, err := decoder.ReadIntN(bytes)
+	n += nn; if err != nil { return n, err }
+	setInt(destination, value)
+	return n, nil
+}
+
+// decodeAndSetUint expects a settable destination.
+func decodeAndSetUint(decoder *Decoder, destination reflect.Value, bytes int) (n int, err error) {
+	value, nn, err := decoder.ReadUintN(bytes)
+	n += nn; if err != nil { return n, err }
+	setInt(destination, value)
+	return n, nil
+}
+
+// decodeAndSetInt expects a settable destination.
+func decodeAndSetFloat(decoder *Decoder, destination reflect.Value, bytes int) (n int, err error) {
+	switch bytes {
+	case 8:
+		value, nn, err := decoder.ReadFloat64()
+		n += nn; if err != nil { return n, err }
+		setFloat(destination, float64(value))
+		return n, nil
+	case 4:
+		value, nn, err := decoder.ReadFloat32()
+		n += nn; if err != nil { return n, err }
+		setFloat(destination, float64(value))
+		return n, nil
+	}
+	return n, errCantAssignf("unsupported bit width float%d", bytes * 8)
+}
+
+// skeletonValue returns a pointer value. In order for it to be set, it must be
+// dereferenced using Elem().
+func skeletonValue(decoder *Decoder, tag Tag) (reflect.Value, error) {
+	typ, err := typeOf(decoder, tag)
+	if err != nil { return reflect.Value { }, err }
+	return reflect.New(typ), nil
+}
+
+// typeOf returns the type of the current tag being decoded. It does not use up
+// the decoder, it only peeks.
+func typeOf(decoder *Decoder, tag Tag) (reflect.Type, error) {
+	switch tag.WithoutCN() {
+	case SI:
+		return reflect.TypeOf(uint8(0)), nil
+	case LI:
+		switch tag.CN() {
+		case 0: return reflect.TypeOf(uint8(0)), nil
+		case 1: return reflect.TypeOf(uint16(0)), nil
+		case 3: return reflect.TypeOf(uint32(0)), nil
+		case 7: return reflect.TypeOf(uint64(0)), nil
+		}
+		return nil, fmt.Errorf("unknown CN %d for LI", tag.CN())
+	case LSI:
+		switch tag.CN() {
+		case 0: return reflect.TypeOf(int8(0)), nil
+		case 1: return reflect.TypeOf(int16(0)), nil
+		case 3: return reflect.TypeOf(int32(0)), nil
+		case 7: return reflect.TypeOf(int64(0)), nil
+		}
+		return nil, fmt.Errorf("unknown CN %d for LSI", tag.CN())
+	case FP:
+		switch tag.CN() {
+		case 3: return reflect.TypeOf(float32(0)), nil
+		case 7: return reflect.TypeOf(float64(0)), nil
+		}
+		return nil, fmt.Errorf("unknown CN %d for FP", tag.CN())
+	case SBA: return reflect.SliceOf(reflect.TypeOf(byte(0))), nil
+	case LBA: return reflect.SliceOf(reflect.TypeOf(byte(0))), nil
+	case OTA:
+		elemTag, dimension, err := peekSlice(decoder, tag)
+		if err != nil { return nil, err }
+		if elemTag.Is(OTA) { panic("peekSlice cannot return OTA") }
+		typ, err := typeOf(decoder, elemTag)
+		if err != nil { return nil, err }
+		for _ = range dimension {
+			typ = reflect.SliceOf(typ)
+		}
+		return typ, nil
+	case KTV: return reflect.TypeOf(dummyMap), nil
+	}
+	return nil, fmt.Errorf("unknown TN %d", tag.TN())
+}
+
+// peekSlice returns the element tag and dimension count of the OTA currently
+// being decoded. It does not use up the decoder, it only peeks.
+func peekSlice(decoder *Decoder, tag Tag) (Tag, int, error) {
+	offset    := 0
+	dimension := 0
+	currentTag := tag
+	for {
+		elem, populated, n, err := peekSliceOnce(decoder, currentTag, offset)
+		if err != nil { return 0, 0, err }
+		currentTag = elem
+		offset     = n
+		dimension += 1
+		if elem.Is(OTA) {
+			if !populated {
+				// default to a large byte array, will be
+				// interpreted as a string.
+				return LBA, dimension + 1, nil
+			}
+		} else {
+			return elem, dimension, nil
+		}
+	}
+}
+
+// peekSliceOnce returns the element tag of the OTA located offset bytes ahead
+// of the current position. It does not use up the decoder, it only peeks. The n
+// return value denotes how far away from 0 it peeked. If the OTA has more than
+// zero items, populated will be set to true.
+func peekSliceOnce(decoder *Decoder, tag Tag, offset int) (elem Tag, populated bool, n int, err error) {
+	lengthStart  := offset
+	lengthEnd    := lengthStart + tag.CN() + 1
+	elemTagStart := lengthEnd
+	elemTagEnd   := elemTagStart + 1
+
+	headerBytes, err := decoder.Peek(elemTagEnd)
+	if err != nil { return 0, false, 0, err }
+
+	elem = Tag(headerBytes[len(headerBytes) - 1])
+	for index := lengthStart; index < lengthEnd; index += 1 {
+		if headerBytes[index] > 0 {
+			populated = true
+			break
+		}
+	}
+	n = elemTagEnd
+
+	return
+}

tape/dynamic_test.go

@@ -0,0 +1,310 @@
+package tape
+
+import "fmt"
+import "bytes"
+import "testing"
+import "reflect"
+import tu "git.tebibyte.media/sashakoshka/hopp/internal/testutil"
+
+type userDefinedInteger int16
+
+func TestEncodeAnyInt(test *testing.T) {
+	err := testEncodeAny(test, uint8(0xCA), LI.WithCN(0), tu.S(0xCA))
+	if err != nil { test.Fatal(err) }
+	err = testEncodeAny(test, 400, LSI.WithCN(3), tu.S(
+		0, 0, 0x1, 0x90,
+	))
+	if err != nil { test.Fatal(err) }
+}
+
+func TestEncodeAnyTable(test *testing.T) {
+	err := testEncodeAny(test, map[uint16] any {
+		0xF3B9: 1,
+		0x0102: 2,
+		0x0000: "hi!",
+		0xFFFF: []uint16 { 0xBEE5, 0x7777 },
+		0x1234: [][]uint16 { []uint16 { 0x5 }, []uint16 { 0x17, 0xAAAA} },
+		0x2345: [][]int16 { []int16 { 0x5 }, []int16 { 0x17, -0xAAA } },
+		0x3456: userDefinedInteger(0x3921),
+	}, KTV.WithCN(0), tu.S(7).AddVar(
+		[]byte {
+			0xF3, 0xB9,
+			byte(LSI.WithCN(3)),
+			0, 0, 0, 1,
+		},
+		[]byte {
+			0x01, 0x02,
+			byte(LSI.WithCN(3)),
+			0, 0, 0, 2,
+		},
+		[]byte {
+			0, 0,
+			byte(SBA.WithCN(3)),
+			'h', 'i', '!',
+		},
+		[]byte {
+			0xFF, 0xFF,
+			byte(OTA.WithCN(0)), 2, byte(LI.WithCN(1)),
+			0xBE, 0xE5, 0x77, 0x77,
+		},
+		[]byte {
+			0x12, 0x34,
+			byte(OTA.WithCN(0)), 2, byte(OTA.WithCN(0)),
+			1, byte(LI.WithCN(1)),
+			0, 0x5,
+			2, byte(LI.WithCN(1)),
+			0,    0x17,
+			0xAA, 0xAA,
+		},
+		[]byte {
+			0x23, 0x45,
+			byte(OTA.WithCN(0)), 2, byte(OTA.WithCN(0)),
+			1, byte(LSI.WithCN(1)),
+			0, 0x5,
+			2, byte(LSI.WithCN(1)),
+			0,    0x17,
+			0xF5, 0x56,
+		},
+		[]byte {
+			0x34, 0x56,
+			byte(LSI.WithCN(1)),
+			0x39, 0x21,
+		},
+	))
+	if err != nil { test.Fatal(err) }
+}
+
+func TestDecodeWrongType(test *testing.T) {
+	datas := [][]byte {
+		/* int8     */ []byte { byte(LSI.WithCN(0)), 0x45 },
+		/* int16    */ []byte { byte(LSI.WithCN(1)), 0x45, 0x67 },
+		/* int32    */ []byte { byte(LSI.WithCN(3)), 0x45, 0x67, 0x89, 0xAB },
+		/* int64    */ []byte { byte(LSI.WithCN(7)), 0x45, 0x67, 0x89, 0xAB, 0xCD, 0xEF, 0x01, 0x23 },
+		/* uint5    */ []byte { byte(SI.WithCN(12)) },
+		/* uint8    */ []byte { byte(LI.WithCN(0)), 0x45 },
+		/* uint16   */ []byte { byte(LI.WithCN(1)), 0x45, 0x67 },
+		/* uint32   */ []byte { byte(LI.WithCN(3)), 0x45, 0x67, 0x89, 0xAB },
+		/* uint64   */ []byte { byte(LI.WithCN(7)), 0x45, 0x67, 0x89, 0xAB, 0xCD, 0xEF, 0x01, 0x23 },
+		/* string   */ []byte { byte(SBA.WithCN(7)), 'p', 'u', 'p', 'e', 'v', 'e', 'r' },
+		/* []byte   */ []byte { byte(SBA.WithCN(5)), 'b', 'l', 'a', 'r', 'g' },
+		/* []string */ []byte {
+			byte(OTA.WithCN(0)), 2, byte(LBA.WithCN(0)),
+			0x08, 0x45, 0x67, 0x89, 0xAB, 0xCD, 0xEF, 0x01, 0x23,
+			0x05, 0x11, 0x11, 0x11, 0x11, 0x11,
+		},
+		/* map[uint16] any */ []byte {
+			byte(KTV.WithCN(0)), 2,
+			0x02, 0x23, byte(LSI.WithCN(1)), 0x45, 0x67,
+			0x02, 0x23, byte(LI.WithCN(3)),  0x45, 0x67, 0x89, 0xAB,
+		},
+	}
+
+	for index, data := range datas {
+		test.Logf("data %2d %v [%s]", index, Tag(data[0]), tu.HexBytes(data[1:]))
+		// integers should only assign to other integers
+		if index > 8 {
+			cas := func(destination any) {
+				n, err := DecodeAny(NewDecoder(bytes.NewBuffer(data[1:])), destination, Tag(data[0]))
+				if err != nil { test.Fatalf("error: %v | n: %d", err, n) }
+				reflectValue := reflect.ValueOf(destination).Elem()
+				if reflectValue.CanInt() {
+					if reflectValue.Int() != 0 {
+						test.Fatalf("destination not zero: %v", reflectValue.Elem().Interface())
+					}
+				} else {
+					if reflectValue.Uint() != 0 {
+						test.Fatalf("destination not zero: %v", reflectValue.Elem().Interface())
+					}
+				}
+				if n != len(data) - 1 {
+					test.Fatalf("n not equal: %d != %d", n, len(data) - 1)
+				}
+			}
+			test.Log("- int8")
+			{ var dest int8;   cas(&dest) }
+			test.Log("- int16")
+			{ var dest int16;  cas(&dest) }
+			test.Log("- int32")
+			{ var dest int32;  cas(&dest) }
+			test.Log("- int64")
+			{ var dest int64;  cas(&dest) }
+			test.Log("- uint8")
+			{ var dest uint8;  cas(&dest) }
+			test.Log("- uint16")
+			{ var dest uint16; cas(&dest) }
+			test.Log("- uint32")
+			{ var dest uint32; cas(&dest) }
+			test.Log("- uint64")
+			{ var dest uint64; cas(&dest) }
+		}
+		arrayCase := func(destination any) {
+			n, err := DecodeAny(NewDecoder(bytes.NewBuffer(data[1:])), destination, Tag(data[0]))
+			if err != nil { test.Fatalf("error: %v | n: %d", err, n) }
+			reflectDestination := reflect.ValueOf(destination)
+			reflectValue := reflectDestination.Elem()
+			if reflectValue.Len() != 0 {
+				test.Fatalf("len(destination) not zero: %v", reflectValue.Interface())
+			}
+			if n != len(data) - 1 {
+				test.Fatalf("n not equal: %d != %d", n, len(data) - 1)
+			}
+		}
+		// SBA/LBA types should only assign to other SBA/LBA types
+		if index != 9 && index != 10 {
+			test.Log("- string")
+			{ var dest string; arrayCase(&dest) }
+			test.Log("- []byte")
+			{ var dest []byte; arrayCase(&dest) }
+		}
+		// arrays should only assign to other arrays
+		if index != 11 {
+			test.Log("- []string")
+			{ var dest []string; arrayCase(&dest) }
+		}
+		// tables should only assign to other tables
+		if index != 12 {
+			test.Log("- map[uint16] any")
+			{ var dest = map[uint16] any { }; arrayCase(&dest) }
+		}
+	}
+}
+
+func TestEncodeDecodeAnyTable(test *testing.T) {
+	err := testEncodeDecodeAny(test, map[uint16] any {
+		0xF3B9: uint32(1),
+		0x0102: uint32(2),
+		0x0103: int64(23432),
+		0x0104: int64(-88777),
+		0x0000: []byte("hi!"),
+		0xFFFF: []uint16 { 0xBEE5, 0x7777 },
+		0x1234: [][]uint16 { []uint16 { 0x5 }, []uint16 { 0x17, 0xAAAA} },
+	}, nil)
+	if err != nil { test.Fatal(err) }
+}
+
+func TestPeekSlice(test *testing.T) {
+	buffer := bytes.NewBuffer([]byte {
+		2, byte(OTA.WithCN(3)),
+		0, 0, 0, 1, byte(LI.WithCN(1)),
+		0, 0x5,
+		2, byte(LI.WithCN(1)),
+		0,    0x17,
+		0xAA, 0xAA,
+	})
+	decoder := NewDecoder(buffer)
+
+	elem, dimension, err := peekSlice(decoder, OTA.WithCN(0))
+	if err != nil { test.Fatal(err) }
+	if elem != LI.WithCN(1) {
+		test.Fatalf("wrong element tag: %v %02X", elem, byte(elem))
+	}
+	if got, correct := dimension, 2; got != correct {
+		test.Fatalf("wrong dimension: %d != %d", got, correct)
+	}
+}
+
+func TestPeekSliceOnce(test *testing.T) {
+	buffer := bytes.NewBuffer([]byte {
+		2, byte(OTA.WithCN(3)),
+		0, 0, 0, 1, byte(LI.WithCN(1)),
+		0, 0x5,
+		2, byte(LI.WithCN(1)),
+		0,    0x17,
+		0xAA, 0xAA,
+	})
+	decoder := NewDecoder(buffer)
+
+	test.Log("--- stage 1")
+	elem, populated, n, err := peekSliceOnce(decoder, OTA.WithCN(0), 0)
+	if err != nil { test.Fatal(err) }
+	if elem != OTA.WithCN(3) {
+		test.Fatal("wrong element tag:", elem)
+	}
+	if !populated {
+		test.Fatal("wrong populated:", populated)
+	}
+	if got, correct := n, 2; got != correct {
+		test.Fatalf("wrong n: %d != %d", got, correct)
+	}
+	
+	test.Log("--- stage 2")
+	elem, populated, n, err = peekSliceOnce(decoder, elem, n)
+	if err != nil { test.Fatal(err) }
+	if elem != LI.WithCN(1) {
+		test.Fatal("wrong element tag:", elem)
+	}
+	if !populated {
+		test.Fatal("wrong populated:", populated)
+	}
+	if got, correct := n, 7; got != correct {
+		test.Fatalf("wrong n: %d != %d", got, correct)
+	}
+}
+
+func encAny(value any) ([]byte, Tag, int, error) {
+	tag, err := TagAny(value)
+	if err != nil { return nil, 0, 0, err }
+	buffer := bytes.Buffer { }
+	encoder := NewEncoder(&buffer)
+	n, err := EncodeAny(encoder, value, tag)
+	if err != nil { return nil, 0, n, err }
+	encoder.Flush()
+	return buffer.Bytes(), tag, n, nil
+}
+
+func decAny(data []byte) (Tag, any, int, error) {
+	destination := map[uint16] any { }
+	tag, err := TagAny(destination)
+	if err != nil { return 0, nil, 0, err }
+	n, err := DecodeAny(NewDecoder(bytes.NewBuffer(data)), &destination, tag)
+	if err != nil { return 0, nil, n, err }
+	return tag, destination, n, nil
+}
+
+func testEncodeAny(test *testing.T, value any, correctTag Tag, correctBytes tu.Snake) error {
+	bytes, tag, n, err := encAny(value)
+	if err != nil { return err }
+	test.Log("n:      ", n)
+	test.Log("tag:    ", tag)
+	test.Log("got:    ", tu.HexBytes(bytes))
+	test.Log("correct:", correctBytes)
+	if tag != correctTag {
+		return fmt.Errorf("tag not equal: %v != %v", tag, correctTag)
+	}
+	if ok, n := correctBytes.Check(bytes); !ok {
+		return fmt.Errorf("bytes not equal at index %d", n)
+	}
+	if n != len(bytes) {
+		return fmt.Errorf("n not equal: %d != %d", n, len(bytes))
+	}
+	return nil
+}
+
+func testEncodeDecodeAny(test *testing.T, value, correctValue any) error {
+	if correctValue == nil { 
+		correctValue = value
+	}
+
+	test.Log("encoding...")
+	bytes, tag, n, err := encAny(value)
+	if err != nil { return err }
+	test.Log("n:  ", n)
+	test.Log("tag:", tag)
+	test.Log("got:", tu.HexBytes(bytes))
+	test.Log("decoding...", tag)
+	if n != len(bytes) {
+		return fmt.Errorf("n not equal: %d != %d", n, len(bytes))
+	}
+	
+	_, decoded, n, err := decAny(bytes)
+	if err != nil { return err }
+	test.Log("got:    ", tu.Describe(decoded))
+	test.Log("correct:", tu.Describe(correctValue))
+	if !reflect.DeepEqual(decoded, correctValue) {
+		return fmt.Errorf("values not equal")
+	}
+	if n != len(bytes) {
+		return fmt.Errorf("n not equal: %d != %d", n, len(bytes))
+	}
+	return nil
+}

tape/encode.go

@@ -0,0 +1,189 @@
+package tape
+
+import "io"
+import "math"
+import "bufio"
+
+// Encodable is any type that can write itself to an encoder.
+type Encodable interface {
+	// Encode sends data to encoder. It returns the amount of bytes written,
+	// and an error if the write stopped early.
+	Encode(encoder *Encoder) (n int, err error)
+}
+	
+// Encoder encodes data to an io.Writer.
+type Encoder struct {
+	bufio.Writer
+}
+
+// NewEncoder creates a new encoder that writes to writer.
+func NewEncoder(writer io.Writer) *Encoder {
+	encoder := &Encoder { }
+	encoder.Reset(writer)
+	return encoder
+}
+
+// WriteInt8 encodes an 8-bit signed integer to the output writer.
+func (this *Encoder) WriteInt8(value int8) (n int, err error) {
+	return this.WriteUint8(uint8(value))
+}
+
+// WriteUint8 encodes an 8-bit unsigned integer to the output writer.
+func (this *Encoder) WriteUint8(value uint8) (n int, err error) {
+	return this.Write([]byte { byte(value) })
+}
+
+// WriteInt16 encodes an 16-bit signed integer to the output writer.
+func (this *Encoder) WriteInt16(value int16) (n int, err error) {
+	return this.WriteUint16(uint16(value))
+}
+
+// WriteUint16 encodes an 16-bit unsigned integer to the output writer.
+func (this *Encoder) WriteUint16(value uint16) (n int, err error) {
+	return this.Write([]byte {
+		byte(value >> 8),
+		byte(value),
+	})
+}
+
+// WriteInt32 encodes an 32-bit signed integer to the output writer.
+func (this *Encoder) WriteInt32(value int32) (n int, err error) {
+	return this.WriteUint32(uint32(value))
+}
+
+// WriteUint32 encodes an 32-bit unsigned integer to the output writer.
+func (this *Encoder) WriteUint32(value uint32) (n int, err error) {
+	return this.Write([]byte {
+		byte(value >> 24),
+		byte(value >> 16),
+		byte(value >> 8),
+		byte(value),
+	})
+}
+
+// WriteInt64 encodes an 64-bit signed integer to the output writer.
+func (this *Encoder) WriteInt64(value int64) (n int, err error) {
+	return this.WriteUint64(uint64(value))
+}
+
+// WriteUint64 encodes an 64-bit unsigned integer to the output writer.
+func (this *Encoder) WriteUint64(value uint64) (n int, err error) {
+	return this.Write([]byte {
+		byte(value >> 56),
+		byte(value >> 48),
+		byte(value >> 40),
+		byte(value >> 32),
+		byte(value >> 24),
+		byte(value >> 16),
+		byte(value >> 8),
+		byte(value),
+	})
+}
+
+// WriteIntN encodes an N-byte signed integer to the output writer.
+func (this *Encoder) WriteIntN(value int64, bytes int) (n int, err error) {
+	return this.WriteUintN(uint64(value), bytes)
+}
+
+// for Write/ReadUintN, increase buffers if go somehow gets support for over 64
+// bit integers. we could also make an expanding int type in goutil to use here,
+// or maybe there is one in the stdlib. keep the int64 versions as well though
+// because its ergonomic.
+	
+// WriteUintN encodes an N-byte unsigned integer to the output writer.
+func (this *Encoder) WriteUintN(value uint64, bytes int) (n int, err error) {
+	// TODO: don't make multiple write calls (without allocating)
+	buffer := [1]byte { }
+	for bytesLeft := bytes; bytesLeft > 0; bytesLeft -- {
+		buffer[0] = byte(value) >> ((bytesLeft - 1) * 8)
+		nn, err := this.Write(buffer[:])
+		n += nn; if err != nil { return n, err }
+	}
+	return n, nil
+}
+
+// WriteFloat16 encodes a 16-bit floating point value to the output writer.
+func (this *Encoder) WriteFloat16(value float32) (n int, err error) {
+	return this.WriteUint16(f32bitsToF16bits(math.Float32bits(value)))
+}
+
+// WriteFloat32 encodes a 32-bit floating point value to the output writer.
+func (this *Encoder) WriteFloat32(value float32) (n int, err error) {
+	return this.WriteUint32(math.Float32bits(value))
+}
+
+// WriteFloat64 encodes a 64-bit floating point value to the output writer.
+func (this *Encoder) WriteFloat64(value float64) (n int, err error) {
+	return this.WriteUint64(math.Float64bits(value))
+}
+
+// WriteTag encodes a [Tag] to the output writer.
+func (this *Encoder) WriteTag(value Tag) (n int, err error) {
+	return this.WriteUint8(uint8(value))
+}
+
+// f32bitsToF16bits returns uint16 (Float16 bits) converted from the specified float32.
+// Conversion rounds to nearest integer with ties to even.
+// Taken from https://github.com/x448/float16/blob/v0.8.4/float16
+//
+// MIT License
+// 
+// Copyright (c) 2019 Montgomery Edwards⁴⁴⁸ and Faye Amacker
+// 
+// Permission is hereby granted, free of charge, to any person obtaining a copy
+// of this software and associated documentation files (the "Software"), to deal
+// in the Software without restriction, including without limitation the rights
+// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+// copies of the Software, and to permit persons to whom the Software is
+// furnished to do so, subject to the following conditions:
+// 
+// The above copyright notice and this permission notice shall be included in all
+// copies or substantial portions of the Software.
+func f32bitsToF16bits(u32 uint32) uint16 {
+	// Translated from Rust to Go by Montgomery Edwards⁴⁴⁸ (github.com/x448).
+	// All 4294967296 conversions with this were confirmed to be correct by x448.
+	// Original Rust implementation is by Kathryn Long (github.com/starkat99) with MIT license.
+
+	sign := u32 & 0x80000000
+	exp := u32 & 0x7f800000
+	coef := u32 & 0x007fffff
+
+	if exp == 0x7f800000 {
+		// NaN or Infinity
+		nanBit := uint32(0)
+		if coef != 0 {
+			nanBit = uint32(0x0200)
+		}
+		return uint16((sign >> 16) | uint32(0x7c00) | nanBit | (coef >> 13))
+	}
+
+	halfSign := sign >> 16
+
+	unbiasedExp := int32(exp>>23) - 127
+	halfExp := unbiasedExp + 15
+
+	if halfExp >= 0x1f {
+		return uint16(halfSign | uint32(0x7c00))
+	}
+
+	if halfExp <= 0 {
+		if 14-halfExp > 24 {
+			return uint16(halfSign)
+		}
+		coef := coef | uint32(0x00800000)
+		halfCoef := coef >> uint32(14-halfExp)
+		roundBit := uint32(1) << uint32(13-halfExp)
+		if (coef&roundBit) != 0 && (coef&(3*roundBit-1)) != 0 {
+			halfCoef++
+		}
+		return uint16(halfSign | halfCoef)
+	}
+
+	uHalfExp := uint32(halfExp) << 10
+	halfCoef := coef >> 13
+	roundBit := uint32(0x00001000)
+	if (coef&roundBit) != 0 && (coef&(3*roundBit-1)) != 0 {
+		return uint16((halfSign | uHalfExp | halfCoef) + 1)
+	}
+	return uint16(halfSign | uHalfExp | halfCoef)
+}

tape/error.go

@@ -0,0 +1,12 @@
+package tape
+
+// Error enumerates common errors in this package.
+type Error string; const (
+	ErrTooLong  Error = "data structure too long"
+	ErrTooLarge Error = "number too large"
+)
+
+// Error implements the error interface.
+func (err Error) Error() string {
+	return string(err)
+}

tape/limits.go

@@ -0,0 +1,26 @@
+package tape
+
+// MaxStructureLength determines how long a TAPE data structure can be. This
+// applies to:
+//
+//  - OTA
+//  - SBA/LBA
+//  - KTV
+//
+// By default it is set at 2^20 (about a million).
+// You shouldn't need to change this. If you do, it should only be set once at
+// the start of the program.
+var MaxStructureLength = 1024 * 1024
+
+// MaxInt is the maximum value an int can hold. This varies depending on the
+// system.
+const MaxInt int = int(^uint(0) >> 1)
+
+// Uint64ToIntSafe casts the input to an int if it can be done without overflow,
+// or returns an error otherwise.
+func Uint64ToIntSafe(input uint64) (int, error) {
+	if input > uint64(MaxInt) {
+		return 0, ErrTooLarge
+	}
+	return int(input), nil
+}

tape/measure.go

@@ -0,0 +1,12 @@
+package tape
+
+// IntBytes returns the number of bytes required to hold a given unsigned
+// integer.
+func IntBytes(value uint64) int {
+	bytes := 0
+	for value > 0 || bytes == 0 {
+		value >>= 8;
+		bytes ++
+	}
+	return bytes
+}

tape/measure_test.go

@@ -0,0 +1,21 @@
+package tape
+
+import "testing"
+
+func TestIntBytes(test *testing.T) {
+	if correct, got := 1, IntBytes(0); correct != got {
+		test.Fatal("wrong:", got)
+	}
+	if correct, got := 1, IntBytes(1); correct != got {
+		test.Fatal("wrong:", got)
+	}
+	if correct, got := 1, IntBytes(16); correct != got {
+		test.Fatal("wrong:", got)
+	}
+	if correct, got := 1, IntBytes(255); correct != got {
+		test.Fatal("wrong:", got)
+	}
+	if correct, got := 2, IntBytes(256); correct != got {
+		test.Fatal("wrong:", got)
+	}
+}

tape/skim.go

@@ -0,0 +1,54 @@
+package tape
+
+import "fmt"
+
+// Skim uses up data from a decoder to "skim" over one value (and all else
+// contained within it) without actually putting the data anywhere.
+func Skim(decoder *Decoder, tag Tag) (n int, err error) {
+	switch tag.WithoutCN() {
+	case SI:
+		// SI: (none)
+		return n, nil
+	case LI, LSI, FP:
+		// LI: <value: IntN>
+		// LSI: <value: IntN>
+		// FP: <value: FloatN>
+		nn, err := decoder.Discard(tag.CN() + 1)
+		n += nn; if err != nil { return n, err }
+	case SBA:
+		// SBA: <data: U8>*
+		nn, err := decoder.Discard(tag.CN())
+		n += nn; if err != nil { return n, err }
+	case LBA:
+		// LBA: <length: UN> <data: U8>*
+		length, nn, err := decoder.ReadUintN(tag.CN() + 1)
+		n += nn; if err != nil { return n, err }
+		nn, err = decoder.Discard(int(length))
+		n += nn; if err != nil { return n, err }
+	case OTA:
+		// OTA: <length: UN> <elementTag: tape.Tag> <values>*
+		length, nn, err := decoder.ReadUintN(tag.CN() + 1)
+		n += nn; if err != nil { return n, err }
+		oneTag, nn, err := decoder.ReadTag()
+		n += nn; if err != nil { return n, err }
+		for _ = range length {
+			nn, err := Skim(decoder, oneTag)
+			n += nn; if err != nil { return n, err }
+		}
+	case KTV:
+		// KTV: <length: UN> (<key: U16> <tag: Tag> <value>)*
+		length, nn, err := decoder.ReadUintN(tag.CN() + 1)
+		n += nn; if err != nil { return n, err }
+		for _ = range length {
+			nn, err := decoder.Discard(2)
+			n += nn; if err != nil { return n, err }
+			itemTag, nn, err := decoder.ReadTag()
+			n += nn; if err != nil { return n, err }
+			nn, err = Skim(decoder, itemTag)
+			n += nn; if err != nil { return n, err }
+		}
+	default:
+		return n, fmt.Errorf("unknown TN %d", tag.TN())
+	}
+	return n, nil
+}

tape/skim_test.go

@@ -0,0 +1,137 @@
+package tape
+
+import "bytes"
+import "testing"
+
+func TestSkimInteger(test *testing.T) {
+	data := []byte {
+		0x12, 0x45, 0x23, 0xF9,
+	}
+	mainDataLen := len(data)
+	// extra junk
+	data = append(data, 0x00, 0x01, 0x02, 0x03,)
+
+	n, err := Skim(NewDecoder(bytes.NewBuffer(data)), LI.WithCN(3))
+	if err != nil {
+		test.Fatal(err)
+	}
+	if got, correct := n, mainDataLen; got != correct {
+		test.Fatalf("n not equal: %d != %d", got, correct)
+	}
+}
+
+func TestSkimArray(test *testing.T) {
+	data := []byte {
+		2, byte(LI.WithCN(1)),
+		0xBE, 0xE5, 0x77, 0x77,
+	}
+	mainDataLen := len(data)
+	// extra junk
+	data = append(data, 0x00, 0x01, 0x02, 0x03,)
+
+	n, err := Skim(NewDecoder(bytes.NewBuffer(data)), OTA.WithCN(0))
+	if err != nil {
+		test.Fatal(err)
+	}
+	if got, correct := n, mainDataLen; got != correct {
+		test.Fatalf("n not equal: %d != %d", got, correct)
+	}
+}
+
+func TestSkimNestedArray(test *testing.T) {
+	data := []byte {
+		2, byte(OTA.WithCN(0)),
+		1, byte(LSI.WithCN(1)),
+		0, 0x5,
+		2, byte(LSI.WithCN(1)),
+		0,    0x17,
+		0xF5, 0x56,
+	}
+	mainDataLen := len(data)
+	// extra junk
+	data = append(data, 0x00, 0x01, 0x02, 0x03,)
+
+	n, err := Skim(NewDecoder(bytes.NewBuffer(data)), OTA.WithCN(0))
+	if err != nil {
+		test.Fatal(err)
+	}
+	if got, correct := n, mainDataLen; got != correct {
+		test.Fatalf("n not equal: %d != %d", got, correct)
+	}
+}
+
+func TestSkimTable(test *testing.T) {
+	data := []byte {
+		2,
+		0xF3, 0xB9,
+		byte(LSI.WithCN(3)),
+		0, 0, 0, 1,
+	
+		0x01, 0x02,
+		byte(LSI.WithCN(3)),
+		0, 0, 0, 2,
+	}
+	mainDataLen := len(data)
+	// extra junk
+	data = append(data, 0x00, 0x01, 0x02, 0x03, 0x00, 0x01, 0x02, 0x03, 0x00, 0x01, 0x02, 0x03)
+
+	n, err := Skim(NewDecoder(bytes.NewBuffer(data)), KTV.WithCN(0))
+	if got, correct := n, mainDataLen; got != correct {
+		test.Fatalf("n not equal: %d != %d ... (%d)", got, correct, len(data))
+	}
+	if err != nil {
+		test.Fatal(err)
+	}
+}
+
+func TestSkimTableComplex(test *testing.T) {
+	data := []byte {
+		7,
+		0xF3, 0xB9,
+		byte(LSI.WithCN(3)),
+		0, 0, 0, 1,
+	
+		0x01, 0x02,
+		byte(LSI.WithCN(3)),
+		0, 0, 0, 2,
+	
+		0, 0,
+		byte(SBA.WithCN(3)),
+		'h', 'i', '!',
+	
+		0xFF, 0xFF,
+		byte(OTA.WithCN(0)), 2, byte(LI.WithCN(1)),
+		0xBE, 0xE5, 0x77, 0x77,
+	
+		0x12, 0x34,
+		byte(OTA.WithCN(0)), 2, byte(OTA.WithCN(0)),
+		1, byte(LI.WithCN(1)),
+		0, 0x5,
+		2, byte(LI.WithCN(1)),
+		0,    0x17,
+		0xAA, 0xAA,
+	
+		0x23, 0x45,
+		byte(OTA.WithCN(0)), 2, byte(OTA.WithCN(0)),
+		1, byte(LSI.WithCN(1)),
+		0, 0x5,
+		2, byte(LSI.WithCN(1)),
+		0,    0x17,
+		0xF5, 0x56,
+	
+		0x34, 0x56,
+		byte(LSI.WithCN(1)),
+		0x39, 0x21,
+	}
+	mainDataLen := len(data)
+	// extra junk
+	data = append(data, 0x00, 0x01, 0x02, 0x03, 0x00, 0x01, 0x02, 0x03, 0x00, 0x01, 0x02, 0x03)
+
+	n, err := Skim(NewDecoder(bytes.NewBuffer(data)), KTV.WithCN(0))
+	if got, correct := n, mainDataLen; got != correct {
+		test.Fatalf("n not equal: %d != %d ... (%d)", got, correct, len(data))
+	}
+	if err != nil {
+		test.Fatal(err)
+	}
+}

tape/tag.go

@@ -0,0 +1,72 @@
+package tape
+
+import "fmt"
+
+// TODO: fix #7
+
+type Tag byte; const (
+	SI      Tag = 0 << 5 // Small integer
+	LI      Tag = 1 << 5 // Large unsigned integer
+	LSI     Tag = 2 << 5 // Large signed integer
+	FP      Tag = 3 << 5 // Floating point
+	SBA     Tag = 4 << 5 // Small byte array
+	LBA     Tag = 5 << 5 // Large byte array
+	OTA     Tag = 6 << 5 // One-tag array
+	KTV     Tag = 7 << 5 // Key-tag-value table
+	TNMask  Tag = 0xE0   // The entire TN bitfield
+	CNMask  Tag = 0x1F   // The entire CN bitfield
+	CNLimit Tag = 32     // All valid CNs are < CNLimit
+)
+
+func (tag Tag) TN() int {
+	return int(tag >> 5)
+}
+
+func (tag Tag) CN() int {
+	return int(tag & CNMask)
+}
+
+func (tag Tag) WithCN(cn int) Tag {
+	return (tag & TNMask) | Tag(cn % 32)
+}
+
+func (tag Tag) WithoutCN() Tag {
+	return tag.WithCN(0)
+}
+
+func (tag Tag) Is(other Tag) bool {
+	return tag.TN() == other.TN()
+}
+
+func (tag Tag) String() string {
+	tn := fmt.Sprint(tag.TN())
+	switch tag.WithoutCN() {
+	case SI:  tn = "SI"
+	case LI:  tn = "LI"
+	case LSI: tn = "LSI"
+	case FP:  tn = "FP"
+	case SBA: tn = "SBA"
+	case LBA: tn = "LBA"
+	case OTA: tn = "OTA"
+	case KTV: tn = "KTV"
+	}
+	return fmt.Sprintf("%s:%d", tn, tag.CN())
+}
+
+// BufferTag returns the appropriate tag for a buffer.
+func BufferTag(value []byte) Tag {
+	return bufferLenTag(len(value))
+}
+
+// StringTag returns the appropriate tag for a string.
+func StringTag(value string) Tag {
+	return bufferLenTag(len(value))
+}
+
+func bufferLenTag(length int) Tag {
+	if length < int(CNLimit) {
+		return SBA.WithCN(length)
+	} else {
+		return LBA.WithCN(IntBytes(uint64(length)))
+	}
+}