adding similarity check (#29)

* adding similarity check

* adding paging

* fixing tests

* docu

* better error handling

* fixing tests

* adjusting page limit

* fixing login screen

* cleanup

* upgrade browser list

* prevent spamming the log

* fixing tests

* removing job listings when removing a job or the user

lib/services/similarity-check/SimilarityCacheEntry.js

@@ -0,0 +1,36 @@
+const stringSimilarity = require('string-similarity');
+
+//if the score is higher than this, it will be considered a match
+const MAX_DICE_INDEX = 0.7;
+
+/**
+ * The similarity check is based on the dice coefficient. => https://en.wikipedia.org/wiki/S%C3%B8rensen%E2%80%93Dice_coefficient
+ *
+ * @type {module.SimilarityCacheEntry}
+ */
+module.exports = class SimilarityCacheEntry {
+  constructor(time) {
+    this.time = time;
+    this.values = [];
+  }
+
+  setCacheEntry = (entry) => {
+    this.values.push(entry);
+  };
+
+  getTime = () => {
+    return this.time;
+  };
+
+  hasSimilarEntries = (value) => {
+    if (this.values.length > 0) {
+      for (let i = 0; i < this.values.length; i++) {
+        const index = stringSimilarity.compareTwoStrings(value, this.values[i]);
+        if (index >= MAX_DICE_INDEX) {
+          return true;
+        }
+      }
+    }
+    return false;
+  };
+};

lib/services/similarity-check/similarityCache.js

@@ -0,0 +1,63 @@
+/**
+ * each job that runs scrapes all provider. This cache holds the titles of the found listing(s) and provides
+ * a similarity check. if this check returns true, it will not be forwarded to the notification adapter, thus
+ * the user won't see any duplicates
+ *
+ * The retention of this cache is per default 5 minutes, but can be smaller if the interval is > 5 mins.
+ *
+ * @type {module.SimilarityCacheEntry|{}}
+ */
+const SimilarityCacheEntry = require('./SimilarityCacheEntry');
+const config = require('../../../conf/config.json');
+
+//5 minutes
+let retention = 5 * 60 * 1000;
+
+const intervalInMs = config.interval * 60 * 1000;
+//an interval below 5 mins sounds crazy, but there are ppl out there doing crazy shit.
+if (intervalInMs <= retention) {
+  retention = Math.floor(intervalInMs / 2);
+}
+
+//jobid -> SimilarityCacheEntry
+const cache = {};
+
+let intervalId;
+
+exports.addCacheEntry = (jobId, value) => {
+  cache[jobId] = cache[jobId] || new SimilarityCacheEntry(Date.now());
+  cache[jobId].setCacheEntry(value);
+};
+
+exports.hasSimilarEntries = (jobId, value) => {
+  if (cache[jobId] == null) {
+    return false;
+  }
+
+  return cache[jobId].hasSimilarEntries(value);
+};
+
+/**
+ * cleanup
+ */
+intervalId = setInterval(() => {
+  const keysToBeRemoved = [];
+  const now = Date.now();
+
+  Object.keys(cache).forEach((key) => {
+    if (cache[key].getTime() + retention < now) {
+      keysToBeRemoved.push(key);
+    }
+  });
+
+  if (keysToBeRemoved.length > 0) {
+    keysToBeRemoved.forEach((key) => delete cache[key]);
+  }
+}, 10000);
+
+/**
+ * mostly used for tests
+ */
+exports.stopCacheCleanup = () => {
+  clearInterval(intervalId);
+};

lib/services/storage/jobStorage.js

@@ -61,12 +61,18 @@ exports.setJobStatus = ({ jobId, status }) => {
 };
 
 exports.removeJob = (jobId) => {
+  listingStorage.removeListings(jobId);
   db.get('jobs')
     .remove((job) => job.id === jobId)
     .write();
 };
 
 exports.removeJobsByUserId = (userId) => {
+  db.get('jobs')
+    .value()
+    .filter((job) => job.userId === userId)
+    .forEach((job) => listingStorage.removeListings(job.id));
+
   db.get('jobs')
     .remove((job) => job.userId === userId)
     .write();

lib/services/storage/listingsStorage.js

@@ -47,3 +47,7 @@ exports.setLastJobExecution = (jobId) => {
   const key = buildKey(jobId, null, 'lastExecution');
   return db.set(key, Date.now()).write();
 };
+
+exports.removeListings = (jobId) => {
+  db.unset(jobId).write();
+};