Сайт ИННОТВ и фильтрация пустых тегов. #8

news_parser.js

@@ -47,8 +47,8 @@
 
     // Работа со строками
 
-    function TrimString(s) {
+    function TrimString(str) {
-        return ( s || '' ).replace( /^\s+|\s+$/g, '' );
+        return ( str || '' ).replace( /^\s+|\s+$/g, '' );
     }
 
     function RemoveBeforeSplitter(a_String, a_Splitter) {
@@ -99,9 +99,12 @@
         let re = new RegExp("(https?:\/\/.*\.(?:png|jpg))");
         for (i in a_Elements) {
             let e = a_Elements[i];
+            if (e.nodeName == 'IMG' && re.test(e.src)) {
+                img_src = e.src;
+            }
             let children = e.querySelectorAll("*");
             for(let i = 0; i < children.length; i++){
-                var c = children[i];
+                let c = children[i];
                 if (c.nodeName == 'IMG' && re.test(c.src)) {
                     img_src = c.src;
                 }
@@ -208,7 +211,7 @@
 
         let elements = clear_element.querySelectorAll('*');
         elements.forEach(function (element) {
-            if (a_SubElementCheckerToRemove(element)) {
+            if (a_SubElementCheckerToRemove && a_SubElementCheckerToRemove(element)) {
                 element.parentNode.removeChild(element);
             }
             else {
@@ -240,14 +243,14 @@
             let elements = clear_element.querySelectorAll(align_tag_name);
             elements.forEach(function (element) {
                 element.style.textAlign = a_TextAlign;
-                if (element.innerHTML == '&nbsp;' || element.innerHTML == '') {
+                if (TrimString(element.innerHTML).replace('&nbsp;', '') == '') {
                     element.parentNode.removeChild(element);
                 }
             });
         });
 
-        let result = clear_element.innerHTML;
+        let result = TrimString(clear_element.innerHTML);
-        if (a_OutTag && a_TextAlign) {
+        if (a_OutTag && a_TextAlign && TrimString(result).replace('&nbsp;', '') != '') {
             result = '<' + a_OutTag + ' style = "text-align:' + a_TextAlign + ';">' + result + '</' + a_OutTag + '>';
         }
         return result;
@@ -275,11 +278,11 @@
     }
 
     // Создание контента для стандартных новостей
+    const title_tag = 'h2';
+    const title_finish_text_func = FinishWorkFuncTemplate(title_tag, 'center')
 
     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 = '';
@@ -374,6 +377,29 @@
                     ),
                 []
                 );
+                
+            if (content.length == 0) {
+                // test: https://russian.rt.com/inotv/2023-10-27/DELFI-Latviya-budet-konfiskovivat-mashini
+                const base_element_title = document.getElementsByClassName('left-column page')[0].getElementsByTagName("h1")[0];
+                const base_element_image = document.getElementsByTagName("figure")[0];
+                const base_element_text = document.getElementsByTagName("article")[0];
+
+                content = title_finish_text_func(base_element_title) +
+                    MakeContentByNews(
+                        base_element_title,
+                        base_element_image,
+                        base_element_text,
+                        '!!!!', 
+                        '.*',
+                        'article-intro|article-body',
+                        ElementCheckerTrue,
+                        SubElementCheckerToRemoveTemplate(
+                            'meta',
+                            'IMG'
+                            ),
+                        []
+                        );
+            }
         }
         else if (location.hostname == 'www.cnews.ru') {
             // test: https://www.cnews.ru/news/top/2023-10-27_rossiyane_sozdali_polnotsennyj