generate: Generator compiles

This is a stupid fucking restriction

.gitignore

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

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

@@ -0,0 +1,109 @@
+# PDL Compiler Specification
+
+Given one or more PDL files representing a protocol, the compiler shall generate
+a Go package named "protocol", which shall contain types for message and type
+definitions, as well as encoding and decoding methods.
+
+## Static Section
+
+The compiler shall write a static section alongside the generated code. It
+shall contain this text:
+
+```go
+// Table is a KTV table with an undefined schema.
+type Table map[uint16] any
+
+// Message is any message that can be sent along this protocol.
+type Message interface {
+	codec.Encodable
+	codec.Decodable
+
+	// Method returns the method code of the message.
+	Method() uint16
+}
+```
+
+## Preamble
+
+At the start of each file but after the package name, the compiler shall emit
+this text:
+
+```go
+/* # Do not edit this package by hand!
+ * 
+ * This file was automatically generated by the Holanet PDL compiler. The
+ * source file is located at <path>
+ * Please edit that file instead, and re-compile it to this location.
+ *
+ * HOPP, TAPE, METADAPT, PDL/0 (c) 2025 holanet.xyz
+ */
+```
+
+Where `<path>` is the path of the protocol definition file relative to the
+generated file.
+
+## Message Definitions
+
+For each defined message, the compiler shall generate a Go type named
+`MessageName`, where `Name` is the name of the message as written in its
+definition. The message shall be encodable, and shall have `Encode` and `Decode`
+methods as described below.
+
+All messages shall satisfy a `Message` interface, which is defined in the
+static section.
+
+## Type Definitions
+
+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
+`EncodeValue`, `DecodeValue`, and `Tag` methods as described below.
+
+## Encoding and Decoding Methods
+
+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` 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 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
+is the real "porcelain" of the generated code. It shall provide methods to
+create and accept transactions. Each transaction shall be a struct which embeds
+a `hopp.Trans`, and shall have methods for sending and receiving messages.
+
+### Sending
+
+To send a message along a transaction, the program shall:
+
+1. Obtain the method code from the message
+3. Obtain a writer from the connection using the method code
+4. Wrap the writer in a `codec.Encoder`
+5. Use the encoder to encode the message
+6. Close the writer
+
+### Receiving
+
+To receiving a message from a transaction, the program shall:
+
+1. Obtain a method code and reader from the connection
+2. Wrap the reader in a `codec.Decoder`
+3. Switch on the method code, and decode the correct message using the decoder
+4. Return the message to the caller as a value
+
+The recieve function must return the message as a value instead of a pointer in
+order to avoid making an allocation. Because of this, the return value must be
+`any` instead of `Message`. The caller must then use a type switch to figure out
+what message was sent.

design/pdl-language.md

@@ -0,0 +1,104 @@
+# PDL Language Definition
+
+PDL allows defining a protocol using HOPP and TAPE.
+
+## Data Types
+
+| 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 | 
+| U5         | SI      |    | 
+| U8         | LI      |  0 | 
+| U16        | LI      |  1 | 
+| U32        | LI      |  3 | 
+| U64        | LI      |  7 | 
+| U128[^2]   | LI      | 15 | 
+| U256[^2]   | LI      | 31 | 
+| F16        | FP      |  1 | 
+| F32        | FP      |  3 | 
+| F64        | FP      |  7 | 
+| F128[^2]   | FP      | 15 | 
+| F256[^2]   | FP      | 31 | 
+| String     | SBA/LBA |  * | UTF-8 string
+| Buffer     | SBA/LBA |  * | Byte array
+| []\<TYPE\> | OTA     |  * | Array of any type[^1]
+| Table      | KTV     |  * | Table with undefined schema
+| {...}      | KTV     |  * | Table with defined schema
+
+[^1]: Excluding SI and SBA. I5 and U5 cannot be used in an array, but String and
+Buffer are simply forced to use their "long" variant.
+
+[^2]: Some systems may lack support for this.
+
+## Tokens
+
+PDL files are divided into tokens, which assemble together into larger language
+structures. They are separated by whitespace.
+
+| Name     | Syntax             | Description
+| -------- | ------------------ | -----------
+| Method   | `M[0-9A-Fa-f]{4}`  | A 16-bit hexadecimal method code.
+| Key      | `[0-9A-Fa-f]{4}`   | A 16-bit hexadecimal table key.
+| Ident    | `[A-Z][A-Za-z0-9]` | An identifier.
+| Comma    | `,`                | A comma separator.
+| LBrace   | `{`                | A left curly brace.
+| RBrace   | `}`                | A right curly brace.
+| LBracket | `[`                | A left square bracket.
+| RBracket | `]`                | A right square bracket.
+
+## Syntax
+
+Types are expressed with an Ident. A table can be used by either writing the
+name of the type (Table), or by defining a schema with curly braces. Arrays must
+be expressed using two matching square brackets before their element type.
+
+A table schema contains comma-separated fields in-between its braces. Each field
+has three parts: the key number (Key), the field name (Ident), and the field
+type. Tables, Arrays, etc. can be nested.
+
+Files directly contain messages and types, which start with a Method token and
+an Ident token respectively. A message consists of the method code (Method), the
+message name (Ident), and the message's root type. This is usually a table, but
+can be anything.
+
+Here is an example of all that:
+
+```
+M0000 Connect {
+	0000 Name String,
+	0001 Password String,
+}
+
+M0001 UserList {
+	0000 Users []User,
+}
+
+User {
+	0000 Name      String,
+	0001 Bio       String,
+	0002 Followers U32,
+}
+```
+
+## EBNF Description
+
+Below is an EBNF description of the language.
+
+```
+<file>    -> (<message> | <typedef)*
+<method>  -> /M[0-9A-Fa-f]{4}/
+<key>     -> /[0-9A-Fa-f]{4}/
+<ident>   -> /[A-Z][A-Za-z0-9]/
+<field>   -> <key> <ident> <type>
+<type> -> <ident>
+        | "[" "]" <type>
+        | "{" (<field> ",")* [<field>] "}"
+<message> -> <method> <ident> <type>
+<typedef> -> <ident> <type>
+```

design/protocol.md

@@ -18,12 +18,10 @@ dependant on which transport is being used.
 A message refers to a block of octets sent within a transaction, paired with an
 unsigned 16-bit method code. The order of messages within a given transaction is
 preserved, but the order of messages accross the entire connection is not
-guaranteed.
+guaranteed. There is no functional limit on the size of a message payload, but
-
+there may be one depending on which
-The message payload must be 65,535 (unsigned 16-bit integer limit) octets or
+[METADAPT sub-protocol](#message-and-transaction-demarcation-protocol-metadapt)
-smaller in length. This does not include the method code. Applications are free
+is in use.
-to send whatever data they wish as the payload, but TAPE is recommended for
-encoding it.
 
 Method codes should be written in upper-case base 16 with the prefix "M" in
 logs, error messages, documentation, etc. For example, the method code 62,206 in
@@ -37,88 +35,78 @@ fucking with you.
 ## Table Pair Encoding (TAPE)
 The Table Pair Encoding (TAPE) scheme is a method for encoding structured data
 within HOPP messages. It defines standard binary encoding methods for common
-data types, as well as a corruption-resistant table structure that maps numeric
+data types, as well as aggregate data types such as tables and arrays. It is
-IDs to values. It is designed to allow applications to be presented with data
+designed to allow applications to be presented with data they are not equipped
-they are not equipped to handle while continuing to function normally. This
+to handle while continuing to function normally. This enables backwards
-enables backwards compatibile application protocol changes.
+compatibile application protocol changes.
 
-### Table Structure
+TAPE expresses types using tags. A tag is 8 bits in size, and is divided into
-A table is divided into two sections: the header, and the values. The header
+two parts: the Type Number (TN), and the Configuration Number (CN). The TN is 3
-begins with the number (U16) of pairs in the table, which is then followed by
+bits, and the CN is 5 bits. Both are interpreted as unsigned integers. Both
-that many tag-offset pairs. A tag-offset pair consists of a numerical (U16) tag,
+sides of the connection must agree on the semantic meaning of the values and
-followed the position (U16) of the value relative to the start of the values
+their arrangement.
-section. The values section contains the value data for each pair, where the
-start of each value is determined by its offset, and the end is determined by
-the offset of the next value, or the end of the message if there is no value
-after it.
 
-Both sections must be in the same order, and because of this, each value offset
+A TAPE structure begins with one root, which consists of a tag followed by a
-must be greater than or equal to the last. If a message has erratic structure
+payload. This is usually an aggregate data structure such as KTV to allow for
-(such as unordered or out-of-bounds offsets), implementations may opt to discard
+several different values.
-only the erratic pairs, as well as the pairs directly before those.
+
+TAPE is based on an encoding method previously developed by silt.
 
 ### Data Value Types
-The table below lists all data value types supported by TAPE.
+The table below lists all data value types supported by TAPE. They are discussed
+in detail in the following sections.
 
-| Name        | Size            | Description                 | Encoding Method
+| TN | Bits | Name | Description
-| ----------- | --------------: | --------------------------- | ---------------
+| -: | ---: | ---- | -----------
-| I8          |               1 | A signed 8-bit integer      | BETC
+|  0 |  000 | SI   | Small integer
-| I16         |               2 | A signed 16-bit integer     | BETC
+|  1 |  001 | LI   | Large integer
-| I32         |               4 | A signed 32-bit integer     | BETC
+|  2 |  010 | FP   | Floating point
-| I64         |               8 | A signed 64-bit integer     | BETC
+|  3 |  011 | SBA  | Small byte array
-| U8          |               1 | An unsigned 8-bit integer   | BEU
+|  4 |  100 | LBA  | Large byte array
-| U16         |               2 | An unsigned 16-bit integer  | BEU
+|  5 |  101 | OTA  | One-tag array
-| U32         |               4 | An unsigned 32-bit integer  | BEU
+|  6 |  110 | KTV  | Key-tag-value table
-| U64         |               8 | An unsigned 64-bit integer  | BEU
+|  7 |  111 | N/A  | Reserved
-| Array[^1]   |         SOP[^2] | An array of any above type  | PASTA
-| String      |             N/A | A UTF-8 string              | UTF-8
-| StringArray | n * 2 + SOP[^2] | An array the String type    | VILA
 
-[^1]: Array types are written as <E>Array, where <E> is the element type. For
+#### Small Integer (SI)
-example, an array of I32 would be written as I32Array. StringArray still follows
+SI encodes an integer of up to 5 bits, which are stored in the CN. It has no
-this rule, even though it is encoded differently from other arrays. Nesting
+payload. Whether the bits are interpreted as unsigned or as signed two's
-arrays inside of arrays is prohibited. This problem can be avoided in most cases
+complement is semantic information and must be agreed upon by both sides of the
-by effectively utilizing the table structure, or by improving the design of
+connection. Thus, the value may range from 0 to 31 if unsigned, and from -16 to
-your protocol.
+17 if signed.
 
-[^2]: SOP (sum of parts) refers to the sum of the size of every item in a data
+#### Large Integer (LI)
-structure.
+LI encodes an integer of up to 256 bits, which are stored in the payload. The CN
+determine the length of the payload in bytes. The integer is big-endian. Whether
+the payload is interpreted as unsigned or as signed two's complement is semantic
+information and must be agreed upon by both sides of the connection. Thus, the
+value may range from 0 to 31 if unsigned, and from -16 to 17 if signed.
 
-### Encoding Methods
+#### Floating Point (FP)
-Below are all encoding methods supported by TAPE.
+FP encodes an IEEE 754 floating point number of up to 256 bits, which are stored
+in the payload. The CN determines the length of the payload in bytes, and it may
+only be one of these values: 16, 32, 64, 128, or 256.
 
-#### BETC
+#### Small Byte Array (SBA)
-Big-Endian, Two's Complement signed integer. The size is defined as the least
+SBA encodes an array of up to 32 bytes, which are stored in the paylod. The
-amount of whole octets which can fit all bits in the integer, regardless if the
+CN determines the length of the payload in bytes.
-bits are on or off. Therefore, the size cannot change at runtime.
 
-#### BEU
+#### Large Byte Array (LBA)
-Big-Endian, Unsigned integer. The size is defined as the least amount of whole
+LBA encodes an array of up to 2^256 bytes, which are stored in the second part
-octets which can fit all bits in the integer, regardless if the bits are on or
+of the payload, directly after the length. The length of the data length field
-off. Therefore, the size cannot change at runtime.
+in bytes is determined by the CN.
 
-#### PASTA
+#### One-Tag Array (OTA)
-Packed Single-Type Array. The size is defined as the size of an individual item
+OTA encodes an array of up to 2^256 items, which are stored in the payload after
-times the number of items. Items are placed one after the other with no gaps
+the length field and the item tag, where the length field comes first. Each item
-in-between them, except as required to align the start of each item to the
+must be the same length, as they all share the same tag. The length of the data
-nearest whole octet. Items should be of the same type and must be of the same
+length field in bytes is determined by the CN.
-size.
 
-#### UTF-8
+#### Key-Tag-Value Table (KTV)
-UTF-8 string. The size is defined as the least amount of whole octets which can
+KTV encodes a table of up to 2^256 key/value pairs, which are stored in the
-fit all bits in the string, regardless if the bits are on or off. The size of
+payload after the length field. The pairs themselves consist of a 16-bit
-this type is not fixed and may change at runtime, so this needs to be accounted
+unsigned big-endian key followed by a tag and then the payload. Pair values can
-for during use.
+be of different types and sizes. The order of the pairs is not significant and
-
+should never be treated as such.
-#### VILA
-Variable Item Length Array. The size is defined as the least amount of whole
-octets which can fit each item plus one U16 per item. The size of this type is
-not fixed and may change at runtime, so this needs to be accounted for during
-use. The amount of items must be greater than zero. Items are each prefixed by
-their size (in octets) encoded as a U16, and they are placed one after the other
-with no gaps in-between them, except as required to align the start of each item
-to the nearest whole octet. Items should be of the same type but do not need to
-be of the same size.
 
 ## Transports
 A transport is a protocol that HOPP connections can run on top of. HOPP
@@ -169,7 +157,6 @@ sun will have expanded to swallow earth by then. Your connection will not last
 that long.
 
 #### Message Chunking
-
 The most significant bit of the payload size field of an MMB is called the Chunk
 Control Bit (CCB). If the CCB of a given MMB is zero, the represented message is
 interpreted as being self-contained and the data is processed immediately. If

dial.go

@@ -1,9 +1,9 @@
 package hopp
 
 import "net"
+import "errors"
 import "context"
 import "crypto/tls"
-import "github.com/quic-go/quic-go"
 
 // Dial opens a connection to a server. The network must be one of "quic",
 // "quic4", (IPv4-only) "quic6" (IPv6-only), or "unix". For now, "quic4" and
@@ -19,9 +19,8 @@ type Dialer struct {
 }
 
 // Dial opens a connection to a server. The network must be one of "quic",
-// "quic4", (IPv4-only) "quic6" (IPv6-only), or "unix". For now, "quic4" and
+// "quic4", (IPv4-only) "quic6" (IPv6-only), or "unix". For now, quic is not
-// "quic6" don't do anything as the quic-go package doesn't seem to support this
+// supported.
-// behavior.
 func (diale Dialer) Dial(ctx context.Context, network, address string) (Conn, error) {
 	switch network {
 	case "quic", "quic4", "quic6": return diale.dialQUIC(ctx, network, address)
@@ -31,12 +30,7 @@ func (diale Dialer) Dial(ctx context.Context, network, address string) (Conn, er
 }
 
 func (diale Dialer) dialQUIC(ctx context.Context, network, address string) (Conn, error) {
-	// sorry i fucking lied to you about the network parameter. for all
+	return nil, errors.New("quic is not yet implemented")
-	// quic-go's bullshit bloat, it doesnt even support that. not even when
-	// instantiating a transport. go figure :/
-	conn, err := quic.DialAddr(ctx, address, tlsConfig(diale.TLSConfig), quicConfig())
-	if err != nil { return nil, err }
-	return AdaptB(quicMultiConn { underlying: conn }), nil
 }
 
 func (diale Dialer) dialUnix(ctx context.Context, network, address string) (Conn, error) {
@@ -60,15 +54,6 @@ func tlsConfig(conf *tls.Config) *tls.Config {
 	return conf
 }
 
-func quicConfig() *quic.Config {
-	return &quic.Config {
-		// TODO: perhaps we might want to put something here
-		// the quic config shouldn't be exported, just set up
-		// automatically. we can't have that strangely built quic-go
-		// package be part of the API, or any third-party packages for
-		// that matter. it must all be abstracted away.
-	}
-}
 
 func quicNetworkToUDPNetwork(network string) (string, error) {
 	switch network {

generate/generate_test.go

@@ -0,0 +1,327 @@
+package generate
+
+// import "fmt"
+import "strings"
+import "testing"
+import "git.tebibyte.media/sashakoshka/goparse"
+
+var testGenerateCorrect =
+`package protocol
+
+/* # Do not edit this package by hand!
+ * 
+ * This file was automatically generated by the Holanet PDL compiler. The
+ * source file is located at input.pdl
+ * Please edit that file instead, and re-compile it to this location.
+ *
+ * HOPP, TAPE, METADAPT, PDL/0 (c) 2025 holanet.xyz
+ */
+
+import "git.tebibyte.media/sashakoshka/hopp/tape"
+
+// Table is a KTV table with an undefined schema.
+type Table map[uint16] any
+
+// Message is any message that can be sent along this protocol.
+type Message interface {
+	tape.Encodable
+	tape.Decodable
+
+	// Method returns the method code of the message.
+	Method() uint16
+}
+
+// User represents the protocol data type User.
+type User struct {
+	Name string
+	Bio string
+	Followers uint32
+}
+
+// EncodeValue encodes the value of this type without the tag. The value is
+// encoded according to the parameters specified by the tag, if possible.
+func (this *User) EncodeValue(encoder *tape.Encoder) (n int, err error) {
+	nn, err := tape.WriteTableHeader(2)
+	n += nn; if err != nil { return n, err }
+	nn, err := encoder.WriteUint16(0x0000)
+	n += nn; if err != nil { return n, err }
+	nn, err := tape.WriteString(encoder, this.Name)
+	n += nn; if err != nil { return n, err }
+	nn, err := encoder.WriteUint16(0x0001)
+	n += nn; if err != nil { return n, err }
+	nn, err := tape.WriteString(encoder, this.Bio)
+	n += nn; if err != nil { return n, err }
+	return n, nil
+}
+
+// Decode replaces the data in this User with information from the decoder.
+func (this *User) Decode(decoder *tape.Decoder) (n int, err error) {
+	pull, nn, err := tape.ReadTableHeader(decoder)
+	n += nn; if err != nil { return n, err }
+
+	for {
+		key, tag, end, nn, err := pull()
+		n += nn; if err != nil { return n, err }
+		if end { break }
+
+		switch key {
+		case 0x0000:
+			value, nn, err := tape.ReadString(decoder)
+			n += nn; if err != nil { return n, err }
+			this.Name = value
+		case 0x0001:
+			value, nn, err := tape.ReadString(decoder)
+			n += nn; if err != nil { return n, err }
+			this.Bio = value 
+		}
+	}
+	
+	return n, nil
+}
+
+// MessageConnect represents the protocol message M0000 Connect.
+type MessageConnect struct {
+	Name string
+	Password string
+}
+
+// Method returns the method code, M0000.
+func (this *MessageConnect) Method() uint16 {
+	return 0x0000
+}
+
+// Encode encodes the message to the encoder.
+func (this *MessageConnect) Encode(encoder *tape.Encoder) (n int, err error) {
+	nn, err := tape.WriteTableHeader(2)
+	n += nn; if err != nil { return n, err }
+	nn, err := encoder.WriteUint16(0x0000)
+	n += nn; if err != nil { return n, err }
+	nn, err := tape.WriteString(encoder, this.Name)
+	n += nn; if err != nil { return n, err }
+	nn, err := encoder.WriteUint16(0x0001)
+	n += nn; if err != nil { return n, err }
+	nn, err := tape.WriteString(encoder, this.Password)
+	n += nn; if err != nil { return n, err }
+	return n, nil
+}
+
+// Decode replaces the data in this message with information from the decoder.
+func (this *MessageConnect) Decode(decoder *tape.Decoder) (n int, err error) {
+	pull, nn, err := tape.ReadTableHeader(decoder)
+	n += nn; if err != nil { return n, err }
+
+	for {
+		key, tag, end, nn, err := pull()
+		n += nn; if err != nil { return n, err }
+		if end { break }
+
+		switch key {
+		case 0x0000:
+			value, nn, err := tape.ReadString(decoder)
+			n += nn; if err != nil { return n, err }
+			this.Name = value
+		case 0x0001:
+			value, nn, err := tape.ReadString(decoder)
+			n += nn; if err != nil { return n, err }
+			this.Password = value 
+		}
+	}
+	
+	return n, nil
+}
+
+// MessageUserList represents the protocol message M0001 UserList.
+type MessageUserList struct {
+	Users []User
+}
+
+// Method returns the method code, M0001.
+func (this *MessageUserList) Method() uint16 {
+	return 0x0001
+}
+
+// TODO methods
+`
+
+func TestGenerate(test *testing.T) {
+	protocol := defaultProtocol()
+	protocol.Messages[0x0000] = Message {
+		Name: "Connect",
+		Type: TypeTableDefined {
+			Fields: map[uint16] Field {
+				0x0000: Field { Name: "Name",      Type: TypeString { } },
+				0x0001: Field { Name: "Password",  Type: TypeString { } },
+			},
+		},
+	}
+	protocol.Messages[0x0001] = Message {
+		Name: "UserList",
+		Type: TypeTableDefined {
+			Fields: map[uint16] Field {
+				0x0000: Field { Name: "Users",  Type: TypeArray { Element: TypeNamed { Name: "User" } } },
+			},
+		},
+	}
+	protocol.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 } },
+		},
+	}
+
+	correct := testGenerateCorrect
+
+	builder := strings.Builder { }
+	generator := Generator { Output: &builder }
+	/* TODO test n: */ _, err := generator.Generate(&protocol)
+	if err != nil { test.Fatal(parse.Format(err)) }
+	got := builder.String()
+
+	test.Log("CORRECT:")
+	test.Log(correct)
+	test.Log("GOT:")
+	test.Log(got)
+
+	if correct != got {
+		test.Error("not equal")
+		for index := range min(len(correct), len(got)) {
+			if correct[index] == got[index] { continue }
+			test.Log("C:", correct[max(0, index - 8):min(len(correct), index + 8)])
+			test.Log("G:", got[max(0, index - 8):min(len(got), index + 8)])
+			break
+		}
+		test.FailNow()
+	}
+}
+
+func TestGenerateRun(test *testing.T) {
+	protocol := defaultProtocol()
+	protocol.Messages[0x0000] = Message {
+		Name: "Connect",
+		Type: TypeTableDefined {
+			Fields: map[uint16] Field {
+				0x0000: Field { Name: "Name",      Type: TypeString { } },
+				0x0001: Field { Name: "Password",  Type: TypeString { } },
+			},
+		},
+	}
+	protocol.Messages[0x0001] = Message {
+		Name: "UserList",
+		Type: TypeTableDefined {
+			Fields: map[uint16] Field {
+				0x0000: Field { Name: "Users",  Type: TypeArray { Element: TypeNamed { Name: "User" } } },
+			},
+		},
+	}
+	protocol.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 }},
+			},
+		},
+	}
+	protocol.Messages[0x0003] = Message {
+		Name: "NestedArray",
+		Type: TypeArray { Element: TypeArray { Element: TypeInt { Bits: 8 } } },
+	}
+	protocol.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 } },
+		},
+	}
+	testGenerateRun(test, &protocol, `
+		// imports
+	`, `
+		// test case
+		log.Println("MessageConnect")
+		messageConnect := MessageConnect {
+			Name: "rarity",
+			Password: "gems",
+		}
+		testEncode(
+			&messageConnect,
+			tu.S(0xC1, 0x02).AddVar(
+				[]byte { 0x00, 0x00, 0x66, 'r', 'a', 'r', 'i', 't', 'y' },
+				[]byte { 0x00, 0x01, 0x64, '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,
+				},
+			},
+		}
+		testEncode(
+			&messageUserList,
+			tu.S(0xC1, 0x01, 0x00, 0x00,
+				0xA1, 0x03, 0xC1,
+					).Add(0x03).AddVar(
+						[]byte { 0x00, 0x00, 0x66, 'r', 'a', 'r', 'i', 't', 'y' },
+						[]byte { 0x00, 0x01, 0x67, 'a', 's', 'd', 'j', 'a', 'd', 's' },
+						[]byte { 0x00, 0x02, 0x23, 0x00, 0x00, 0x03, 0x24 },
+					).Add(0x03).AddVar(
+						[]byte { 0x00, 0x00, 0x69, 'd', 'e', 'e', 'z', ' ', 'n', 'u', 't', 's' },
+						[]byte { 0x00, 0x01, 0x64, 'l', 'o', 'g', 'y' },
+						[]byte { 0x00, 0x02, 0x23, 0x00, 0x00, 0x80, 0x00 },
+					).Add(0x03).AddVar(
+						[]byte { 0x00, 0x00, 0x69, 'c', 'r', 'e', 'e', 'k', 'f', 'l', 'o', 'w' },
+						[]byte { 0x00, 0x01, 0x6C, '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: 45.389,
+			Y: 294.1,
+			Z: 384729384.234892034,
+		}
+		testEncode(
+			&messagePulse,
+			tu.S(0xC1, 0x05).AddVar(
+				[]byte { 0x00, 0x00, 0x09 },
+				[]byte { 0x00, 0x01, 0x21, 0xCA, 0xDF },
+				[]byte { 0x00, 0x02, 0x41, 0x51, 0xAC },
+				[]byte { 0x00, 0x03, 0x43, 0x43, 0x93, 0x0C, 0xCD },
+				[]byte { 0x00, 0x04, 0x47, 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),
+		}
+		testEncode(
+			&messageNestedArray,
+			tu.S(0xA1, // TODO
+				))
+	`)
+}

generate/lex.go

@@ -0,0 +1,230 @@
+package generate
+
+import "io"
+import "bufio"
+import "unicode"
+import "unicode/utf8"
+import "git.tebibyte.media/sashakoshka/goparse"
+
+const (
+	TokenMethod parse.TokenKind = iota
+	TokenKey
+	TokenIdent
+	TokenComma
+	TokenLBrace
+	TokenRBrace
+	TokenLBracket
+	TokenRBracket
+)
+
+var tokenNames = map[parse.TokenKind] string {
+	TokenMethod:   "Method",
+	TokenKey:      "Key",
+	TokenIdent:    "Ident",
+	TokenComma:    "Comma",
+	TokenLBrace:   "LBrace",
+	TokenRBrace:   "RBrace",
+	TokenLBracket: "LBracket",
+	TokenRBracket: "RBracket",
+}
+
+func Lex(fileName string, reader io.Reader) (parse.Lexer, error) {
+	lex := &lexer {
+		fileName: fileName,
+		lineScanner: bufio.NewScanner(reader),
+	}
+	lex.nextRune()
+	return lex, nil
+}
+
+type lexer struct {
+	fileName    string
+	lineScanner *bufio.Scanner
+	rune        rune
+	line        string
+	lineFood    string
+	
+	offset   int
+	row      int
+	column   int
+
+	eof bool
+
+}
+
+func (this *lexer) Next() (parse.Token, error) {
+	token, err := this.nextInternal()
+	if err == io.EOF { err = this.errUnexpectedEOF() }
+	return token, err
+}
+
+func (this *lexer) nextInternal() (token parse.Token, err error) {
+	err = this.skipWhitespace()
+	token.Position = this.pos()
+	if this.eof {
+		token.Kind = parse.EOF
+		err = nil
+		return
+	}
+	if err != nil { return }
+
+	appendRune := func () {
+		token.Value += string(this.rune)
+		err = this.nextRune()
+	}
+
+	doNumber := func () {
+		for isDigit(this.rune) {
+			appendRune()
+			if this.eof { err = nil; return }
+			if err != nil { return }
+		}
+	}
+
+	defer func () {
+		newPos := this.pos()
+		newPos.End -- // TODO figure out why tf we have to do this
+		token.Position = token.Position.Union(newPos)
+	} ()
+
+	switch {
+	// Method
+	case this.rune == 'M':
+		token.Kind = TokenMethod
+		err = this.nextRune()
+		if err != nil { return }
+		doNumber()
+		if this.eof { err = nil; return }
+	// Key
+	case isDigit(this.rune):
+		token.Kind = TokenKey
+		doNumber()
+		if this.eof { err = nil; return }
+	// Ident
+	case unicode.IsUpper(this.rune):
+		token.Kind = TokenIdent
+		for unicode.IsLetter(this.rune) || isDigit(this.rune) {
+			appendRune()
+			if this.eof { err = nil; return }
+			if err != nil { return }
+		}
+	// Comma
+	case this.rune == ',':
+		token.Kind = TokenComma
+		appendRune()
+		if this.eof { err = nil; return }
+	// LBrace
+	case this.rune == '{':
+		token.Kind = TokenLBrace
+		appendRune()
+		if this.eof { err = nil; return }
+	// RBrace
+	case this.rune == '}':
+		token.Kind = TokenRBrace
+		appendRune()
+		if this.eof { err = nil; return }
+	// LBracket
+	case this.rune == '[':
+		token.Kind = TokenLBracket
+		appendRune()
+		if this.eof { err = nil; return }
+	// RBracket
+	case this.rune == ']':
+		token.Kind = TokenRBracket
+		appendRune()
+		if this.eof { err = nil; return }
+	case unicode.IsPrint(this.rune):
+		err = parse.Errorf (
+			this.pos(), "unexpected rune '%c'",
+			this.rune)
+	default:
+		err = parse.Errorf (
+			this.pos(), "unexpected rune %U",
+			this.rune)
+	}
+
+	return
+}
+
+func (this *lexer) nextRune() error {
+	if this.lineFood == "" {
+		ok := this.lineScanner.Scan()
+		if ok {
+			this.line     = this.lineScanner.Text()
+			this.lineFood = this.line
+			this.rune     = '\n'
+			this.column   = 0
+			this.row      ++
+		} else {
+			err := this.lineScanner.Err()
+			if err == nil {
+				this.eof = true
+				return io.EOF
+			} else {
+				return err
+			}
+		}
+	} else {
+		var ch   rune
+		var size int
+		for ch == 0 && this.lineFood != "" {
+			ch, size = utf8.DecodeRuneInString(this.lineFood)
+			this.lineFood = this.lineFood[size:]
+		}
+		this.rune = ch
+		this.column ++
+	}
+
+	return nil
+}
+
+func (this *lexer) skipWhitespace() error {
+	err := this.skipComment()
+	if err != nil { return err }
+	for isWhitespace(this.rune) {
+		err := this.nextRune()
+		if err != nil { return err }
+		err = this.skipComment()
+		if err != nil { return err }
+	}
+	return nil
+}
+
+func (this *lexer) skipComment() error {
+	if this.rune == ';' {
+		for this.rune != '\n' {
+			err := this.nextRune()
+			if err != nil { return err }
+		}
+	}
+	return nil
+}
+
+func (this *lexer) pos() parse.Position {
+	return parse.Position {
+		File:   this.fileName,
+		Line:   this.lineScanner.Text(),
+		Row:    this.row    - 1,
+		Start:  this.column - 1,
+		End:    this.column,
+	}
+}
+
+func (this *lexer) errUnexpectedEOF() error {
+	return parse.Errorf(this.pos(), "unexpected EOF")
+}
+
+func isWhitespace(char rune) bool {
+	switch char {
+	case ' ', '\t', '\r', '\n': return true
+	default: return false
+	}
+}
+
+func isDigit(char rune) bool {
+	return char >= '0' && char <= '9'
+}
+
+func isHexDigit(char rune) bool {
+	return isDigit(char) || char >= 'a' && char <= 'f' || char >= 'A' && char <= 'F'
+}

generate/lex_test.go

@@ -0,0 +1,54 @@
+package generate
+
+import "strings"
+import "testing"
+import "git.tebibyte.media/sashakoshka/goparse"
+
+func TestLex(test *testing.T) {
+	lexer, err := Lex("test.pdl", strings.NewReader(`
+	M0001 User {
+		0000 Name      String,
+		0001 Users []User,
+		0002 Followers U32,
+	}`))
+	if err != nil { test.Fatal(parse.Format(err)) }
+
+	correctTokens := []parse.Token {
+		tok(TokenMethod,   "0001"),
+		tok(TokenIdent,    "User"),
+		tok(TokenLBrace,   "{"),
+		tok(TokenKey,      "0000"),
+		tok(TokenIdent,    "Name"),
+		tok(TokenIdent,    "String"),
+		tok(TokenComma,    ","),
+		tok(TokenKey,      "0001"),
+		tok(TokenIdent,    "Users"),
+		tok(TokenLBracket, "["),
+		tok(TokenRBracket, "]"),
+		tok(TokenIdent,    "User"),
+		tok(TokenComma,    ","),
+		tok(TokenKey,      "0002"),
+		tok(TokenIdent,    "Followers"),
+		tok(TokenIdent,    "U32"),
+		tok(TokenComma,    ","),
+		tok(TokenRBrace,   "}"),
+		tok(parse.EOF,     ""),
+	}
+
+	for index, correct := range correctTokens {
+		got, err := lexer.Next()
+		if err != nil { test.Fatal(parse.Format(err)) }
+		if got.Kind != correct.Kind || got.Value != correct.Value {
+			test.Logf("token %d mismatch", index)
+			test.Log("GOT:", tokenNames[got.Kind], got.Value)
+			test.Fatal("CORRECT:", tokenNames[correct.Kind], correct.Value)
+		}
+	}
+}
+
+func tok(kind parse.TokenKind, value string) parse.Token {
+	return parse.Token {
+		Kind:  kind,
+		Value: value,
+	}
+}

generate/misc_test.go

@@ -0,0 +1,73 @@
+package generate
+
+import "os"
+import "fmt"
+import "os/exec"
+import "testing"
+import "path/filepath"
+
+func testGenerateRun(test *testing.T, protocol *Protocol, imports string, testCase string) {
+	// reset data directory
+	dir := "test/generate-run"
+	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 "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)
+			}
+		}
+		`
+	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", "./generate/test/generate-run")
+	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

@@ -0,0 +1,207 @@
+package generate
+
+import "io"
+import "strconv"
+import "git.tebibyte.media/sashakoshka/goparse"
+
+func Parse(lx parse.Lexer) (*Protocol, error) {
+	protocol := defaultProtocol()
+	par := parser {
+		Parser: parse.Parser {
+			Lexer: lx,
+			TokenNames: tokenNames,
+		},
+		protocol: &protocol,
+	}
+	err := par.parse()
+	if err != nil { return nil, err }
+	return par.protocol, nil
+}
+
+func defaultProtocol() Protocol {
+	return Protocol {
+		Messages: make(map[uint16] Message),
+		Types: map[string] Type { },
+	}
+}
+
+func ParseReader(reader io.Reader) (*Protocol, error) {
+	lx, err := Lex("test.pdl", reader)
+	if err != nil { return nil, err }
+	return Parse(lx)
+}
+
+type parser struct {
+	parse.Parser
+	protocol *Protocol
+}
+
+func (this *parser) parse() error {
+	err := this.Next()
+	if err != nil { return err }
+	for this.Token.Kind != parse.EOF {
+		err = this.parseTopLevel()
+		if err != nil { return err }
+	}
+	return nil
+}
+
+func (this *parser) parseTopLevel() error {
+	err := this.ExpectDesc("message or typedef", TokenMethod, TokenIdent)
+	if err != nil { return err }
+	if this.EOF() { return nil }
+
+	switch this.Kind() {
+	case TokenMethod: return this.parseMessage()
+	case TokenIdent:  return this.parseTypedef()
+	}
+	panic("bug")
+}
+
+func (this *parser) parseMessage() error {
+	err := this.Expect(TokenMethod)
+	if err != nil { return err }
+	method, err := this.parseHexNumber(this.Value(), 0xFFFF)
+	if err != nil { return err }
+	err = this.ExpectNext(TokenIdent)
+	if err != nil { return err }
+	name := this.Value()
+	err = this.Next()
+	if err != nil { return err }
+	typ, err := this.parseType()
+	if err != nil { return err }
+	this.protocol.Messages[uint16(method)] = Message {
+		Name: name,
+		Type: typ,
+	}
+	return nil
+}
+
+func (this *parser) parseTypedef() error {
+	err := this.Expect(TokenIdent)
+	if err != nil { return err }
+	name := this.Value()
+	err = this.Next()
+	if err != nil { return err }
+	typ, err := this.parseType()
+	if err != nil { return err }
+	this.protocol.Types[name] = typ
+	return nil
+}
+
+func (this *parser) parseType() (Type, error) {
+	err := this.ExpectDesc("type", TokenIdent, TokenLBracket, TokenLBrace)
+	if err != nil { return nil, err }
+
+	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()
+	case TokenLBrace:
+		return this.parseTypeTable()
+	}
+	panic("bug")
+}
+
+func (this *parser) parseTypeNamed() (TypeNamed, error) {
+	err := this.Expect(TokenIdent)
+	if err != nil { return TypeNamed { }, err }
+	name := this.Value()
+	err = this.Next()
+	if err != nil { return TypeNamed { }, err }
+	return TypeNamed { Name: name }, nil
+}
+
+func (this *parser) parseTypeArray() (TypeArray, error) {
+	err := this.Expect(TokenLBracket)
+	if err != nil { return TypeArray { }, err }
+	err = this.ExpectNext(TokenRBracket)
+	if err != nil { return TypeArray { }, err }
+	err = this.Next()
+	if err != nil { return TypeArray { }, err }
+	typ, err := this.parseType()
+	if err != nil { return TypeArray { }, err }
+	return TypeArray { Element: typ }, nil
+}
+
+func (this *parser) parseTypeTable() (TypeTableDefined, error) {
+	err := this.Expect(TokenLBrace)
+	if err != nil { return TypeTableDefined { }, err }
+	err = this.Next()
+	if err != nil { return TypeTableDefined { }, err }
+	typ := TypeTableDefined {
+		Fields: make(map[uint16] Field),
+	}
+	for {
+		err := this.ExpectDesc("table field", TokenKey, TokenRBrace)
+		if err != nil { return TypeTableDefined { }, err }
+		if this.Is(TokenRBrace) {
+			break
+		}
+		key, field, err := this.parseField()
+		if err != nil { return TypeTableDefined { }, err }
+		typ.Fields[key] = field
+		err = this.Expect(TokenComma, TokenRBrace)
+		if err != nil { return TypeTableDefined { }, err }
+		if this.Is(TokenRBrace) {
+			break
+		}
+		err = this.Next()
+		if err != nil { return TypeTableDefined { }, err }
+	}
+	err = this.Next()
+	if err != nil { return TypeTableDefined { }, err }
+	return typ, nil
+}
+
+func (this *parser) parseField() (uint16, Field, error) {
+	err := this.Expect(TokenKey)
+	if err != nil { return 0, Field { }, err }
+	key, err := this.parseHexNumber(this.Value(), 0xFFFF)
+	if err != nil { return 0, Field { }, err }
+	err = this.ExpectNext(TokenIdent)
+	if err != nil { return 0, Field { }, err }
+	name := this.Value()
+	err = this.Next()
+	if err != nil { return 0, Field { }, err }
+	typ, err := this.parseType()
+	if err != nil { return 0, Field { }, err }
+	return uint16(key), Field {
+		Name: name,
+		Type: typ,
+	}, nil
+}
+
+func (this *parser) parseHexNumber(input string, maxValue int64) (int64, error) {
+	number, err := strconv.ParseInt(input, 16, 64)
+	if err != nil {
+		return 0, parse.Errorf(this.Pos(), "%v", err)
+	}
+	if maxValue > 0 && number > maxValue {
+		return 0, parse.Errorf(this.Pos(), "value too large (max %X)", maxValue)
+	}
+	return number, nil
+}

generate/parse_test.go

@@ -0,0 +1,68 @@
+package generate
+
+import "fmt"
+import "strings"
+import "testing"
+import "git.tebibyte.media/sashakoshka/goparse"
+
+func TestParse(test *testing.T) {
+	correct := defaultProtocol()
+	correct.Messages[0x0000] = Message {
+		Name: "Connect",
+		Type: TypeTableDefined {
+			Fields: map[uint16] Field {
+				0x0000: Field { Name: "Name",      Type: TypeString { } },
+				0x0001: Field { Name: "Password",  Type: TypeString { } },
+			},
+		},
+	}
+	correct.Messages[0x0001] = Message {
+		Name: "UserList",
+		Type: TypeTableDefined {
+			Fields: map[uint16] Field {
+				0x0000: Field { Name: "Users",  Type: TypeArray { Element: TypeNamed { Name: "User" } } },
+			},
+		},
+	}
+	correct.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 } },
+		},
+	}
+	test.Log("CORRECT:", &correct)
+	
+	got, err := ParseReader(strings.NewReader(`
+	M0000 Connect {
+		0000 Name String,
+		0001 Password String,
+	}
+	
+	M0001 UserList {
+		0000 Users []User,
+	}
+	
+	User {
+		0000 Name      String,
+		0001 Bio       String,
+		0002 Followers U32,
+	}
+	`))
+	if err != nil { test.Fatal(parse.Format(err)) }
+	test.Log("GOT:    ", got)
+
+	correctStr := fmt.Sprint(&correct)
+	gotStr     := fmt.Sprint(got)
+
+	if correctStr != gotStr {
+		test.Error("not equal")
+		for index := range min(len(correctStr), len(gotStr)) {
+			if correctStr[index] == gotStr[index] { continue }
+			test.Log("C:", correctStr[max(0, index - 8):min(len(correctStr), index + 8)])
+			test.Log("G:", gotStr[max(0, index - 8):min(len(gotStr), index + 8)])
+			break
+		}
+		test.FailNow()
+	}
+}

generate/protocol.go

@@ -1,244 +1,107 @@
 package generate
 
-import "io"
 import "fmt"
-import "errors"
+import "maps"
-import "strconv"
+import "slices"
-import "strings"
+import "crypto/md5"
-import "github.com/gomarkdown/markdown"
-import "github.com/gomarkdown/markdown/ast"
-import "github.com/gomarkdown/markdown/parser"
 
-// Protocol describes a protocol.
 type Protocol struct {
-	Messages []Message
+	Messages map[uint16] Message
+	Types    map[string] Type
 }
 
-// Message describes a protocol message.
 type Message struct {
-	Doc    string
+	Name string
-	Method uint16
+	Type Type
-	Name   string
+}
-	Fields []Field
+
+type Type interface {
+	fmt.Stringer
+}
+
+type TypeInt struct {
+	Bits   int
+	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
 }
 
-// Field describes a named value within a message.
 type Field struct {
-	Doc      string
+	Name string
-	Tag      uint16
+	Type Type
-	Name     string
-	Optional bool
-	Type     string
 }
 
-// ParseReader parses a protocol definition from a reader.
+func (field Field) String() string {
-func ParseReader(reader io.Reader) (*Protocol, error) {
+	return fmt.Sprintf("%s %v", field.Name, field.Type)
-	data, err := io.ReadAll(reader)
-	if err != nil { return nil, err }
-	protocol := new(Protocol)
-	err = protocol.UnmarshalText(data)
-	if err != nil { return nil, err }
-	return protocol, nil
 }
 
-// UnmarshalText unmarshals markdown-formatted text data into the protocol.
+type TypeNamed struct {
-func (this *Protocol) UnmarshalText(text []byte) error {
+	Name string
-	var state int; const (
-		stateIdle = iota
-		stateMessage
-		stateMessageDoc
-		stateMessageField
-	)
-
-	var message *Message
-	addMessage := func(method uint16, name string) {
-		this.Messages = append(this.Messages, Message {
-			Method: method,
-			Name:   name,
-		})
-		message = &this.Messages[len(this.Messages) - 1]
-	}
-
-	root := markdown.Parse(text, parser.New())
-	for _, node := range root.GetChildren() {
-		if node, ok := node.(*ast.Heading); ok {
-			if node.Level == 2 {
-				if removeBreaks(flatten(node)) == "Messages" {
-					state = stateMessage
-					continue
-				}
-			}
-
-			if node.Level > 3 {
-				state = stateIdle
-				continue
-			}
-
-			if state != stateIdle && node.Level == 3 {
-				heading := removeBreaks(flatten(node))
-				method, name, err := splitMessageHeading(heading)
-				if err != nil { return err }
-				addMessage(method, name)
-				state = stateMessageDoc
-			}
-		}
-
-		if state == stateIdle { continue }
-		if message == nil     { continue }
-
-		// TODO when we are adding text content to the doc comment, it
-		// might be wise to do stuff like indent lists and quotes so
-		// that go doc renders them correctly
-		switch node := node.(type) {
-		case *ast.Paragraph:
-			if message.Doc != "" { message.Doc += "\n\n" }
-			message.Doc += removeBreaks(flatten(node))
-		case *ast.BlockQuote:
-			if message.Doc != "" { message.Doc += "\n\n> " }
-			message.Doc += removeBreaks(flatten(node))
-		case *ast.List:
-			// FIXME format the list
-			if message.Doc != "" { message.Doc += "\n\n" }
-			message.Doc += removeBreaks(flatten(node))
-		case *ast.Table:
-			fields, err := processFieldTable(node)
-			if err != nil { return err}
-			message.Fields = append(message.Fields, fields...)
-		}
-	}
-	
-	return nil
 }
 
-func processFieldTable(node *ast.Table) ([]Field, error) {
+func (typ TypeNamed) String() string {
-	fields := []Field { }
+	return typ.Name
-	children := node.GetChildren()
-	if len(children) != 2 {
-		return nil, errors.New("malformed field table")
-	}
-
-	// get columns
-	columns := []string { }
-	if header, ok := children[0].(*ast.TableHeader); ok {
-		children := header.GetChildren()
-		if len(children) != 1 {
-			return nil, errors.New("malformed field table header")
-		}
-		if row, ok := header.Children[0].(*ast.TableRow); ok {
-			for _, cell := range row.GetChildren() {
-				if cell, ok := cell.(*ast.TableCell); ok {
-					columns = append(columns, flatten(cell))
-				}
-			}
-		} else {
-			return nil, errors.New("malformed field table header")
-		}
-		for index, column := range columns {
-			columns[index] = strings.ToLower(column)
-		}
-	} else {
-		return nil, errors.New("malformed field table: no header")
-	}
-
-	// get data
-	if body, ok := children[1].(*ast.TableBody); ok {
-		for _, node := range body.GetChildren() {
-		if row, ok := node.(*ast.TableRow); ok {
-			children := row.GetChildren()
-			if len(children) != len(columns) {
-				return nil, errors.New (
-					"malformed field table row: wrong " +
-					"number of columns")
-			}
-
-			field := Field { }
-		
-			for index, node := range children {
-			if cell, ok := node.(*ast.TableCell); ok {
-				text := flatten(cell)
-				switch columns[index] {
-				case "tag":
-					tag, err := parseTag(text)
-					if err != nil { return nil, err }
-					field.Tag = tag
-				case "name":
-					field.Name = text
-				case "required":
-					field.Optional = !parseBool(text)
-				case "optional":
-					field.Optional = parseBool(text)
-				case "type":
-					field.Type = text
-				case "comment", "purpose", "documentation":
-					field.Doc = text
-				}
-			}}
-
-			fields = append(fields, field)
-		}}
-	} else {
-		return nil, errors.New("malformed field table: no body")
-	}
-	return fields, nil
 }
 
-type nodeFlattener struct {
+func HashType(typ Type) [16]byte {
-	text string
+	// 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
-func (this *nodeFlattener) String() string { return this.text }
+	// (sorta)
-func (this *nodeFlattener) Visit(node ast.Node, entering bool) ast.WalkStatus {
+	return md5.Sum([]byte(typ.String()))
-	if entering {
-		if node := node.AsLeaf(); node != nil {
-			this.text += string(node.Literal)
-		}
-	}
-	return ast.GoToNext
-}
-func flatten(node ast.Node) string {
-	flattener := new(nodeFlattener)
-	ast.Walk(node, flattener)
-	return flattener.text
-}
-
-
-func removeBreaks(text string) string {
-	text = strings.ReplaceAll(text, "\n", " ")
-	text = strings.ReplaceAll(text, "\r", "")
-	return text
-}
-
-func parseBool(text string) bool {
-	switch(strings.ToLower(text)) {
-	case "yes": return true
-	case "no": return false
-	case "true": return true
-	case "false": return false
-	}
-	return false
-}
-
-func parseTag(text string) (uint16, error) {
-	tag, err := strconv.ParseUint(text, 10, 16)
-	if err != nil {
-		return 0, fmt.Errorf("malformed tag '%s': %w", text, err)
-	}
-	return uint16(tag), nil
-}
-
-func splitMessageHeading(text string) (uint16, string, error) {
-	text = strings.TrimSpace(text)
-	methodText, name, ok := strings.Cut(text, " ")
-	if !ok {
-		return 0, "", fmt.Errorf(
-			"malformed message heading '%s': no message name",
-			text)
-	}
-	method, err := strconv.ParseUint(methodText, 16, 16)
-	if err != nil {
-		return 0, "", fmt.Errorf(
-			"malformed method number '%s': %w",
-			methodText, err)
-	}
-	name = strings.TrimSpace(name)
-	return uint16(method), name, nil
 }

go.mod

@@ -4,19 +4,5 @@ go 1.23.0
 
 require (
 	git.tebibyte.media/sashakoshka/go-util v0.9.1
-	github.com/gomarkdown/markdown v0.0.0-20241205020045-f7e15b2f3e62
+	git.tebibyte.media/sashakoshka/goparse v0.2.0
-	github.com/quic-go/quic-go v0.48.2
-)
-
-require (
-	github.com/go-task/slim-sprig v0.0.0-20230315185526-52ccab3ef572 // indirect
-	github.com/google/pprof v0.0.0-20210407192527-94a9f03dee38 // indirect
-	github.com/onsi/ginkgo/v2 v2.9.5 // indirect
-	go.uber.org/mock v0.4.0 // indirect
-	golang.org/x/crypto v0.26.0 // indirect
-	golang.org/x/exp v0.0.0-20240506185415-9bf2ced13842 // indirect
-	golang.org/x/mod v0.17.0 // indirect
-	golang.org/x/net v0.28.0 // indirect
-	golang.org/x/sys v0.23.0 // indirect
-	golang.org/x/tools v0.21.1-0.20240508182429-e35e4ccd0d2d // indirect
 )

go.sum

@@ -1,60 +1,4 @@
 git.tebibyte.media/sashakoshka/go-util v0.9.1 h1:eGAbLwYhOlh4aq/0w+YnJcxT83yPhXtxnYMzz6K7xGo=
 git.tebibyte.media/sashakoshka/go-util v0.9.1/go.mod h1:0Q1t+PePdx6tFYkRuJNcpM1Mru7wE6X+it1kwuOH+6Y=
-github.com/chzyer/logex v1.1.10/go.mod h1:+Ywpsq7O8HXn0nuIou7OrIPyXbp3wmkHB+jjWRnGsAI=
+git.tebibyte.media/sashakoshka/goparse v0.2.0 h1:uQmKvOCV2AOlCHEDjg9uclZCXQZzq2PxaXfZ1aIMiQI=
-github.com/chzyer/readline v0.0.0-20180603132655-2972be24d48e/go.mod h1:nSuG5e5PlCu98SY8svDHJxuZscDgtXS6KTTbou5AhLI=
+git.tebibyte.media/sashakoshka/goparse v0.2.0/go.mod h1:tSQwfuD+EujRoKr6Y1oaRy74ZynatzkRLxjE3sbpCmk=
-github.com/chzyer/test v0.0.0-20180213035817-a1ea475d72b1/go.mod h1:Q3SI9o4m/ZMnBNeIyt5eFwwo7qiLfzFZmjNmxjkiQlU=
-github.com/davecgh/go-spew v1.1.0/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
-github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c=
-github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
-github.com/go-logr/logr v1.2.4 h1:g01GSCwiDw2xSZfjJ2/T9M+S6pFdcNtFYsp+Y43HYDQ=
-github.com/go-logr/logr v1.2.4/go.mod h1:jdQByPbusPIv2/zmleS9BjJVeZ6kBagPoEUsqbVz/1A=
-github.com/go-task/slim-sprig v0.0.0-20230315185526-52ccab3ef572 h1:tfuBGBXKqDEevZMzYi5KSi8KkcZtzBcTgAUUtapy0OI=
-github.com/go-task/slim-sprig v0.0.0-20230315185526-52ccab3ef572/go.mod h1:9Pwr4B2jHnOSGXyyzV8ROjYa2ojvAY6HCGYYfMoC3Ls=
-github.com/golang/protobuf v1.5.3 h1:KhyjKVUg7Usr/dYsdSqoFveMYd5ko72D+zANwlG1mmg=
-github.com/golang/protobuf v1.5.3/go.mod h1:XVQd3VNwM+JqD3oG2Ue2ip4fOMUkwXdXDdiuN0vRsmY=
-github.com/gomarkdown/markdown v0.0.0-20241205020045-f7e15b2f3e62 h1:pbAFUZisjG4s6sxvRJvf2N7vhpCvx2Oxb3PmS6pDO1g=
-github.com/gomarkdown/markdown v0.0.0-20241205020045-f7e15b2f3e62/go.mod h1:JDGcbDT52eL4fju3sZ4TeHGsQwhG9nbDV21aMyhwPoA=
-github.com/google/go-cmp v0.6.0 h1:ofyhxvXcZhMsU5ulbFiLKl/XBFqE1GSq7atu8tAmTRI=
-github.com/google/go-cmp v0.6.0/go.mod h1:17dUlkBOakJ0+DkrSSNjCkIjxS6bF9zb3elmeNGIjoY=
-github.com/google/pprof v0.0.0-20210407192527-94a9f03dee38 h1:yAJXTCF9TqKcTiHJAE8dj7HMvPfh66eeA2JYW7eFpSE=
-github.com/google/pprof v0.0.0-20210407192527-94a9f03dee38/go.mod h1:kpwsk12EmLew5upagYY7GY0pfYCcupk39gWOCRROcvE=
-github.com/ianlancetaylor/demangle v0.0.0-20200824232613-28f6c0f3b639/go.mod h1:aSSvb/t6k1mPoxDqO4vJh6VOCGPwU4O0C2/Eqndh1Sc=
-github.com/onsi/ginkgo/v2 v2.9.5 h1:+6Hr4uxzP4XIUyAkg61dWBw8lb/gc4/X5luuxN/EC+Q=
-github.com/onsi/ginkgo/v2 v2.9.5/go.mod h1:tvAoo1QUJwNEU2ITftXTpR7R1RbCzoZUOs3RonqW57k=
-github.com/onsi/gomega v1.27.6 h1:ENqfyGeS5AX/rlXDd/ETokDz93u0YufY1Pgxuy/PvWE=
-github.com/onsi/gomega v1.27.6/go.mod h1:PIQNjfQwkP3aQAH7lf7j87O/5FiNr+ZR8+ipb+qQlhg=
-github.com/pmezard/go-difflib v1.0.0 h1:4DBwDE0NGyQoBHbLQYPwSUPoCMWR5BEzIk/f1lZbAQM=
-github.com/pmezard/go-difflib v1.0.0/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
-github.com/quic-go/quic-go v0.48.2 h1:wsKXZPeGWpMpCGSWqOcqpW2wZYic/8T3aqiOID0/KWE=
-github.com/quic-go/quic-go v0.48.2/go.mod h1:yBgs3rWBOADpga7F+jJsb6Ybg1LSYiQvwWlLX+/6HMs=
-github.com/stretchr/objx v0.1.0/go.mod h1:HFkY916IF+rwdDfMAkV7OtwuqBVzrE8GR6GFx+wExME=
-github.com/stretchr/testify v1.6.1/go.mod h1:6Fq8oRcR53rry900zMqJjRRixrwX3KX962/h/Wwjteg=
-github.com/stretchr/testify v1.9.0 h1:HtqpIVDClZ4nwg75+f6Lvsy/wHu+3BoSGCbBAcpTsTg=
-github.com/stretchr/testify v1.9.0/go.mod h1:r2ic/lqez/lEtzL7wO/rwa5dbSLXVDPFyf8C91i36aY=
-go.uber.org/mock v0.4.0 h1:VcM4ZOtdbR4f6VXfiOpwpVJDL6lCReaZ6mw31wqh7KU=
-go.uber.org/mock v0.4.0/go.mod h1:a6FSlNadKUHUa9IP5Vyt1zh4fC7uAwxMutEAscFbkZc=
-golang.org/x/crypto v0.26.0 h1:RrRspgV4mU+YwB4FYnuBoKsUapNIL5cohGAmSH3azsw=
-golang.org/x/crypto v0.26.0/go.mod h1:GY7jblb9wI+FOo5y8/S2oY4zWP07AkOJ4+jxCqdqn54=
-golang.org/x/exp v0.0.0-20240506185415-9bf2ced13842 h1:vr/HnozRka3pE4EsMEg1lgkXJkTFJCVUX+S/ZT6wYzM=
-golang.org/x/exp v0.0.0-20240506185415-9bf2ced13842/go.mod h1:XtvwrStGgqGPLc4cjQfWqZHG1YFdYs6swckp8vpsjnc=
-golang.org/x/mod v0.17.0 h1:zY54UmvipHiNd+pm+m0x9KhZ9hl1/7QNMyxXbc6ICqA=
-golang.org/x/mod v0.17.0/go.mod h1:hTbmBsO62+eylJbnUtE2MGJUyE7QWk4xUqPFrRgJ+7c=
-golang.org/x/net v0.28.0 h1:a9JDOJc5GMUJ0+UDqmLT86WiEy7iWyIhz8gz8E4e5hE=
-golang.org/x/net v0.28.0/go.mod h1:yqtgsTWOOnlGLG9GFRrK3++bGOUEkNBoHZc8MEDWPNg=
-golang.org/x/sync v0.8.0 h1:3NFvSEYkUoMifnESzZl15y791HH1qU2xm6eCJU5ZPXQ=
-golang.org/x/sync v0.8.0/go.mod h1:Czt+wKu1gCyEFDUtn0jG5QVvpJ6rzVqr5aXyt9drQfk=
-golang.org/x/sys v0.0.0-20191204072324-ce4227a45e2e/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.23.0 h1:YfKFowiIMvtgl1UERQoTPPToxltDeZfbj4H7dVUCwmM=
-golang.org/x/sys v0.23.0/go.mod h1:/VUhepiaJMQUp4+oa/7Zr1D23ma6VTLIYjOOTFZPUcA=
-golang.org/x/text v0.17.0 h1:XtiM5bkSOt+ewxlOE/aE/AKEHibwj/6gvWMl9Rsh0Qc=
-golang.org/x/text v0.17.0/go.mod h1:BuEKDfySbSR4drPmRPG/7iBdf8hvFMuRexcpahXilzY=
-golang.org/x/time v0.5.0 h1:o7cqy6amK/52YcAKIPlM3a+Fpj35zvRj2TP+e1xFSfk=
-golang.org/x/time v0.5.0/go.mod h1:3BpzKBy/shNhVucY/MWOyx10tF3SFh9QdLuxbVysPQM=
-golang.org/x/tools v0.21.1-0.20240508182429-e35e4ccd0d2d h1:vU5i/LfpvrRCpgM/VPfJLg5KjxD3E+hfT1SH+d9zLwg=
-golang.org/x/tools v0.21.1-0.20240508182429-e35e4ccd0d2d/go.mod h1:aiJjzUbINMkxbQROHiO6hDPo2LHcIPhhQsa9DLh0yGk=
-google.golang.org/protobuf v1.33.0 h1:uNO2rsAINq/JlFpSdYEKIZ0uKD/R9cpdv0T+yoGwGmI=
-google.golang.org/protobuf v1.33.0/go.mod h1:c6P6GXX6sHbq/GpV6MGZEdwhWPcYBgnhAHhKbcUYpos=
-gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
-gopkg.in/yaml.v3 v3.0.0-20200313102051-9f266ea9e77c/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=
-gopkg.in/yaml.v3 v3.0.1 h1:fxVm/GzAzEWqLHuvctI91KS9hhNmmWOoWu0XTYJS7CA=
-gopkg.in/yaml.v3 v3.0.1/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=

internal/testutil/testutil.go

@@ -0,0 +1,161 @@
+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
+}
+
+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) }
+}

listen.go

@@ -1,9 +1,8 @@
 package hopp
 
 import "net"
-import "context"
+import "errors"
 import "crypto/tls"
-import "github.com/quic-go/quic-go"
 
 // Listener is an object which listens for incoming HOPP connections.
 type Listener interface {
@@ -17,7 +16,8 @@ type Listener interface {
 }
 
 // Listen listens for incoming HOPP connections. The network must be one of
-// "quic", "quic4", (IPv4-only) "quic6" (IPv6-only), or "unix".
+// "quic", "quic4", (IPv4-only) "quic6" (IPv6-only), or "unix". For now, quic is
+// not supported.
 func Listen(network, address string) (Listener, error) {
 	switch network {
 	case "quic", "quic4", "quic6": return ListenQUIC(network, address, nil)
@@ -30,19 +30,8 @@ func Listen(network, address string) (Listener, error) {
 // The network must be one of "quic", "quic4", (IPv4-only) or "quic6"
 // (IPv6-only).
 func ListenQUIC(network, address string, tlsConf *tls.Config) (Listener, error) {
-	tlsConf   = tlsConfig(tlsConf)
+	// tlsConf   = tlsConfig(tlsConf)
-	quicConf := quicConfig()
+	return nil, errors.New("quic is not yet implemented")
-	udpNetwork, err := quicNetworkToUDPNetwork(network)
-	if err != nil { return nil, err }
-	addr, err := net.ResolveUDPAddr(udpNetwork, address)
-	if err != nil { return nil, err }
-	udpListener, err := net.ListenUDP(udpNetwork, addr)
-	if err != nil { return nil, err }
-	quicListener, err := quic.Listen(udpListener, tlsConf, quicConf)
-	if err != nil { return nil, err }
-	return &listenerQUIC {
-		underlying: quicListener,
-	}, nil
 }
 
 // ListenUnix listens for incoming HOPP connections using a Unix domain socket
@@ -58,24 +47,6 @@ func ListenUnix(network, address string) (Listener, error) {
 	}, nil
 }
 
-type listenerQUIC struct {
-	underlying *quic.Listener
-}
-
-func (this *listenerQUIC) Accept() (Conn, error) {
-	conn, err := this.underlying.Accept(context.Background())
-	if err != nil { return nil, err }
-	return AdaptB(quicMultiConn { underlying: conn }), nil
-}
-
-func (this *listenerQUIC) Close() error {
-	return this.underlying.Close()
-}
-
-func (this *listenerQUIC) Addr() net.Addr {
-	return this.underlying.Addr()
-}
-
 type listenerUnix struct {
 	underlying *net.UnixListener
 }

quicwrap.go

@@ -1,54 +0,0 @@
-package hopp
-
-import "net"
-import "context"
-import "github.com/quic-go/quic-go"
-
-var _ MultiConn = quicMultiConn { }
-type quicMultiConn struct {
-	underlying quic.Connection
-}
-
-func (conn quicMultiConn) Close() error {
-	return conn.underlying.CloseWithError(0, "good bye")
-}
-
-func (conn quicMultiConn) LocalAddr() net.Addr {
-	return conn.underlying.LocalAddr()
-}
-
-func (conn quicMultiConn) RemoteAddr() net.Addr {
-	return conn.underlying.RemoteAddr()
-}
-
-func (conn quicMultiConn) AcceptStream(ctx context.Context) (Stream, error) {
-	strea, err := conn.underlying.AcceptStream(ctx)
-	if err != nil { return nil, err }
-	return quicStream { underlying: strea }, nil
-}
-
-func (conn quicMultiConn) OpenStream() (Stream, error) {
-	strea, err := conn.underlying.OpenStream()
-	if err != nil { return nil, err }
-	return quicStream { underlying: strea }, nil
-}
-
-type quicStream struct {
-	underlying quic.Stream
-}
-
-func (strea quicStream) Read(buffer []byte) (n int, err error) {
-	return strea.underlying.Read(buffer)
-}
-
-func (strea quicStream) Write(buffer []byte) (n int, err error) {
-	return strea.underlying.Read(buffer)
-}
-
-func (strea quicStream) Close() error {
-	return strea.underlying.Close()
-}
-
-func (strea quicStream) ID() int64 {
-	return int64(strea.underlying.StreamID())
-}

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 +1,392 @@
+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
+
+// 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: return EncodeAny(encoder, []byte(reflectValue.String()), tag)
+	}
+	if reflectValue.CanConvert(reflect.TypeOf(dummyBuffer)) {
+		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:
+		// return encodeAnySlice(encoder, reflect.ValueOf(value).Slice(0, reflectType.Len()).Interface(), tag)
+	case reflect.Map:
+		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.
+func decodeAny(decoder *Decoder, destination reflect.Value, tag Tag) (n int, err error) {
+	errWrongDestinationType := func(expected string) error {
+		panic(fmt.Errorf(
+		// return fmt.Errorf(
+			"expected %s destination, not %v",
+			expected, destination))
+	}
+	
+	switch tag.WithoutCN() {
+	case SI:
+		// SI: (none)
+		err = setInt(destination, uint64(tag.CN()))
+		if err != nil { return n, err }
+	case LI:
+		// LI: <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>*
+		buffer := make([]byte, tag.CN())
+		nn, err := decoder.Read(buffer)
+		n += nn; if err != nil { return n, err }
+		err = setByteArray(destination, buffer)
+		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 }
+		buffer := make([]byte, length)
+		nn, err = decoder.Read(buffer)
+		n += nn; if err != nil { return n, err }
+		err = setByteArray(destination, buffer)
+		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 }
+		if destination.Kind() != reflect.Slice {
+			return n, errWrongDestinationType("slice")
+		}
+		if destination.Cap() < int(length) {
+			destination.Grow(int(length) - destination.Cap())
+		}
+		destination.SetLen(int(length))
+		for index := range length {
+			nn, err := decodeAny(decoder, destination.Index(int(index)), oneTag)
+			n += nn; if err != nil { return n, err }
+		}
+	case KTV:
+		// KTV: <length: UN> (<key: U16> <tag: Tag> <value>)*
+		table := destination
+		if table.Type() != reflect.TypeOf(dummyMap) {
+			return n, errWrongDestinationType("map[uint16] any")
+		}
+		length, nn, err := decoder.ReadUintN(tag.CN() + 1)
+		n += nn; if err != nil { return n, err }
+		table.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 }
+			table.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) {
+	// TODO use reflection for all of this to ignore type names
+	// primitives
+	switch value := value.(type) {
+	case int,   uint:   return LI.WithCN(3), nil
+	case int8,  uint8:  return LI.WithCN(0), nil
+	case int16, uint16: return LI.WithCN(1), nil
+	case int32, uint32: return LI.WithCN(3), nil
+	case int64, uint64: return LI.WithCN(7), nil
+	case string:        return bufferLenTag(len(value)), nil
+	case []byte:        return bufferLenTag(len(value)), nil
+	}
+	
+	// aggregates
+	reflectType := reflect.TypeOf(value)
+	switch reflectType.Kind() {
+	case reflect.Slice: return OTA.WithCN(IntBytes(uint64(reflect.ValueOf(value).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(reflect.ValueOf(value).Len())) - 1), nil
+		}
+		return 0, fmt.Errorf("cannot encode map key %T, key must be uint16", value)
+	}
+	return 0, fmt.Errorf("cannot get tag of type %T", value)
+}
+
+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()).Interface())
+	if err != nil { return n, err }
+	for index := 0; index < reflectValue.Len(); index += 1 {
+		item := reflectValue.Index(index).Interface()
+		itemTag, err := TagAny(item)
+		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() {
+		key   := iter.Key().Interface().(uint16)
+		value := iter.Value().Interface()
+		nn, err = encoder.WriteUint16(key)
+		n += nn; if err != nil { return n, err }
+		itemTag, err := TagAny(value)
+		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
+}
+
+// setInt expects a settable destination.
+func setInt(destination reflect.Value, value uint64) error {
+	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:
+		return fmt.Errorf("cannot assign integer to %T", destination.Interface())
+	}
+	return nil
+}
+
+// setFloat expects a settable destination.
+func setFloat(destination reflect.Value, value float64) error {
+	if !destination.CanFloat() {
+		return fmt.Errorf("cannot assign float to %T", destination.Interface())
+	}
+	destination.Set(reflect.ValueOf(value).Convert(destination.Type()))
+	return nil
+}
+
+// setByteArrayexpects a settable destination.
+func setByteArray(destination reflect.Value, value []byte) error {
+	typ := destination.Type()
+	if typ.Kind() != reflect.Slice {
+		return fmt.Errorf("cannot assign %T to ", value)
+	}
+	if typ.Elem() != reflect.TypeOf(byte(0)) {
+		return fmt.Errorf("cannot convert %T to *[]byte", value)
+	}
+	destination.Set(reflect.ValueOf(value))
+	return nil
+}
+
+// decodeAndSetInt expects a settable destination.
+func decodeAndSetInt(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 }
+	return n, setInt(destination, value)
+}
+
+// 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 }
+		return n, setFloat(destination, float64(value))
+	case 4:
+		value, nn, err := decoder.ReadFloat32()
+		n += nn; if err != nil { return n, err }
+		return n, setFloat(destination, float64(value))
+	}
+	return n, fmt.Errorf("cannot decode 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 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 +1,195 @@
+package tape
+
+import "fmt"
+import "bytes"
+import "testing"
+import "reflect"
+import tu "git.tebibyte.media/sashakoshka/hopp/internal/testutil"
+
+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, LI.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} },
+	}, KTV.WithCN(0), tu.S(5).AddVar(
+		[]byte {
+			0xF3, 0xB9,
+			byte(LI.WithCN(3)),
+			0, 0, 0, 1,
+		},
+		[]byte {
+			0x01, 0x02,
+			byte(LI.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,
+		},
+	))
+	if err != nil { test.Fatal(err) }
+}
+
+func TestEncodeDecodeAnyTable(test *testing.T) {
+	err := testEncodeDecodeAny(test, map[uint16] any {
+		0xF3B9: uint32(1),
+		0x0102: uint32(2),
+		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")
+	}
+	if ok, n := correctBytes.Check(bytes); !ok {
+		return fmt.Errorf("bytes not equal: %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/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/pairs.go

@@ -1,83 +0,0 @@
-package tape
-
-import "iter"
-
-// DecodePairs decodes message tag/value pairs from a byte slice. It returns an
-// iterator over all pairs, where the first value is the tag and the second is
-// the value. If data yielded by the iterator is retained, it must be copied
-// first.
-func DecodePairs(data []byte) (iter.Seq2[uint16, []byte], error) {
-	// determine section bounds
-	if len(data) < 2 { return nil, ErrDataTooLarge }
-	length16, _ := DecodeI16[uint16](data[0:2])
-	data = data[2:]
-	length := int(length16)
-	headerSize := length * 4
-	if len(data) < headerSize { return nil, ErrDataTooLarge }
-	valuesData := data[headerSize:]
-
-	// ensure the value buffer is big enough
-	var valuesSize int
-	for index := range length {
-		offset := index * 4
-		end, _ := DecodeI16[uint16](data[offset + 2:offset + 4])
-		valuesSize = int(end)
-	}
-	if valuesSize > len(valuesData) {
-		return nil, ErrDataTooLarge
-	}
-
-	// return iterator
-	return func(yield func(uint16, []byte) bool) {
-		start := uint16(0)
-		for index := range length {
-			offset := index * 4
-			key , _ := DecodeI16[uint16](data[offset + 0:offset + 2])
-			end,  _ := DecodeI16[uint16](data[offset + 2:offset + 4])
-			// if nextValuesOffset < len(valuesData) {
-				if !yield(key, valuesData[start:end]) {
-					return
-				}
-			// } else {
-				// if !yield(key, nil) {
-					// return
-				// }
-			// }
-			start = end
-		}
-	}, nil
-}
-
-// EncodePairs encodes message tag/value pairs into a byte slice.
-func EncodePairs(pairs map[uint16] []byte) ([]byte, error) {
-	// determine section bounds
-	headerSize := 2 + len(pairs) * 4
-	valuesSize := 0
-	for _, value := range pairs {
-		valuesSize += len(value)
-	}
-
-	// generate data
-	buffer := make([]byte, headerSize + valuesSize)
-	length16, ok := U16CastSafe(len(pairs))
-	if !ok { return nil, ErrDataTooLarge }
-	EncodeI16[uint16](buffer[0:2], length16)
-	index  := 0
-	end    := headerSize
-	for key, value := range pairs {
-		start := end
-		end += len(value)
-		tagOffset := 2 + index * 4
-		end16, ok := U16CastSafe(end - headerSize)
-		if !ok { return nil, ErrDataTooLarge }
-		
-		// write tag and length
-		EncodeI16[uint16](buffer[tagOffset + 0:tagOffset + 2], key)
-		EncodeI16[uint16](buffer[tagOffset + 2:tagOffset + 4], end16)
-
-		// write value
-		copy(buffer[start:end], value)
-		index ++
-	}
-	return buffer, nil
-}

tape/pairs_test.go

@@ -1,62 +0,0 @@
-package tape
-
-import "slices"
-import "testing"
-
-func TestDecodePairs(test *testing.T) {
-	pairs := map[uint16] []byte {
-		3894: []byte("foo"),
-		7:    []byte("br"),
-	}
-	got, err := DecodePairs([]byte {
-		  0,   2,
-		  0,   7,   0,  2,
-		 15,  54,   0,  5,
-		 98, 114,
-		102, 111, 111})
-	if err != nil { test.Fatal(err) }
-	length := 0
-	for key, value := range got {
-		test.Log(key, value)
-		if !slices.Equal(pairs[key], value) { test.Fatal("not equal") }
-		length ++
-	}
-	test.Log("length")
-	if length != len(pairs) { test.Fatal("wrong length") }
-}
-
-func TestEncodePairs(test *testing.T) {
-	pairs := map[uint16] []byte {
-		3894: []byte("foo"),
-		7:    []byte("br"),
-	}
-	got, err := EncodePairs(pairs)
-	if err != nil { test.Fatal(err) }
-	test.Log(got)
-	valid := slices.Equal(got, []byte {
-		  0,   2,
-		 15,  54,   0,  3,
-		  0,   7,   0,  5,
-		102, 111, 111,
-		 98, 114}) ||
-	slices.Equal(got, []byte {
-		  0,   2,
-		  0,   7,   0,  2,
-		 15,  54,   0,  5,
-		 98, 114,
-		102, 111, 111})
-	if !valid { test.Fatal("not equal") }
-}
-
-func FuzzDecodePairs(fuzz *testing.F) {
-	fuzz.Add([]byte {
-		  0,   2,
-		  0,   7,   0,  2,
-		 15,  54,   0,  5,
-		 98, 114,
-		102, 111, 111})
-	fuzz.Fuzz(func(t *testing.T, buffer []byte) {
-		// ensure it does not panic :P
-		DecodePairs(buffer)
-	})
-}

tape/tag.go

@@ -0,0 +1,68 @@
+package tape
+
+import "fmt"
+
+type Tag byte; const (
+	SI      Tag = 0 << 5 // Small integer
+	LI      Tag = 1 << 5 // Large integer
+	FP      Tag = 2 << 5 // Floating point
+	SBA     Tag = 3 << 5 // Small byte array
+	LBA     Tag = 4 << 5 // Large byte array
+	OTA     Tag = 5 << 5 // One-tag array
+	KTV     Tag = 6 << 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 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)))
+	}
+}

tape/types.go

@@ -1,311 +0,0 @@
-// Package tape implements Table Pair Encoding.
-package tape
-
-import "fmt"
-
-const dataMaxSize = 0xFFFF
-const uint16Max = 0xFFFF
-
-// Error enumerates common errors in this package.
-type Error string; const (
-	ErrWrongBufferLength Error = "wrong buffer length"
-	ErrDataTooLarge      Error = "data too large"
-)
-
-// Error implements the error interface.
-func (err Error) Error() string {
-	return string(err)
-}
-
-// Int8 is any 8-bit integer.
-type Int8  interface { ~uint8  | ~int8 }
-// Int16 is any 16-bit integer.
-type Int16 interface { ~uint16 | ~int16 }
-// Int32 is any 32-bit integer.
-type Int32 interface { ~uint32 | ~int32 }
-// Int64 is any 64-bit integer.
-type Int64 interface { ~uint64 | ~int64 }
-// String is any string.
-type String interface { ~string }
-
-// DecodeI8 decodes an 8 bit integer from the given data.
-func DecodeI8[T Int8](data []byte) (T, error) {
-	if len(data) != 1 { return 0, fmt.Errorf("decoding int8: %w", ErrWrongBufferLength) }
-	return T(data[0]), nil
-}
-
-// EncodeI8 encodes an 8 bit integer into the given buffer.
-func EncodeI8[T Int8](buffer []byte, value T) error {
-	if len(buffer) != 1 { return fmt.Errorf("encoding int8: %w", ErrWrongBufferLength) }
-	buffer[0] = byte(value)
-	return nil
-}
-
-// DecodeI16 decodes a 16 bit integer from the given data.
-func DecodeI16[T Int16](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 Int16](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
-}
-
-// DecodeI32 decodes a 32 bit integer from the given data.
-func DecodeI32[T Int32](data []byte) (T, error) {
-	if len(data) != 4 { return 0, fmt.Errorf("decoding int32: %w", ErrWrongBufferLength) }
-	return T(data[0]) << 24 |
-		T(data[1]) << 16 |
-		T(data[2]) <<  8 |
-		T(data[3]), nil
-}
-
-// EncodeI32 encodes a 32 bit integer into the given buffer.
-func EncodeI32[T Int32](buffer []byte, value T) error {
-	if len(buffer) != 4 { return fmt.Errorf("encoding int32: %w", ErrWrongBufferLength) }
-	buffer[0] = byte(value >> 24)
-	buffer[1] = byte(value >> 16)
-	buffer[2] = byte(value >> 8)
-	buffer[3] = byte(value)
-	return nil
-}
-
-// DecodeI64 decodes a 64 bit integer from the given data.
-func DecodeI64[T Int64](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 Int64](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
-}
-
-// DecodeString decodes a string from the given data.
-func DecodeString[T String](data []byte) (T, error) {
-	return T(data), nil
-}
-
-// EncodeString encodes a string into the given buffer.
-func EncodeString[T String](data []byte, value T) error {
-	if len(data) != len(value) { return fmt.Errorf("encoding string: %w", ErrWrongBufferLength) }
-	copy(data, value)
-	return nil
-}
-
-// StringSize returns the size of a string. Returns 0 and an error if the size
-// is too large.
-func StringSize[T String](value T) (int, error) {
-	if len(value) > dataMaxSize { return 0, ErrDataTooLarge }
-	return len(value), nil
-}
-
-// DecodeStringArray decodes a packed string array from the given data.
-func DecodeStringArray[T String](data []byte) ([]T, error) {
-	result := []T { }
-	for len(data) > 0 {
-		if len(data) < 2 { return nil, fmt.Errorf("decoding []string: %w", ErrWrongBufferLength) }
-		itemSize16, _ := DecodeI16[uint16](data[:2])
-		itemSize := int(itemSize16)
-		data = data[2:]
-		if len(data) < itemSize { return nil, fmt.Errorf("decoding []string: %w", ErrWrongBufferLength) }
-		result = append(result, T(data[:itemSize]))
-		data = data[itemSize:]
-	}
-	return result, nil
-}
-
-// EncodeStringArray encodes a packed string array into the given buffer.
-func EncodeStringArray[T String](buffer []byte, value []T) error {
-	for _, item := range value {
-		length, err := StringSize(item)
-		if err != nil { return err }
-		if len(buffer) < 2 + length { return fmt.Errorf("encoding []string: %w", ErrWrongBufferLength) }
-		EncodeI16(buffer[:2], uint16(length))
-		buffer = buffer[2:]
-		copy(buffer, item)
-		buffer = buffer[length:]
-	}
-	if len(buffer) > 0 { return fmt.Errorf("encoding []string: %w", ErrWrongBufferLength) }
-	return nil
-}
-
-// StringArraySize returns the size of a packed string array. Returns 0 and an
-// error if the size is too large.
-func StringArraySize[T String](value []T) (int, error) {
-	total := 0
-	for _, item := range value {
-		total += 2 + len(item)
-	}
-	if total > dataMaxSize { return 0, ErrDataTooLarge }
-	return total, nil
-}
-
-// DecodeI8Array decodes a packed array of 8 bit integers from the given data.
-func DecodeI8Array[T Int8](data []byte) ([]T, error) {
-	result := make([]T, len(data))
-	for index, item := range data {
-		result[index] = T(item)
-	}
-	return result, nil
-}
-
-// EncodeI8Array encodes a packed array of 8 bit integers into the given buffer.
-func EncodeI8Array[T Int8](buffer []byte, value []T) error {
-	if len(buffer) != len(value) { return fmt.Errorf("encoding []int8: %w", ErrWrongBufferLength) }
-	for index, item := range value {
-		buffer[index] = byte(item)
-	}
-	return nil
-}
-
-// I8ArraySize returns the size of a packed 8 bit integer array. Returns 0 and
-// an error if the size is too large.
-func I8ArraySize[T Int8](value []T) (int, error) {
-	total := len(value)
-	if total > dataMaxSize { return 0, ErrDataTooLarge }
-	return total, nil
-}
-
-// DecodeI16Array decodes a packed array of 16 bit integers from the given data.
-func DecodeI16Array[T Int16](data []byte) ([]T, error) {
-	if len(data) % 2 != 0 { return nil, fmt.Errorf("decoding []int16: %w", ErrWrongBufferLength) }
-	length := len(data) / 2
-	result := make([]T, length)
-	for index := range length {
-		offset := index * 2
-		result[index] = T(data[offset]) << 8 | T(data[offset + 1])
-	}
-	return result, nil
-}
-
-// EncodeI16Array encodes a packed array of 16 bit integers into the given buffer.
-func EncodeI16Array[T Int16](buffer []byte, value []T) error {
-	if len(buffer) != len(value) * 2 { return fmt.Errorf("encoding []int16: %w", ErrWrongBufferLength) }
-	for _, item := range value {
-		buffer[0] = byte(item >> 8)
-		buffer[1] = byte(item)
-		buffer = buffer[2:]
-	}
-	return nil
-}
-
-// I16ArraySize returns the size of a packed 16 bit integer array. Returns 0 and
-// an error if the size is too large.
-func I16ArraySize[T Int16](value []T) (int, error) {
-	total := len(value) * 2
-	if total > dataMaxSize { return 0, ErrDataTooLarge }
-	return total, nil
-}
-
-// DecodeI32Array decodes a packed array of 32 bit integers from the given data.
-func DecodeI32Array[T Int32](data []byte) ([]T, error) {
-	if len(data) % 4 != 0 { return nil, fmt.Errorf("decoding []int32: %w", ErrWrongBufferLength) }
-	length := len(data) / 4
-	result := make([]T, length)
-	for index := range length {
-		offset := index * 4
-		result[index] =
-			T(data[offset + 0]) << 24 |
-			T(data[offset + 1]) << 16 |
-			T(data[offset + 2]) <<  8 |
-			T(data[offset + 3])
-	}
-	return result, nil
-}
-
-// EncodeI32Array encodes a packed array of 32 bit integers into the given buffer.
-func EncodeI32Array[T Int32](buffer []byte, value []T) error {
-	if len(buffer) != len(value) * 4 { return fmt.Errorf("encoding []int32: %w", ErrWrongBufferLength) }
-	for _, item := range value {
-		buffer[0] = byte(item >> 24)
-		buffer[1] = byte(item >> 16)
-		buffer[2] = byte(item >> 8)
-		buffer[3] = byte(item)
-		buffer = buffer[4:]
-	}
-	return nil
-}
-
-// I32ArraySize returns the size of a packed 32 bit integer array. Returns 0 and
-// an error if the size is too large.
-func I32ArraySize[T Int32](value []T) (int, error) {
-	total := len(value) * 4
-	if total > dataMaxSize { return 0, ErrDataTooLarge }
-	return total, nil
-}
-
-// DecodeI64Array decodes a packed array of 32 bit integers from the given data.
-func DecodeI64Array[T Int64](data []byte) ([]T, error) {
-	if len(data) % 8 != 0 { return nil, fmt.Errorf("decoding []int64: %w", ErrWrongBufferLength) }
-	length := len(data) / 8
-	result := make([]T, length)
-	for index := range length {
-		offset := index * 8
-		result[index] =
-			T(data[offset + 0]) << 56 |
-			T(data[offset + 1]) << 48 |
-			T(data[offset + 2]) << 40 |
-			T(data[offset + 3]) << 32 |
-			T(data[offset + 4]) << 24 |
-			T(data[offset + 5]) << 16 |
-			T(data[offset + 6]) <<  8 |
-			T(data[offset + 7])
-	}
-	return result, nil
-}
-
-// EncodeI64Array encodes a packed array of 64 bit integers into the given buffer.
-func EncodeI64Array[T Int64](buffer []byte, value []T) error {
-	if len(buffer) != len(value) * 8 { return fmt.Errorf("encoding []int64: %w", ErrWrongBufferLength) }
-	for _, item := range value {
-		buffer[0] = byte(item >> 56)
-		buffer[1] = byte(item >> 48)
-		buffer[2] = byte(item >> 40)
-		buffer[3] = byte(item >> 32)
-		buffer[4] = byte(item >> 24)
-		buffer[5] = byte(item >> 16)
-		buffer[6] = byte(item >> 8)
-		buffer[7] = byte(item)
-		buffer = buffer[8:]
-	}
-	return nil
-}
-
-// I64ArraySize returns the size of a packed 64 bit integer array. Returns 0 and
-// an error if the size is too large.
-func I64ArraySize[T Int64](value []T) (int, error) {
-	total := len(value) * 8
-	if total > dataMaxSize { return 0, ErrDataTooLarge }
-	return total, nil
-}
-
-// U16CastSafe safely casts an integer to a uint16. If an overflow or underflow
-// occurs, it will return (0, false).
-func U16CastSafe(n int) (uint16, bool) {
-	if n < uint16Max && n >= 0 {
-		return uint16(n), true
-	} else {
-		return 0, false
-	}
-}

tape/types_test.go

@@ -1,292 +0,0 @@
-package tape
-
-import "slices"
-import "errors"
-import "testing"
-import "math/rand"
-
-const largeNumberNTestRounds = 2048
-const randStringBytes        = "-abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ"
-
-func TestI8(test *testing.T) {
-	var buffer [16]byte
-	err := EncodeI8[uint8](buffer[:], 5)
-	if err.Error() != "encoding int8: wrong buffer length" { test.Fatal(err) }
-	err = EncodeI8[uint8](buffer[:0], 5)
-	if err.Error() != "encoding int8: wrong buffer length" { test.Fatal(err) }
-	_, err = DecodeI8[uint8](buffer[:])
-	if err.Error() != "decoding int8: wrong buffer length" { test.Fatal(err) }
-	_, err = DecodeI8[uint8](buffer[:0])
-	if err.Error() != "decoding int8: wrong buffer length" { test.Fatal(err) }
-	
-	for number := range uint8(255) {
-		err := EncodeI8[uint8](buffer[:1], number)
-		if err != nil { test.Fatal(err) }
-		decoded, err := DecodeI8[uint8](buffer[:1])
-		if err != nil { test.Fatal(err) }
-		if decoded != number {
-			test.Fatalf("%d != %d", decoded, number)
-		}
-	}
-}
-
-func TestI16(test *testing.T) {
-	var buffer [16]byte
-	err := EncodeI16[uint16](buffer[:], 5)
-	if err.Error() != "encoding int16: wrong buffer length" { test.Fatal(err) }
-	err = EncodeI16[uint16](buffer[:0], 5)
-	if err.Error() != "encoding int16: wrong buffer length" { test.Fatal(err) }
-	_, err = DecodeI16[uint16](buffer[:])
-	if err.Error() != "decoding int16: wrong buffer length" { test.Fatal(err) }
-	_, err = DecodeI16[uint16](buffer[:0])
-	if err.Error() != "decoding int16: wrong buffer length" { test.Fatal(err) }
-	
-	for _ = range largeNumberNTestRounds {
-		number := uint16(rand.Int())
-		err := EncodeI16[uint16](buffer[:2], number)
-		if err != nil { test.Fatal(err) }
-		decoded, err := DecodeI16[uint16](buffer[:2])
-		if err != nil { test.Fatal(err) }
-		if decoded != number {
-			test.Fatalf("%d != %d", decoded, number)
-		}
-	}
-}
-
-func TestI32(test *testing.T) {
-	var buffer [16]byte
-	err := EncodeI32[uint32](buffer[:], 5)
-	if err.Error() != "encoding int32: wrong buffer length" { test.Fatal(err) }
-	err = EncodeI32[uint32](buffer[:0], 5)
-	if err.Error() != "encoding int32: wrong buffer length" { test.Fatal(err) }
-	_, err = DecodeI32[uint32](buffer[:])
-	if err.Error() != "decoding int32: wrong buffer length" { test.Fatal(err) }
-	_, err = DecodeI32[uint32](buffer[:0])
-	if err.Error() != "decoding int32: wrong buffer length" { test.Fatal(err) }
-	
-	for _ = range largeNumberNTestRounds {
-		number := uint32(rand.Int())
-		err := EncodeI32[uint32](buffer[:4], number)
-		if err != nil { test.Fatal(err) }
-		decoded, err := DecodeI32[uint32](buffer[:4])
-		if err != nil { test.Fatal(err) }
-		if decoded != number {
-			test.Fatalf("%d != %d", decoded, number)
-		}
-	}
-}
-
-func TestI64(test *testing.T) {
-	var buffer [16]byte
-	err := EncodeI64[uint64](buffer[:], 5)
-	if err.Error() != "encoding int64: wrong buffer length" { test.Fatal(err) }
-	err = EncodeI64[uint64](buffer[:0], 5)
-	if err.Error() != "encoding int64: wrong buffer length" { test.Fatal(err) }
-	_, err = DecodeI64[uint64](buffer[:])
-	if err.Error() != "decoding int64: wrong buffer length" { test.Fatal(err) }
-	_, err = DecodeI64[uint64](buffer[:0])
-	if err.Error() != "decoding int64: wrong buffer length" { test.Fatal(err) }
-	
-	for _ = range largeNumberNTestRounds {
-		number := uint64(rand.Int())
-		err := EncodeI64[uint64](buffer[:8], number)
-		if err != nil { test.Fatal(err) }
-		decoded, err := DecodeI64[uint64](buffer[:8])
-		if err != nil { test.Fatal(err) }
-		if decoded != number {
-			test.Fatalf("%d != %d", decoded, number)
-		}
-	}
-}
-
-func TestString(test *testing.T) {
-	var buffer [16]byte
-	err := EncodeString[string](buffer[:], "hello")
-	if !errIs(err, ErrWrongBufferLength, "encoding string: wrong buffer length") { test.Fatal(err) }
-	err = EncodeString[string](buffer[:0], "hello")
-	if !errIs(err, ErrWrongBufferLength, "encoding string: wrong buffer length") { test.Fatal(err) }
-	_, err = DecodeString[string](buffer[:])
-	if err != nil { test.Fatal(err) }
-	_, err = DecodeString[string](buffer[:0])
-	if err != nil { test.Fatal(err) }
-	
-	for _ = range largeNumberNTestRounds {
-		length := rand.Intn(16)
-		str := randString(length)
-		err := EncodeString[string](buffer[:length], str)
-		if err != nil { test.Fatal(err) }
-		decoded, err := DecodeString[string](buffer[:length])
-		if err != nil { test.Fatal(err) }
-		if decoded != str {
-			test.Fatalf("%s != %s", decoded, str)
-		}
-	}
-}
-
-func TestI8Array(test *testing.T) {
-	var buffer [64]byte
-	err := EncodeI8Array[uint8](buffer[:], []uint8 { 0, 4, 50, 19 })
-	if !errIs(err, ErrWrongBufferLength, "encoding []int8: wrong buffer length") { test.Fatal(err) }
-	err = EncodeI8Array[uint8](buffer[:0], []uint8 { 0, 4, 50, 19 })
-	if !errIs(err, ErrWrongBufferLength, "encoding []int8: wrong buffer length") { test.Fatal(err) }
-	_, err = DecodeI8Array[uint8](buffer[:])
-	if err != nil { test.Fatal(err) }
-	_, err = DecodeI8Array[uint8](buffer[:0])
-	if err != nil { test.Fatal(err) }
-	
-	for _ = range largeNumberNTestRounds {
-		array := randInts[uint8](rand.Intn(16))
-		length, _ := I8ArraySize(array)
-		if length != len(array) { test.Fatalf("%d != %d", length, len(array)) }
-		err := EncodeI8Array[uint8](buffer[:length], array)
-		if err != nil { test.Fatal(err) }
-		decoded, err := DecodeI8Array[uint8](buffer[:length])
-		if err != nil { test.Fatal(err) }
-		if !slices.Equal(decoded, array) {
-			test.Fatalf("%v != %v", decoded, array)
-		}
-	}
-}
-
-func TestI16Array(test *testing.T) {
-	var buffer [128]byte
-	err := EncodeI16Array[uint16](buffer[:], []uint16 { 0, 4, 50, 19 })
-	if !errIs(err, ErrWrongBufferLength, "encoding []int16: wrong buffer length") { test.Fatal(err) }
-	err = EncodeI16Array[uint16](buffer[:0], []uint16 { 0, 4, 50, 19 })
-	if !errIs(err, ErrWrongBufferLength, "encoding []int16: wrong buffer length") { test.Fatal(err) }
-	_, err = DecodeI16Array[uint16](buffer[:])
-	if err != nil { test.Fatal(err) }
-	_, err = DecodeI16Array[uint16](buffer[:0])
-	if err != nil { test.Fatal(err) }
-	
-	for _ = range largeNumberNTestRounds {
-		array := randInts[uint16](rand.Intn(16))
-		length, _ := I16ArraySize(array)
-		if length != 2 * len(array) { test.Fatalf("%d != %d", length, 2 * len(array)) }
-		err := EncodeI16Array[uint16](buffer[:length], array)
-		if err != nil { test.Fatal(err) }
-		decoded, err := DecodeI16Array[uint16](buffer[:length])
-		if err != nil { test.Fatal(err) }
-		if !slices.Equal(decoded, array) {
-			test.Fatalf("%v != %v", decoded, array)
-		}
-	}
-}
-
-func TestI32Array(test *testing.T) {
-	var buffer [256]byte
-	err := EncodeI32Array[uint32](buffer[:], []uint32 { 0, 4, 50, 19 })
-	if !errIs(err, ErrWrongBufferLength, "encoding []int32: wrong buffer length") { test.Fatal(err) }
-	err = EncodeI32Array[uint32](buffer[:0], []uint32 { 0, 4, 50, 19 })
-	if !errIs(err, ErrWrongBufferLength, "encoding []int32: wrong buffer length") { test.Fatal(err) }
-	_, err = DecodeI32Array[uint32](buffer[:])
-	if err != nil { test.Fatal(err) }
-	_, err = DecodeI32Array[uint32](buffer[:0])
-	if err != nil { test.Fatal(err) }
-	
-	for _ = range largeNumberNTestRounds {
-		array := randInts[uint32](rand.Intn(16))
-		length, _ := I32ArraySize(array)
-		if length != 4 * len(array) { test.Fatalf("%d != %d", length, 4 * len(array)) }
-		err := EncodeI32Array[uint32](buffer[:length], array)
-		if err != nil { test.Fatal(err) }
-		decoded, err := DecodeI32Array[uint32](buffer[:length])
-		if err != nil { test.Fatal(err) }
-		if !slices.Equal(decoded, array) {
-			test.Fatalf("%v != %v", decoded, array)
-		}
-	}
-}
-
-func TestI64Array(test *testing.T) {
-	var buffer [512]byte
-	err := EncodeI64Array[uint64](buffer[:], []uint64 { 0, 4, 50, 19 })
-	if !errIs(err, ErrWrongBufferLength, "encoding []int64: wrong buffer length") { test.Fatal(err) }
-	err = EncodeI64Array[uint64](buffer[:0], []uint64 { 0, 4, 50, 19 })
-	if !errIs(err, ErrWrongBufferLength, "encoding []int64: wrong buffer length") { test.Fatal(err) }
-	_, err = DecodeI64Array[uint64](buffer[:])
-	if err != nil { test.Fatal(err) }
-	_, err = DecodeI64Array[uint64](buffer[:0])
-	if err != nil { test.Fatal(err) }
-	
-	for _ = range largeNumberNTestRounds {
-		array := randInts[uint64](rand.Intn(16))
-		length, _ := I64ArraySize(array)
-		if length != 8 * len(array) { test.Fatalf("%d != %d", length, 8 * len(array)) }
-		err := EncodeI64Array[uint64](buffer[:length], array)
-		if err != nil { test.Fatal(err) }
-		decoded, err := DecodeI64Array[uint64](buffer[:length])
-		if err != nil { test.Fatal(err) }
-		if !slices.Equal(decoded, array) {
-			test.Fatalf("%v != %v", decoded, array)
-		}
-	}
-}
-
-func TestStringArray(test *testing.T) {
-	var buffer [8192]byte
-	err := EncodeStringArray[string](buffer[:], []string { "0", "4", "50", "19" })
-	if !errIs(err, ErrWrongBufferLength, "encoding []string: wrong buffer length") { test.Fatal(err) }
-	err = EncodeStringArray[string](buffer[:0], []string { "0", "4", "50", "19" })
-	if !errIs(err, ErrWrongBufferLength, "encoding []string: wrong buffer length") { test.Fatal(err) }
-	_, err = DecodeStringArray[string](buffer[:0])
-	if err != nil { test.Fatal(err) }
-	
-	for _ = range largeNumberNTestRounds {
-		array := randStrings[string](rand.Intn(16), 16)
-		length, _ := StringArraySize(array)
-		// TODO test length
-		err := EncodeStringArray[string](buffer[:length], array)
-		if err != nil { test.Fatal(err) }
-		decoded, err := DecodeStringArray[string](buffer[:length])
-		if err != nil { test.Fatal(err) }
-		if !slices.Equal(decoded, array) {
-			test.Fatalf("%v != %v", decoded, array)
-		}
-	}
-}
-
-func TestU16CastSafe(test *testing.T) {
-	number, ok := U16CastSafe(90_000)
-	if ok { test.Fatalf("false positive: %v, %v", number, ok) }
-	number, ok = U16CastSafe(-478)
-	if ok { test.Fatalf("false positive: %v, %v", number, ok) }
-	number, ok = U16CastSafe(3870)
-	if !ok { test.Fatalf("false negative: %v, %v", number, ok) }
-	if got, correct := number, uint16(3870); got != correct {
-		test.Fatalf("not equal: %v %v", got, correct)
-	}
-	number, ok = U16CastSafe(0)
-	if !ok { test.Fatalf("false negative: %v, %v", number, ok) }
-	if got, correct := number, uint16(0); got != correct {
-		test.Fatalf("not equal: %v %v", got, correct)
-	}
-}
-
-func randString(length int) string {
-	buffer := make([]byte, length)
-	for index := range buffer {
-		buffer[index] = randStringBytes[rand.Intn(len(randStringBytes))]
-	}
-	return string(buffer)
-}
-
-func randInts[T interface { ~uint | ~uint8 | ~uint16 | ~uint32 | ~uint64 }] (length int) []T {
-	buffer := make([]T, length)
-	for index := range buffer {
-		buffer[index] = T(rand.Int())
-	}
-	return buffer
-}
-
-func randStrings[T interface { ~string }] (length, maxItemLength int) []T {
-	buffer := make([]T, length)
-	for index := range buffer {
-		buffer[index] = T(randString(rand.Intn(maxItemLength)))
-	}
-	return buffer
-}
-
-func errIs(err error, wraps error, description string) bool {
-	return err != nil && (wraps == nil || errors.Is(err, wraps)) && err.Error() == description
-}