Реализованы CNEWS #7

1 year ago · 79b00c6702
1 changed files with 57 additions and 11 deletions
--- a/news_parser.js
+++ b/news_parser.js
@ -7,6 +7,7 @@
 // @match        https://tass.ru/*
 // @match        https://ria.ru/*
 // @match        https://rg.ru/*
+// @match        https://www.cnews.ru/*
 // @match        https://russian.rt.com/*
 // @match        https://zakonvremeni.ru/*
 // @icon         https://icons.duckduckgo.com/ip2/zakonvremeni.ru.ico
@ -142,6 +143,10 @@
        return true;
    }

+    function ElementCheckerFalse(a_Element) {
+        return false;
+    }
+
    function ElementCheckerRia(a_Element) {
        if (a_Element.dataset.type == 'article' || a_Element.dataset.type == 'banner' || a_Element.dataset.type == 'media') {
            return false;
@ -157,6 +162,22 @@
        return false;
    }

+    function SubElementCheckerToRemoveTemplate(a_Classes, a_NodeNames) {
+        function SubElementCheckerToRemove(a_Element) {
+            if (!a_Element || !a_Element.parentNode) {
+                return false;
+            }
+            if (CheckRegExp(GetElementClassName, '.*(' + a_Classes + ').*', a_Element)) {
+                return true;
+            }
+            if ((a_NodeNames && CheckRegExp(GetNodeName, '(' + a_NodeNames + ')', a_Element))) {
+                return true;
+            }
+            return false;
+        }
+        return SubElementCheckerToRemove
+    }
+
    // Обработка элементов

    function GrubTextFuncTemplate() {
@ -182,15 +203,12 @@
        parent.removeChild(a_Element);
    }

-    function GetClearHtml(a_Element, a_OutTag, a_TextAlign) {
+    function GetClearHtml(a_Element, a_OutTag, a_TextAlign, a_SubElementCheckerToRemove) {
        let clear_element = a_Element.cloneNode(true);

        let elements = clear_element.querySelectorAll('*');
        elements.forEach(function (element) {
-            let for_rt_com = 'read-more|article__cover';
-            let for_rg_com = 'portal|rg-incut|article-img|Section';
-            let delete_in_rg = element.nodeName == 'RG-VIDEO' || element.nodeName == 'RG-INCUT';
-            if (element && (element.parentNode && CheckRegExp(GetElementClassName, '.*(' + for_rt_com + '|' + for_rg_com + ').*', element) || delete_in_rg)) {
+            if (a_SubElementCheckerToRemove(element)) {
                element.parentNode.removeChild(element);
            }
            else {
@ -236,13 +254,13 @@
    }


-    function FinishWorkFuncTemplate(a_OutTag, a_TextAlign, a_ClearTextFunc) {
+    function FinishWorkFuncTemplate(a_OutTag, a_TextAlign, a_ClearTextFunc, a_SubElementCheckerToRemove) {
        function FinishWorkFunc(a_Element, a_ElementIndex) {
            let out_tag = a_OutTag;
            if (a_Element && CheckRegExp(GetElementClassName, 'Title_title.*', a_Element)) {
                out_tag = 'h2';
            }
-            let a_Content = GetClearHtml(a_Element, out_tag, a_TextAlign)
+            let a_Content = GetClearHtml(a_Element, out_tag, a_TextAlign, a_SubElementCheckerToRemove)
            if (a_ClearTextFunc) {
                a_Content = a_ClearTextFunc(a_Content, a_ElementIndex);
            }
@ -258,14 +276,14 @@

    // Создание контента для стандартных новостей

-    function MakeContentByNews(a_BaseElementTitle, a_BaseElementImage, a_BaseElementText, a_TitleRegExpElementPattern, a_ImageRegExpElementPattern, a_TextRegExpElementPattern, a_ElementChecker, a_ClearTextPatterns) {
+    function MakeContentByNews(a_BaseElementTitle, a_BaseElementImage, a_BaseElementText, a_TitleRegExpElementPattern, a_ImageRegExpElementPattern, a_TextRegExpElementPattern, a_ElementChecker, a_SubElementCheckerToRemove, a_ClearTextPatterns) {
        const title_tag = 'h2';
        const p_tag = 'p';
        const title_finish_text_func = FinishWorkFuncTemplate(title_tag, 'center')
        const grub_func = GrubTextFuncTemplate();

        let content = '';
-        const paragraph_finish_text_func = FinishWorkFuncTemplate(p_tag, 'justify', ClearTextFuncTemplate(a_ClearTextPatterns, true));
+        const paragraph_finish_text_func = FinishWorkFuncTemplate(p_tag, 'justify', ClearTextFuncTemplate(a_ClearTextPatterns, true), a_SubElementCheckerToRemove);
        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);
@ -291,6 +309,7 @@
                'Image_wrapper_.*',
                '(Paragraph_paragraph|Title_title).*',
                ElementCheckerTrue,
+                ElementCheckerFalse,
                ['. ']
                );
        }
@ -313,6 +332,7 @@
                'photoview__open',
                'article__block',
                ElementCheckerRia,
+                ElementCheckerFalse,
                ['</strong>']
                );
        }
@ -328,7 +348,11 @@
                'PageArticleContent_title.*',
                'PageArticleContent_image.*',
                '(PageContentCommonStyling_text|PageArticleContent_lead).*',
-                ElementCheckerRia,
+                ElementCheckerTrue,
+                SubElementCheckerToRemoveTemplate(
+                    'portal|rg-incut|article-img|Section',
+                    'RG-VIDEO|RG-INCUT'
+                    ),
                []
                );
        }
@ -344,7 +368,29 @@
                'article__heading',
                'article__cover article__cover_article-page',
                'article__text',
-                ElementCheckerRia,
+                ElementCheckerTrue,
+                SubElementCheckerToRemoveTemplate(
+                    'read-more|article__cover'
+                    ),
+                []
+                );
+        }
+        else if (location.hostname == 'www.cnews.ru') {
+            // test: https://www.cnews.ru/news/top/2023-10-27_rossiyane_sozdali_polnotsennyj
+            const base_element = document.getElementsByClassName('news_containere')[0];
+
+            content = MakeContentByNews(
+                base_element,
+                base_element,
+                document,
+                '!!!', // Нет названия
+                'img-block',
+                'news_container',
+                ElementCheckerTrue,
+                SubElementCheckerToRemoveTemplate(
+                    'article-top-author|article-menu_base|d-flex|img-block|NewsBodyLeftInclude|mobile-zone|other-news-note|cnLike|article-bottom-info|banner|comments_all',
+                    'NOINDEX|BR'
+                    ),
                []
                );
        }