tallship/FoundKey
0
0
Fork 0
forked from FoundKeyGang/FoundKey
Code Issues Pull requests Packages Projects Releases Wiki Activity

server: refactor fromHtml attribute handling

Browse source 
Also try to recognize owncast hashtag links.
This commit is contained in:
Johann150 2022-12-04 03:43:22 +01:00
parent cda9197700
commit 97052b1f61
Signed by untrusted user: Johann150
GPG key ID: 9EE6577A2A06F8F1
1 changed files with 22 additions and 14 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

36
packages/backend/src/mfm/from-html.ts
View file

@ -7,6 +7,16 @@ const treeAdapter = parse5.defaultTreeAdapter;
const urlRegex = /^https?:\/\/[\w\/:%#@$&?!()\[\]~.,=+\-]+/;
const urlRegexFull = /^https?:\/\/[\w\/:%#@$&?!()\[\]~.,=+\-]+$/;
function getAttr(node: TreeAdapter.Node, attr: string): string {
return node.attrs.find(({ name }) => name === attr)?.value;
}
function attrHas(node: TreeAdapter.Node, attr: string, value: string): boolean {
const attrValue = getAttr(node, attr);
if (!attrValue) return false;
return new RegExp('\\b' + value + '\\b').test(attrValue);
}
export function fromHtml(html: string, quoteUri?: string | null): string {
const dom = parse5.parseFragment(
// some AP servers like Pixelfed use br tags as well as newlines
@ -59,19 +69,18 @@ export function fromHtml(html: string, quoteUri?: string | null): string {
case 'a':
{
const txt = getText(node);
const rel = node.attrs.find(x => x.name === 'rel');
const href = node.attrs.find(x => x.name === 'href');
const href = getAttr(node, 'href');
// hashtags
if (txt.startsWith('#') && href && /\btag\b/.test(rel?.value)) {
if (txt.startsWith('#') && href && (attrHas(node, 'rel', 'tag') || attrHas(node, 'class', 'hashtag')) {
text += txt;
// mentions
} else if (txt.startsWith('@') && !(rel && rel.value.match(/^me /))) {
} else if (txt.startsWith('@') && !attrHas(node, 'rel', 'me')) {
const part = txt.split('@');
if (part.length === 2 && href) {
// restore the host name part
const acct = `${txt}@${(new URL(href.value)).hostname}`;
const acct = `${txt}@${(new URL(href)).hostname}`;
text += acct;
} else if (part.length === 3) {
text += txt;
@ -85,17 +94,17 @@ export function fromHtml(html: string, quoteUri?: string | null): string {
if (!href) {
return txt;
}
if (!txt || txt === href.value) { // #6383: Missing text node
if (href.value.match(urlRegexFull)) {
return href.value;
if (!txt || txt === href) { // #6383: Missing text node
if (href.match(urlRegexFull)) {
return href;
} else {
return `<${href.value}>`;
return `<${href}>`;
}
}
if (href.value.match(urlRegex) && !href.value.match(urlRegexFull)) {
return `[${txt}](<${href.value}>)`; // #6846
if (href.match(urlRegex) && !href.match(urlRegexFull)) {
return `[${txt}](<${href}>)`; // #6846
} else {
return `[${txt}](${href.value})`;
return `[${txt}](${href})`;
}
};
@ -204,8 +213,7 @@ export function fromHtml(html: string, quoteUri?: string | null): string {
case 'span':
{
const nodeClass = node.attrs.find(({ name }) => name === 'class')?.value;
if (/\bquote-inline\b/.test(nodeClass) && quoteUri && getText(node).trim() === `RE: ${quoteUri}`) {
if (attrHas(node, 'class', 'quote-inline') && quoteUri && getText(node).trim() === `RE: ${quoteUri}`) {
// embedded quote thingy for backwards compatibility, don't show it
} else {
appendChildren(node.childNodes);