431 lines
15 KiB
TypeScript
431 lines
15 KiB
TypeScript
// Text syntax reader.
|
|
|
|
import type { Value } from './values';
|
|
import { DecodeError, ShortPacket } from './codec';
|
|
import { Dictionary, Set } from './dictionary';
|
|
import { strip } from './strip';
|
|
import { Bytes, unhexDigit } from './bytes';
|
|
import { Decoder, DecoderState, neverEmbeddedTypeDecode } from './decoder';
|
|
import { Record } from './record';
|
|
import { Annotated, newPosition, Position, updatePosition } from './annotated';
|
|
import { Double, DoubleFloat, FloatType } from './float';
|
|
import { stringify } from './text';
|
|
import { embed, GenericEmbedded, EmbeddedTypeDecode } from './embedded';
|
|
|
|
export interface ReaderStateOptions {
|
|
includeAnnotations?: boolean;
|
|
name?: string | Position;
|
|
}
|
|
|
|
export interface ReaderOptions<T> extends ReaderStateOptions {
|
|
embeddedDecode?: EmbeddedTypeDecode<T>;
|
|
}
|
|
|
|
const MAX_SAFE_INTEGERn = BigInt(Number.MAX_SAFE_INTEGER);
|
|
const MIN_SAFE_INTEGERn = BigInt(Number.MIN_SAFE_INTEGER);
|
|
|
|
export const NUMBER_RE: RegExp = /^([-+]?\d+)((\.\d+([eE][-+]?\d+)?)|([eE][-+]?\d+))?$/;
|
|
// Groups:
|
|
// 1 - integer part and sign
|
|
// 2 - decimal part and exponent
|
|
|
|
export class ReaderState {
|
|
buffer: string;
|
|
pos: Position;
|
|
index: number;
|
|
discarded = 0;
|
|
options: ReaderStateOptions;
|
|
|
|
constructor(buffer: string, options: ReaderStateOptions) {
|
|
this.buffer = buffer;
|
|
switch (typeof options.name) {
|
|
case 'undefined': this.pos = newPosition(); break;
|
|
case 'string': this.pos = newPosition(options.name); break;
|
|
case 'object': this.pos = { ... options.name }; break;
|
|
}
|
|
this.index = 0;
|
|
this.options = options;
|
|
}
|
|
|
|
error(message: string, pos: Position): never {
|
|
throw new DecodeError(message, { ... pos });
|
|
}
|
|
|
|
get includeAnnotations(): boolean {
|
|
return this.options.includeAnnotations ?? false;
|
|
}
|
|
|
|
copyPos(): Position {
|
|
return { ... this.pos };
|
|
}
|
|
|
|
write(data: string) {
|
|
if (this.atEnd()) {
|
|
this.buffer = data;
|
|
} else {
|
|
this.buffer = this.buffer.substring(this.index) + data;
|
|
}
|
|
this.discarded += this.index;
|
|
this.index = 0;
|
|
}
|
|
|
|
atEnd(): boolean {
|
|
return (this.index >= this.buffer.length);
|
|
}
|
|
|
|
peek(): string {
|
|
if (this.atEnd()) throw new ShortPacket("Short term", this.pos);
|
|
return this.buffer[this.index];
|
|
}
|
|
|
|
advance(): number {
|
|
const n = this.index++;
|
|
updatePosition(this.pos, this.buffer[n]);
|
|
return n;
|
|
}
|
|
|
|
nextchar(): string {
|
|
if (this.atEnd()) throw new ShortPacket("Short term", this.pos);
|
|
return this.buffer[this.advance()];
|
|
}
|
|
|
|
nextcharcode(): number {
|
|
if (this.atEnd()) throw new ShortPacket("Short term", this.pos);
|
|
return this.buffer.charCodeAt(this.advance());
|
|
}
|
|
|
|
skipws(skipCommas = false) {
|
|
while (true) {
|
|
if (this.atEnd()) break;
|
|
const c = this.peek();
|
|
if (!(isSpace(c) || (skipCommas && c === ','))) break;
|
|
this.advance();
|
|
}
|
|
}
|
|
|
|
readHex2(): number {
|
|
const x1 = unhexDigit(this.nextcharcode(), DecodeError);
|
|
const x2 = unhexDigit(this.nextcharcode(), DecodeError);
|
|
return (x1 << 4) | x2;
|
|
}
|
|
|
|
readHex4(): number {
|
|
const x1 = unhexDigit(this.nextcharcode(), DecodeError);
|
|
const x2 = unhexDigit(this.nextcharcode(), DecodeError);
|
|
const x3 = unhexDigit(this.nextcharcode(), DecodeError);
|
|
const x4 = unhexDigit(this.nextcharcode(), DecodeError);
|
|
return (x1 << 12) | (x2 << 8) | (x3 << 4) | x4;
|
|
}
|
|
|
|
readHexBinary(): Bytes {
|
|
const acc: number[] = [];
|
|
while (true) {
|
|
this.skipws();
|
|
if (this.peek() === '"') {
|
|
this.advance();
|
|
return Bytes.from(acc);
|
|
}
|
|
acc.push(this.readHex2());
|
|
}
|
|
}
|
|
|
|
readHexFloat(): DoubleFloat {
|
|
const pos = this.copyPos();
|
|
if (this.nextchar() !== '"') {
|
|
this.error("Missing open-double-quote in hex-encoded floating-point number", pos);
|
|
}
|
|
const bs = this.readHexBinary();
|
|
if (bs.length !== 8) this.error("Incorrect number of bytes in hex-encoded Double", pos);
|
|
return DoubleFloat.fromBytes(bs);
|
|
}
|
|
|
|
readBase64Binary(): Bytes {
|
|
let acc = '';
|
|
while (true) {
|
|
this.skipws();
|
|
const c = this.nextchar();
|
|
if (c === ']') break;
|
|
acc = acc + c;
|
|
}
|
|
return Bytes.fromBase64(acc);
|
|
}
|
|
|
|
requireDelimiter(prefix: string): void {
|
|
if (this.delimiterFollows()) return;
|
|
this.error(`Delimiter must follow ${prefix}`, this.pos);
|
|
}
|
|
|
|
delimiterFollows(): boolean {
|
|
if (this.atEnd()) return true;
|
|
const ch = this.peek();
|
|
return ('(){}[]<>";,@#:|'.indexOf(ch) !== -1) || isSpace(ch);
|
|
}
|
|
|
|
readRawSymbolOrNumber<T>(acc: string): Value<T> {
|
|
while (!this.delimiterFollows()) acc = acc + this.nextchar();
|
|
const m = NUMBER_RE.exec(acc);
|
|
if (m) {
|
|
if (m[2] === void 0) {
|
|
let v = BigInt(m[1]);
|
|
if (v <= MIN_SAFE_INTEGERn || v >= MAX_SAFE_INTEGERn) {
|
|
return v;
|
|
} else {
|
|
return Number(v);
|
|
}
|
|
} else {
|
|
return Double(parseFloat(acc));
|
|
}
|
|
} else {
|
|
return Symbol.for(acc);
|
|
}
|
|
}
|
|
|
|
readStringlike<E, R>(xform: (ch: string) => E,
|
|
finish: (acc: E[]) => R,
|
|
terminator: string,
|
|
hexescape: string,
|
|
hex: () => E): R
|
|
{
|
|
let acc: E[] = [];
|
|
while (true) {
|
|
const ch = this.nextchar();
|
|
switch (ch) {
|
|
case terminator:
|
|
return finish(acc);
|
|
case '\\': {
|
|
const ch = this.nextchar();
|
|
switch (ch) {
|
|
case hexescape: acc.push(hex()); break;
|
|
|
|
case terminator:
|
|
case '\\':
|
|
case '/':
|
|
acc.push(xform(ch)); break;
|
|
|
|
case 'b': acc.push(xform('\x08')); break;
|
|
case 'f': acc.push(xform('\x0c')); break;
|
|
case 'n': acc.push(xform('\x0a')); break;
|
|
case 'r': acc.push(xform('\x0d')); break;
|
|
case 't': acc.push(xform('\x09')); break;
|
|
|
|
default:
|
|
this.error(`Invalid escape code \\${ch}`, this.pos);
|
|
}
|
|
break;
|
|
}
|
|
default:
|
|
acc.push(xform(ch));
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
|
|
readString(terminator: string): string {
|
|
return this.readStringlike(x => x, xs => xs.join(''), terminator, 'u', () => {
|
|
const n1 = this.readHex4();
|
|
if ((n1 >= 0xd800) && (n1 <= 0xdfff)) {
|
|
if ((this.nextchar() === '\\') && (this.nextchar() === 'u')) {
|
|
const n2 = this.readHex4();
|
|
if ((n2 >= 0xdc00) && (n2 <= 0xdfff) && (n1 <= 0xdbff)) {
|
|
return String.fromCharCode(n1, n2);
|
|
}
|
|
}
|
|
this.error('Invalid surrogate pair', this.pos);
|
|
}
|
|
return String.fromCharCode(n1);
|
|
});
|
|
}
|
|
|
|
readLiteralBinary(): Bytes {
|
|
return this.readStringlike(
|
|
x => {
|
|
const v = x.charCodeAt(0);
|
|
if (v >= 256) this.error(`Invalid code point ${v} in literal binary`, this.pos);
|
|
return v;
|
|
},
|
|
Bytes.from,
|
|
'"',
|
|
'x',
|
|
() => this.readHex2());
|
|
}
|
|
}
|
|
|
|
export const genericEmbeddedTypeDecode: EmbeddedTypeDecode<GenericEmbedded> = {
|
|
decode(s: DecoderState): GenericEmbedded {
|
|
return new GenericEmbedded(new Decoder(s, this).next());
|
|
},
|
|
|
|
fromValue(v: Value<GenericEmbedded>, options: ReaderStateOptions): GenericEmbedded {
|
|
return new GenericEmbedded(options.includeAnnotations ? v : strip(v));
|
|
},
|
|
};
|
|
|
|
export class Reader<T> {
|
|
state: ReaderState;
|
|
embeddedType: EmbeddedTypeDecode<T>;
|
|
|
|
constructor(state: ReaderState, embeddedType: EmbeddedTypeDecode<T>);
|
|
constructor(buffer: string, options?: ReaderOptions<T>);
|
|
constructor(
|
|
state_or_buffer: (ReaderState | string) = '',
|
|
embeddedType_or_options?: (EmbeddedTypeDecode<T> | ReaderOptions<T>))
|
|
{
|
|
if (state_or_buffer instanceof ReaderState) {
|
|
this.state = state_or_buffer;
|
|
this.embeddedType = embeddedType_or_options as EmbeddedTypeDecode<T>;
|
|
} else {
|
|
const options = (embeddedType_or_options as ReaderOptions<T>) ?? {};
|
|
this.state = new ReaderState(state_or_buffer, options);
|
|
this.embeddedType = options.embeddedDecode ?? neverEmbeddedTypeDecode;
|
|
}
|
|
}
|
|
|
|
write(data: string) {
|
|
this.state.write(data);
|
|
}
|
|
|
|
readCommentLine(): Value<T> {
|
|
const startPos = this.state.copyPos();
|
|
let acc = '';
|
|
while (true) {
|
|
const c = this.state.nextchar();
|
|
if (c === '\n' || c === '\r') {
|
|
return this.wrap(acc, startPos);
|
|
}
|
|
acc = acc + c;
|
|
}
|
|
}
|
|
|
|
wrap(v: Value<T>, pos: Position): Value<T> {
|
|
if (this.state.includeAnnotations && !Annotated.isAnnotated(v)) {
|
|
v = new Annotated(v, pos);
|
|
}
|
|
return v;
|
|
}
|
|
|
|
annotateNextWith(v: Value<T>): Value<T> {
|
|
this.state.skipws();
|
|
if (this.state.atEnd()) {
|
|
throw new DecodeError("Trailing annotations and comments are not permitted",
|
|
this.state.pos);
|
|
}
|
|
const u = this.next();
|
|
if (this.state.includeAnnotations) (u as Annotated<T>).annotations.unshift(v);
|
|
return u;
|
|
}
|
|
|
|
readToEnd(): Array<Value<T>> {
|
|
const acc = [];
|
|
while (true) {
|
|
this.state.skipws();
|
|
if (this.state.atEnd()) return acc;
|
|
acc.push(this.next());
|
|
}
|
|
}
|
|
|
|
next(): Value<T> {
|
|
this.state.skipws();
|
|
const startPos = this.state.copyPos();
|
|
const unwrapped = ((): Value<T> => {
|
|
const c = this.state.nextchar();
|
|
switch (c) {
|
|
case '"':
|
|
return this.state.readString('"');
|
|
case '|':
|
|
return Symbol.for(this.state.readString('|'));
|
|
case ';':
|
|
this.state.error('Semicolon is reserved syntax', startPos);
|
|
case '@':
|
|
return this.annotateNextWith(this.next());
|
|
case ':':
|
|
this.state.error('Unexpected key/value separator between items', startPos);
|
|
case '#': {
|
|
const c = this.state.nextchar();
|
|
switch (c) {
|
|
case ' ': case '\t': return this.annotateNextWith(this.readCommentLine());
|
|
case '\n': case '\r': return this.annotateNextWith('');
|
|
case 'f': this.state.requireDelimiter('#f'); return false;
|
|
case 't': this.state.requireDelimiter('#t'); return true;
|
|
case '{': return this.readSet();
|
|
case '"': return this.state.readLiteralBinary();
|
|
case 'x': switch (this.state.nextchar()) {
|
|
case '"': return this.state.readHexBinary();
|
|
case 'd': return this.state.readHexFloat();
|
|
default: this.state.error('Invalid #x syntax', startPos);
|
|
}
|
|
case '[': return this.state.readBase64Binary();
|
|
case ':': return embed(this.embeddedType.fromValue(
|
|
new Reader<GenericEmbedded>(this.state, genericEmbeddedTypeDecode).next(),
|
|
this.state.options));
|
|
default:
|
|
this.state.error(`Invalid # syntax: ${c}`, startPos);
|
|
}
|
|
}
|
|
case '<': {
|
|
const label = this.next();
|
|
const fields = this.readSequence('>', false);
|
|
return Record(label, fields);
|
|
}
|
|
case '[': return this.readSequence(']', true);
|
|
case '{': return this.readDictionary();
|
|
case '>': this.state.error('Unexpected >', startPos);
|
|
case ']': this.state.error('Unexpected ]', startPos);
|
|
case '}': this.state.error('Unexpected }', startPos);
|
|
case ',': this.state.error('Unexpected ,', startPos);
|
|
default:
|
|
return this.state.readRawSymbolOrNumber(c);
|
|
}
|
|
})();
|
|
return this.wrap(unwrapped, startPos);
|
|
}
|
|
|
|
seq<S>(skipCommas: boolean, acc: S, update: (v: Value<T>, acc: S) => void, ch: string): S {
|
|
while (true) {
|
|
this.state.skipws(skipCommas);
|
|
if (this.state.peek() === ch) {
|
|
this.state.advance();
|
|
return acc;
|
|
}
|
|
update(this.next(), acc);
|
|
}
|
|
}
|
|
|
|
readSequence(ch: string, skipCommas: boolean): Array<Value<T>> {
|
|
return this.seq(skipCommas, [] as Array<Value<T>>, (v, acc) => acc.push(v), ch);
|
|
}
|
|
|
|
readDictionary(): Dictionary<T> {
|
|
return this.seq(true,
|
|
new Dictionary<T>(),
|
|
(k, acc) => {
|
|
this.state.skipws();
|
|
switch (this.state.peek()) {
|
|
case ':':
|
|
if (acc.has(k)) this.state.error(
|
|
`Duplicate key: ${stringify(k)}`, this.state.pos);
|
|
this.state.advance();
|
|
acc.set(k, this.next());
|
|
break;
|
|
default:
|
|
this.state.error('Missing key/value separator', this.state.pos);
|
|
}
|
|
},
|
|
'}');
|
|
}
|
|
|
|
readSet(): Set<T> {
|
|
return this.seq(true,
|
|
new Set<T>(),
|
|
(v, acc) => {
|
|
if (acc.has(v)) this.state.error(
|
|
`Duplicate value in set: ${stringify(v)}`, this.state.pos);
|
|
acc.add(v);
|
|
},
|
|
'}');
|
|
}
|
|
}
|
|
|
|
function isSpace(s: string): boolean {
|
|
return ' \t\n\r'.indexOf(s) !== -1;
|
|
}
|