trilium/static/js/html2notecase.js

156 lines
5 KiB
JavaScript
Raw Normal View History

2017-06-12 04:04:07 +08:00
function html2notecase(contents, note) {
2017-08-24 08:35:47 +08:00
// remove any possible extra newlines which might be inserted - all relevant new lines should be only in <br> and <p>
contents = contents.replace(/(?:\r\n|\r|\n)/, '');
2017-09-05 06:55:09 +08:00
// some editors insert this font stuff which messes up parsing by doubling newlines
contents = contents.replace(/<font[^>]*>/g, '');
contents = contents.replace(/<\/font>/g, '');
contents = contents.replace(/<br><\/p>/g, '\n');
contents = contents.replace(/<p><br><\/p>/g, '\n');
2017-06-12 04:04:07 +08:00
contents = contents.replace(/<br \/>/g, '\n');
contents = contents.replace(/<br>/g, '\n');
contents = contents.replace(/<\/p>/g, '\n');
contents = contents.replace(/<p>/g, '');
2017-08-14 09:42:10 +08:00
contents = contents.replace(/&nbsp;/g, ' ');
2017-06-12 04:04:07 +08:00
let index = 0;
note.formatting = [];
note.links = [];
note.images = [];
while (index < contents.length) {
let curContent = contents.substr(index);
if (contents[index] === '<') {
let found = false;
2017-08-30 10:25:58 +08:00
const endOfTag = curContent.indexOf('>');
2017-06-12 04:04:07 +08:00
if (endOfTag === -1) {
2017-06-12 04:04:07 +08:00
console.log("Can't find the end of the tag");
}
2017-08-30 10:25:58 +08:00
const curTag = curContent.substr(0, endOfTag + 1);
2017-06-12 04:04:07 +08:00
for (tagId in tags) {
2017-08-30 10:25:58 +08:00
const tag = tags[tagId];
2017-06-12 04:04:07 +08:00
if (contents.substr(index, tag.length) === tag) {
2017-06-12 04:04:07 +08:00
found = true;
note.formatting.push({
note_id: note.detail.note_id,
note_offset: index,
fmt_tag: tagId,
fmt_color: '',
fmt_font: '',
fmt_value: 100
});
contents = contents.substr(0, index) + contents.substr(index + tag.length);
break;
}
}
if (curTag.substr(0, 4) === "<img") {
2017-06-12 04:04:07 +08:00
//console.log("Found img tag");
let dataImagePos = curTag.indexOf('data:image/');
if (dataImagePos !== -1) {
2017-06-12 04:04:07 +08:00
let imageType = curTag.substr(dataImagePos + 11, 3);
//console.log("image type: " + imageType);
let dataStart = curTag.substr(dataImagePos + 22);
let endOfDataPos = dataStart.indexOf('"');
if (endOfDataPos !== -1) {
2017-08-30 10:25:58 +08:00
const imageData = dataStart.substr(0, endOfDataPos);
2017-06-12 04:04:07 +08:00
note.images.push({
note_id: note.detail.note_id,
note_offset: index,
is_png: imageType === "png",
2017-06-12 04:04:07 +08:00
image_data: imageData
});
contents = contents.substr(0, index) + contents.substr(index + curTag.length);
found = true;
}
}
}
const linkMatch = /^<a[^>]+?href="([^"]+?)"[^>]*?>([^<]+?)<\/a>/.exec(curContent);
2017-06-12 04:04:07 +08:00
2017-08-30 10:25:58 +08:00
if (linkMatch !== null) {
const targetUrl = linkMatch[1];
const linkText = linkMatch[2];
2017-08-28 02:39:26 +08:00
const newLink = {
2017-06-12 04:04:07 +08:00
note_id: note.detail.note_id,
note_offset: index,
2017-08-28 02:39:26 +08:00
lnk_text: linkText
};
const noteIdMatch = /app#([A-Za-z0-9]{22})/.exec(targetUrl);
if (noteIdMatch !== null) {
newLink.target_note_id = noteIdMatch[1];
}
else {
newLink.target_url = targetUrl;
}
note.links.push(newLink);
2017-06-12 04:04:07 +08:00
2017-08-30 10:25:58 +08:00
contents = contents.substr(0, index) + linkText + contents.substr(index + linkMatch[0].length);
2017-06-12 04:04:07 +08:00
2017-09-05 06:51:07 +08:00
// we'll skip the link text so that it's not processed twice (second time by link auto detection)
index += linkText.length;
2017-06-12 04:04:07 +08:00
found = true;
}
if (!found) {
contents = contents.substr(0, index) + contents.substr(index + endOfTag + 1);
}
}
else {
2017-08-23 08:19:46 +08:00
let linkMatch = /^(https?|ftp|file):\/\/[-A-Z0-9+&@#\/%?=~_|!:,.;]*[-A-Z0-9+&@#\/%=~_|]/i.exec(curContent);
if (linkMatch !== null) {
2017-08-30 10:25:58 +08:00
const targetUrl = linkMatch[0];
2017-08-28 02:39:26 +08:00
2017-08-30 10:25:58 +08:00
const newLink = {
note_id: note.detail.note_id,
note_offset: index,
2017-08-28 02:39:26 +08:00
lnk_text: targetUrl
};
const noteIdMatch = /app#([A-Za-z0-9]{22})/.exec(targetUrl);
if (noteIdMatch !== null) {
newLink.target_note_id = noteIdMatch[1];
}
else {
newLink.target_url = targetUrl;
}
2017-08-28 02:39:26 +08:00
note.links.push(newLink);
2017-08-23 08:19:46 +08:00
2017-08-28 02:39:26 +08:00
index += targetUrl.length;
}
else {
index++;
}
2017-06-12 04:04:07 +08:00
}
}
contents = contents.trim();
2017-06-12 04:04:07 +08:00
note.detail.note_text = contents;
}