Add Expresso.pt (json)

README.md

@@ -785,6 +785,7 @@ Grouped in options:\
 
 ##### Portugal
 [Correio da Manhã](https://www.cmjornal.pt) -
+[Expresso](https://expresso.pt) -
 [Record](https://www.record.pt) -
 [Sábado](https://www.sabado.pt)
 

background.js

@@ -28,6 +28,7 @@ var restrictions = {
   'elespanol.com': /^((?!\/cronicaglobal\.elespanol\.com\/).)*$/,
   'espn.com': /^((?!espn\.com\/watch).)*$/,
   'esquire.com': /^((?!\/classic\.esquire\.com\/).)*$/,
+  'expresso.pt': /^((?!\/tribuna\.expresso\.pt\/).)*$/,
   'foreignaffairs.com': /^((?!\/reader\.foreignaffairs\.com\/).)*$/,
   'ft.com': /^((?!\/cn\.ft\.com\/).)*$/,
   'hilltimes.com': /^((?!\.hilltimes\.com\/slideshow\/).)*$/,

changelog.txt

@@ -6,6 +6,7 @@ Post-release
 Add BusinessTimes.com.sg & StraitsTimes.com (SPH Media)
 Add Dagsavisen.no
 Add Dn.se (fetch from archive.is)
+Add Expresso.pt (json)
 Add McPherson Media Group (au; opt-in to custom sites)
 Add The Lamp Magazine
 Fix Augsburger-allgemeine.de (Googlebot)

contentScript.js

@@ -1618,6 +1618,101 @@ else if (matchDomain(es_epiberica_domains) || matchDomain(es_epiberica_custom_do
   }
 }
 
+else if (matchDomain('expresso.pt')) {
+  if (!window.location.hostname.startsWith('amp.')) {
+    let article_sel = 'div.article-content';
+    let paywall = document.querySelector(article_sel + ' > div.g-premium-blocker');
+    if (paywall && dompurify_loaded) {
+      removeDOMElement(paywall);
+      let article = document.querySelector(article_sel);
+      if (article) {
+        let url = window.location.href.split(/[#\?]/)[0];
+        fetch(url)
+        .then(response => {
+          if (response.ok) {
+            response.text().then(html => {
+              if (html.match(/window\.__INITIAL_DATA__\s?=\s?/)) {
+                try {
+                  article.innerHTML = '';
+                  let json = JSON.parse(html.split(/window\.__INITIAL_DATA__\s?=\s?/)[1].split(';window.')[0].replace(/":undefined([,}])/g, "\":\"undefined\"$1")).nodes;
+                  let pars;
+                  for (let elem in json) {
+                    let item = json[elem];
+                    if (item.type === 'Layout' && item.nodes[0].type === 'MainBody') {
+                      pars = item.nodes[0].nodes[0].data.content.contents;
+                      break;
+                    }
+                  }
+                  let parser = new DOMParser();
+                  for (let par of pars) {
+                    let par_new;
+                    if (par.html) {
+                      let doc = parser.parseFromString('<div>' + DOMPurify.sanitize(par.html) + '</div>', 'text/html');
+                      par_new = doc.querySelector('div');
+                    } else if (par.type === 'PICTURE') {
+                      if (par.urlOriginal) {
+                        par_new = document.createElement('figure');
+                        let img = document.createElement('img');
+                        img.src = par.urlOriginal;
+                        img.style = 'width:100%';
+                        par_new.appendChild(img);
+                        if (par.caption) {
+                          let caption = document.createElement('p');
+                          caption.innerText = par.caption;
+                          par_new.appendChild(caption);
+                        }
+                      }
+                    } else if (par.link && par.title) {
+                      if (par.contents) {
+                        par_new = document.createElement('div');
+                        for (let elem of par.contents) {
+                          let elem_new;
+                          if (elem.html) {
+                            let doc = parser.parseFromString('<div>' + DOMPurify.sanitize(elem.html) + '</div>', 'text/html');
+                            elem_new = doc.querySelector('div');
+                          } else if (elem.urlOriginal) {
+                            elem_new = document.createElement('figure');
+                            let img = document.createElement('img');
+                            img.src = elem.urlOriginal;
+                            img.style = 'width:100%';
+                            elem_new.appendChild(img);
+                            if (elem.caption) {
+                              let caption = document.createElement('p');
+                              caption.innerText = elem.caption;
+                              elem_new.appendChild(caption);
+                            }
+                          }
+                          if (elem_new)
+                            par_new.appendChild(elem_new);
+                        }
+                      } else {
+                        par_new = document.createElement('p');
+                        let art_link = document.createElement('a');
+                        art_link.innerText = par.title;
+                        art_link.href = par.link;
+                        par_new.appendChild(art_link);
+                      }
+                    }
+                    if (par_new)
+                      article.appendChild(par_new);
+                    else
+                      console.log(par);
+                  }
+                } catch (err) {
+                  console.log(err);
+                }
+              }
+            });
+          }
+        }).catch(function (err) {
+          false;
+        });
+      }
+    }
+  } else
+    ampToHtml();
+}
+
 else if (matchDomain(['lavanguardia.com', 'mundodeportivo.com'])) {
   let ads = document.querySelectorAll('span.content-ad, span.hidden-ad, span.ad-unit, div.ad-div');
   hideDOMElement(...ads);

custom/manifest.json

@@ -51,5 +51,5 @@
     "webRequestBlocking",
     "*://*/*"
   ],
-  "version": "3.5.3.7"
+  "version": "3.5.3.8"
 }

manifest.json

@@ -261,6 +261,7 @@
     "*://*.examiner.com.au/*",
     "*://*.expansion.com/*",
     "*://*.expressnews.com/*",
+    "*://*.expresso.pt/*",
     "*://*.farodevigo.es/*",
     "*://*.fastcompany.com/*",
     "*://*.faz.net/*",
@@ -845,5 +846,5 @@
     "*://archive.vn/*",
     "*://webcache.googleusercontent.com/*"
   ],
-  "version": "3.5.3.7"
+  "version": "3.5.3.8"
 }

sites.js

@@ -692,6 +692,11 @@ var defaultSites = {
     domain: "exame.com",
     allow_cookies: 1
   },
+  "Expresso.pt": {
+    domain: "expresso.pt",
+    allow_cookies: 1,
+    cs_dompurify: 1
+  },
   "Fast Company": {
     domain: "fastcompany.com",
     allow_cookies: 1,
@@ -3022,4 +3027,4 @@ var de_westfalen_medien_nofix_domains = ['muensterschezeitung.de', 'westfalen-bl
 var fr_groupe_ebra_nofix_domains = ['bienpublic.com', 'dna.fr', 'estrepublicain.fr', 'lalsace.fr', 'ledauphine.com', 'lejsl.com', 'leprogres.fr', 'republicain-lorrain.fr', 'vosgesmatin.fr'];
 var fr_indigo_nofix_domains = ['africaintelligence.com', 'africaintelligence.fr', 'glitz.paris', 'intelligenceonline.com', 'intelligenceonline.fr', 'lalettre.fr'];
 var it_gedi_nofix_domains = ['gelocal.it', 'limesonline.com'];
-var nofix_sites = ['11freunde.de', 'aamulehti.fi', 'abendblatt.de', 'aftenposten.no', 'aftonbladet.se', 'allgaeuer-zeitung.de', 'asahi.com', 'asiatimes.com', 'autosport.com', 'aviationweek.com', 'badische-zeitung.de', 'bhaskar.com', 'bloomberglaw.com', 'bloombergtax.com', 'bnef.com', 'bnn.de', 'borsen.dk', 'businessinsider.de', 'businessinsider.jp', 'businesslive.co.za', 'caixin.com', 'caixinglobal.com', 'caravanmagazine.in', 'catalyst-journal.com', 'chegg.com', 'codesports.com.au', 'compactmag.com', 'courrierinternational.com', 'coursehero.com',  'deutsche-wirtschafts-nachrichten.de', 'die-glocke.de', 'dn.no', 'elordenmundial.com', 'entrepreneur.com', 'epw.in', 'expresso.pt', 'falter.at', 'finance.si', 'franc-tireur.fr', 'ftchinese.com', 'ftchineselive.com', 'gamestar.de', 'geo.de', 'golem.de', 'gp.se', 'gva.be', 'handelsblatt.com', 'hbrarabic.com', 'hbrchina.org', 'hbrfrance.fr', 'heise.de', 'hs.fi', 'ilsole24ore.com', 'information.dk', 'investors.com', 'iltalehti.fi', 'jacobin.com', 'jeuneafrique.com', 'jungefreiheit.de', 'kleinezeitung.at', 'laverita.info', 'lavie.fr', 'lavozdegalicia.es', 'law360.co.uk', 'law360.com', 'le1hebdo.fr', 'leconomiste.com', 'lefilmfrancais.com', 'lemonde.fr', 'lequipe.fr', 'lesjours.fr', 'letemps.ch', 'liberation.fr', 'libertiesjournal.com', 'main-echo.de', 'mainpost.de', 'manager-magazin.de', 'medianama.com', 'mediapart.fr', 'milanofinanza.it', 'mittelbayerische.de', 'monde-diplomatique.fr', 'mondediplo.com', 'money.it', 'moneycontrol.com', 'moodys.com', 'morningstar.com', 'motorsport.com', 'moz.de', 'nachrichten.at', 'nationaljournal.com', 'nature.com', 'nbr.co.nz', 'newcriterion.com', 'news24.com', 'newslaundry.com', 'nn.de', 'nwzonline.de', 'observador.pt', 'ouest-france.fr', 'philonomist.com', 'pnp.de', 'politicopro.com', 'politiken.dk', 'pressreader.com', 'publico.pt', 'quillette.com', 'rbc.ru', 'republic.ru', 'rheinpfalz.de', 'risk.net', 'rnz.de', 'saechsische.de', 'sciencedirect.com', 'springer.com', 'statnews.com', 'stern.de', 'stimme.de', 'streetinsider.com', 'substack.com', 'suedkurier.de', 'swp.de', 'taxation.co.uk', 'taxjournal.com', 'techcrunch.com', 'the-ken.com', 'theinformation.com', 'theinitium.com', 'themorningcontext.com', 'theparisreview.org', 'thestar.com.my', 'thewirechina.com', 'timeslive.co.za', 'weltwoche.ch', 'weltwoche.de', 'wissenschaft.de', 'worldpoliticsreview.com', 'wz.de', 'ynet.co.il', 'zaobao.com.sg'].concat(de_funke_medien_nofix_domains, de_rp_aachen_medien_nofix_domains, de_westfalen_medien_nofix_domains, fr_groupe_ebra_nofix_domains, fr_indigo_nofix_domains, it_gedi_nofix_domains);
+var nofix_sites = ['11freunde.de', 'aamulehti.fi', 'abendblatt.de', 'aftenposten.no', 'aftonbladet.se', 'allgaeuer-zeitung.de', 'asahi.com', 'asiatimes.com', 'autosport.com', 'aviationweek.com', 'badische-zeitung.de', 'bhaskar.com', 'bloomberglaw.com', 'bloombergtax.com', 'bnef.com', 'bnn.de', 'borsen.dk', 'businessinsider.de', 'businessinsider.jp', 'businesslive.co.za', 'caixin.com', 'caixinglobal.com', 'caravanmagazine.in', 'catalyst-journal.com', 'chegg.com', 'codesports.com.au', 'compactmag.com', 'courrierinternational.com', 'coursehero.com',  'deutsche-wirtschafts-nachrichten.de', 'die-glocke.de', 'dn.no', 'elordenmundial.com', 'entrepreneur.com', 'epw.in', 'falter.at', 'finance.si', 'franc-tireur.fr', 'ftchinese.com', 'ftchineselive.com', 'gamestar.de', 'geo.de', 'golem.de', 'gp.se', 'gva.be', 'handelsblatt.com', 'hbrarabic.com', 'hbrchina.org', 'hbrfrance.fr', 'heise.de', 'hs.fi', 'ilsole24ore.com', 'information.dk', 'investors.com', 'iltalehti.fi', 'jacobin.com', 'jeuneafrique.com', 'jungefreiheit.de', 'kleinezeitung.at', 'laverita.info', 'lavie.fr', 'lavozdegalicia.es', 'law360.co.uk', 'law360.com', 'le1hebdo.fr', 'leconomiste.com', 'lefilmfrancais.com', 'lemonde.fr', 'lequipe.fr', 'lesjours.fr', 'letemps.ch', 'liberation.fr', 'libertiesjournal.com', 'main-echo.de', 'mainpost.de', 'manager-magazin.de', 'medianama.com', 'mediapart.fr', 'milanofinanza.it', 'mittelbayerische.de', 'monde-diplomatique.fr', 'mondediplo.com', 'money.it', 'moneycontrol.com', 'moodys.com', 'morningstar.com', 'motorsport.com', 'moz.de', 'nachrichten.at', 'nationaljournal.com', 'nature.com', 'nbr.co.nz', 'newcriterion.com', 'news24.com', 'newslaundry.com', 'nn.de', 'nwzonline.de', 'observador.pt', 'ouest-france.fr', 'philonomist.com', 'pnp.de', 'politicopro.com', 'politiken.dk', 'pressreader.com', 'publico.pt', 'quillette.com', 'rbc.ru', 'republic.ru', 'rheinpfalz.de', 'risk.net', 'rnz.de', 'saechsische.de', 'sciencedirect.com', 'springer.com', 'statnews.com', 'stern.de', 'stimme.de', 'streetinsider.com', 'substack.com', 'suedkurier.de', 'swp.de', 'taxation.co.uk', 'taxjournal.com', 'techcrunch.com', 'the-ken.com', 'theinformation.com', 'theinitium.com', 'themorningcontext.com', 'theparisreview.org', 'thestar.com.my', 'thewirechina.com', 'timeslive.co.za', 'weltwoche.ch', 'weltwoche.de', 'wissenschaft.de', 'worldpoliticsreview.com', 'wz.de', 'ynet.co.il', 'zaobao.com.sg'].concat(de_funke_medien_nofix_domains, de_rp_aachen_medien_nofix_domains, de_westfalen_medien_nofix_domains, fr_groupe_ebra_nofix_domains, fr_indigo_nofix_domains, it_gedi_nofix_domains);