Move some of preserves implementation to preserves/private

This commit is contained in:
Emery Hemingway 2023-12-19 22:07:44 +02:00
parent 126365d164
commit adadcc181a
6 changed files with 681 additions and 666 deletions

View File

@ -1,12 +1,14 @@
# SPDX-FileCopyrightText: ☭ Emery Hemingway
# SPDX-License-Identifier: Unlicense
import std/[base64, endians, hashes, options, sets, sequtils, streams, strutils, tables, typetraits]
import std/[options, sets, sequtils, strutils, tables, typetraits]
import ./preserves/private/macros
from std/algorithm import sort
from std/json import escapeJson, escapeJsonUnquoted
import ./preserves/private/dot
import ./preserves/private/[encoding, decoding, dot, parsing, texts, values]
export encoding, decoding, parsing, texts, values
when defined(tracePreserves):
when defined(posix):
@ -16,194 +18,6 @@ when defined(tracePreserves):
else:
template trace(args: varargs[untyped]) = discard
type
PreserveKind* = enum
pkBoolean, pkFloat, pkDouble, pkSignedInteger, pkString, pkByteString, pkSymbol,
pkRecord, pkSequence, pkSet, pkDictionary, pkEmbedded
const
atomKinds* = {pkBoolean, pkFloat, pkDouble, pkSignedInteger, pkString, pkByteString, pkSymbol}
compoundKinds* = {pkRecord, pkSequence, pkSet, pkDictionary}
type Symbol* = distinct string
proc `<`*(x, y: Symbol): bool {.borrow.}
proc `==`*(x, y: Symbol): bool {.borrow.}
proc hash*(s: Symbol): Hash {.borrow.}
proc len*(s: Symbol): int {.borrow.}
proc `$`*(s: Symbol): string =
let sym = string s
if sym.len > 0 and sym[0] in {'A'..'z'} and not sym.anyIt(char(it) in { '\x00'..'\x19', '"', '\\', '|' }):
result = sym
else:
result = newStringOfCap(sym.len shl 1)
result.add('|')
for c in sym:
case c
of '\\':
result.add("\\\\")
of '/':
result.add("\\/")
of '\x08':
result.add("\\b")
of '\x0c':
result.add("\\f")
of '\x0a':
result.add("\\n")
of '\x0d':
result.add("\\r")
of '\x09':
result.add("\\t")
of '|':
result.add("\\|")
else:
result.add(c)
result.add('|')
type
Preserve*[E] = object
embedded*: bool
## Flag to mark embedded Preserves
case kind*: PreserveKind
of pkBoolean:
bool*: bool
of pkFloat:
float*: float32
of pkDouble:
double*: float64
of pkSignedInteger:
int*: BiggestInt
of pkString:
string*: string
of pkByteString:
bytes*: seq[byte]
of pkSymbol:
symbol*: Symbol
of pkRecord:
record*: seq[Preserve[E]] # label is last
of pkSequence:
sequence*: seq[Preserve[E]]
of pkSet:
set*: seq[Preserve[E]]
# TODO: HashSet
of pkDictionary:
dict*: seq[DictEntry[E]]
# TODO: Tables
of pkEmbedded:
embed*: E
DictEntry[E] = tuple[key: Preserve[E], val: Preserve[E]]
func `==`*[A, B](x: Preserve[A]; y: Preserve[B]): bool =
## Check `x` and `y` for equivalence.
if x.kind == y.kind and x.embedded == y.embedded:
case x.kind
of pkBoolean:
result = x.bool == y.bool
of pkFloat:
result = x.float == y.float
of pkDouble:
result = x.double == y.double
of pkSignedInteger:
result = x.int == y.int
of pkString:
result = x.string == y.string
of pkByteString:
result = x.bytes == y.bytes
of pkSymbol:
result = x.symbol == y.symbol
of pkRecord:
result = x.record.len == y.record.len
for i in 0..x.record.high:
if not result: break
result = result and (x.record[i] == y.record[i])
of pkSequence:
for i, val in x.sequence:
if y.sequence[i] != val: return false
result = true
of pkSet:
result = x.set.len == y.set.len
for i in 0..x.set.high:
if not result: break
result = result and (x.set[i] == y.set[i])
of pkDictionary:
result = x.dict.len == y.dict.len
for i in 0..x.dict.high:
if not result: break
result = result and
(x.dict[i].key == y.dict[i].key) and
(x.dict[i].val == y.dict[i].val)
of pkEmbedded:
when A is B:
when A is void:
result = true
else:
result = x.embed == y.embed
proc `<`(x, y: string | seq[byte]): bool =
for i in 0 .. min(x.high, y.high):
if x[i] < y[i]: return true
if x[i] != y[i]: return false
x.len < y.len
proc `<`*[A, B](x: Preserve[A]; y: Preserve[B]): bool =
## Preserves have a total order over values. Check if `x` is ordered before `y`.
if x.embedded != y.embedded:
result = y.embedded
elif x.kind != y.kind:
result = x.kind < y.kind
else:
case x.kind
of pkBoolean:
result = (not x.bool) and y.bool
of pkFloat:
result = x.float < y.float
of pkDouble:
result = x.double < y.double
of pkSignedInteger:
result = x.int < y.int
of pkString:
result = x.string < y.string
of pkByteString:
result = x.bytes < y.bytes
of pkSymbol:
result = x.symbol < y.symbol
of pkRecord:
if x.record[x.record.high] < y.record[y.record.high]: return true
for i in 0..<min(x.record.high, y.record.high):
if x.record[i] < y.record[i]: return true
if x.record[i] == y.record[i]: return false
result = x.record.len < y.record.len
of pkSequence:
for i in 0..min(x.sequence.high, y.sequence.high):
if x.sequence[i] < y.sequence[i]: return true
if x.sequence[i] != y.sequence[i]: return false
result = x.sequence.len < y.sequence.len
of pkSet:
for i in 0..min(x.set.high, y.set.high):
if x.set[i] < y.set[i]: return true
if x.set[i] != y.set[i]: return false
result = x.set.len < y.set.len
of pkDictionary:
for i in 0..min(x.dict.high, y.dict.high):
if x.dict[i].key < y.dict[i].key: return true
if x.dict[i].key == y.dict[i].key:
if x.dict[i].val < y.dict[i].val: return true
if x.dict[i].val != y.dict[i].val: return false
result = x.dict.len < y.dict.len
of pkEmbedded:
when (not A is void) and (A is B):
result = x.embed < y.embed
func cmp*[E](x, y: Preserve[E]): int =
## Compare by Preserves total ordering.
if x == y: 0
elif x < y: -1
else: 1
proc sort*[E](pr: var Preserve[E]) = sort(pr.sequence, cmp)
## Sort a Preserves array by total ordering.
proc sortDict[E](pr: var Preserve[E]) =
sort(pr.dict) do (x, y: DictEntry[E]) -> int:
cmp(x.key, y.key)
@ -223,64 +37,6 @@ proc cannonicalize*[E](pr: var Preserve[E]) =
else:
discard
proc hash*[E](pr: Preserve[E]): Hash =
## Produce a `Hash` of `pr` for use with a `HashSet` or `Table`.
var h = hash(pr.kind.int) !& hash(pr.embedded)
case pr.kind
of pkBoolean:
h = h !& hash(pr.bool)
of pkFloat:
h = h !& hash(pr.float)
of pkDouble:
h = h !& hash(pr.double)
of pkSignedInteger:
h = h !& hash(pr.int)
of pkString:
h = h !& hash(pr.string)
of pkByteString:
h = h !& hash(pr.bytes)
of pkSymbol:
h = h !& hash(string pr.symbol)
of pkRecord:
for val in pr.record:
h = h !& hash(val)
of pkSequence:
for val in pr.sequence:
h = h !& hash(val)
of pkSet:
for val in pr.set.items:
h = h !& hash(val)
of pkDictionary:
for (key, val) in pr.dict.items:
h = h !& hash(key) !& hash(val)
of pkEmbedded:
when E is void:
h = h !& hash(pr.embed)
else:
if pr.embed.isNil:
h = h !& hash(false)
else:
h = h !& hash(pr.embed)
!$h
proc `[]`*(pr: Preserve; i: int): Preserve =
## Select an indexed value from ``pr``.
## Only valid for records and sequences.
case pr.kind
of pkRecord: pr.record[i]
of pkSequence: pr.sequence[i]
else:
raise newException(ValueError, "Preserves value is not indexable")
proc `[]=`*(pr: var Preserve; i: Natural; val: Preserve) =
## Assign an indexed value into ``pr``.
## Only valid for records and sequences.
case pr.kind
of pkRecord: pr.record[i] = val
of pkSequence: pr.sequence[i] = val
else:
raise newException(ValueError, "`Preserves value is not indexable")
proc getOrDefault(pr: Preserve; key: Preserve): Preserve =
## Retrieves the value of `pr[key]` if `pr` is a dictionary containing `key`
## or returns the `#f` Preserves value.
@ -302,21 +58,6 @@ proc pop*(pr: var Preserve; key: Preserve; val: var Preserve): bool =
delete(pr.dict, i, i)
return true
proc incl*(pr: var Preserve; key: Preserve) =
## Include `key` in the Preserves set `pr`.
for i in 0..pr.set.high:
if key < pr.set[i]:
insert(pr.set, [key], i)
return
pr.set.add(key)
proc excl*(pr: var Preserve; key: Preserve) =
## Exclude `key` from the Preserves set `pr`.
for i in 0..pr.set.high:
if pr.set[i] == key:
delete(pr.set, i..i)
break
proc `[]`*(pr, key: Preserve): Preserve {.deprecated: "use step instead".} =
## Select a value by `key` from `pr`.
## Works for sequences, records, and dictionaries.
@ -362,17 +103,6 @@ func step*[E](pr: Preserve[E]; key: Symbol): Option[Preserve[E]] =
result = some(v)
break
proc `[]=`*(pr: var Preserve; key, val: Preserve) =
## Insert `val` by `key` in the Preserves dictionary `pr`.
for i in 0..pr.dict.high:
if key < pr.dict[i].key:
insert(pr.dict, [(key, val, )], i)
return
elif key == pr.dict[i].key:
pr.dict[i].val = val
return
pr.dict.add((key, val, ))
proc mget*(pr: var Preserve; key: Preserve): var Preserve =
## Select a value by `key` from the Preserves dictionary `pr`.
if pr.isDictionary:
@ -551,262 +281,6 @@ proc unembed*[E](pr: Preserve[E]): E =
raise newException(ValueError, "not an embedded value")
pr.embed
proc writeVarint(s: Stream; n: Natural) =
var n = n
while n > 0x7f:
s.write(uint8 n or 0x80)
n = n shr 7
s.write(uint8 n and 0x7f)
proc readVarint(s: Stream): uint =
var
shift = 0
c = uint s.readUint8
while (c and 0x80) == 0x80:
result = result or ((c and 0x7f) shl shift)
inc(shift, 7)
c = uint s.readUint8
result = result or (c shl shift)
proc write*[E](str: Stream; pr: Preserve[E]) =
## Write the binary-encoding of a Preserves value to a stream.
if pr.embedded: str.write(0x86'u8)
case pr.kind:
of pkBoolean:
case pr.bool
of false: str.write(0x80'u8)
of true: str.write(0x81'u8)
of pkFloat:
str.write("\x87\x04")
when system.cpuEndian == bigEndian:
str.write(pr.float)
else:
var be: float32
swapEndian32(be.addr, pr.float.unsafeAddr)
str.write(be)
of pkDouble:
str.write("\x87\x08")
when system.cpuEndian == bigEndian:
str.write(pr.double)
else:
var be: float64
swapEndian64(be.addr, pr.double.unsafeAddr)
str.write(be)
of pkSignedInteger:
if pr.int == 0:
str.write("\xb0\x00")
else:
var bitCount = 1'u8
if pr.int < 0:
while ((not pr.int) shr bitCount) != 0:
inc(bitCount)
else:
while (pr.int shr bitCount) != 0:
inc(bitCount)
var byteCount = (bitCount + 8) div 8
str.write(0xb0'u8)
str.writeVarint(byteCount)
proc write(n: uint8; i: BiggestInt) =
if n > 1:
write(n.pred, i shr 8)
str.write(i.uint8)
write(byteCount, pr.int)
of pkString:
str.write(0xb1'u8)
str.writeVarint(pr.string.len)
str.write(pr.string)
of pkByteString:
str.write(0xb2'u8)
str.writeVarint(pr.bytes.len)
str.write(cast[string](pr.bytes))
of pkSymbol:
str.write(0xb3'u8)
str.writeVarint(pr.symbol.len)
str.write(string pr.symbol)
of pkRecord:
assert(pr.record.len > 0)
str.write(0xb4'u8)
str.write(pr.record[pr.record.high])
for i in 0..<pr.record.high:
str.write(pr.record[i])
str.write(0x84'u8)
of pkSequence:
str.write(0xb5'u8)
for e in pr.sequence:
str.write(e)
str.write(0x84'u8)
of pkSet:
str.write(0xb6'u8)
for val in pr.set.items:
str.write(val)
str.write(0x84'u8)
of pkDictionary:
str.write(0xb7'u8)
for (key, value) in pr.dict.items:
str.write(key)
str.write(value)
str.write(0x84'u8)
of pkEmbedded:
str.write(0x86'u8)
str.write(pr.embed.toPreserve)
proc encode*[E](pr: Preserve[E]): seq[byte] =
## Return the binary-encoding of a Preserves value.
let s = newStringStream()
s.write pr
result = cast[seq[byte]](move s.data)
proc decodePreserves*(s: Stream; E = void): Preserve[E] =
## Decode a Preserves value from a binary-encoded stream.
if s.atEnd: raise newException(IOError, "End of Preserves stream")
const endMarker = 0x84
let tag = s.readUint8()
case tag
of 0x80: result = Preserve[E](kind: pkBoolean, bool: false)
of 0x81: result = Preserve[E](kind: pkBoolean, bool: true)
of 0x85:
discard decodePreserves(s, E)
result = decodePreserves(s, E)
of 0x86:
result = decodePreserves(s, E)
result.embedded = true
of 0x87:
let n = s.readUint8()
case n
of 4:
when system.cpuEndian == bigEndian:
result = Preserve[E](kind: pkFloat, float: s.readFloat32())
else:
result = Preserve[E](kind: pkFloat)
var be = s.readFloat32()
swapEndian32(result.float.addr, be.addr)
of 8:
when system.cpuEndian == bigEndian:
result = Preserve[E](kind: pkDouble, double: s.readFloat64())
else:
result = Preserve[E](kind: pkDouble)
var be = s.readFloat64()
swapEndian64(result.double.addr, be.addr)
else:
raise newException(IOError, "unhandled IEEE754 value of " & $n & " bytes")
of 0xb1:
var data = newString(s.readVarint())
if data.len > 0:
let n = s.readData(unsafeAddr data[0], data.len)
if n != data.len:
raise newException(IOError, "short read")
result = Preserve[E](kind: pkString, string: data)
of 0xb2:
var data = newSeq[byte](s.readVarint())
if data.len > 0:
let n = s.readData(addr data[0], data.len)
if n != data.len:
raise newException(IOError, "short read")
result = Preserve[E](kind: pkByteString, bytes: data)
of 0xb3:
var data = newString(s.readVarint())
if data.len > 0:
let n = s.readData(addr data[0], data.len)
if n != data.len:
raise newException(IOError, "short read")
result = Preserve[E](kind: pkSymbol, symbol: Symbol data)
of 0xb4:
result = Preserve[E](kind: pkRecord)
var label = decodePreserves(s, E)
while s.peekUint8() != endMarker:
result.record.add decodePreserves(s, E)
result.record.add(move label)
discard s.readUint8()
of 0xb5:
result = Preserve[E](kind: pkSequence)
while s.peekUint8() != endMarker:
result.sequence.add decodePreserves(s, E)
discard s.readUint8()
of 0xb6:
result = Preserve[E](kind: pkSet)
while s.peekUint8() != endMarker:
incl(result, decodePreserves(s, E))
discard s.readUint8()
of 0xb7:
result = Preserve[E](kind: pkDictionary)
while s.peekUint8() != endMarker:
result[decodePreserves(s, E)] = decodePreserves(s, E)
discard s.readUint8()
of 0xb0:
var len = s.readVarint()
result = Preserve[E](kind: pkSignedInteger)
if len > 0:
if (s.peekUint8() and 0x80) == 0x80:
result.int = BiggestInt -1
while len > 0:
result.int = (result.int shl 8) + s.readUint8().BiggestInt
dec(len)
of endMarker:
raise newException(ValueError, "invalid Preserves stream")
else:
raise newException(ValueError, "invalid Preserves tag byte 0x" & tag.toHex(2))
proc decodePreserves*(s: string; E = void): Preserve[E] =
## Decode a string of binary-encoded Preserves.
decodePreserves(s.newStringStream, E)
proc decodePreserves*(s: seq[byte]; E = void): Preserve[E] =
## Decode a byte-string of binary-encoded Preserves.
decodePreserves(cast[string](s), E)
type BufferedDecoder* = object
## Type for buffering binary Preserves before decoding.
stream: StringStream
appendPosition, decodePosition, maxSize: int
proc newBufferedDecoder*(maxSize = 4096): BufferedDecoder =
## Create a new `newBufferedDecoder`.
runnableExamples:
var
buf = newBufferedDecoder()
bin = encode(parsePreserves("<foobar>"))
buf.feed(bin[0..2])
buf.feed(bin[3..bin.high])
var (success, pr) = decode(buf)
assert success
assert $pr == "<foobar>"
BufferedDecoder(
stream: newStringStream(newStringOfCap(maxSize)),
maxSize: maxSize,
)
proc feed*(dec: var BufferedDecoder; buf: pointer; len: int) =
assert len > 0
if dec.maxSize > 0 and dec.maxSize < (dec.appendPosition + len):
raise newException(IOError, "BufferedDecoder at maximum buffer size")
dec.stream.setPosition(dec.appendPosition)
dec.stream.writeData(buf, len)
inc(dec.appendPosition, len)
assert dec.appendPosition == dec.stream.getPosition()
proc feed*[T: byte|char](dec: var BufferedDecoder; data: openarray[T]) =
if data.len > 0:
dec.feed(unsafeAddr data[0], data.len)
proc decode*(dec: var BufferedDecoder; E = void): (bool, Preserve[E]) =
## Decode from `dec`. If decoding fails the internal position of the
## decoder does not advance.
if dec.appendPosition > 0:
assert(dec.decodePosition < dec.appendPosition)
dec.stream.setPosition(dec.decodePosition)
try:
result[1] = decodePreserves(dec.stream, E)
result[0] = true
dec.decodePosition = dec.stream.getPosition()
if dec.decodePosition == dec.appendPosition:
dec.stream.setPosition(0)
dec.stream.data.setLen(0)
dec.appendPosition = 0
dec.decodePosition = 0
except IOError:
discard
template preservesRecord*(label: string) {.pragma.}
## Serialize this object or tuple as a record.
## See ``toPreserve``.
@ -1413,131 +887,3 @@ proc getOrDefault*[T, V](pr: Preserve[T]; key: string; default: V): V =
if fromPreserve(result, v): return
else: break
default
type TextMode* = enum textPreserves, textJson
proc writeText*[E](stream: Stream; pr: Preserve[E]; mode = textPreserves) =
## Encode Preserves to a `Stream` as text.
if pr.embedded: write(stream, "#!")
case pr.kind:
of pkBoolean:
case pr.bool
of false: write(stream, "#f")
of true: write(stream, "#t")
of pkFloat:
write(stream, $pr.float)
write(stream, 'f')
of pkDouble:
write(stream, $pr.double)
of pkSignedInteger:
write(stream, $pr.int)
of pkString:
write(stream, escapeJson(pr.string))
of pkByteString:
if pr.bytes.allIt(char(it) in {' '..'!', '#'..'~'}):
write(stream, "#\"")
write(stream, cast[string](pr.bytes))
write(stream, '"')
else:
if pr.bytes.len > 64:
write(stream, "#[") #]#
write(stream, base64.encode(pr.bytes))
write(stream, ']')
else:
const alphabet = "0123456789abcdef"
write(stream, "#x\"")
for b in pr.bytes:
write(stream, alphabet[int(b shr 4)])
write(stream, alphabet[int(b and 0xf)])
write(stream, '"')
of pkSymbol:
let sym = pr.symbol.string
if sym.len > 0 and sym[0] in {'A'..'z'} and not sym.anyIt(char(it) in { '\x00'..'\x19', '"', '\\', '|' }):
write(stream, sym)
else:
write(stream, '|')
for c in sym:
case c
of '\\':
write(stream, "\\\\")
of '/':
write(stream, "\\/")
of '\x08':
write(stream, "\\b")
of '\x0c':
write(stream, "\\f")
of '\x0a':
write(stream, "\\n")
of '\x0d':
write(stream, "\\r")
of '\x09':
write(stream, "\\t")
of '|':
write(stream, "\\|")
else:
write(stream, c)
write(stream, '|')
of pkRecord:
assert(pr.record.len > 0)
write(stream, '<')
writeText(stream, pr.record[pr.record.high], mode)
for i in 0..<pr.record.high:
write(stream, ' ')
writeText(stream, pr.record[i], mode)
write(stream, '>')
of pkSequence:
write(stream, '[')
var insertSeperator: bool
case mode
of textPreserves:
for val in pr.sequence:
if insertSeperator: write(stream, ' ')
else: insertSeperator = true
writeText(stream, val, mode)
of textJson:
for val in pr.sequence:
if insertSeperator: write(stream, ',')
else: insertSeperator = true
writeText(stream, val, mode)
write(stream, ']')
of pkSet:
write(stream, "#{")
var insertSeperator: bool
for val in pr.set.items:
if insertSeperator: write(stream, ' ')
else: insertSeperator = true
writeText(stream, val, mode)
write(stream, '}')
of pkDictionary:
write(stream, '{')
var insertSeperator: bool
case mode
of textPreserves:
for (key, value) in pr.dict.items:
if insertSeperator: write(stream, ' ')
else: insertSeperator = true
writeText(stream, key, mode)
write(stream, ": ")
writeText(stream, value, mode)
of textJson:
for (key, value) in pr.dict.items:
if insertSeperator: write(stream, ',')
else: insertSeperator = true
writeText(stream, key, mode)
write(stream, ':')
writeText(stream, value, mode)
write(stream, '}')
of pkEmbedded:
write(stream, "#!")
when compiles($pr.embed) and not E is void:
write(stream, $pr.embed)
else:
write(stream, "")
proc `$`*[E](pr: Preserve[E]): string =
## Generate the textual representation of ``pr``.
var stream = newStringStream()
writeText(stream, pr, textPreserves)
result = move stream.data
include ./preserves/private/parse

View File

@ -0,0 +1,166 @@
# SPDX-FileCopyrightText: ☭ Emery Hemingway
# SPDX-License-Identifier: Unlicense
import std/[endians, streams, strutils]
import ./values
proc readVarint(s: Stream): uint =
var
shift = 0
c = uint s.readUint8
while (c and 0x80) == 0x80:
result = result or ((c and 0x7f) shl shift)
inc(shift, 7)
c = uint s.readUint8
result = result or (c shl shift)
proc decodePreserves*(s: Stream; E = void): Preserve[E] =
## Decode a Preserves value from a binary-encoded stream.
if s.atEnd: raise newException(IOError, "End of Preserves stream")
const endMarker = 0x84
let tag = s.readUint8()
case tag
of 0x80: result = Preserve[E](kind: pkBoolean, bool: false)
of 0x81: result = Preserve[E](kind: pkBoolean, bool: true)
of 0x85:
discard decodePreserves(s, E)
result = decodePreserves(s, E)
of 0x86:
result = decodePreserves(s, E)
result.embedded = true
of 0x87:
let n = s.readUint8()
case n
of 4:
when system.cpuEndian == bigEndian:
result = Preserve[E](kind: pkFloat, float: s.readFloat32())
else:
result = Preserve[E](kind: pkFloat)
var be = s.readFloat32()
swapEndian32(result.float.addr, be.addr)
of 8:
when system.cpuEndian == bigEndian:
result = Preserve[E](kind: pkDouble, double: s.readFloat64())
else:
result = Preserve[E](kind: pkDouble)
var be = s.readFloat64()
swapEndian64(result.double.addr, be.addr)
else:
raise newException(IOError, "unhandled IEEE754 value of " & $n & " bytes")
of 0xb1:
var data = newString(s.readVarint())
if data.len > 0:
let n = s.readData(unsafeAddr data[0], data.len)
if n != data.len:
raise newException(IOError, "short read")
result = Preserve[E](kind: pkString, string: data)
of 0xb2:
var data = newSeq[byte](s.readVarint())
if data.len > 0:
let n = s.readData(addr data[0], data.len)
if n != data.len:
raise newException(IOError, "short read")
result = Preserve[E](kind: pkByteString, bytes: data)
of 0xb3:
var data = newString(s.readVarint())
if data.len > 0:
let n = s.readData(addr data[0], data.len)
if n != data.len:
raise newException(IOError, "short read")
result = Preserve[E](kind: pkSymbol, symbol: Symbol data)
of 0xb4:
result = Preserve[E](kind: pkRecord)
var label = decodePreserves(s, E)
while s.peekUint8() != endMarker:
result.record.add decodePreserves(s, E)
result.record.add(move label)
discard s.readUint8()
of 0xb5:
result = Preserve[E](kind: pkSequence)
while s.peekUint8() != endMarker:
result.sequence.add decodePreserves(s, E)
discard s.readUint8()
of 0xb6:
result = Preserve[E](kind: pkSet)
while s.peekUint8() != endMarker:
incl(result, decodePreserves(s, E))
discard s.readUint8()
of 0xb7:
result = Preserve[E](kind: pkDictionary)
while s.peekUint8() != endMarker:
result[decodePreserves(s, E)] = decodePreserves(s, E)
discard s.readUint8()
of 0xb0:
var len = s.readVarint()
result = Preserve[E](kind: pkSignedInteger)
if len > 0:
if (s.peekUint8() and 0x80) == 0x80:
result.int = BiggestInt -1
while len > 0:
result.int = (result.int shl 8) + s.readUint8().BiggestInt
dec(len)
of endMarker:
raise newException(ValueError, "invalid Preserves stream")
else:
raise newException(ValueError, "invalid Preserves tag byte 0x" & tag.toHex(2))
proc decodePreserves*(s: string; E = void): Preserve[E] =
## Decode a string of binary-encoded Preserves.
decodePreserves(s.newStringStream, E)
proc decodePreserves*(s: seq[byte]; E = void): Preserve[E] =
## Decode a byte-string of binary-encoded Preserves.
decodePreserves(cast[string](s), E)
type BufferedDecoder* = object
## Type for buffering binary Preserves before decoding.
stream: StringStream
appendPosition, decodePosition, maxSize: int
proc newBufferedDecoder*(maxSize = 4096): BufferedDecoder =
## Create a new `newBufferedDecoder`.
runnableExamples:
var
buf = newBufferedDecoder()
bin = encode(parsePreserves("<foobar>"))
buf.feed(bin[0..2])
buf.feed(bin[3..bin.high])
var (success, pr) = decode(buf)
assert success
assert $pr == "<foobar>"
BufferedDecoder(
stream: newStringStream(newStringOfCap(maxSize)),
maxSize: maxSize,
)
proc feed*(dec: var BufferedDecoder; buf: pointer; len: int) =
assert len > 0
if dec.maxSize > 0 and dec.maxSize < (dec.appendPosition + len):
raise newException(IOError, "BufferedDecoder at maximum buffer size")
dec.stream.setPosition(dec.appendPosition)
dec.stream.writeData(buf, len)
inc(dec.appendPosition, len)
assert dec.appendPosition == dec.stream.getPosition()
proc feed*[T: byte|char](dec: var BufferedDecoder; data: openarray[T]) =
if data.len > 0:
dec.feed(unsafeAddr data[0], data.len)
proc decode*(dec: var BufferedDecoder; E = void): (bool, Preserve[E]) =
## Decode from `dec`. If decoding fails the internal position of the
## decoder does not advance.
if dec.appendPosition > 0:
assert(dec.decodePosition < dec.appendPosition)
dec.stream.setPosition(dec.decodePosition)
try:
result[1] = decodePreserves(dec.stream, E)
result[0] = true
dec.decodePosition = dec.stream.getPosition()
if dec.decodePosition == dec.appendPosition:
dec.stream.setPosition(0)
dec.stream.data.setLen(0)
dec.appendPosition = 0
dec.decodePosition = 0
except IOError:
discard

View File

@ -0,0 +1,100 @@
# SPDX-FileCopyrightText: ☭ Emery Hemingway
# SPDX-License-Identifier: Unlicense
import std/[endians, options, sets, sequtils, streams, tables, typetraits]
import ./values
proc writeVarint(s: Stream; n: Natural) =
var n = n
while n > 0x7f:
s.write(uint8 n or 0x80)
n = n shr 7
s.write(uint8 n and 0x7f)
proc write*[E](str: Stream; pr: Preserve[E]) =
## Write the binary-encoding of a Preserves value to a stream.
if pr.embedded: str.write(0x86'u8)
case pr.kind:
of pkBoolean:
case pr.bool
of false: str.write(0x80'u8)
of true: str.write(0x81'u8)
of pkFloat:
str.write("\x87\x04")
when system.cpuEndian == bigEndian:
str.write(pr.float)
else:
var be: float32
swapEndian32(be.addr, pr.float.unsafeAddr)
str.write(be)
of pkDouble:
str.write("\x87\x08")
when system.cpuEndian == bigEndian:
str.write(pr.double)
else:
var be: float64
swapEndian64(be.addr, pr.double.unsafeAddr)
str.write(be)
of pkSignedInteger:
if pr.int == 0:
str.write("\xb0\x00")
else:
var bitCount = 1'u8
if pr.int < 0:
while ((not pr.int) shr bitCount) != 0:
inc(bitCount)
else:
while (pr.int shr bitCount) != 0:
inc(bitCount)
var byteCount = (bitCount + 8) div 8
str.write(0xb0'u8)
str.writeVarint(byteCount)
proc write(n: uint8; i: BiggestInt) =
if n > 1:
write(n.pred, i shr 8)
str.write(i.uint8)
write(byteCount, pr.int)
of pkString:
str.write(0xb1'u8)
str.writeVarint(pr.string.len)
str.write(pr.string)
of pkByteString:
str.write(0xb2'u8)
str.writeVarint(pr.bytes.len)
str.write(cast[string](pr.bytes))
of pkSymbol:
str.write(0xb3'u8)
str.writeVarint(pr.symbol.len)
str.write(string pr.symbol)
of pkRecord:
assert(pr.record.len > 0)
str.write(0xb4'u8)
str.write(pr.record[pr.record.high])
for i in 0..<pr.record.high:
str.write(pr.record[i])
str.write(0x84'u8)
of pkSequence:
str.write(0xb5'u8)
for e in pr.sequence:
str.write(e)
str.write(0x84'u8)
of pkSet:
str.write(0xb6'u8)
for val in pr.set.items:
str.write(val)
str.write(0x84'u8)
of pkDictionary:
str.write(0xb7'u8)
for (key, value) in pr.dict.items:
str.write(key)
str.write(value)
str.write(0x84'u8)
of pkEmbedded:
str.write(0x86'u8)
str.write(pr.embed.toPreserve)
proc encode*[E](pr: Preserve[E]): seq[byte] =
## Return the binary-encoding of a Preserves value.
let s = newStringStream()
s.write pr
result = cast[seq[byte]](move s.data)

View File

@ -1,16 +1,12 @@
# SPDX-FileCopyrightText: 2021 ☭ Emery Hemingway
# SPDX-FileCopyrightText: ☭ Emery Hemingway
# SPDX-License-Identifier: Unlicense
# this module is included in ../../preserves.nim
import std/[parseutils, unicode]
when isMainModule:
import std/strutils
from std/sequtils import insert
import std/[base64, parseutils, strutils, unicode]
from std/sequtils import insert
import npeg
import ../pegs
import ./decoding, ./values
type
Value = Preserve[void]

View File

@ -0,0 +1,160 @@
# SPDX-FileCopyrightText: ☭ Emery Hemingway
# SPDX-License-Identifier: Unlicense
import std/[base64, json, options, sets, sequtils, streams, strutils, tables, typetraits]
import ./values
proc `$`*(s: Symbol): string =
let sym = string s
if sym.len > 0 and sym[0] in {'A'..'z'} and not sym.anyIt(char(it) in { '\x00'..'\x19', '"', '\\', '|' }):
result = sym
else:
result = newStringOfCap(sym.len shl 1)
result.add('|')
for c in sym:
case c
of '\\':
result.add("\\\\")
of '/':
result.add("\\/")
of '\x08':
result.add("\\b")
of '\x0c':
result.add("\\f")
of '\x0a':
result.add("\\n")
of '\x0d':
result.add("\\r")
of '\x09':
result.add("\\t")
of '|':
result.add("\\|")
else:
result.add(c)
result.add('|')
type TextMode* = enum textPreserves, textJson
proc writeText*[E](stream: Stream; pr: Preserve[E]; mode = textPreserves) =
## Encode Preserves to a `Stream` as text.
if pr.embedded: write(stream, "#!")
case pr.kind:
of pkBoolean:
case pr.bool
of false: write(stream, "#f")
of true: write(stream, "#t")
of pkFloat:
write(stream, $pr.float)
write(stream, 'f')
of pkDouble:
write(stream, $pr.double)
of pkSignedInteger:
write(stream, $pr.int)
of pkString:
write(stream, escapeJson(pr.string))
of pkByteString:
if pr.bytes.allIt(char(it) in {' '..'!', '#'..'~'}):
write(stream, "#\"")
write(stream, cast[string](pr.bytes))
write(stream, '"')
else:
if pr.bytes.len > 64:
write(stream, "#[") #]#
write(stream, base64.encode(pr.bytes))
write(stream, ']')
else:
const alphabet = "0123456789abcdef"
write(stream, "#x\"")
for b in pr.bytes:
write(stream, alphabet[int(b shr 4)])
write(stream, alphabet[int(b and 0xf)])
write(stream, '"')
of pkSymbol:
let sym = pr.symbol.string
if sym.len > 0 and sym[0] in {'A'..'z'} and not sym.anyIt(char(it) in { '\x00'..'\x19', '"', '\\', '|' }):
write(stream, sym)
else:
write(stream, '|')
for c in sym:
case c
of '\\':
write(stream, "\\\\")
of '/':
write(stream, "\\/")
of '\x08':
write(stream, "\\b")
of '\x0c':
write(stream, "\\f")
of '\x0a':
write(stream, "\\n")
of '\x0d':
write(stream, "\\r")
of '\x09':
write(stream, "\\t")
of '|':
write(stream, "\\|")
else:
write(stream, c)
write(stream, '|')
of pkRecord:
assert(pr.record.len > 0)
write(stream, '<')
writeText(stream, pr.record[pr.record.high], mode)
for i in 0..<pr.record.high:
write(stream, ' ')
writeText(stream, pr.record[i], mode)
write(stream, '>')
of pkSequence:
write(stream, '[')
var insertSeperator: bool
case mode
of textPreserves:
for val in pr.sequence:
if insertSeperator: write(stream, ' ')
else: insertSeperator = true
writeText(stream, val, mode)
of textJson:
for val in pr.sequence:
if insertSeperator: write(stream, ',')
else: insertSeperator = true
writeText(stream, val, mode)
write(stream, ']')
of pkSet:
write(stream, "#{")
var insertSeperator: bool
for val in pr.set.items:
if insertSeperator: write(stream, ' ')
else: insertSeperator = true
writeText(stream, val, mode)
write(stream, '}')
of pkDictionary:
write(stream, '{')
var insertSeperator: bool
case mode
of textPreserves:
for (key, value) in pr.dict.items:
if insertSeperator: write(stream, ' ')
else: insertSeperator = true
writeText(stream, key, mode)
write(stream, ": ")
writeText(stream, value, mode)
of textJson:
for (key, value) in pr.dict.items:
if insertSeperator: write(stream, ',')
else: insertSeperator = true
writeText(stream, key, mode)
write(stream, ':')
writeText(stream, value, mode)
write(stream, '}')
of pkEmbedded:
write(stream, "#!")
when compiles($pr.embed) and not E is void:
write(stream, $pr.embed)
else:
write(stream, "")
proc `$`*[E](pr: Preserve[E]): string =
## Generate the textual representation of ``pr``.
var stream = newStringStream()
writeText(stream, pr, textPreserves)
result = move stream.data

View File

@ -0,0 +1,247 @@
# SPDX-FileCopyrightText: ☭ Emery Hemingway
# SPDX-License-Identifier: Unlicense
import std/[hashes, options, sets, sequtils, tables]
type
PreserveKind* = enum
pkBoolean, pkFloat, pkDouble, pkSignedInteger, pkString, pkByteString, pkSymbol,
pkRecord, pkSequence, pkSet, pkDictionary, pkEmbedded
const
atomKinds* = {pkBoolean, pkFloat, pkDouble, pkSignedInteger, pkString, pkByteString, pkSymbol}
compoundKinds* = {pkRecord, pkSequence, pkSet, pkDictionary}
type Symbol* = distinct string
proc `<`*(x, y: Symbol): bool {.borrow.}
proc `==`*(x, y: Symbol): bool {.borrow.}
proc hash*(s: Symbol): Hash {.borrow.}
proc len*(s: Symbol): int {.borrow.}
type
Preserve*[E] = object
embedded*: bool
## Flag to mark embedded Preserves
case kind*: PreserveKind
of pkBoolean:
bool*: bool
of pkFloat:
float*: float32
of pkDouble:
double*: float64
of pkSignedInteger:
int*: BiggestInt
of pkString:
string*: string
of pkByteString:
bytes*: seq[byte]
of pkSymbol:
symbol*: Symbol
of pkRecord:
record*: seq[Preserve[E]] # label is last
of pkSequence:
sequence*: seq[Preserve[E]]
of pkSet:
set*: seq[Preserve[E]]
# TODO: HashSet
of pkDictionary:
dict*: seq[DictEntry[E]]
# TODO: Tables
of pkEmbedded:
embed*: E
DictEntry*[E] = tuple[key: Preserve[E], val: Preserve[E]]
func `==`*[A, B](x: Preserve[A]; y: Preserve[B]): bool =
## Check `x` and `y` for equivalence.
if x.kind == y.kind and x.embedded == y.embedded:
case x.kind
of pkBoolean:
result = x.bool == y.bool
of pkFloat:
result = x.float == y.float
of pkDouble:
result = x.double == y.double
of pkSignedInteger:
result = x.int == y.int
of pkString:
result = x.string == y.string
of pkByteString:
result = x.bytes == y.bytes
of pkSymbol:
result = x.symbol == y.symbol
of pkRecord:
result = x.record.len == y.record.len
for i in 0..x.record.high:
if not result: break
result = result and (x.record[i] == y.record[i])
of pkSequence:
for i, val in x.sequence:
if y.sequence[i] != val: return false
result = true
of pkSet:
result = x.set.len == y.set.len
for i in 0..x.set.high:
if not result: break
result = result and (x.set[i] == y.set[i])
of pkDictionary:
result = x.dict.len == y.dict.len
for i in 0..x.dict.high:
if not result: break
result = result and
(x.dict[i].key == y.dict[i].key) and
(x.dict[i].val == y.dict[i].val)
of pkEmbedded:
when A is B:
when A is void:
result = true
else:
result = x.embed == y.embed
proc `<`(x, y: string | seq[byte]): bool =
for i in 0 .. min(x.high, y.high):
if x[i] < y[i]: return true
if x[i] != y[i]: return false
x.len < y.len
proc `<`*[A, B](x: Preserve[A]; y: Preserve[B]): bool =
## Preserves have a total order over values. Check if `x` is ordered before `y`.
if x.embedded != y.embedded:
result = y.embedded
elif x.kind != y.kind:
result = x.kind < y.kind
else:
case x.kind
of pkBoolean:
result = (not x.bool) and y.bool
of pkFloat:
result = x.float < y.float
of pkDouble:
result = x.double < y.double
of pkSignedInteger:
result = x.int < y.int
of pkString:
result = x.string < y.string
of pkByteString:
result = x.bytes < y.bytes
of pkSymbol:
result = x.symbol < y.symbol
of pkRecord:
if x.record[x.record.high] < y.record[y.record.high]: return true
for i in 0..<min(x.record.high, y.record.high):
if x.record[i] < y.record[i]: return true
if x.record[i] == y.record[i]: return false
result = x.record.len < y.record.len
of pkSequence:
for i in 0..min(x.sequence.high, y.sequence.high):
if x.sequence[i] < y.sequence[i]: return true
if x.sequence[i] != y.sequence[i]: return false
result = x.sequence.len < y.sequence.len
of pkSet:
for i in 0..min(x.set.high, y.set.high):
if x.set[i] < y.set[i]: return true
if x.set[i] != y.set[i]: return false
result = x.set.len < y.set.len
of pkDictionary:
for i in 0..min(x.dict.high, y.dict.high):
if x.dict[i].key < y.dict[i].key: return true
if x.dict[i].key == y.dict[i].key:
if x.dict[i].val < y.dict[i].val: return true
if x.dict[i].val != y.dict[i].val: return false
result = x.dict.len < y.dict.len
of pkEmbedded:
when (not A is void) and (A is B):
result = x.embed < y.embed
func cmp*[E](x, y: Preserve[E]): int =
## Compare by Preserves total ordering.
if x == y: 0
elif x < y: -1
else: 1
proc sort*[E](pr: var Preserve[E]) = sort(pr.sequence, cmp)
## Sort a Preserves array by total ordering.
proc hash*[E](pr: Preserve[E]): Hash =
## Produce a `Hash` of `pr` for use with a `HashSet` or `Table`.
var h = hash(pr.kind.int) !& hash(pr.embedded)
case pr.kind
of pkBoolean:
h = h !& hash(pr.bool)
of pkFloat:
h = h !& hash(pr.float)
of pkDouble:
h = h !& hash(pr.double)
of pkSignedInteger:
h = h !& hash(pr.int)
of pkString:
h = h !& hash(pr.string)
of pkByteString:
h = h !& hash(pr.bytes)
of pkSymbol:
h = h !& hash(string pr.symbol)
of pkRecord:
for val in pr.record:
h = h !& hash(val)
of pkSequence:
for val in pr.sequence:
h = h !& hash(val)
of pkSet:
for val in pr.set.items:
h = h !& hash(val)
of pkDictionary:
for (key, val) in pr.dict.items:
h = h !& hash(key) !& hash(val)
of pkEmbedded:
when E is void:
h = h !& hash(pr.embed)
else:
if pr.embed.isNil:
h = h !& hash(false)
else:
h = h !& hash(pr.embed)
!$h
proc `[]`*(pr: Preserve; i: int): Preserve =
## Select an indexed value from ``pr``.
## Only valid for records and sequences.
case pr.kind
of pkRecord: pr.record[i]
of pkSequence: pr.sequence[i]
else:
raise newException(ValueError, "Preserves value is not indexable")
proc `[]=`*(pr: var Preserve; i: Natural; val: Preserve) =
## Assign an indexed value into ``pr``.
## Only valid for records and sequences.
case pr.kind
of pkRecord: pr.record[i] = val
of pkSequence: pr.sequence[i] = val
else:
raise newException(ValueError, "`Preserves value is not indexable")
proc `[]=`*(pr: var Preserve; key, val: Preserve) =
## Insert `val` by `key` in the Preserves dictionary `pr`.
for i in 0..pr.dict.high:
if key < pr.dict[i].key:
insert(pr.dict, [(key, val, )], i)
return
elif key == pr.dict[i].key:
pr.dict[i].val = val
return
pr.dict.add((key, val, ))
proc incl*(pr: var Preserve; key: Preserve) =
## Include `key` in the Preserves set `pr`.
for i in 0..pr.set.high:
if key < pr.set[i]:
insert(pr.set, [key], i)
return
pr.set.add(key)
proc excl*(pr: var Preserve; key: Preserve) =
## Exclude `key` from the Preserves set `pr`.
for i in 0..pr.set.high:
if pr.set[i] == key:
delete(pr.set, i..i)
break