Add rewrite NodeJS app

It helps to rewrite/reformat and make consistent the `baggers.js` file. ```sh node rewrite/rewrite.js rewrite/rewrite.js ``` It will generate a file named `rewrite/baggers-new.js` which is git ignored. The script: * moves `tags` under each talk, which means that every speaker will have then to modify its talks if tags are incorrects * normalized the `name` by uppercasing each first letter and lowercasing the other letters. * normalized website titles. We check for `LinkedIn`, `GitHub`, `Podcast`, `Google+`, `Slideshare`, `Hopwork`, `Book`, `InfoQ` and `Viadeo`. Other titles are set to `Web` * contact is now an object which contains two fields: * `type`: facultative. No data mean `mail`. If no contact was provided, it fallbacks to `twitter`. * `url`: the email address is provided, the twitter handle otherwise For example, `Gregory Estrade` will be modified from: ```json { "since": "2015-09-24", "name": "Gregory Estrade", "bio": "Directeur R&D de Lyra-Network, software & hardware hacker, reverse-engineer, philosophe agile auto-proclamé. Ou Disruptive Innovation Lead Strategist, Instant Expert in Everything, ça dépend. Membre de l'association Silicium de préservation du patrimoine micro-informatique.", "picture": "http://torlus.github.io/hat_pro.png", "websites": [ { "title": "Home Page", "href": "http://lvt.tl/" }, { "title": "GitHub", "href": "https://github.com/Torlus" }, { "title": "Silicium", "href": "http://www.silicium.org/" } ], "twitter": "Torlus", "contact": "gregory.estrade@gmail.com", "mail": "gregory.estrade@gmail.com", "location": "Toulouse", "sessions": [ { "title": "Les nouveaux enjeux du paiement.", "summary": "Etat de l'art, sécurité, les moyens de paiement et leur avenir, les enjeux du click-and-mortar, SEPA, marketplaces." }, { "title": "IoT, software & hardware hacking, FPGA.", "summary": "Vous aimez bricoler, prototyper? Moi aussi. Invitez-moi pour de l'aide ou du conseil sur un sujet de votre choix." }, { "title": "Café philosophique, agilité, histoire de l'informatique.", "summary": "Ayant eu depuis 30 ans les mains sur un clavier (En commençant sur un TRS-80) et étant membre de l'association Silicium, je vous propose de deviser autour de tout ce qui fait l'actualité, et qui tend à montrer que quoi qu'on fasse, l'histoire vient souvent à se répéter." } ], "tags": [ "Software development", "Hardware hacking", "Optimisation", "Telecom & Réseau", "FPGA", "Philosophie", "Agilité" ], "cities": [ "Toulouse" ] } ``` to: ```json { "since": "2015-09-24", "name": "Gregory Estrade", "bio": "Directeur R&D de Lyra-Network, software & hardware hacker, reverse-engineer, philosophe agile auto-proclamé. Ou Disruptive Innovation Lead Strategist, Instant Expert in Everything, ça dépend. Membre de l'association Silicium de préservation du patrimoine micro-informatique.", "picture": "http://torlus.github.io/hat_pro.png", "websites": [ { "name": "Web", "url": "http://lvt.tl/" }, { "name": "GitHub", "url": "https://github.com/Torlus" }, { "name": "Web", "url": "http://www.silicium.org/" } ], "twitter": "Torlus", "location": "Toulouse", "sessions": [ { "title": "Les nouveaux enjeux du paiement.", "abstract": "Etat de l'art, sécurité, les moyens de paiement et leur avenir, les enjeux du click-and-mortar, SEPA, marketplaces.", "tags": [ "Software development", "Hardware hacking", "Optimisation", "Telecom & Réseau", "FPGA", "Philosophie", "Agilité" ] }, { "title": "IoT, software & hardware hacking, FPGA.", "abstract": "Vous aimez bricoler, prototyper? Moi aussi. Invitez-moi pour de l'aide ou du conseil sur un sujet de votre choix.", "tags": [ "Software development", "Hardware hacking", "Optimisation", "Telecom & Réseau", "FPGA", "Philosophie", "Agilité" ] }, { "title": "Café philosophique, agilité, histoire de l'informatique.", "abstract": "Ayant eu depuis 30 ans les mains sur un clavier (En commençant sur un TRS-80) et étant membre de l'association Silicium, je vous propose de deviser autour de tout ce qui fait l'actualité, et qui tend à montrer que quoi qu'on fasse, l'histoire vient souvent à se répéter.", "tags": [ "Software development", "Hardware hacking", "Optimisation", "Telecom & Réseau", "FPGA", "Philosophie", "Agilité" ] } ], "cities": [ "Toulouse" ], "contact": { "url": "gregory.estrade@gmail.com" } } ``` Related to brownbaglunch#65.
dadoonet · Jan 21, 2016 · 43e8909 · 43e8909
1 parent 5c991d0
commit 43e8909
Show file tree

Hide file tree

Showing 3 changed files with 250 additions and 0 deletions.
diff --git a/.gitignore b/.gitignore
@@ -0,0 +1,3 @@
+.idea
+node_modules/
+baggers-new.js
diff --git a/rewrite/package.json b/rewrite/package.json
@@ -0,0 +1,30 @@
+{
+  "name": "rewrite",
+  "version": "1.0.0",
+  "description": "Rewrite Data",
+  "main": "rewrite.js",
+  "scripts": {
+    "test": "echo \"Error: no test specified\" && exit 1"
+  },
+  "repository": {
+    "type": "git",
+    "url": "git+https://github.com/brownbaglunch/bblfr_data.git"
+  },
+  "keywords": [
+    "bbl",
+    "brownbaglunch"
+  ],
+  "author": "David Pilato",
+  "license": "Apache-2.0",
+  "bugs": {
+    "url": "https://github.com/brownbaglunch/bblfr_data/issues"
+  },
+  "homepage": "https://github.com/brownbaglunch/bblfr_data#readme",
+  "dependencies": {
+    "async": ">= 1.5.1",
+    "express": "^4.13.3",
+    "moment": ">= 2.11.0",
+    "request": "^2.67.0",
+    "sync-request": ">= 2.1.0"
+  }
+}
diff --git a/rewrite/rewrite.js b/rewrite/rewrite.js
@@ -0,0 +1,217 @@
+#!/usr/bin/env node
+
+var https = require('https');
+var fs = require('fs');
+var moment = require('moment');
+var request = require('request');
+var path = require('path');
+
+console.log("*** starting");
+
+var now = moment();
+
+var data = fs.readFileSync(path.resolve(__dirname, '../baggers.js')).toString();
+
+// Remove the "var data = " part and
+script = data.replace("var data = ", "").replace(/;([^;]*)$/, '');
+bblfrData = JSON.parse(script);
+
+// Store cities as a map so we will be able to replace cities by coordinates
+var cities = {};
+bblfrData.cities.forEach(function(obj) {
+    cities[obj.name] = obj;
+} );
+
+var webSiteExists = function(name, url) {
+    request(url, {method: 'HEAD', timeout: 10000}, function (err, res, body){
+        if (res !== undefined) {
+            if (res.statusCode == 404 || res.statusCode >= 500) {
+                console.log(name, 'Res incorrect URL', url, 'got', res.statusCode);
+            }
+        } else {
+            console.log(name, 'Err incorrect URL', url, 'got', err);
+        }
+    });
+};
+
+// Read speakers
+var websiteTypes = {};
+var speakers = [];
+bblfrData.baggers.forEach(function(obj) {
+    var sessions = [];
+
+    obj.sessions.forEach(function(session) {
+        sessions.push({
+            "title": session.title,
+            "abstract": session.summary,
+            "tags": obj.tags
+        });
+    });
+
+    obj.sessions = sessions;
+    delete obj.tags;
+
+    // Generate the new contact
+    var contact;
+    if (obj.contact === undefined) {
+        contact = {
+            type: 'twitter',
+            url: obj.twitter
+        }
+    } else {
+        contact = {
+            url: obj.mail
+        }
+    }
+
+    if (obj.since === undefined || obj.since === '') {
+        // For audit purpose
+        console.log('no date for', obj.name);
+    }
+
+    delete obj.mail;
+    delete obj.contact;
+
+    obj.contact = contact;
+
+    // Rewrite names
+    obj.name = obj.name.replace(/\w\S*/g, function(txt){return txt.charAt(0).toUpperCase() + txt.substr(1).toLowerCase();});
+
+    // Rewrite website links
+    var websites = [];
+    if (obj.websites === undefined) {
+        // For audit purpose
+        console.log('no website for', obj.name);
+    } else {
+        obj.websites.forEach(function(website) {
+            // webSiteExists(obj.name, website.href);
+
+            var name = website.title.toLowerCase();
+            switch (name) {
+                case 'linkedin':
+                case 'linked in':
+                case 'profil linkedin':
+                    name = 'LinkedIn';
+                    break;
+                case 'github':
+                    name = 'GitHub';
+                    break;
+                case 'podcast':
+                    name = 'Podcast';
+                    break;
+                case 'google+':
+                    name = 'Google+';
+                    break;
+                case 'slideshare':
+                    name = 'Slideshare';
+                    break;
+                case 'hopwork':
+                    name = 'Hopwork';
+                    break;
+                case 'home page':
+                case 'silicium':
+                case 'ippon':
+                case 'code distillers':
+                case 'loxodata':
+                case 'use case driven':
+                case 'one.hundred.acorns':
+                case 'n-fluent.net':
+                case 'pragmafabrik':
+                case 'homepage':
+                case 'iergo':
+                case 'easymock':
+                case 'objenesis':
+                case 'faveod':
+                case 'arolla':
+                case 'agileit':
+                case 'websbite':
+                case 'codeurs en seine':
+                case 'lean kanban france':
+                case 'lesfurets.com':
+                case 'duchess france':
+                case 'affini-tech':
+                case 'random code':
+                case 'about me':
+                case 'ux-fr':
+                case 'carbon it':
+                case 'agilexperience':
+                case 'human coders':
+                case 'human coders formations':
+                case 'pragmatic entrepreneurs':
+                case 'conseil en business model':
+                case 'bloc-notes sur l\'expérience utilisateur':
+                case 'mes projets':
+                case 'tous sur java, c++, la sécurité et la poo':
+                    name = 'Web';
+                    break;
+                case 'les bouquins que j\'ai écrit':
+                    name = 'Book';
+                    break;
+                case 'infoq':
+                    name = 'InfoQ';
+                    break;
+                case 'viadeo':
+                    name = 'Viadeo';
+                    break;
+                default:
+                    if (name.toLowerCase().indexOf('blog') > -1) {
+                        name = 'Web';
+                    } else if (name.toLowerCase().indexOf('site') > -1) {
+                        name = 'Web';
+                    } else if (website.href.toLowerCase().indexOf('linkedin') > -1) {
+                        name = 'LinkedIn';
+                    } else if (website.href.toLowerCase().indexOf('about.me') > -1) {
+                        name = 'AboutMe';
+                    } else {
+                        name = 'Web';
+                    }
+            }
+
+            // Audit HTTP link
+
+
+            websites.push({
+                "name": name,
+                "url": website.href
+            });
+
+            obj.websites = websites;
+
+            // For audit purpose
+            if (websiteTypes[name] !== undefined) {
+                var nb = websiteTypes[name].nb;
+                nb++;
+                websiteTypes[name].nb = nb;
+                websiteTypes[name].list.push(website.href);
+            } else {
+                websiteTypes[name] =  {
+                    nb: 1,
+                    list: [ website.href ]
+                };
+            }
+
+        });
+    }
+
+    speakers.push(obj);
+
+//        console.log('speaker', obj.name, JSON.stringify(obj, null, 2));
+});
+
+// For audit purpose
+//console.log(websiteTypes);
+//console.log('different website types', Object.keys(websiteTypes).length);
+
+// Create the new final object
+var newData = {
+    speakers: speakers,
+    cities: cities
+};
+
+
+var finalString = 'var data = ' + JSON.stringify(newData, null, 2) + ";";
+
+fs.writeFileSync(path.resolve(__dirname, 'baggers-new.js'), finalString);
+
+console.log('file generated');
+