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

More cleanup in conversion

Browse source
This commit is contained in:
Domenic Denicola 2015-05-09 01:28:43 +02:00
commit ba1e7b956f
2 changed files with 11 additions and 7 deletions
Download patch file Download diff file Expand all files Collapse all files

12
lib/convert.js
View file

@ -80,14 +80,16 @@ function getBodyXml(headingEl, contentEl) {
contentEl.removeChild(contentEl.lastElementChild); contentEl.removeChild(contentEl.lastElementChild);
} }
// Remove redundant dir="ltr" and align="LEFT" and style="text-align: left;" // Remove redundant attributes
Array.prototype.forEach.call(contentEl.children, function (child) { Array.prototype.forEach.call(contentEl.children, function (child) {
if (child.getAttribute("dir") === "ltr") { if (child.getAttribute("dir") === "ltr") {
child.removeAttribute("dir"); child.removeAttribute("dir");
} }
if ((child.getAttribute("align") || "").toLowerCase() === "left") {
child.removeAttribute("align"); // Only ever appears with align="LEFT" (useless) or align="CENTER" overridden by style="text-align: left;" (also
} // useless)
child.removeAttribute("align");
if (child.getAttribute("style") === "text-align:left;") { if (child.getAttribute("style") === "text-align:left;") {
child.removeAttribute("style"); child.removeAttribute("style");
} }
@ -101,6 +103,8 @@ function getBodyXml(headingEl, contentEl) {
} }
}); });
// TODO: remove redundant <span id>s inside <p>s
// Synthesize a <body> tag to serialize // Synthesize a <body> tag to serialize
const bodyEl = contentEl.ownerDocument.createElement("body"); const bodyEl = contentEl.ownerDocument.createElement("body");
const h1El = contentEl.ownerDocument.createElement("h1"); const h1El = contentEl.ownerDocument.createElement("h1");

6
lib/worm-scraper.js
View file

@ -21,13 +21,13 @@ Promise.resolve()
// return download(START_CHAPTER_URL, cachePath); // return download(START_CHAPTER_URL, cachePath);
}) })
.then(function () { .then(function () {
// return rimraf(chaptersPath); return rimraf(chaptersPath);
}) })
.then(function () { .then(function () {
// return mkdirp(chaptersPath); return mkdirp(chaptersPath);
}) })
.then(function () { .then(function () {
// return convert(cachePath, chaptersPath); return convert(cachePath, chaptersPath);
}) })
.then(function () { .then(function () {
return extras(contentPath, chaptersPath); return extras(contentPath, chaptersPath);