Convert src/compiler/parse/utils/html.ts to JavaScript

pull/8569/head
Simon Holthausen 3 years ago
parent 561ecf28ff
commit 61977f1d92

@ -150,6 +150,7 @@
"prettier-plugin-svelte": "^2.10.0", "prettier-plugin-svelte": "^2.10.0",
"puppeteer": "^19.8.5", "puppeteer": "^19.8.5",
"rollup": "^3.20.2", "rollup": "^3.20.2",
"rollup-plugin-dts": "^5.3.0",
"source-map": "^0.7.4", "source-map": "^0.7.4",
"source-map-support": "^0.5.21", "source-map-support": "^0.5.21",
"tiny-glob": "^0.2.9", "tiny-glob": "^0.2.9",

@ -115,6 +115,9 @@ devDependencies:
rollup: rollup:
specifier: ^3.20.2 specifier: ^3.20.2
version: 3.20.2 version: 3.20.2
rollup-plugin-dts:
specifier: ^5.3.0
version: 5.3.0(rollup@3.20.2)(typescript@5.0.4)
source-map: source-map:
specifier: ^0.7.4 specifier: ^0.7.4
version: 0.7.4 version: 0.7.4
@ -2736,6 +2739,20 @@ packages:
glob: 7.2.3 glob: 7.2.3
dev: true dev: true
/rollup-plugin-dts@5.3.0(rollup@3.20.2)(typescript@5.0.4):
resolution: {integrity: sha512-8FXp0ZkyZj1iU5klkIJYLjIq/YZSwBoERu33QBDxm/1yw5UU4txrEtcmMkrq+ZiKu3Q4qvPCNqc3ovX6rjqzbQ==}
engines: {node: '>=v14'}
peerDependencies:
rollup: ^3.0.0
typescript: ^4.1 || ^5.0
dependencies:
magic-string: 0.30.0
rollup: 3.20.2
typescript: 5.0.4
optionalDependencies:
'@babel/code-frame': 7.21.4
dev: true
/rollup@3.20.2: /rollup@3.20.2:
resolution: {integrity: sha512-3zwkBQl7Ai7MFYQE0y1MeQ15+9jsi7XxfrqwTb/9EK8D9C9+//EBR4M+CuA1KODRaNbFez/lWxA5vhEGZp4MUg==} resolution: {integrity: sha512-3zwkBQl7Ai7MFYQE0y1MeQ15+9jsi7XxfrqwTb/9EK8D9C9+//EBR4M+CuA1KODRaNbFez/lWxA5vhEGZp4MUg==}
engines: {node: '>=14.18.0', npm: '>=8.0.0'} engines: {node: '>=14.18.0', npm: '>=8.0.0'}

@ -1,11 +1,14 @@
import entities from './entities'; import entities from './entities.js';
const windows_1252 = [ const windows_1252 = [
8364, 129, 8218, 402, 8222, 8230, 8224, 8225, 710, 8240, 352, 8249, 338, 141, 381, 143, 144, 8216, 8364, 129, 8218, 402, 8222, 8230, 8224, 8225, 710, 8240, 352, 8249, 338, 141, 381, 143, 144, 8216,
8217, 8220, 8221, 8226, 8211, 8212, 732, 8482, 353, 8250, 339, 157, 382, 376 8217, 8220, 8221, 8226, 8211, 8212, 732, 8482, 353, 8250, 339, 157, 382, 376
]; ];
function reg_exp_entity(entity_name: string, is_attribute_value: boolean) { /**
* @param {string} entity_name
* @param {boolean} is_attribute_value
*/
function reg_exp_entity(entity_name, is_attribute_value) {
// https://html.spec.whatwg.org/multipage/parsing.html#named-character-reference-state // https://html.spec.whatwg.org/multipage/parsing.html#named-character-reference-state
// doesn't decode the html entity which not ends with ; and next character is =, number or alphabet in attribute value. // doesn't decode the html entity which not ends with ; and next character is =, number or alphabet in attribute value.
if (is_attribute_value && !entity_name.endsWith(';')) { if (is_attribute_value && !entity_name.endsWith(';')) {
@ -14,94 +17,88 @@ function reg_exp_entity(entity_name: string, is_attribute_value: boolean) {
return entity_name; return entity_name;
} }
function get_entity_pattern(is_attribute_value: boolean) { /**
* @param {boolean} is_attribute_value
*/
function get_entity_pattern(is_attribute_value) {
const reg_exp_num = '#(?:x[a-fA-F\\d]+|\\d+)(?:;)?'; const reg_exp_num = '#(?:x[a-fA-F\\d]+|\\d+)(?:;)?';
const reg_exp_entities = Object.keys(entities).map((entity_name) => const reg_exp_entities = Object.keys(entities).map((entity_name) => reg_exp_entity(entity_name, is_attribute_value));
reg_exp_entity(entity_name, is_attribute_value)
);
const entity_pattern = new RegExp(`&(${reg_exp_num}|${reg_exp_entities.join('|')})`, 'g'); const entity_pattern = new RegExp(`&(${reg_exp_num}|${reg_exp_entities.join('|')})`, 'g');
return entity_pattern; return entity_pattern;
} }
const entity_pattern_content = get_entity_pattern(false); const entity_pattern_content = get_entity_pattern(false);
const entity_pattern_attr_value = get_entity_pattern(true); const entity_pattern_attr_value = get_entity_pattern(true);
export function decode_character_references(html: string, is_attribute_value: boolean) { /**
* @param {string} html
* @param {boolean} is_attribute_value
*/
export function decode_character_references(html, is_attribute_value) {
const entity_pattern = is_attribute_value ? entity_pattern_attr_value : entity_pattern_content; const entity_pattern = is_attribute_value ? entity_pattern_attr_value : entity_pattern_content;
return html.replace(entity_pattern, (match, entity) => { return html.replace(entity_pattern, (match, entity) => {
let code; let code;
// Handle named entities // Handle named entities
if (entity[0] !== '#') { if (entity[0] !== '#') {
code = entities[entity]; code = entities[entity];
} else if (entity[1] === 'x') { }
else if (entity[1] === 'x') {
code = parseInt(entity.substring(2), 16); code = parseInt(entity.substring(2), 16);
} else { }
else {
code = parseInt(entity.substring(1), 10); code = parseInt(entity.substring(1), 10);
} }
if (!code) { if (!code) {
return match; return match;
} }
return String.fromCodePoint(validate_code(code)); return String.fromCodePoint(validate_code(code));
}); });
} }
const NUL = 0; const NUL = 0;
// some code points are verboten. If we were inserting HTML, the browser would replace the illegal // some code points are verboten. If we were inserting HTML, the browser would replace the illegal
// code points with alternatives in some cases - since we're bypassing that mechanism, we need // code points with alternatives in some cases - since we're bypassing that mechanism, we need
// to replace them ourselves // to replace them ourselves
// //
// Source: http://en.wikipedia.org/wiki/Character_encodings_in_HTML#Illegal_characters // Source: http://en.wikipedia.org/wiki/Character_encodings_in_HTML#Illegal_characters
function validate_code(code: number) {
/**
* @param {number} code
*/
function validate_code(code) {
// line feed becomes generic whitespace // line feed becomes generic whitespace
if (code === 10) { if (code === 10) {
return 32; return 32;
} }
// ASCII range. (Why someone would use HTML entities for ASCII characters I don't know, but...) // ASCII range. (Why someone would use HTML entities for ASCII characters I don't know, but...)
if (code < 128) { if (code < 128) {
return code; return code;
} }
// code points 128-159 are dealt with leniently by browsers, but they're incorrect. We need // code points 128-159 are dealt with leniently by browsers, but they're incorrect. We need
// to correct the mistake or we'll end up with missing € signs and so on // to correct the mistake or we'll end up with missing € signs and so on
if (code <= 159) { if (code <= 159) {
return windows_1252[code - 128]; return windows_1252[code - 128];
} }
// basic multilingual plane // basic multilingual plane
if (code < 55296) { if (code < 55296) {
return code; return code;
} }
// UTF-16 surrogate halves // UTF-16 surrogate halves
if (code <= 57343) { if (code <= 57343) {
return NUL; return NUL;
} }
// rest of the basic multilingual plane // rest of the basic multilingual plane
if (code <= 65535) { if (code <= 65535) {
return code; return code;
} }
// supplementary multilingual plane 0x10000 - 0x1ffff // supplementary multilingual plane 0x10000 - 0x1ffff
if (code >= 65536 && code <= 131071) { if (code >= 65536 && code <= 131071) {
return code; return code;
} }
// supplementary ideographic plane 0x20000 - 0x2ffff // supplementary ideographic plane 0x20000 - 0x2ffff
if (code >= 131072 && code <= 196607) { if (code >= 131072 && code <= 196607) {
return code; return code;
} }
return NUL; return NUL;
} }
// based on http://developers.whatwg.org/syntax.html#syntax-tag-omission // based on http://developers.whatwg.org/syntax.html#syntax-tag-omission
const disallowed_contents = new Map([ const disallowed_contents = new Map([
['li', new Set(['li'])], ['li', new Set(['li'])],
@ -109,11 +106,7 @@ const disallowed_contents = new Map([
['dd', new Set(['dt', 'dd'])], ['dd', new Set(['dt', 'dd'])],
[ [
'p', 'p',
new Set( new Set('address article aside blockquote div dl fieldset footer form h1 h2 h3 h4 h5 h6 header hgroup hr main menu nav ol p pre section table ul'.split(' '))
'address article aside blockquote div dl fieldset footer form h1 h2 h3 h4 h5 h6 header hgroup hr main menu nav ol p pre section table ul'.split(
' '
)
)
], ],
['rt', new Set(['rt', 'rp'])], ['rt', new Set(['rt', 'rp'])],
['rp', new Set(['rt', 'rp'])], ['rp', new Set(['rt', 'rp'])],
@ -126,15 +119,22 @@ const disallowed_contents = new Map([
['td', new Set(['td', 'th', 'tr'])], ['td', new Set(['td', 'th', 'tr'])],
['th', new Set(['td', 'th', 'tr'])] ['th', new Set(['td', 'th', 'tr'])]
]); ]);
// can this be a child of the parent element, or does it implicitly // can this be a child of the parent element, or does it implicitly
// close it, like `<li>one<li>two`? // close it, like `<li>one<li>two`?
export function closing_tag_omitted(current: string, next?: string) {
/**
* @param {string} current
* @param {string} next
*/
export function closing_tag_omitted(current, next) {
if (disallowed_contents.has(current)) { if (disallowed_contents.has(current)) {
if (!next || disallowed_contents.get(current).has(next)) { if (!next || disallowed_contents.get(current).has(next)) {
return true; return true;
} }
} }
return false; return false;
} }

Loading…
Cancel
Save