Skip to content

Commit 7fd8a75

Browse files
hoothinhoothin
committed
Update DownloadAllContent.user.js
1 parent 72a8701 commit 7fd8a75

1 file changed

Lines changed: 55 additions & 40 deletions

File tree

‎DownloadAllContent/DownloadAllContent.user.js‎

Lines changed: 55 additions & 40 deletions
Original file line numberDiff line numberDiff line change
@@ -4,7 +4,7 @@
44
// @name:zh-TW 怠惰小説下載器
55
// @name:ja 怠惰者小説ダウンロードツール
66
// @namespace hoothin
7-
// @version 2.8.1
7+
// @version 2.8.2
88
// @description Fetch and download main textual content from the current page, provide special support for novels
99
// @description:zh-CN 通用网站内容抓取工具，可批量抓取任意站点的小说、论坛内容等并保存为TXT文档
1010
// @description:zh-TW 通用網站內容抓取工具，可批量抓取任意站點的小說、論壇內容等並保存為TXT文檔
@@ -1311,52 +1311,67 @@ if (window.top != window.self) {
13111311
}
13121312
if(!largestContent)return i18n.error+" : NO TEXT CONTENT";
13131313
var retainImage=!!GM_getValue("retainImage");
1314-
var childlist=pageData.querySelectorAll(largestContent.nodeName);//+(largestContent.className?"."+largestContent.className.replace(/(^\s*)|(\s*$)/g, '').replace(/\s+/g, '.'):""));
1315-
function getRightStr(ele, noTextEnable){
1316-
if(retainImage){
1317-
[].forEach.call(ele.querySelectorAll("img[src]"), img => {
1318-
let imgTxtNode=document.createTextNode(`![img](${canonicalUri(img.getAttribute("src"), url || location.href)})`);
1319-
img.parentNode.replaceChild(imgTxtNode, img);
1314+
function getContentByLargest() {
1315+
var childlist=pageData.querySelectorAll(largestContent.nodeName);//+(largestContent.className?"."+largestContent.className.replace(/(^\s*)|(\s*$)/g, '').replace(/\s+/g, '.'):""));
1316+
function getRightStr(ele, noTextEnable){
1317+
[].forEach.call(ele.querySelectorAll("a[href]"), a => {
1318+
a.parentNode && a.parentNode.removeChild(a);
13201319
});
1321-
}
1322-
let childNodes=ele.childNodes,cStr="\r\n",hasText=false;
1323-
[].forEach.call(ele.querySelectorAll("a[href]"), a => {
1324-
a.parentNode && a.parentNode.removeChild(a);
1325-
});
1326-
for(let j=0;j<childNodes.length;j++){
1327-
let childNode=childNodes[j];
1328-
if(childNode.nodeType==3 && childNode.data && !/^[\s\-\_\?\>\|]*$/.test(childNode.data))hasText=true;
1329-
if(childNode.innerHTML){
1330-
childNode.innerHTML=childNode.innerHTML.replace(/\<\s*br\s*\>/gi,"\r\n").replace(/\n+/gi,"\n").replace(/\r+/gi,"\r");
1320+
if(retainImage){
1321+
[].forEach.call(ele.querySelectorAll("img[src]"), img => {
1322+
let imgTxtNode=document.createTextNode(`![img](${canonicalUri(img.getAttribute("src"), url || location.href)})`);
1323+
img.parentNode.replaceChild(imgTxtNode, img);
1324+
});
13311325
}
1332-
let content=childNode.textContent;
1333-
if(content){
1334-
if(!content.trim())continue;
1335-
cStr+=content.replace(/[\uFEFF\xA0 ]+/g," ").replace(/([^\r]|^)\n([^\r]|$)/gi,"$1\r\n$2");
1326+
let childNodes=ele.childNodes,cStr="\r\n",hasText=false;
1327+
for(let j=0;j<childNodes.length;j++){
1328+
let childNode=childNodes[j];
1329+
if(childNode.nodeType==3 && childNode.data && !/^[\s\-\_\?\>\|]*$/.test(childNode.data))hasText=true;
1330+
if(childNode.innerHTML){
1331+
childNode.innerHTML=childNode.innerHTML.replace(/\<\s*br\s*\>/gi,"\r\n").replace(/\n+/gi,"\n").replace(/\r+/gi,"\r");
1332+
}
1333+
let content=childNode.textContent;
1334+
if(content){
1335+
if(!content.trim())continue;
1336+
cStr+=content.replace(/[\uFEFF\xA0 ]+/g," ").replace(/([^\r]|^)\n([^\r]|$)/gi,"$1\r\n$2");
1337+
}
1338+
if(childNode.nodeType!=3 && !/^(I|A|STRONG|B|FONT|IMG)$/.test(childNode.nodeName))cStr+="\r\n";
13361339
}
1337-
if(childNode.nodeType!=3 && !/^(I|A|STRONG|B|FONT|IMG)$/.test(childNode.nodeName))cStr+="\r\n";
1340+
if(hasText || noTextEnable || ele==largestContent)rStr+=cStr+"\r\n";
13381341
}
1339-
if(hasText || noTextEnable || ele==largestContent)rStr+=cStr+"\r\n";
1340-
}
1341-
var sameDepthChildren=[];
1342-
for(i=0;i<childlist.length;i++){
1343-
var child=childlist[i];
1344-
if(getDepth(child)==getDepth(largestContent)){
1345-
if(largestContent.className != child.className)continue;
1346-
sameDepthChildren.push(child);
1342+
var sameDepthChildren=[];
1343+
for(i=0;i<childlist.length;i++){
1344+
var child=childlist[i];
1345+
if(getDepth(child)==getDepth(largestContent)){
1346+
if(largestContent.className != child.className)continue;
1347+
sameDepthChildren.push(child);
1348+
}
13471349
}
1350+
var minLength = largestNum>>2;
1351+
var tooShort = sameDepthChildren.length <= 3;
1352+
sameDepthChildren.forEach(child => {
1353+
if(tooShort && child.innerText.length < minLength) return;
1354+
if((largestContent.className && largestContent.className == child.className) || largestContent.parentNode == child.parentNode){
1355+
getRightStr(child, true);
1356+
}else {
1357+
getRightStr(child, false);
1358+
}
1359+
});
1360+
rStr = rStr.replace(/[\n\r]+/g,"\n\r");
13481361
}
1349-
var minLength = largestNum>>2;
1350-
var tooShort = sameDepthChildren.length <= 3;
1351-
sameDepthChildren.forEach(child => {
1352-
if(tooShort && child.innerText.length < minLength) return;
1353-
if((largestContent.className && largestContent.className == child.className) || largestContent.parentNode == child.parentNode){
1354-
getRightStr(child, true);
1355-
}else {
1356-
getRightStr(child, false);
1362+
getContentByLargest();
1363+
if (rStr.length < 100) {
1364+
let articles = pageData.querySelectorAll("article");
1365+
if (articles && articles.length == 1) {
1366+
largestContent = articles[0];
1367+
largestNum = largestContent.innerText.length;
1368+
if (largestNum > 100) {
1369+
rStr = "";
1370+
getContentByLargest();
1371+
}
13571372
}
1358-
});
1359-
return rStr.replace(/[\n\r]+/g,"\n\r");
1373+
}
1374+
return rStr;
13601375
}
13611376

13621377
function getI18n(key, args){

0 commit comments

Comments
 (0)