Merge pull request #4726 from paulexyz/insta-import-utf8

fix: Instagram import broken UTF8 characters
This commit is contained in:
daniel 2023-11-02 04:27:19 -06:00 committed by GitHub
commit e6301bfa51
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 14 additions and 4 deletions

View File

@ -348,8 +348,16 @@
}, 500);
},
// Facebook and Instagram are encoding UTF8 characters in a weird way in their json
// here is a good explanation what's going wrong https://sorashi.github.io/fix-facebook-json-archive-encoding
fixFacebookEncoding(string) {
const replaced = string.replace(/\\u00([a-f0-9]{2})/g, (x) => String.fromCharCode(parseInt(x.slice(2), 16)));
const buffer = Array.from(replaced, (c) => c.charCodeAt(0));
return new TextDecoder().decode(new Uint8Array(buffer));
},
filterPostMeta(media) {
let json = JSON.parse(media);
let json = JSON.parse(this.fixFacebookEncoding(media));
let res = json.filter(j => {
let ids = j.media.map(m => m.uri).filter(m => {
if(this.config.allow_video_posts == true) {
@ -396,12 +404,14 @@
this.filterPostMeta(media);
let imgs = await Promise.all(entries.filter(entry => {
return entry.filename.startsWith('media/posts/') && (entry.filename.endsWith('.png') || entry.filename.endsWith('.jpg') || entry.filename.endsWith('.mp4'));
return (entry.filename.startsWith('media/posts/') || entry.filename.startsWith('media/other/')) && (entry.filename.endsWith('.png') || entry.filename.endsWith('.jpg') || entry.filename.endsWith('.mp4'));
})
.map(async entry => {
if(
entry.filename.startsWith('media/posts/') &&
(
entry.filename.startsWith('media/posts/') ||
entry.filename.startsWith('media/other/')
) && (
entry.filename.endsWith('.png') ||
entry.filename.endsWith('.jpg') ||
entry.filename.endsWith('.mp4')