Added price history log.

Rental crawler fix

See merge request saburly/marketalarm/web!80

.gitignore

@@ -2,4 +2,4 @@ node_modules/
 .env
 .idea/
 .eslintrc
-.vscode/
+.vscode/

app/crawler/savers/postgres.js

@@ -1,6 +1,7 @@
 const moment = require("moment");
 
 const { bulkUpsertRealEstates } = require("../../helpers/db/realEstate");
+const { bulkUpsertPriceHistory } = require("../../helpers/db/priceHistory");
 
 class PostgresSaver {
   connect() {
@@ -11,6 +12,21 @@ class PostgresSaver {
 
   async save(results) {
     const savedRecords = await bulkUpsertRealEstates(results);
+    //Extruding data for price history table
+    const resultPrices = savedRecords.map(realEstate => {
+      //Null values canot be recognized by ignore duplicates in sequalize
+      //Value price = 0 indicates 'cijena na upit'
+      const priceTmp =
+        realEstate.dataValues.price === null ? 0 : realEstate.dataValues.price;
+
+      return {
+        realEstateId: realEstate.dataValues.id,
+        price: priceTmp,
+        createdAt: realEstate.dataValues.createdAt,
+        updatedAt: realEstate.dataValues.updatedAt
+      };
+    });
+    const savedPrices = await bulkUpsertPriceHistory(resultPrices);
 
     if (Array.isArray(savedRecords)) {
       const newRealEstates = [];

app/crawler/specificCrawlers/rental.js

@@ -312,7 +312,7 @@ class RentalCrawler {
 
         let numberOfRooms =
             parseInt(extractedData["re_realEstates_roomsNO"]) +
-              parseInt(extractedData["re_realEstates_bedroomNO"]) || null,
+              parseInt(extractedData["re_realEstates_bedNO"]) || null,
           numberOfFloors =
             parseInt(extractedData["re_realEstates_floorsNO"]) ||
             this.getNumberOfFloorsFromFloorId(extractedData["re_floorNO_id"]),
@@ -352,7 +352,9 @@ class RentalCrawler {
             realEstatePropertiesFromInfrastructure.phoneConnection,
           cableTV = realEstatePropertiesFromInfrastructure.cableTV,
           internet = realEstatePropertiesFromInfrastructure.internet,
-          basementAttic = realEstatePropertiesFromSpaces.basementAttic,
+          basementAttic =
+            realEstatePropertiesFromSpaces.basementAttic ||
+            this.checkBasemAtticFromFloors(extractedData["re_floorNO_id"]),
           storeRoom = realEstatePropertiesFromSpaces.storeRoom,
           videoSurveillance =
             realEstatePropertiesFromDescriptions.videoSurveillance ||
@@ -397,9 +399,7 @@ class RentalCrawler {
         );
         if (!publishedDateMoment.isValid()) {
           throw {
-            message: `Invalid published date : ${
-              extractedData["re_realEstates_inserted"]
-            }`
+            message: `Invalid published date : ${extractedData["re_realEstates_inserted"]}`
           };
         }
 
@@ -410,9 +410,7 @@ class RentalCrawler {
         );
         if (!renewedDateMoment.isValid()) {
           throw {
-            message: `Invalid renewed date : ${
-              extractedData["re_realEstates_edited"]
-            }`
+            message: `Invalid renewed date : ${extractedData["re_realEstates_edited"]}`
           };
         }
 
@@ -782,8 +780,42 @@ class RentalCrawler {
     if (floorIds.length === 0) {
       return null;
     }
+    let noOfFloors = floorIds.length;
+    // Floors of 'suteren', 'podrum', 'tavan' and 'potkrovlje' are not counted
+    floorIds.forEach(id => {
+      if (
+        parseInt(id) === 1 ||
+        parseInt(id) === 2 ||
+        parseInt(id) === 12 ||
+        parseInt(id) === 14
+      ) {
+        noOfFloors--;
+      }
+    });
+    return noOfFloors;
+  }
 
-    return floorIds.length;
+  checkBasemAtticFromFloors(floorsIdText) {
+    // floorIdText can be array of numbers, separated by comma or number
+    const floorIds = floorsIdText.split(",");
+
+    let check = false;
+
+    if (floorIds.length === 0) {
+      check = false;
+    }
+    //If floors 'suteren', 'podrum', 'tavan' and 'potkrovlje' exists then tag for basement-attic is true
+    floorIds.forEach(id => {
+      if (
+        parseInt(id) === 1 ||
+        parseInt(id) === 2 ||
+        parseInt(id) === 12 ||
+        parseInt(id) === 14
+      ) {
+        check = true;
+      }
+    });
+    return check;
   }
 
   async sleep(ms) {

app/helpers/db/priceHistory.js

@@ -0,0 +1,20 @@
+"use strict";
+const db = require("../../models/index");
+const sequelize = require("sequelize");
+
+const bulkUpsertPriceHistory = async priceHistoryData => {
+  try {
+    const order = [["realEstateId", "desc"]];
+
+    return await db.PriceHistory.bulkCreate(priceHistoryData, {
+      order,
+      ignoreDuplicates: true
+    });
+  } catch (e) {
+    console.log("Error bulk upserting priceHistory : ", e);
+  }
+};
+
+module.exports = {
+  bulkUpsertPriceHistory
+};

app/migrations/20200121000524-add-priceHistory-table.js

@@ -0,0 +1,42 @@
+"use strict";
+
+module.exports = {
+  up: (queryInterface, Sequelize) => {
+    const tableFields = {
+      id: {
+        type: Sequelize.BIGINT,
+        autoIncrement: true,
+        allowNull: false,
+        primaryKey: true
+      },
+      realEstateId: {
+        type: Sequelize.BIGINT,
+        allowNull: false,
+        unique: "uniquePriceRealEstate",
+        references: {
+          model: "RealEstates",
+          key: "id"
+        },
+        onUpdate: "CASCADE",
+        onDelete: "SET NULL"
+      },
+      price: {
+        type: Sequelize.REAL,
+        unique: "uniquePriceRealEstate"
+      },
+      createdAt: {
+        type: Sequelize.DATE,
+        defaultValue: Sequelize.literal("NOW()")
+      },
+      updatedAt: {
+        type: Sequelize.DATE,
+        defaultValue: Sequelize.literal("NOW()")
+      }
+    };
+    return queryInterface.createTable("PriceHistory", tableFields);
+  },
+
+  down: queryInterface => {
+    return queryInterface.dropTable("PriceHistory", {});
+  }
+};

app/migrations/20200121094500-add-constraint-priceHistory.js

@@ -0,0 +1,10 @@
+"use strict";
+module.exports = {
+  up: (queryInterface, Sequelize) =>
+    queryInterface.addConstraint("PriceHistory", ["realEstateId", "price"], {
+      type: "unique",
+      name: "uniquePriceRealEstate"
+    }),
+  down: queryInterface =>
+    queryInterface.removeConstraint("PriceHistory", "uniquePriceRealEstate")
+};

app/models/priceHistory.js

@@ -0,0 +1,44 @@
+"use strict";
+
+module.exports = (sequalize, DataTypes) => {
+  const PriceHistory = sequalize.define(
+    "PriceHistory",
+    {
+      id: {
+        type: DataTypes.BIGINT,
+        autoIncrement: true,
+        primaryKey: true,
+        allowNull: false
+      },
+      realEstateId: {
+        type: DataTypes.BIGINT,
+        allowNull: false,
+        unique: "uniquePriceRealEstate",
+        references: {
+          model: "RealEstates",
+          key: "id"
+        },
+        onUpdate: "CASCADE",
+        onDelete: "SET NULL"
+      },
+      price: {
+        type: DataTypes.REAL,
+        unique: "uniquePriceRealEstate"
+      }
+    },
+    {
+      freezeTableName: true
+    }
+  );
+
+  PriceHistory.associate = models => {
+    PriceHistory.hasMany(models.RealEstate, {
+      foreignKey: "id",
+      sourceKey: "realEstateId",
+      targetKey: "id",
+      as: "realEstates"
+    });
+  };
+
+  return PriceHistory;
+};

app/seeders/20200121150443-initial-price-history.js

@@ -0,0 +1,34 @@
+"use strict";
+
+const { RealEstate } = require("../models");
+
+module.exports = {
+  async up(queryInterface, Sequelize) {
+    //Reading initial data from RealEstate table in db
+    const realEstateInitialData = await RealEstate.findAll();
+    //Extruding data for table PriceHistory
+    const priceHistoryInitialData = realEstateInitialData.map(realEstate => {
+      //Null values canot be recognized by ignore duplicates in sequalize
+      //Value price = 0 indicates 'cijena na upit'
+      const priceTmp =
+        realEstate.dataValues.price === null ? 0 : realEstate.dataValues.price;
+
+      return {
+        realEstateId: realEstate.dataValues.id,
+        price: priceTmp,
+        createdAt: realEstate.dataValues.createdAt,
+        updatedAt: realEstate.dataValues.updatedAt
+      };
+    });
+
+    return queryInterface.bulkInsert(
+      "PriceHistory",
+      priceHistoryInitialData,
+      {}
+    );
+  },
+
+  async down(queryInterface, Sequelize) {
+    return queryInterface.bulkDelete("PriceHistory", null, {});
+  }
+};

package.json

@@ -8,6 +8,7 @@
     "start": "node ./index.js",
     "start-mon": "nodemon ./index.js",
     "migrate": "cd app && npx sequelize db:migrate",
+    "seed": "cd app && npx sequelize db:seed:all",
     "setup": "docker build -t marketalerts .  && docker run -e POSTGRES_USER=docker -e POSTGRES_PASSWORD=docker -e POSTGRES_DB=marketalerts --name pg_marketalerts -d -p 5432:5432 marketalerts && sleep 10 && npm run migrate",
     "docker-start": "docker start pg_marketalerts",
     "docker-stop": "docker stop pg_marketalerts",

test/rentalScrapeTest.js

@@ -13,5 +13,5 @@ if (urlToScrape) {
   })();
 } else {
   console.log("No URL to scrape. Use like this : ");
-  console.log("npm run test-olx-scraper -- URL_TO_SCRAPE");
+  console.log("npm run test-rental-scraper -- URL_TO_SCRAPE");
 }