Merge pull request #9 from ntkog/feature/improvement_regex_regexgen

Recovered old randomize, because randomPosition work worse. Other adjustments

config/elections.json

@@ -2,36 +2,39 @@
  "words" : {
     "pp": [
       "PP",
+      "@populares",
       "pablocasado_",
       "Casado",
       "ppopular",
       "populares"
     ],
     "psoe": [
       "PSOE",
+      "@PSOE",
       "sanchezcastejon",
       "Sanchez",
       "socialistas"
     ],
     "podemos": [
-      "Podemos",
+      "@ahorapodemos",
       "pablo_iglesias_",
       "ahorapodemos",
       "unidaspodemos",
       "unidas podemos"
     ],
     "ciudadanos": [
       "Ciudadanos",
+      "@CiudadanosCs",
       "Cs",
       "Albert_Rivera",
       "Rivera",
       "ciudadanoscs"
     ],
     "vox": [
       "VOX",
+      "@vox_es",
       "santi_abascal",
-      "Abascal",
-      "vox_es"
+      "Abascal"
     ]
   },
   "ws" : {

lib/helpers/elections.js

@@ -1,10 +1,11 @@
 const WORDS = require('../../config/elections.json').words;
 const WORD_SEPARATOR = ",";
 const PARTIES = Object.keys(WORDS);
+const regexgen = require('regexgen');
 
-let rePartiesMap = PARTIES.reduce((old,cur,i,arr) => {
-    old[cur] = new RegExp(`\\b(${WORDS[cur].map(el => el.toLowerCase()).join("|")})\\b`, "ig")
-    return old;
+const rePartiesMap = PARTIES.reduce((old,cur,i,arr) => {
+  old[cur] = regexgen(WORDS[cur], "ig");
+  return old;
 }, {})
 
 let PARTIES_EXTENDED = [
@@ -14,9 +15,8 @@ let PARTIES_EXTENDED = [
     old.push(...cur);
     return old; },
   []);
-console.log(PARTIES_EXTENDED);
 
-const reParties = new RegExp(`\\b(${Object.keys(PARTIES_EXTENDED).join("|")})\\b`, "ig");
+const reParties = regexgen(PARTIES_EXTENDED, "ig");
 var reTrackingWords;
 
 function mapping_parties (t) {
@@ -25,9 +25,6 @@ function mapping_parties (t) {
   let obj = PARTIES.reduce((old,cur,i,arr) => {
     let test = rePartiesMap[cur].test(t.text)
     old[cur] = test;
-    // if (test) {
-    //   markWords.add(cur.toLowerCase());
-    // }
     return old;
   }, {});
   obj.markWords = [...markWords];
@@ -36,10 +33,10 @@ function mapping_parties (t) {
 
 function _setRegexWords (wordsStr){
   let trackingWords = wordsStr.split(WORD_SEPARATOR).filter(el => !reParties.test(el));
-  reTrackingWords = new RegExp(`\\b(${trackingWords.map(el => el.toLowerCase()).join("|")})\\b`, "ig");
+  reTrackingWords = regexgen(trackingWords, "ig");
 }
 
 module.exports =  function(words2track) {
-    _setRegexWords(words2track);
-    return mapping_parties;
+  _setRegexWords(words2track);
+  return mapping_parties;
 }

package.json

@@ -32,9 +32,10 @@
     "jsonpath": "^1.0.1",
     "lowdb": "^1.0.0",
     "node-fetch": "^2.4.0",
+    "polygon-lookup": "ntkog/polygon-lookup",
+    "regexgen": "^1.3.0",
     "twitter-stream-api": "^0.5.2",
     "websocket-stream": "^5.5.0",
-    "ws": "^6.2.1",
-    "polygon-lookup": "ntkog/polygon-lookup"
+    "ws": "^6.2.1"
   }
 }

src/locationUtils.js

@@ -5,16 +5,45 @@ const spain = require('../data/spain-boundaries.json')
 const ccaa = require('../data/spanish-ccaa-boundaries.json')
 const LOOKUP_CCAA = new PolygonLookup(ccaa);
 const LOOKUP_SPAIN = new PolygonLookup(spain);
+const SPAIN_EXTENT = [-18.1607999245551,27.6377369098232,4.32788540952535,43.7899967308538];
 
 function randomize(location){
-  let {xmin,ymin,xmax,ymax} = location.boundingbox;
-  let [lon,lat] = randomPosition([xmin,ymin,xmax,ymax]);
+  let lat, lon;
+  try{
+    let i = 1,
+    ymin = parseFloat(location.boundingbox.ymin),
+    ymax = parseFloat(location.boundingbox.ymax),
+    xmin = parseFloat(location.boundingbox.xmin),
+    xmax = parseFloat(location.boundingbox.xmax);
+    yDiff = ymax - ymin,
+    xDiff = xmax - xmin;
+    do {
+      lat = getRandomArbitrary(ymin, ymax);
+      lon = getRandomArbitrary(xmin, xmax);
+      ymin += yDiff;
+      ymax -= yDiff;
+      xmin += xDiff;
+      xmax -= xDiff;
+      i++;
+    } while(!_isInsideSpain(lon, lat) && i < 3);
+    if(i === 3){
+      throw "Geolocation attempts exceeded";
+    }
+  } catch(err) {
+    console.log(`${err}\nRandomzing location: ${JSON.stringify(location)}`.red);
+    lon = lat = 0;
+  }
+
   return {
-    lat : parseFloat(lat),
-    lon : parseFloat(lon)
-  };
+    lat: lat,
+    lon: lon
+  }
 }
 
+function getRandomArbitrary(min, max) {
+  return Math.random() * (max - min) + min;
+};
+
 function _isInsideSpain(lon, lat){
   let candidate = LOOKUP_SPAIN.search(lon,lat);
   return candidate !== undefined;