4
0
Fork 0
Wiki Issues Projects Code Pull Requests Releases Activity
Browse Source

Тип переменых т работа с rg.ru

master
Alexei 1 year ago
parent
85c1ec5e71
commit
d9055b0894
1 changed files with 82 additions and 49 deletions
Whitespace
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Split View
Diff Options
Show Stats Download Patch File Download Diff File
  1. 129
      news_parser.js

129
news_parser.js
Unescape View File

@ -6,6 +6,7 @@
// @author AlexeiBv+mirocod@narod.ru
// @match https://tass.ru/*
// @match https://ria.ru/*
// @match https://rg.ru/*
// @match https://zakonvremeni.ru/*
// @icon https://icons.duckduckgo.com/ip2/zakonvremeni.ru.ico
// @grant none
@ -22,12 +23,20 @@
return a_Element.className;
}
function FindElementsByRegExp(a_GenElementNameFunc, a_RegExpPattern, a_ElementParent) {
function GetNodeName(a_Element) {
return a_Element.nodeName;
}
function CheckRegExp(a_GetElementNameFunc, a_RegExpPattern, a_Element) {
let re = new RegExp("(?:^|\\s)" + a_RegExpPattern + "(?!\\S)");
return re.test(a_GetElementNameFunc(a_Element));
}
function FindElementsByRegExp(a_GetElementNameFunc, a_RegExpPattern, a_ElementParent) {
a_ElementParent || (a_ElementParent=document);
var descendants = a_ElementParent.getElementsByTagName('*'), i=-1, e, result=[];
var re = new RegExp("(?:^|\\s)" + a_RegExpPattern + "(?!\\S)");
let descendants = a_ElementParent.getElementsByTagName('*'), i=-1, e, result=[];
while (e = descendants[++i]) {
if (re.test(a_GenElementNameFunc(e))){
if (CheckRegExp(a_GetElementNameFunc, a_RegExpPattern, e)){
result.push(e);
}
}
@ -41,7 +50,7 @@
}
function RemoveBeforeSplitter(a_String, a_Splitter) {
var index = a_String.indexOf(a_Splitter)
let index = a_String.indexOf(a_Splitter)
if (index != -1) {
return a_String.substring(index + a_Splitter.length);
}
@ -49,9 +58,9 @@
}
function RemoveAfterSplitter(a_String, a_Splitter, a_SaveSplitter) {
var index = a_String.indexOf(a_Splitter)
let index = a_String.indexOf(a_Splitter)
if (index != -1) {
var spl_len = a_Splitter.length
let spl_len = a_Splitter.length
if (!a_SaveSplitter) {
spl_len = 0;
}
@ -61,15 +70,15 @@
}
function ClearUrl(a_Url) {
var separator = '?';
const separator = '?';
return RemoveAfterSplitter(a_Url, separator, false);
}
function ClearTextFuncTemplate(a_RemoveBeforeList) {
function ClearTextFunc(a_Content) {
var content = a_Content;
let content = a_Content;
for (let i = 0; i < a_RemoveBeforeList.length; i++) {
var r = a_RemoveBeforeList[i];
let r = a_RemoveBeforeList[i];
content = RemoveBeforeSplitter(content, r);
}
return content;
@ -80,12 +89,12 @@
// Работа с контейнерами
function GetImageInContainers(a_Elements, a_TextAlign) {
var i;
var img_src = '';
var re = new RegExp("(https?:\/\/.*\.(?:png|jpg))");
let i;
let img_src = '';
let re = new RegExp("(https?:\/\/.*\.(?:png|jpg))");
for (i in a_Elements) {
var e = a_Elements[i];
var children = e.querySelectorAll("*");
let e = a_Elements[i];
let children = e.querySelectorAll("*");
for(let i = 0; i < children.length; i++){
var c = children[i];
if (c.nodeName == 'IMG' && re.test(c.src)) {
@ -100,11 +109,11 @@
}
function GetContentInContainers(a_Elements, a_GrubTextFunc, a_FinishWorkFunc) {
var result = '';
let result = '';
for (var i in a_Elements) {
var e = a_Elements[i];
let e = a_Elements[i];
var content = '';
let content = '';
if (e.querySelectorAll) {
var children = e.querySelectorAll("*");
if (children.length == 0 || e.innerText) {
@ -112,7 +121,7 @@
}
else {
for (let i = 0; i < children.length; i++) {
var c = children[i];
let c = children[i];
content += a_GrubTextFunc(c);
}
}
@ -131,9 +140,9 @@
// Фильтрация элементов
function FIlterElements(a_Elements, a_ElementChecker) {
var result = [];
let result = [];
for (let i = 0; i < a_Elements.length; i++) {
var e = a_Elements[i];
let e = a_Elements[i];
if (a_ElementChecker(e)) {
result.push(e);
}
@ -180,14 +189,23 @@
a_Content = a_ClearTextFunc(a_Content);
}
if (a_Element && a_Element.dataset && a_Element.dataset.type == 'list') {
var childrens = FindElementsByRegExp(GetElementClassName, 'article__list-item', a_Element);
let childrens = FindElementsByRegExp(GetElementClassName, 'article__list-item', a_Element);
let content = '';
for (let i = 0; i < childrens.length; i++) {
var c = childrens[i];
let c = childrens[i];
content += '<li>' + GrubTextFuncTemplate()(c) + '</li>';
}
a_Content = '<ul>' + content + '</ul>';
}
if (a_Element && CheckRegExp(GetElementClassName, 'PageContentCommonStyling_text.*', a_Element)) {
let content = '';
let childrens = FindElementsByRegExp(GetNodeName, 'P', a_Element);
for (let i = 0; i < childrens.length; i++) {
let c = childrens[i];
content += FinishWorkFunc(GrubTextFuncTemplate()(c), c);
}
return content
}
if (a_OutTag && a_TextAlign) {
a_Content = '<' + a_OutTag + ' style = "text-align:' + a_TextAlign + ';">' + a_Content + '</' + a_OutTag + '>';
@ -204,13 +222,13 @@
// Создание контента для стандартных новостей
function MakeContentByNews(a_BaseElementTitle, a_BaseElementImage, a_BaseElementText, a_TitleRegExpElementPattern, a_ImageRegExpElementPattern, a_TextRegExpElementPattern, a_ElementChecker, a_ClearTextPatterns) {
var title_tag = 'h2';
var p_tag = 'p';
var title_finish_text_func = FinishWorkFuncTemplate(title_tag, 'center')
var grub_func = GrubTextFuncTemplate();
const title_tag = 'h2';
const p_tag = 'p';
const title_finish_text_func = FinishWorkFuncTemplate(title_tag, 'center')
const grub_func = GrubTextFuncTemplate();
var content = '';
var paragraph_finish_text_func = FinishWorkFuncTemplate(p_tag, 'justify', ClearTextFuncTemplate(a_ClearTextPatterns));
let content = '';
const paragraph_finish_text_func = FinishWorkFuncTemplate(p_tag, 'justify', ClearTextFuncTemplate(a_ClearTextPatterns));
content += GetContentInContainers(FindElementsByRegExp(GetElementClassName, a_TitleRegExpElementPattern, a_BaseElementTitle), grub_func, title_finish_text_func);
content += GetImageInContainers(FindElementsByRegExp(GetElementClassName, a_ImageRegExpElementPattern, a_BaseElementImage), 'center');
content += GetContentInContainers(FIlterElements(FindElementsByRegExp(GetElementClassName, a_TextRegExpElementPattern, a_BaseElementText), a_ElementChecker), grub_func, paragraph_finish_text_func);
@ -220,13 +238,13 @@
// Создание контента для сайта
function MakeContent() {
var content = '';
var source_add = true;
var zero_tag_func = FinishWorkFuncTemplate()
var grub_text_func = GrubTextFuncTemplate()
let content = '';
let source_add = true;
const zero_tag_func = FinishWorkFuncTemplate()
const grub_text_func = GrubTextFuncTemplate()
if (location.hostname == 'tass.ru') {
let base_element = document.getElementById('content_box');
const base_element = document.getElementById('content_box');
content = MakeContentByNews(
base_element,
base_element,
@ -239,12 +257,12 @@
);
}
else if (location.hostname == 'ria.ru') {
let base_element = document.getElementsByClassName('article__header')[0];
var base_element_text = document.getElementsByClassName('article__body')[0];
var tire = ['–', '—', '‒', '―', '⸺', '⸻'];
var clear_text = [];
for (var i in tire) {
var t = tire[i];
const base_element = document.getElementsByClassName('article__header')[0];
const base_element_text = document.getElementsByClassName('article__body')[0];
const tire = ['–', '—', '‒', '―', '⸺', '⸻'];
let clear_text = [];
for (let i in tire) {
let t = tire[i];
clear_text.push(t + ' РИА Новости. ');
}
@ -259,17 +277,32 @@
clear_text
);
}
else if (location.hostname == 'rg.ru') {
const base_element = document.getElementsByClassName('article__header')[0];
const base_element_text = document.getElementsByClassName('article__body')[0];
content = MakeContentByNews(
document,
document,
base_element_text,
'PageArticleContent_title.*',
'PageArticleContent_image.*',
'(PageContentCommonStyling_text|PageArticleContent_lead).*',
ElementCheckerRia,
[]
);
}
else if (location.hostname == 'zakonvremeni.ru') {
let base_element = document.getElementsByClassName('item-page')[0];
var title = GetContentInContainers(FindElementsByRegExp(GetElementClassName, 'page-header', base_element), grub_text_func);
var parent_category = GetContentInContainers(FindElementsByRegExp(GetElementClassName, 'parent-category-name', base_element), grub_text_func);
var category = GetContentInContainers(FindElementsByRegExp(GetElementClassName, 'category-name', base_element), grub_text_func);
var page = RemoveAfterSplitter(TrimString(document.getElementsByClassName('item-page')[0].querySelector('[itemprop=articleBody]').textContent), '.', true);
const base_element = document.getElementsByClassName('item-page')[0];
const title = GetContentInContainers(FindElementsByRegExp(GetElementClassName, 'page-header', base_element), grub_text_func);
const parent_category = GetContentInContainers(FindElementsByRegExp(GetElementClassName, 'parent-category-name', base_element), grub_text_func);
const category = GetContentInContainers(FindElementsByRegExp(GetElementClassName, 'category-name', base_element), grub_text_func);
const page = RemoveAfterSplitter(TrimString(document.getElementsByClassName('item-page')[0].querySelector('[itemprop=articleBody]').textContent), '.', true);
content = title + '\n' + parent_category + ' ' + category + '\n\n' + page + '\n' + document.URL;
source_add = false;
}
var result = '';
let result = '';
if (content.length > 0) {
result = '<textarea id = "news_content" rows="10" cols="100">' + content;
if (source_add) {
@ -280,8 +313,8 @@
return result;
}
var content = MakeContent();
var logo = document.createElement("div");
let content = MakeContent();
let logo = document.createElement("div");
logo.innerHTML = '<div style="margin: 0pt auto; width: 800px; text-align: center;">' + content + '</div>';
document.body.insertBefore(logo, document.body.firstChild);

Write Preview
Loading…
Cancel
Save