save-as-ebook/web-extension/extractHtml.js
2016-09-02 12:08:57 +03:00

292 lines
9.3 KiB
JavaScript

var allImages = [];
var extractedImages = [];
var maxNrOfElements = 20000;
var allowedTags = [
'address', 'article', 'aside', 'footer', 'header', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6',
'hgroup', 'nav', 'section', 'dd', 'div', 'dl', 'dt', 'figcaption', 'figure', 'hr', 'li',
'main', 'ol', 'p', 'pre', 'ul', 'a', 'abbr', 'b', 'bdi', 'bdo', 'br', 'cite', 'code', 'data',
'dfn', 'em', 'i', 'img', 'kbd', 'mark', 'q', 'rp', 'rt', 'rtc', 'ruby', 's', 'samp', 'small', 'span',
'strong', 'sub', 'sup', 'time', 'u', 'var', 'wbr', 'del', 'ins', 'caption', 'col', 'colgroup',
'table', 'tbody', 'td', 'tfoot', 'th', 'thead', 'tr',
'math', 'maction', 'menclose', 'merror', 'mfenced', 'mfrac', 'mglyph', 'mi', 'mlabeledtr', 'mmultiscripts', 'mn', 'mo', 'mover', 'mpadded', 'mphantom', 'mroot',
'mrow', 'ms', 'mspace', 'msqrt', 'mstyle', 'msub', 'msup', 'msubsup', 'mtable', 'mtd', 'mtext', 'mtr', 'munder', 'munderover', 'msgroup', 'mlongdiv', 'mscarries',
'mscarry', 'mstack'
];
//////
function getImageSrc(srcTxt) {
if (!srcTxt) {
return '';
}
var isB64Img = isBase64Img(srcTxt);
var fileExtension = getFileExtension(srcTxt);
var newImgFileName = 'img-' + (Math.floor(Math.random()*1000000*Math.random()*100000)) + '.' + fileExtension;
if (isB64Img) {
extractedImages.push({
filename: newImgFileName, // TODO name
data: getBase64ImgData(srcTxt)
});
} else {
allImages.push({
originalUrl: getImgDownloadUrl(srcTxt),
filename: newImgFileName, // TODO name
});
}
return '../images/' + newImgFileName;
}
function generateRandomTag() {
var text = "";
var possible = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz";
for(var i = 0; i < 5; i++)
text += possible.charAt(Math.floor(Math.random() * possible.length));
return text;
}
function formatPreCodeElements($jQueryElement) {
$jQueryElement.find('pre').each(function (i, pre) {
$(pre).replaceWith('<pre>' + pre.innerText + '</pre>');
});
$jQueryElement.find('code').each(function (i, pre) {
$(pre).replaceWith('<code>' + pre.innerText + '</code>');
});
}
function preProcess($htmlObject) {
$htmlObject.find('script[type="math/mml"]').each(function (i, el) {
$(el).replaceWith('<span>' + el.innerHTML + '</span>');
});
$htmlObject.find('script, style, svg, canvas, noscript, iframe').remove();
$htmlObject.find('*:empty').not('img').remove();
$htmlObject.find('*[class^="mjx-chtml"]').remove(); // MathJax formatting
formatPreCodeElements($htmlObject);
}
function force($content, withError) {
try {
var tagOpen = '@@@' + generateRandomTag();
var tagClose = '###' + generateRandomTag();
var startEl = '<object>';
var endEl = '</object>';
if (withError) {
$content = $(content);
preProcess($content);
}
$content.find('img').each(function (index, elem) {
$(elem).replaceWith(startEl + tagOpen + 'img src="' + getImageSrc($(elem).attr('src').trim()) + '"' + tagClose + tagOpen + '/img' + tagClose + endEl);
});
$content.find('a').each(function (index, elem) {
$(elem).replaceWith(startEl + tagOpen + 'a href="' + getHref($(elem).attr('href').trim()) + '"' + tagClose + $(elem).html() + tagOpen + '/a' + tagClose + endEl);
});
if ($('*').length < maxNrOfElements) {
allowedTags.forEach(function (tagName) {
var tmpElems = $content.find(tagName);
while (tmpElems.length > 0) {
$tmpElem = $(tmpElems[0]);
$tmpElem.replaceWith(startEl + tagOpen + tagName + tagClose + $tmpElem.html() + tagOpen + '/' + tagName + tagClose + endEl);
tmpElems = $content.find(tagName);
}
});
}
var contentString = $content.text();
var tagOpenRegex = new RegExp(tagOpen, 'gi');
var tagCloseRegex = new RegExp(tagClose, 'gi');
contentString = contentString.replace(tagOpenRegex, '<');
contentString = contentString.replace(tagCloseRegex, '>');
contentString = contentString.replace(/&nbsp;/gi, '&#160;');
return contentString;
} catch (e) {
console.log('Error:', e);
}
}
function sanitize(rawContentString) {
allImages = [];
extractedImages = [];
var srcTxt = '';
var dirty = null;
try {
var wdirty = $.parseHTML(rawContentString);
$wdirty = $(wdirty);
preProcess($wdirty);
if ($('*').length > maxNrOfElements) {
return force($wdirty, false);
}
dirty = '<div>' + $wdirty.html() + '</div>';
var results = '';
var lastFragment = '';
var lastTag = '';
HTMLParser(dirty, {
start: function(tag, attrs, unary) {
lastTag = tag;
if (allowedTags.indexOf(tag) < 0) {
return;
}
var tattrs = null;
if (tag === 'img') {
tattrs = attrs.filter(function(attr) {
return attr.name === 'src';
}).map(function(attr) {
return getImageSrc(decodeHtmlEntity(attr.value).trim());
});
lastFragment = tattrs.length === 0 ? '<img></img>' : '<img src="' + tattrs[0] + '" alt=""></img>';
} else if (tag === 'a') {
tattrs = attrs.filter(function(attr) {
return attr.name === 'href';
}).map(function(attr) {
return getHref(decodeHtmlEntity(attr.value).trim());
});
lastFragment = tattrs.length === 0 ? '<a>' : '<a href="' + tattrs[0] + '">';
} else {
lastFragment = '<' + tag + '>';
}
results += lastFragment;
lastFragment = '';
},
end: function(tag) {
if (allowedTags.indexOf(tag) < 0 || tag === 'img') {
return;
}
results += "</" + tag + ">\n";
},
chars: function(text) {
if (lastTag !== '' && allowedTags.indexOf(lastTag) < 0) {
return;
}
results += text;
},
comment: function(text) {
// results += "<!--" + text + "-->";
}
});
results = results.replace(/&nbsp;/gi, '&#160;');
return results;
} catch (e) {
console.log('Error:', e);
return force(dirty, true);
}
}
function getContent(htmlContent) {
try {
var tmp = document.createElement('div');
tmp.appendChild(htmlContent.cloneNode(true));
var dirty = '<div>' + tmp.innerHTML + '</div>';
return sanitize(dirty);
} catch (e) {
console.log('Error:', e);
return '';
}
}
/////
function getPageUrl(url) {
return url.toLowerCase().replace(/\s+/g,'_').replace(/[^a-z0-9_]/g,'') + Math.floor(Math.random() * 10000) + '.xhtml';
}
function getPageTitle(title) { //TODO
return title;
}
function getSelectedNodes() {
// if (document.selection) {
// return document.selection.createRange().parentElement();
// return document.selection.createRange();
// }
var selection = window.getSelection();
var docfrag = [];
for (var i = 0; i < selection.rangeCount; i++) {
docfrag.push(selection.getRangeAt(i).cloneContents());
}
return docfrag;
}
/////
function deferredAddZip(url, filename) {
var deferred = $.Deferred();
JSZipUtils.getBinaryContent(url, function(err, data) {
if (err) {
// deferred.reject(err); TODO
console.log('Error:', err);
deferred.resolve();
} else {
extractedImages.push({
filename: filename,
data: base64ArrayBuffer(data)
});
deferred.resolve();
}
});
return deferred;
}
chrome.runtime.onMessage.addListener(function(request, sender, sendResponse) {
var imgsPromises = [];
var result = {};
var pageSrc = '';
var tmpContent = '';
if (request.type === 'extract-page') {
pageSrc = document.getElementsByTagName('body')[0];
tmpContent = getContent(pageSrc);
} else if (request.type === 'extract-selection') {
pageSrc = getSelectedNodes();
pageSrc.forEach(function (page) {
tmpContent += getContent(page);
});
} else if (request.type === 'echo') {
sendResponse({
echo: true
});
return;
}
if (tmpContent.trim() === '') {
return;
}
allImages.forEach(function (tmpImg) {
imgsPromises.push(deferredAddZip(tmpImg.originalUrl, tmpImg.filename));
});
$.when.apply($, imgsPromises).done(function() {
result = {
url: getPageUrl(document.title),
title: getPageTitle(document.title),
baseUrl: getCurrentUrl(),
images: extractedImages,
content: tmpContent
};
sendResponse(result);
}).fail(function(e) {
console.log('Error:', e);
});
return true;
});