forked from Fediversity/fediversity.eu
224 lines
7.1 KiB
JavaScript
224 lines
7.1 KiB
JavaScript
|
import { Scalar } from '../nodes/Scalar.js';
|
||
|
import { resolveEnd } from './resolve-end.js';
|
||
|
|
||
|
function resolveFlowScalar(scalar, strict, onError) {
|
||
|
const { offset, type, source, end } = scalar;
|
||
|
let _type;
|
||
|
let value;
|
||
|
const _onError = (rel, code, msg) => onError(offset + rel, code, msg);
|
||
|
switch (type) {
|
||
|
case 'scalar':
|
||
|
_type = Scalar.PLAIN;
|
||
|
value = plainValue(source, _onError);
|
||
|
break;
|
||
|
case 'single-quoted-scalar':
|
||
|
_type = Scalar.QUOTE_SINGLE;
|
||
|
value = singleQuotedValue(source, _onError);
|
||
|
break;
|
||
|
case 'double-quoted-scalar':
|
||
|
_type = Scalar.QUOTE_DOUBLE;
|
||
|
value = doubleQuotedValue(source, _onError);
|
||
|
break;
|
||
|
/* istanbul ignore next should not happen */
|
||
|
default:
|
||
|
onError(scalar, 'UNEXPECTED_TOKEN', `Expected a flow scalar value, but found: ${type}`);
|
||
|
return {
|
||
|
value: '',
|
||
|
type: null,
|
||
|
comment: '',
|
||
|
range: [offset, offset + source.length, offset + source.length]
|
||
|
};
|
||
|
}
|
||
|
const valueEnd = offset + source.length;
|
||
|
const re = resolveEnd(end, valueEnd, strict, onError);
|
||
|
return {
|
||
|
value,
|
||
|
type: _type,
|
||
|
comment: re.comment,
|
||
|
range: [offset, valueEnd, re.offset]
|
||
|
};
|
||
|
}
|
||
|
function plainValue(source, onError) {
|
||
|
let badChar = '';
|
||
|
switch (source[0]) {
|
||
|
/* istanbul ignore next should not happen */
|
||
|
case '\t':
|
||
|
badChar = 'a tab character';
|
||
|
break;
|
||
|
case ',':
|
||
|
badChar = 'flow indicator character ,';
|
||
|
break;
|
||
|
case '%':
|
||
|
badChar = 'directive indicator character %';
|
||
|
break;
|
||
|
case '|':
|
||
|
case '>': {
|
||
|
badChar = `block scalar indicator ${source[0]}`;
|
||
|
break;
|
||
|
}
|
||
|
case '@':
|
||
|
case '`': {
|
||
|
badChar = `reserved character ${source[0]}`;
|
||
|
break;
|
||
|
}
|
||
|
}
|
||
|
if (badChar)
|
||
|
onError(0, 'BAD_SCALAR_START', `Plain value cannot start with ${badChar}`);
|
||
|
return foldLines(source);
|
||
|
}
|
||
|
function singleQuotedValue(source, onError) {
|
||
|
if (source[source.length - 1] !== "'" || source.length === 1)
|
||
|
onError(source.length, 'MISSING_CHAR', "Missing closing 'quote");
|
||
|
return foldLines(source.slice(1, -1)).replace(/''/g, "'");
|
||
|
}
|
||
|
function foldLines(source) {
|
||
|
/**
|
||
|
* The negative lookbehind here and in the `re` RegExp is to
|
||
|
* prevent causing a polynomial search time in certain cases.
|
||
|
*
|
||
|
* The try-catch is for Safari, which doesn't support this yet:
|
||
|
* https://caniuse.com/js-regexp-lookbehind
|
||
|
*/
|
||
|
let first, line;
|
||
|
try {
|
||
|
first = new RegExp('(.*?)(?<![ \t])[ \t]*\r?\n', 'sy');
|
||
|
line = new RegExp('[ \t]*(.*?)(?:(?<![ \t])[ \t]*)?\r?\n', 'sy');
|
||
|
}
|
||
|
catch (_) {
|
||
|
first = /(.*?)[ \t]*\r?\n/sy;
|
||
|
line = /[ \t]*(.*?)[ \t]*\r?\n/sy;
|
||
|
}
|
||
|
let match = first.exec(source);
|
||
|
if (!match)
|
||
|
return source;
|
||
|
let res = match[1];
|
||
|
let sep = ' ';
|
||
|
let pos = first.lastIndex;
|
||
|
line.lastIndex = pos;
|
||
|
while ((match = line.exec(source))) {
|
||
|
if (match[1] === '') {
|
||
|
if (sep === '\n')
|
||
|
res += sep;
|
||
|
else
|
||
|
sep = '\n';
|
||
|
}
|
||
|
else {
|
||
|
res += sep + match[1];
|
||
|
sep = ' ';
|
||
|
}
|
||
|
pos = line.lastIndex;
|
||
|
}
|
||
|
const last = /[ \t]*(.*)/sy;
|
||
|
last.lastIndex = pos;
|
||
|
match = last.exec(source);
|
||
|
return res + sep + (match?.[1] ?? '');
|
||
|
}
|
||
|
function doubleQuotedValue(source, onError) {
|
||
|
let res = '';
|
||
|
for (let i = 1; i < source.length - 1; ++i) {
|
||
|
const ch = source[i];
|
||
|
if (ch === '\r' && source[i + 1] === '\n')
|
||
|
continue;
|
||
|
if (ch === '\n') {
|
||
|
const { fold, offset } = foldNewline(source, i);
|
||
|
res += fold;
|
||
|
i = offset;
|
||
|
}
|
||
|
else if (ch === '\\') {
|
||
|
let next = source[++i];
|
||
|
const cc = escapeCodes[next];
|
||
|
if (cc)
|
||
|
res += cc;
|
||
|
else if (next === '\n') {
|
||
|
// skip escaped newlines, but still trim the following line
|
||
|
next = source[i + 1];
|
||
|
while (next === ' ' || next === '\t')
|
||
|
next = source[++i + 1];
|
||
|
}
|
||
|
else if (next === '\r' && source[i + 1] === '\n') {
|
||
|
// skip escaped CRLF newlines, but still trim the following line
|
||
|
next = source[++i + 1];
|
||
|
while (next === ' ' || next === '\t')
|
||
|
next = source[++i + 1];
|
||
|
}
|
||
|
else if (next === 'x' || next === 'u' || next === 'U') {
|
||
|
const length = { x: 2, u: 4, U: 8 }[next];
|
||
|
res += parseCharCode(source, i + 1, length, onError);
|
||
|
i += length;
|
||
|
}
|
||
|
else {
|
||
|
const raw = source.substr(i - 1, 2);
|
||
|
onError(i - 1, 'BAD_DQ_ESCAPE', `Invalid escape sequence ${raw}`);
|
||
|
res += raw;
|
||
|
}
|
||
|
}
|
||
|
else if (ch === ' ' || ch === '\t') {
|
||
|
// trim trailing whitespace
|
||
|
const wsStart = i;
|
||
|
let next = source[i + 1];
|
||
|
while (next === ' ' || next === '\t')
|
||
|
next = source[++i + 1];
|
||
|
if (next !== '\n' && !(next === '\r' && source[i + 2] === '\n'))
|
||
|
res += i > wsStart ? source.slice(wsStart, i + 1) : ch;
|
||
|
}
|
||
|
else {
|
||
|
res += ch;
|
||
|
}
|
||
|
}
|
||
|
if (source[source.length - 1] !== '"' || source.length === 1)
|
||
|
onError(source.length, 'MISSING_CHAR', 'Missing closing "quote');
|
||
|
return res;
|
||
|
}
|
||
|
/**
|
||
|
* Fold a single newline into a space, multiple newlines to N - 1 newlines.
|
||
|
* Presumes `source[offset] === '\n'`
|
||
|
*/
|
||
|
function foldNewline(source, offset) {
|
||
|
let fold = '';
|
||
|
let ch = source[offset + 1];
|
||
|
while (ch === ' ' || ch === '\t' || ch === '\n' || ch === '\r') {
|
||
|
if (ch === '\r' && source[offset + 2] !== '\n')
|
||
|
break;
|
||
|
if (ch === '\n')
|
||
|
fold += '\n';
|
||
|
offset += 1;
|
||
|
ch = source[offset + 1];
|
||
|
}
|
||
|
if (!fold)
|
||
|
fold = ' ';
|
||
|
return { fold, offset };
|
||
|
}
|
||
|
const escapeCodes = {
|
||
|
'0': '\0', // null character
|
||
|
a: '\x07', // bell character
|
||
|
b: '\b', // backspace
|
||
|
e: '\x1b', // escape character
|
||
|
f: '\f', // form feed
|
||
|
n: '\n', // line feed
|
||
|
r: '\r', // carriage return
|
||
|
t: '\t', // horizontal tab
|
||
|
v: '\v', // vertical tab
|
||
|
N: '\u0085', // Unicode next line
|
||
|
_: '\u00a0', // Unicode non-breaking space
|
||
|
L: '\u2028', // Unicode line separator
|
||
|
P: '\u2029', // Unicode paragraph separator
|
||
|
' ': ' ',
|
||
|
'"': '"',
|
||
|
'/': '/',
|
||
|
'\\': '\\',
|
||
|
'\t': '\t'
|
||
|
};
|
||
|
function parseCharCode(source, offset, length, onError) {
|
||
|
const cc = source.substr(offset, length);
|
||
|
const ok = cc.length === length && /^[0-9a-fA-F]+$/.test(cc);
|
||
|
const code = ok ? parseInt(cc, 16) : NaN;
|
||
|
if (isNaN(code)) {
|
||
|
const raw = source.substr(offset - 2, length + 2);
|
||
|
onError(offset - 2, 'BAD_DQ_ESCAPE', `Invalid escape sequence ${raw}`);
|
||
|
return raw;
|
||
|
}
|
||
|
return String.fromCodePoint(code);
|
||
|
}
|
||
|
|
||
|
export { resolveFlowScalar };
|