"use strict";
const path = require("path");
const fs = require("mz/fs");
const mkdirp = require("mkdirp-then");
const rimraf = require("rimraf-then");
const jsdom = require("jsdom");
const download = require("./download.js");
require("./track-rejections.js");
const START_CHAPTER_URL = "https://parahumans.wordpress.com/2011/06/11/1-1/";
const cachePath = path.resolve("cache");
const outPath = path.resolve("out");
const contentPath = path.resolve(outPath, "OEBPS");
rimraf(outPath)
.then(function () {
return mkdirp(contentPath);
})
.then(function () {
return download(START_CHAPTER_URL, cachePath);
})
.then(readAllRawChapterDocs)
.then(function (rawChapterDocs) {
console.log("Extracting content into EPUB chapter files");
return Promise.all(rawChapterDocs.map(function (rawChapterDoc, i) {
const output = getChapterString(rawChapterDoc);
const destFilename = path.resolve(contentPath, `chapter${i + 1}.xhtml`);
return fs.writeFile(destFilename, output);
}));
})
.then(function () {
console.log("All done!");
});
function readAllRawChapterDocs() {
return fs.readdir(cachePath).then(function (filenames) {
const htmlFiles = filenames.filter(function (f) { return f.endsWith(".html"); });
return Promise.all(htmlFiles.map(readRawChapterDoc));
});
}
function readRawChapterDoc(filename) {
const filePath = path.resolve(cachePath, filename);
return fs.readFile(filePath, { encoding: "utf-8" }).then(function (contents) {
return jsdom.jsdom(contents);
});
}
function getChapterString(rawChapterDoc) {
const title = rawChapterDoc.querySelector("h1.entry-title").textContent;
const body = cleanContentEl(rawChapterDoc.querySelector(".entry-content")).innerHTML;
return `
${title}
${title}
${body}
`;
}
function cleanContentEl(el) {
// Remove Next Chapter and Previous Chapter s
el.removeChild(el.firstElementChild);
el.removeChild(el.lastElementChild);
// Remove redundant dir="ltr"
Array.prototype.forEach.call(el.children, function (child) {
if (child.getAttribute("dir") === "ltr") {
child.removeAttribute("dir");
}
});
return el;
}