jyapayne/worm-scraper
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Modernize DOM manipulation code

Browse source 
jsdom, and JavaScript, have gotten nicer since this was first written.
This commit is contained in:
Domenic Denicola 2020-10-26 21:19:56 -04:00
commit fb93322823
1 changed files with 16 additions and 20 deletions
Download patch file Download diff file Expand all files Collapse all files

36
lib/convert-worker.js
View file

@ -34,21 +34,21 @@ ${body}
function getBodyXML(chapter, book, contentEl) {
// Remove initial Next Chapter and Previous Chapter <p>
contentEl.removeChild(contentEl.firstElementChild);
contentEl.firstElementChild.remove();
// Remove everything after the last <p> (e.g. analytics <div>s)
const lastP = contentEl.querySelector("p:last-of-type");
while (contentEl.lastElementChild !== lastP) {
contentEl.removeChild(contentEl.lastElementChild);
contentEl.lastElementChild.remove();
}
// Remove empty <p>s or Last Chapter/Next Chapter <p>s
while (isEmptyOrGarbage(contentEl.lastElementChild)) {
contentEl.removeChild(contentEl.lastElementChild);
contentEl.lastElementChild.remove();
}
// Remove redundant attributes and style
Array.prototype.forEach.call(contentEl.children, child => {
for (const child of contentEl.children) {
if (child.getAttribute("dir") === "ltr") {
child.removeAttribute("dir");
}
@ -63,46 +63,42 @@ function getBodyXML(chapter, book, contentEl) {
if (child.getAttribute("style") === "text-align:left;padding-left:30px;") {
child.setAttribute("style", "padding-left:30px;");
}
});
}
// Remove empty <em>s and <i>s
// Remove style attributes from them, as they're always messed up.
const ems = contentEl.querySelectorAll("em, i");
Array.prototype.forEach.call(ems, em => {
for (const em of contentEl.querySelectorAll("em, i")) {
if (em.textContent.trim() === "") {
const replacement = contentEl.ownerDocument.createTextNode(" ");
em.parentNode.replaceChild(replacement, em);
em.replaceWith(contentEl.ownerDocument.createTextNode(" "));
} else {
em.removeAttribute("style");
}
});
}
// In https://parahumans.wordpress.com/2013/01/05/monarch-16-13/ there are some <address>s that should be <p>s O_o
const addresses = contentEl.querySelectorAll("address");
Array.prototype.forEach.call(addresses, address => {
for (const address of contentEl.querySelectorAll("address")) {
const p = contentEl.ownerDocument.createElement("p");
p.innerHTML = address.innerHTML;
address.parentNode.replaceChild(p, address);
});
address.replaceWith(p);
}
// Every <span> except underline ones is pointless at best and frequently messed up. (Weird font size, line spacing,
// etc.)
const spans = contentEl.querySelectorAll("span");
Array.prototype.forEach.call(spans, span => {
for (const span of contentEl.querySelectorAll("span")) {
if (span.getAttribute("style") === "text-decoration:underline;") {
return;
continue;
}
if (span.textContent.trim() === "") {
span.parentNode.removeChild(span);
span.remove();
} else {
const docFrag = contentEl.ownerDocument.createDocumentFragment();
while (span.firstChild) {
docFrag.appendChild(span.firstChild);
}
span.parentNode.replaceChild(docFrag, span);
span.replaceWith(docFrag);
}
});
}
// In Ward, CloudFlare email protection obfuscates the email addresses:
// https://usamaejaz.com/cloudflare-email-decoding/