Accurate word counter for Gutenberg texts
当前为
// ==UserScript==
// @name Gutencount
// @namespace https://vox.quartertone.net/
// @version 1.1
// @description Accurate word counter for Gutenberg texts
// @author quartertone
// @match *://*.gutenberg.org/*
// @grant none
// @license gpl-3.0
// ==/UserScript==
(function() {
'use strict';
let loc = document.location.toString();
let sel = window.getSelection() + "";
let monkey = true; //tampermonkey flag
function wx(t, x, r = "") {
// match gutenberg ebook text between START and END markers
// accuracy of word count depends on presence of extraneous text such as "transcriber's notes", book summaries, page numbers etc
return wc(t.match(/(?<=\*{3} START.*?\*{3}).*(?=\*{3} END.*?\*{3})/s)[0].trim(), x, r);
}
function wc(t, x, r = "") {
// trim whitespace, count non-space things and place in alert box
// TODO: need to include — (long hyhen)
// let w = t.trim().match(/[\*\S]+/g).length;
// let w = t.trim().match(/[^\s\*—-]+/g).length;
let w = t.trim().split(/--|[\s\*—]+/).length;
if (r) return w; // return the word count as a number if requested; this skips the alert
// alert(w + "\ words,\ " + t.length + "\ chars\nin\ " + x);
let alrt = w + "\ words,\ " + t.length + "\ chars\nin\ " + x;
if (monkey) {cornerbox(alrt);} else {alert(alrt);}
}
function fwx(l, a) {
// fetch link => get text from result => send text to wx function
fetch(l).then((r) => r.text()).then((t) => {
let c = wx(t, "ebook", a);
if (a) {
a.innerHTML += "wc: " + c;
}
}).catch((e) => { }); // alert(e));
}
if (sel.length > 0) { // selection
wc(sel, "selection");
} else if (loc.match(/gutenberg.org\/(files|cache\/epub)\/\d+/)) {
// gutenberg text
wx(document.body.innerText, "ebook");
} else if (loc.match(/gutenberg.org\/ebooks\/(\d+)/)) {
// gutenberg main ebook page
fwx(loc + ".txt.utf-8", document.querySelector("#cover"));
} else if (loc.match(/gutenberg.org\/ebooks\/(subject|search)/)) {
// gutenberg search or subject page
for (const a of document.querySelectorAll(".booklink>a>span:nth-child(2)")) {
// append word counts to each book section
// second CHILD of the <a> is the "content" span
fwx(a.parentElement.href + ".txt.utf-8", a);
}
} else { // general webpage
wc(document.body.innerText, "webpage");
}
function cornerbox(wcount) {
console.log(wcount);
let xcount = document.getElementById("xcount") || document.createElement("div");
xcount.id = "xcount";
xcount.style = "position:fixed;top:0;right:0;width:10em;height:2em;background:#333c;color:#fff;z-index:10000;padding:0.5em;text-align:right;";
xcount.innerHTML = "";
xcount.ondblclick = function () {
this.remove();
};
document.body.appendChild(xcount);
xcount.innerHTML = wcount;
}
})();