confidencial-ni-node - npm Package Compare versions

Comparing version 0.0.9 to 0.0.10

lib/confidencial.js

@@ -8,3 +8,5 @@ // Load modules
 var async = require('async');
-var jquery = fs.readFileSync(require('path').resolve(__dirname, 'jquery.min.js')).toString();
+var jquery = fs.readFileSync(
+  require('path').resolve(__dirname, 'jquery.min.js')
+).toString();
 
@@ -31,196 +33,180 @@ // confidencial categories
 
-
 // article scraping data
 var parseArticleOptions = {
-    domain: "http://www.confidencial.com.ni/",
+    domain: 'http://www.confidencial.com.ni/',
     elements: [
-        {
-            name: 'title',
-            sel: function ($) {
-                var result = $('#articleheader h2').text().trim();
-                return ( !S(result).isEmpty() ) ? result : '';
-            }
-        },
-        {
-            name: 'title_sub',
-            sel: function($) {
-                var result = $('#articleheader h3').text().trim();
-                return ( !S(result).isEmpty() ) ? result : '';
-            }
-        },
-        {
-            name: 'title_paragraph',
-            sel: function($) {
-                var result = $('#articleheader p.bold').text().trim();
-                return ( !S(result).isEmpty() ) ? result : '';
-            }
-        },
-        {
-            name: 'author',
-            sel: function($) {
-                var result = $('#articleheader p.authorname').text().trim().split("|")[0].trim();
-                return ( !S(result).isEmpty() ) ? result : '';
-            }
-        },
-        {
-            name: 'date',
-            sel: function($) {
-                var result = $('#articleheader p.authorname').text();
-                if( !S(result).isEmpty() ){
-                    result = result.match( /(\d{1,2}\/\d{1,2}\/\d{4})/g );
-                }
+      {
+        name: 'title',
+        sel: function ($) {
+          var result = $('#articleheader h2').text().trim();
+          return (!S(result).isEmpty()) ? result : '';
+        }
+      },
+      {
+        name: 'title_sub',
+        sel: function($) {
+          var result = $('#articleheader h3').text().trim();
+          return (!S(result).isEmpty()) ? result : '';
+        }
+      },
+      {
+        name: 'title_paragraph',
+        sel: function($) {
+          var result = $('#articleheader p.bold').text().trim();
+          return (!S(result).isEmpty()) ? result : '';
+        }
+      },
+      {
+        name: 'author',
+        sel: function($) {
+          var result = $('#articleheader p.authorname').text().
+            trim().split('|')[0].trim();
+          return (!S(result).isEmpty()) ? result : '';
+        }
+      },
+      {
+        name: 'date',
+        sel: function($) {
+          var result = $('#articleheader p.authorname').text();
+          if (!S(result).isEmpty()) {
+            result = result.match(/(\d{1,2}\/\d{1,2}\/\d{4})/g);
+          }
+          return (!S(result).isEmpty()) ? S(result).left(9).s : '';
+        }
+      },
+      {
+        name: 'images',
+        sel: function($) {
+          var result = [];
+          var array = $('article img').map(function() {
+            return $(this).attr('src');
+          }).get();
+          array.forEach(function(item) {
+            result.push(parseArticleOptions.domain + item);
+          });
+          return (!S(result).isEmpty()) ? result : [];
+        }
+      },
+      {
+        name: 'category',
+        sel: function($) {
+          var result = $('#quicknav').text();
+          if (!S(result).isEmpty()) {
+            result = S(result).trim().s;
+            result = S(result).replaceAll('Confidencial', '').s;
+            result = S(result).replaceAll('Leer artículo', '').s;
+            result = S(result).replaceAll('»', '').s;
+            result = S(result).trim().s;
+          }
+          return (!S(result).isEmpty()) ? S(result).trim().s : '';
+        }
+      },
+      {
+        name: 'content',
+        sel: function($) {
+          var newResult = '';
+          var result = $('article div.content_article div.text_article').html();
+          if (!S(result).isEmpty()) {
+            result = result.trim();
+            newResult = S(result).stripTags('p,br,strong').s;
+            newResult = S(newResult).collapseWhitespace().s;
+            newResult = S(newResult).replaceAll('<br />', '\n').s;
+            newResult = S(newResult).replaceAll('<br /> <br />', '\n').s;
+            newResult = S(newResult).replaceAll('<br /><br />', '\n').s;
+            newResult = S(newResult).replaceAll('<p>', '').s;
+            newResult = S(newResult).replaceAll('</p>', '\n').s;
+            newResult = S(newResult).replaceAll('&nbsp;', '').s;
+            newResult = S(newResult.replace(/<a.*href="(.*?)".*>(.*?)<\/a>/gi, '$2 $1')).trim().s;
+            newResult = newResult.replace(/<(?:.|\n)*?>/gm, '');
+            if (S(newResult).endsWith('\n')) {
 
-                return ( !S(result).isEmpty() ) ? S(result).left(9).s : '';
             }
-        },
-        {
-            name: 'images',
-            sel: function($) {
-                var result = [];
-                var array = $("article img").map(function() {
-                    return $(this).attr("src");
-                }).get();
-                array.forEach(function(item){
-                    result.push( parseArticleOptions.domain + item );
-                });
-                return ( !S(result).isEmpty() ) ? result : [];
-            }
-        },
-        {
-            name: 'category',
-            sel: function($) {
-                var result = $("#quicknav").text();
-                if( !S(result).isEmpty() ){
-                    result = S(result).trim().s;
-                    result = S(result).replaceAll('Confidencial', '').s;
-                    result = S(result).replaceAll('Leer artículo', '').s;
-                    result = S(result).replaceAll('»', '').s;
-                    result = S(result).trim().s;
-
-                }
-                return ( !S(result).isEmpty() ) ? S(result).trim().s : '';
-            }
-        },
-        {
-            name: 'content',
-            sel: function($) {
-                var new_result = "";
-                var result = $("article div.content_article div.text_article").html();
-                if( !S(result).isEmpty() ){
-                    result = result.trim();
-                    new_result = S( result ).stripTags('p,br,strong').s;
-
-                    new_result = S( new_result ).collapseWhitespace().s;
-                    new_result = S( new_result ).replaceAll('<br />', '\n').s;
-                    new_result = S( new_result ).replaceAll('<br /> <br />', '\n').s;
-                    new_result = S( new_result ).replaceAll('<br /><br />', '\n').s;
-                    new_result = S( new_result ).replaceAll('<p>', '').s;
-                    new_result = S( new_result ).replaceAll('</p>', '\n').s;
-                    new_result = S( new_result ).replaceAll('&nbsp;', '').s;
-
-                    new_result = S( new_result.replace(/<a.*href="(.*?)".*>(.*?)<\/a>/gi, "$2 $1") ).trim().s;
-
-                    new_result = new_result.replace(/<(?:.|\n)*?>/gm, '');
-
-                    if( S( new_result).endsWith('\n') ){
-
-                    }
-                }
-                return ( !S(result).isEmpty() ) ? S(new_result).trim().s : '';
-            }
+          }
+          return (!S(result).isEmpty()) ? S(newResult).trim().s : '';
         }
+      }
     ]
-};
+  };
 
-
 // simple function to compare if exist item in array
 function inArray(needle, haystack) {
-    var length = haystack.length;
-    for(var i = 0; i < length; i++) {
-        if(haystack[i] == needle) return true;
+  var length = haystack.length;
+  for (var i = 0; i < length; i++) {
+    if (haystack[i] == needle) {
+      return true;
     }
-    return false;
+  }
+  return false;
 }
 
-
 // function to parse data
 function parse(site, callback) {
-    jsdom.env({
-        url: site,
-        scripts: ["http://code.jquery.com/jquery.js"],
-        done: function (errors, window) {
-            var $ = window.$;
-            callback($, errors);
-        }
-    });
+  jsdom.env({
+    url: site,
+    scripts: ['http://code.jquery.com/jquery.js'],
+    done: function (errors, window) {
+      var $ = window.$;
+      callback($, errors);
+    }
+  });
 };
 
-
 // Global function to get all links from all categories
-module.exports.getAllLinks = function(callback){
-    function out(){
-        var result = [];
+module.exports.getAllLinks = function(callback) {
+  function out() {
+    var result = [];
 
-        var fetch = function(url,cb){
-            request(url, function(err,response,body){
-                if ( err ){
-                    cb( err );
-                } else {
-                    cb( null, body ); // First param indicates error, null=> no error
-                }
-            });
+    var fetch = function(url, cb) {
+      request(url, function(err, response, body) {
+        if (err) {
+          cb(err);
         }
+        else {
+          cb(null, body); // First param indicates error, null=> no error
+        }
+      });
+    }
 
-        async.map(categories, fetch, function(err, results){
-            if ( err){
-                console.log(err);
-                // either file1, file2 or file3 has raised an error, so you should not use results and handle the error
-            } else {
-                results.forEach(function(category){
-                    $ = cheerio.load(category);
-                    var links = $('.article h3 a,article h2 a'); //use your CSS selector here
-                    $(links).each(function(i, link){
-
-                        var article_url = 'http://confidencial.com.ni/' + $(link).attr('href');
-
-                        if( !inArray( article_url, result ) ){
-
-                            result.push( article_url );
-                        }
-                    });
-
-                });
-
-                callback(result);
+    async.map(categories, fetch, function(err, results) {
+      if (err) {
+        console.log(err);
+        // either file1, file2 or file3 has raised an error, so you should not use results and handle the error
+      } else {
+        results.forEach(function(category) {
+          $ = cheerio.load(category);
+          var links = $('.article h3 a,article h2 a'); //use your CSS selector here
+          $(links).each(function(i, link) {
+            var articleUrl = 'http://confidencial.com.ni/' +
+              $(link).attr('href');
+            if (!inArray(articleUrl, result)) {
+              result.push(articleUrl);
             }
+          });
         });
-    }
-
-    out();
+        callback(result);
+      }
+    });
+  }
+  out();
 }
 
-
 // global function to get data from article url
-module.exports.getArticle = function(site, cb){
+module.exports.getArticle = function(site, cb) {
+  function getId(url) {
+    var id = url.split('/');
+    return id[4];
+  }
 
-    function getId(url){
-        var id = url.split('/');
-        return id[4];
-    }
-
-    function out(url) {
-        parse(url, function ($, err) {
-            var result = {};
-            result.id = getId(url);
-            result.url = url;
-            parseArticleOptions.elements.forEach(function (elem) {
-                result[elem.name] = elem.sel($);
-            });
-
-            cb(result, err);
-        });
-    }
-
-    out(site);
-
+  function out(url) {
+    parse(url, function ($, err) {
+      var result = {};
+      result.id = getId(url);
+      result.url = url;
+      parseArticleOptions.elements.forEach(function (elem) {
+        result[elem.name] = elem.sel($);
+      });
+      cb(result, err);
+    });
+  }
+  out(site);
 }

package.json

 {
     "name": "confidencial-ni-node"
     , "description": "Web scraping http://www.confidencial.com.ni/"
-    , "version": "0.0.9"
+    , "version": "0.0.10"
     , "author": "Paulo McNally <[e-mail]>"
@@ -24,2 +24,2 @@ , "keywords": ["nicaragua", "diario", "news"]
 }
-}
+}

No alert changes

Improved metrics

Number of package files

9

increased by12.5%

Lines of code

574

increased by1.77%

Worsened metrics

Total package byte prevSize

103001

decreased by-0.73%

No dependency changes