From 975cb3cf1620127530635ef2c7128db8e40f1ae0 Mon Sep 17 00:00:00 2001
From: Davor Spasovski <davor@spasovski.net>
Date: Tue, 22 Apr 2014 12:35:42 -0400
Subject: [PATCH] add fetchdb task ported for commonplace

---
 .gitignore                                    |   4 +
 Gruntfile.js                                  |  39 +++++
 hearth/db/archives/.git-keep                  |   0
 hearth/downloads/icons/.git-keep              |   0
 hearth/downloads/screenshots-thumbs/.git-keep |   0
 lib/db-transformer.js                         | 151 ++++++++++++++++++
 lib/db.js                                     | 129 +++++++++++++++
 lib/utils.js                                  |  33 ++++
 package.json                                  |   9 +-
 settings.js                                   |  18 +++
 10 files changed, 382 insertions(+), 1 deletion(-)
 create mode 100644 Gruntfile.js
 create mode 100644 hearth/db/archives/.git-keep
 create mode 100644 hearth/downloads/icons/.git-keep
 create mode 100644 hearth/downloads/screenshots-thumbs/.git-keep
 create mode 100644 lib/db-transformer.js
 create mode 100644 lib/db.js
 create mode 100644 lib/utils.js
 create mode 100644 settings.js

diff --git a/.gitignore b/.gitignore
index 1d04e83f..fac62e39 100644
--- a/.gitignore
+++ b/.gitignore
@@ -40,3 +40,7 @@ hearth/media/include.css
 hearth/media/include.js
 hearth/media/js/include.js
 hearth/templates.js
+hearth/db/**/*.json
+hearth/downloads/**/*.gif
+hearth/downloads/**/*.jp*
+hearth/downloads/**/*.png
diff --git a/Gruntfile.js b/Gruntfile.js
new file mode 100644
index 00000000..1ebe1d47
--- /dev/null
+++ b/Gruntfile.js
@@ -0,0 +1,39 @@
+var db = require('./lib/db');
+var utils = require('./lib/utils');
+var settings = require('./settings');
+
+module.exports = function(grunt) {
+  grunt.initConfig({
+    fetchdb: {
+      options: {
+        data_dest: settings.db_dir + '/latest.json'
+      }
+    }
+  });
+
+  // Always show stack traces when Grunt prints out an uncaught exception.
+  grunt.option('stack', true);
+
+  grunt.registerTask('fetchdb', 'Fetches JSON from API, downloads ' +
+                                'icons/screenshots, and transforms data to ' +
+                                'static JSON file to disk', function() {
+    var done = this.async();
+    var options = this.options();
+    db.fetchLatest(options.data_dest).then(function() {
+      grunt.log.writeln(
+        'File ' + utils.color('cyan', options.data_dest) + ' created.');
+      done();
+    }, function(err) {
+      grunt.log.writeln(utils.color('red',
+        'File ' + options.file_dest + ' failed to be created: ' + err));
+      done();
+    }).catch(function(err) {
+      grunt.log.writeln(utils.color('red', 'lib/db failed: ' + err));
+      done();
+    });
+
+  });
+
+  grunt.registerTask('default', ['fetchdb']);
+};
+
diff --git a/hearth/db/archives/.git-keep b/hearth/db/archives/.git-keep
new file mode 100644
index 00000000..e69de29b
diff --git a/hearth/downloads/icons/.git-keep b/hearth/downloads/icons/.git-keep
new file mode 100644
index 00000000..e69de29b
diff --git a/hearth/downloads/screenshots-thumbs/.git-keep b/hearth/downloads/screenshots-thumbs/.git-keep
new file mode 100644
index 00000000..e69de29b
diff --git a/lib/db-transformer.js b/lib/db-transformer.js
new file mode 100644
index 00000000..bdf3fffd
--- /dev/null
+++ b/lib/db-transformer.js
@@ -0,0 +1,151 @@
+var fs = require('fs');
+var path = require('path');
+
+var _ = require('lodash');
+var request = require('request');
+var Promise = require('es6-promise').Promise;
+
+var utils = require('../lib/utils');
+
+
+function generateFilename(url) {
+  // Generates a pretty filename from a remote URL, turning
+  // `97310.png?modified=1366438278` to `97310.png`.
+  return path.basename(url).split('?')[0];
+}
+
+module.exports = function(settings, data) {
+  return new Promise(function(resolveDB, rejectDB) {
+    var images = {};
+
+    data = data.apps.map(function(app) {
+      app._id = app.id;
+      app.icon = app.icons['64'];
+
+      // Collect a list of image URLs (to later download to disk).
+      // key = URL, value = directory name
+      images[app.icon] = 'icons';
+      if (app.previews.length > 0) {
+        images[app.previews[0].thumbnail_url] = 'screenshots-thumbs';
+      }
+
+      // Flatten object of localised name to one key for easy searching.
+      app.name_search = [];
+      Object.keys(app.name).forEach(function(locale) {
+        app.name_search.push(app.name[locale]);
+      });
+      app.name_search = app.name_search.join(' ').replace(/\(|\)/g, '');
+
+      // Flatten object of localised name to one key for easy searching.
+      app.description_search = [];
+      Object.keys(app.description).forEach(function(locale) {
+        app.description_search.push(app.description[locale]);
+      });
+      app.description_search = app.description_search.join(' ').replace(/\(|\)/g, '');
+
+      // Transform from Zamboni categories to Tarako categories.
+      // TODO: figure out a way to use transformMap from src/media/js/categories.js
+      // so we don't need to dupe this constant.
+      var categoryMap = {
+        'games': 'games',
+        'utilities': 'tools',
+        'reference': 'tools',
+        'productivity': 'tools',
+        'education': 'tools',
+        'business': 'tools',
+        'travel': 'lifestyle',
+        'sports': 'lifestyle',
+        'social': 'lifestyle',
+        'shopping': 'lifestyle',
+        'photo-video': 'lifestyle',
+        'news-weather': 'lifestyle',
+        'music': 'lifestyle',
+        'maps-navigation': 'lifestyle',
+        'lifestyle': 'lifestyle',
+        'health-fitness': 'lifestyle',
+        'entertainment': 'lifestyle',
+        'books': 'lifestyle'
+      };
+      var zcategories = app.categories;
+      app.categories = [];
+      _.forEach(zcategories, function (zcat) {
+        var dcat = categoryMap[zcat];
+        if (app.categories.indexOf(dcat) === -1) {
+          app.categories.push(dcat);
+        }
+      });
+
+      return _.pick(app, [
+        '_id',
+        'author',
+        'categories',
+        'content_ratings',
+        'description',
+        'description_search',
+        'icon',
+        'is_packaged',
+        'homepage',
+        'manifest_url',
+        'name',
+        'name_search',
+        'previews',
+        'privacy_policy',
+        'ratings',
+        'slug',
+        'status',
+        'support_email',
+        'support_url',
+      ]);
+    });
+
+    console.log('Transformed data');
+
+    console.log('Fetching images to save to disk');
+
+    var promises = [];
+
+    _.uniq(Object.keys(images)).forEach(function(url) {
+      promises.push(new Promise(function(resolve, reject) {
+        console.log('Fetching', url);
+        var fn = path.join(settings.downloads_dir, images[url], generateFilename(url));
+
+        // Update filename.
+        images[url] = path.relative(settings.frontend_dir, fn);
+
+        var req = request(url).pipe(fs.createWriteStream(fn));
+        var body = '';
+        req.on('data', function (data) {
+          body += data;
+        });
+        req.on('close', function() {
+          // Cachebust these URLs with `?h=<contenthash>`.
+          var hash = utils.computeHash(body);
+          images[url] = utils.cachebust(images[url], hash);
+          resolve();
+        }).on('error', function() {
+          reject();
+        });
+      }));
+    });
+
+    Promise.all(promises).then(function() {
+      data = data.map(function (app) {
+        app.icon = path.join('/', images[app.icon]);
+        if (app.previews.length > 0) {
+          app.previews[0].thumbnail_url = path.join('/', images[app.previews[0].thumbnail_url]);
+        }
+        return app;
+      });
+
+      fs.writeFile(settings.appcache_media,
+        JSON.stringify(_.values(images).sort(), null, 2));
+
+      console.log('Successfully saved all images to disk');
+      resolveDB(data);
+    }, function(err) {
+      console.error('Failed to save images to disk:', err);
+      rejectDB(err);
+    });
+  });
+};
+
diff --git a/lib/db.js b/lib/db.js
new file mode 100644
index 00000000..a5fc74f6
--- /dev/null
+++ b/lib/db.js
@@ -0,0 +1,129 @@
+var fs = require('fs');
+var path = require('path');
+
+var Promise = require('es6-promise').Promise;
+var request = require('request');
+
+var settings = require('../settings');
+var utils = require('../lib/utils');
+
+
+var db_dir = path.join(__dirname, '..', settings.db_dir);
+
+
+module.exports.fetch = fetch = function fetch(dest, preloaded) {
+  return new Promise(function (resolve, reject) {
+    var now = Date.now();
+
+    var fnOriginal = path.join(db_dir, 'original.json');
+    var fnTransformed = dest;
+    var fnArchivedOriginal = path.join(db_dir, 'archives', now + '-original.json');
+    var fnPreloadedHash = path.join(db_dir, 'preloaded-hash.json');
+
+    request(settings.db_url, function (err, res, body) {
+      if (err) {
+        console.error(err);
+        reject(err);
+        return;
+      }
+
+      fs.writeFile(fnOriginal, body);
+      fs.writeFile(fnArchivedOriginal, body);
+
+      var bodyJSON = JSON.parse(body);
+
+      settings.db_transformer(bodyJSON).then(function success(data) {
+        var bodyTransformed = JSON.stringify(data);
+        var hash = utils.computeHash(bodyTransformed);
+        var fnArchivedTransformed = path.join(db_dir, 'archives', hash + '.json');
+
+        var promises = [
+          new Promise(function (resolveFile) {
+            fs.writeFile(fnTransformed, bodyTransformed, function () {
+              console.log('Successfully wrote database to disk', fnTransformed);
+              resolveFile();
+            })
+          }),
+          new Promise(function (resolveFile) {
+            fs.writeFile(fnArchivedTransformed, bodyTransformed, function () {
+              console.log('Successfully wrote archived database to disk', fnArchivedTransformed);
+              resolveFile();
+            });
+          })
+        ];
+        if (preloaded) {
+          promises.push(new Promise(function (resolveFile) {
+            fs.writeFile(fnPreloadedHash, hash, function () {
+              console.log('Successfully wrote database hash to disk', fnPreloadedHash);
+              resolveFile();
+            });
+          }));
+        }
+        Promise.all(promises).then(function () {
+          console.log('Successfully wrote all database files to disk');
+          resolve(data);
+        });
+      }, function error(err) {
+        console.error('settings.db_transformer rejected:', err);
+        reject(err);
+      }).catch(function (err) {
+        console.error('settings.db_transformer errored:', err);
+        reject(err);
+      });
+    });
+  });
+};
+
+module.exports.fetchPreloaded = function fetchPreloaded(dest) {
+  return fetch(dest || path.join(db_dir, 'preloaded.json'), true);
+};
+
+module.exports.fetchLatest = function fetchLatest(dest) {
+  return new Promise(function (resolve, reject) {
+    var latest = fetch(dest || path.join(db_dir, 'latest.json'));
+    latest.then(function (latestDocs) {
+      // (1) Look up the hash of last DB file that was included in the appcache.
+      // (2) Find the archived DB file that matches that hash.
+      // (3) Read the file and remove the apps that were already present.
+      // (4) Write that file to disk: `latest-since-<previousHash>.json`.
+      // (5) The front end uses this static JSON file to fetch.
+      fs.readFile(path.join(db_dir, 'preloaded-hash.json'), function (err, hash) {
+        if (err) {
+          console.error(err);
+          reject(err);
+          return;
+        }
+        var fnPreviousDB = path.join(db_dir, 'archives', hash + '.json');
+        fs.readFile(fnPreviousDB, function (err, previousDB) {
+          if (err) {
+            console.error(err);
+            reject(err);
+            return;
+          }
+
+          var oldDocs = JSON.parse(previousDB);
+
+          var idsToExclude = oldDocs.map(function (doc) {
+            return doc._id.toString();
+          });
+
+          // Build an array of objects that weren't in the last DB that we appcached.
+          var newDocs = latestDocs.filter(function (doc) {
+            if (idsToExclude.indexOf((doc._id || '').toString()) === -1) {
+              return doc;
+            }
+          });
+
+          var fnLatestSince = path.join(db_dir,
+            'latest-since-' + hash + '.json');
+          fs.writeFile(fnLatestSince, JSON.stringify(newDocs), function () {
+            resolve(newDocs);
+          });
+        });
+      });
+    }, function (err) {
+      console.log('reject');
+      reject(err);
+    });
+  });
+};
diff --git a/lib/utils.js b/lib/utils.js
new file mode 100644
index 00000000..76f3efff
--- /dev/null
+++ b/lib/utils.js
@@ -0,0 +1,33 @@
+var crypto = require('crypto');
+
+
+var colors = {
+  cyan: '\x1B[36m',
+  red: '\x1B[31m'
+};
+
+module.exports.color = function color(whichColor, text) {
+  return colors[whichColor] + text + '\x1B[39m';
+};
+
+module.exports.cachebust = function cachebust(fn, hash) {
+  var extPos = fn.lastIndexOf('.');
+  return fn.substr(0, extPos) + '.hash_' + hash + fn.substr(extPos);
+};
+
+module.exports.computeHash = function computeHash(contents) {
+  var hasher = crypto.createHash('sha256');
+  hasher.update(contents, 'binary');
+  return hasher.digest('hex').substr(0, 7);
+};
+
+module.exports.baseurl = function baseurl(url) {
+  return url.split('?')[0];
+};
+
+module.exports.urlparams = function urlparams(url, qs) {
+  if (url.indexOf('?') === -1) {
+    return url + '?' + qs;
+  }
+  return url + '&' + qs;
+};
diff --git a/package.json b/package.json
index 1793cd12..15c4d4ab 100644
--- a/package.json
+++ b/package.json
@@ -13,5 +13,12 @@
     "node": ">= 0.10.x",
     "npm": ">= 1.1.x"
   },
-  "dependencies": {}
+  "dependencies": {
+    "request": "~2.34.0",
+    "es6-promise": "~0.1.1",
+    "lodash": "~2.4.1"
+  },
+  "devDependencies": {
+    "grunt": "~0.4.4"
+  }
 }
diff --git a/settings.js b/settings.js
new file mode 100644
index 00000000..c3873920
--- /dev/null
+++ b/settings.js
@@ -0,0 +1,18 @@
+var dbTransformer = require('./lib/db-transformer');
+
+var settings = {
+  debug: true,
+  db_url: 'https://marketplace-dev.allizom.org/api/v1/fireplace/collection/curated/?region=restofworld',
+  frontend_dir: 'hearth',
+  use_data_uris: true
+};
+
+settings.db_dir = settings.frontend_dir + '/db';
+settings.downloads_dir = settings.frontend_dir + '/downloads';
+
+settings.db_transformer = function(data) {
+  return dbTransformer(settings, data);
+};
+
+module.exports = settings;
+